Quantitative analysis of microplastics in Nile tilapia from a recirculating aquaculture system using pyrolysis–gas chromatography–mass spectrometry

Abstract

Microplastic (MP) ingestion through fish consumption is a concern for human exposure. While the presence of plastic particles in fish tissues has been documented worldwide, information on microplastic concentrations in edible tissues, especially those smaller than 10 µm, remains scarce. Spectrometric techniques provide a complementary analytical tool to measure MP mass for human exposure studies without intrinsic size limitations; however, their application to fish analysis is limited. In this study, we utilized pyrolysis gas chromatography–mass spectrometry (Py-GC–MS) for the identification and quantification of MPs in fish muscle tissues. Two sample preparation methods, pressurized liquid extraction, and chemical digestion, were tested for compatibility with Py-GC–MS analysis. An analytical method using chemical digestion was validated for analyzing particles ≥ 0.7 µm for 4 polymer types: polypropylene, polyethylene, polystyrene, and polymethyl methacrylate. The developed method was applied to 24 adult Nile tilapia (Oreochromis niloticus) samples from a recirculating aquaculture system. MPs were detected in 42% of the samples, with an average concentration of 0.14 ± 0.32 µg/g, while high variations within subsamples were observed. Our findings reveal trace amounts of MPs in edible fish tissues from aquaculture, highlighting the potential risk of microplastic ingestion through fish consumption. This underscores the need for further risk assessments to evaluate the impact on human health and to develop appropriate mitigation measures.

Graphical Abstract

Background

The omnipresence of microplastics (MPs) has been reported in major environmental compartments including marine and freshwater [1, 2], terrestrial [3], and air [4]. Moreover, studies have revealed the occurrence of MPs in consumer products including personal care products [5], drinking water [6, 7], beverages, and foodstuffs [8,9,10]. The human body is inevitably exposed to MPs through ingestion, inhalation, and dermal contact [11]. Recent studies into human matrices have provided evidence of MP exposure, with traces found in blood [12, 13], tissues and organs [14,15,16], breast milk [17], stool [18], and testicle [19] samples.

The increasing prevalence of MPs in the environment has led to growing research into their potential for human ingestion through food consumption. The presence of MPs in fish has been studied as being both recipients of MPs from aquatic environments [20] and pathways of human exposure through food consumption [21]. Limited studies have examined MPs in the edible portions of fish, focusing on organs typically removed before consumption instead, such as the gastrointestinal (GI) tracts and other organs [8, 22]. Plastic particles in edible fish fillets have been reported for a variety of fish species, with some extending to sizes larger than 5000 µm. MPs in fish mussels were found in a broad concentration range from 0 to 10⁵ MP particle/g (hereafter referred to as MPs/g) of tissue with sizes ranging from 0.45 to 5000 µm (Table S1—supplement). Studies investigating both organs and fillets of fish revealed lower concentrations [23,24,25] and smaller particle sizes [26] in the muscle tissues compared to the GI tract. The majority of studies investigating MPs in edible fish tissues analyzed particle sizes larger than 10 µm, due to intrinsic methodological size limitations (Table S1—supplement). Given that particles smaller than 10 µm are more likely to access organs and the bloodstream, their detection, and quantification are crucial for evaluating potential risks to human health [8].

The analysis of MPs in edible fish tissues is commonly conducted using spectroscopic techniques such as the micro Fourier-transform infrared (µ-FTIR) and µ-Raman spectroscopy [22, 27]. These methods yield valuable insights into particle size, shape, and concentration in terms of particle numbers, enabling polymer identification down to 10 and 1 µm, respectively. However, they require long analysis times and generate substantial amounts of data. Complementary to spectroscopic tools, thermo-analytical techniques such as pyrolysis gas chromatography coupled with mass spectrometry (Py-GC–MS) allow for mass concentration analysis of MPs without inherent particle size limitations. It provides a rapid identification and quantification of polymers with high sensitivity. The mass concentrations obtained provide useful data for human exposure studies and modeling MP fate.

The use of Py-GC–MS for the determination of MPs in biota samples is still in the developmental phase. Py-GC–MS analysis requires efficient matrix removal to prevent interferences with the responses of targeted polymers. Therefore, employing validated methods with defined recoveries and evaluating matrix effects is crucial to mitigate potential under- and over-estimations [28]. For the analysis of MPs with Py-GC–MS, different extraction and sample treatment techniques have been implemented including pressurized liquid extraction (PLE) [29,30,31,32], chemical digestion [33,34,35], and enzymatic-chemical digestion [36,37,38], microwave-assisted extraction [39], and a combination of these techniques [31, 36, 37, 40]. As method development continues, there remains limited understanding regarding mass concentrations in fish and other biota samples. To date, the quantification of MPs in biota using Py-GC–MS has been conducted in only a few studies, which include analyses of bivalves [38, 39], benthic organisms (acorn worm, annelid, mollusk) [33], wild animals (otter, harbor seal, sawbill duck, flounder, common guillemot) [36], and seafood (oysters, prawns, squid, crabs, and sardines) [40].

The present study aims to develop and apply a protocol to analyze MPs ≥ 0.7 µm in fish muscle tissues by using Py-GC–MS. For the extraction of MPs from the fillet samples, two sample treatment methods, PLE and chemical digestion with potassium hydroxide (KOH), were evaluated. We developed and validated a rapid and high-throughput analytical method utilizing Py-GC–MS for identifying and quantifying 4 high-production volume polymers, polypropylene (PP), polyethylene (PE), polystyrene (PS), and polymethyl methacrylate (PMMA), within edible fish tissues. The developed method was applied to 24 Nile tilapia samples sourced from a recirculating aquaculture system (RAS) in Sweden. Our findings contribute to understanding the mass concentrations of MPs in fish for particles down to 0.7 µm in diameter, and thus to the knowledge of MP exposure of humans via fish consumption.

Materials and methods

Contamination control

Special attention was given to prevent any contamination within the two laboratories where dissection and sample treatment were done. The use of plastic materials was avoided as much as possible, and 100% cotton lab coats were worn. Anything exposed to air, including samples, solvents, tweezers, etc., was covered with plastics-free aluminum foil. To prevent any external contamination through materials, the stainless steel pyrolysis cups were heated with a torch and the microfiber glass filters (0.7 µm, GF/F, Whatman, United Kingdom) were heated in the muffle oven at 500 °C. All solvents and liquids used for the sample treatment were first filtered through these filters. The demineralized water used in the laboratory was prefiltered with an in-line closed filtration system by placing a 0.7-µm glass filter in a stainless steel filter disk (Sterlitech Corporation, USA). All the glassware was first rinsed with filtered water and then heated in the muffle oven at 525 °C before use (Carbolite -Gero, United Kingdom). All surfaces were cleaned using 70% ethanol and filtered water before use. The dissection of the fish was done in a precleaned fume hood and the sample preparation was performed in a laminar flow cabinet.

Fish culture and dissection

Adult Nile tilapia (Oreochromis niloticus) were obtained from an RAS facility at Gårdsfisk AB in Åhus, Sweden. The facility implemented plastic components made of PE including tanks, pipes, and biofilters. The tanks have a capacity of 2000 cubic meters and maintain a constant water temperature of approximately 30 °C. The fish were fed with Tilapia Grower pellets (Skretting, Nutreco N.V., The Netherlands) on a daily regimen of around 2–3% of body weight.

Euthanized male and female fish were received on ice from the RAS facility in July 2021. The dissection of the fish was done at the laboratory at the University of Gothenburg on the following day. As the fish were obtained post-mortem, no ethical applications were required. The sampling involved the examination of 12 male and 12 female tilapia, characterized by an average body weight of 598 ± 226 g and an average length of 290 ± 40 cm (total body length: head to the longest tail point) (Table S2—supplement). Fillet samples were collected from each fish. The samples were stored in aluminum foil, transferred to dry ice to maintain sample integrity, and transported to Vrije Universiteit Amsterdam, the Netherlands. Samples were kept in the freezer at − 20 °C until further analysis.

Evaluation of different sample preparation methods

Two sample treatment methods were tested for their suitability to analyze MPs in fish fillets using Py-GC–MS: PLE and chemical digestion with KOH. This evaluation was conducted on 6 compounds: PE, PET (polyethylene terephthalate), PMMA, PP, PS, and PVC (polyvinyl chloride).

PLE offers a rapid and automated extraction process where the MPs are extracted from the matrix under critical pressure and temperature with a selective solvent. To analyze MPs in tilapia muscle, we have adopted the method of Ribeiro et al. [40] with minor changes: (i) 1g of freeze-dried fish was directly extracted in 5-mL stainless steel extraction cells without prior digestion and filtration; (ii) a pre-extraction with methanol was performed to remove matrix interferences followed by the extraction of MPs using dichloromethane; and (iii) double-shot pyrolysis (thermal desorption followed by pyrolysis) was used to remove volatile compounds in the sample. These modifications aimed to lower the methodological size limitation, reduce sample dilution, and remove interfering signals. The parameters of PLE were listed in Table S3 in the supplementary material.

In addition, chemical digestion with 10% KOH was evaluated. This digestion method is widely applied in the literature to analyze MPs in fish considering its high efficiency in digestion and low impact on polymer integrity (Table S1—supplement). Here, our focus was on obtaining a liquid that was filterable through a 0.7 µm (⌀ 8 mm) microfiber glass filter to collect MP particles down to 0.7 µm from the edible fish tissues. We evaluated the digestion of freeze-dried fish fillets with KOH and with the combination of KOH and NaClO based on the method described by Süssmann et al. [41] which demonstrated good digestion efficiency for fish samples. In this study, the combination of KOH and NaClO yielded better digestion efficiency. To improve the clarity and filterability of the digested samples, additional solvents were added to the samples prior to filtration. The optimized method is provided in the sample preparation section.

Sample preparation

Fish fillet samples (30–121 g ww/fillet, 77 ± 1% water content) were washed with filtered water and freeze-dried (Christ Alpha 1–4, Germany) in precleaned glassware. The freeze-dried fish fillet was crushed and mixed with a surgical knife. All samples were analyzed in triplicate. 250 mg (dw) freeze-dried fish was weighed into a beaker and 20 mL of 10% KOH (Honeywell, USA) and 6 mL of NaClO (6–14% active chlorine, Supelco, USA) were added. The resulting suspension was homogenized using an ultrasonic probe (Fisher Scientific, USA) for 80 s. After sonication, samples were incubated in the shaking water bath overnight at 50 °C and 55 rpm. The next day, the strong alkaline digestion solution was neutralized with acetic acid (99.8%, Sigma Aldrich, USA) to enhance isopropanol solubilization of the remaining matrix including fatty acids, lipids, and proteins, and prevent emulsification. After cooling down to room temperature, the solution was diluted with 20 mL isopropanol (99.95%, Biosolve, France). The resulting clear solution was filtered through a 0.7-µm microfiber glass filter (GF/F, Whatman, United Kingdom) and concentrated onto an 8 mm diameter area using a customized vacuum filtration unit. After the whole sample was filtered, the sample vial was rinsed with 5 mL of acetic acid: water (1:1, v/v) solution and 5 mL of ethanol (99.9%, Biosolve, France), respectively. The filter with the residue was oxidized with 2 mL of H₂O₂ (30%, Supelco, USA) without vacuum for 10 min. Finally, the glass filtration unit and the filter were rinsed with 5 mL of demineralized water followed by 5 mL ethanol. The concentrated sample on the filter (⌀ 8 mm) was cut out, folded into the pyrolysis cup, and dried in the oven at 50 °C (Binder, Emergo, Landsmeer, the Netherlands) prior to the analysis using Py-GC–MS. The sample preparation is represented in Fig. 1. An example of the post-filtration filter is included in the supplementary material (Fig. S1).

Fig. 1

Schematic representation of the sample treatment

Py-GC–MS analysis

The samples were analyzed using a multi-shot pyrolysis unit, EGA/PY-3030D, coupled with an auto-sampler AS-1020E (Frontier Laboratories, Saikon, Japan). The pyrolyzer was connected to an Agilent 6890 gas chromatograph equipped with an Agilent DB-5HT column (30 m × 0.25 mm × 0.25 µm), and a 5975C mass spectrometer (Santa Clara CA, USA). Double shot analysis was performed and data were collected in selected ion monitoring (SIM) mode. First, thermal desorption was performed from 100 to 300 °C (by 50 °C/min) to eliminate the volatile compounds. Subsequently, the sample was pyrolyzed at 600 °C for 18 s. Helium was used as the carrier gas at a flow rate of 3.0 mL/min with a split ratio of 1:10. The GC oven temperature program consisted of an initial period at 40 °C for 2 min, followed by a ramp from 40 to 360 °C at a rate of 20 °C per minute, and finally, a 1-min hold at 360 °C. The inlet and transfer line temperatures were set at 300 °C.

Six polymers were initially targeted including PE, PET, PMMA, PP, PS, and PVC. However, due to inconsistent recoveries and matrix interferences, PET and PVC were eliminated from the target list (see section Method validation and performance). The characteristic pyrolysis products used for the identification and quantification of the target polymers are listed in Table S4 and the mass spectra of the quantifier compounds are shown in Figs. S3 and S4 in the supplementary material. For the quantification, external calibration curves for each indicator compound were generated using a polymer mixture obtained from pressurized liquid extraction with a Thermo ScientificTM ASETM 350 Accelerator Solvent Extractor (083146, Waltham, MA, USA) as described in Sefiloglu et al. [1]. Prior to Py-GC–MS analyses, 0.25 µm of poly(4-fluoro styrene) was added into each sample cup as an internal standard and the solvent was evaporated in the oven at 50 °C. Each indicator compound was confirmed by ensuring its retention time was within 0.1 min of the reference standards' average and that the relative ratios of its characteristic fragment ions were within ± 30% of those determined by the reference standards. The data are processed using Agilent Mass Hunter software.

Quality assurance and control

Recovery experiments were conducted using particles smaller than 10 µm. For PP, PE, PS, PVC, and PMMA commercial polymer suspensions within 1.0–6.0 µm in diameter were employed (Table S5—supplement). For PET, a standard was prepared using PLE as no commercial suspension was available in this size range. 250 mg of freeze-dried tilapia fillet was spiked with a known amount of target polymers (0.4–4 µg) in triplicate. Spiked samples, along with 3 unspiked tilapia samples were treated, filtered, and transferred to the pyrolysis cups. The unspiked samples were then spiked with the same amount of polymers in the cup, and the solvent was evaporated in the oven set at 50 °C (referred to as control samples from this point on). In addition, reference samples were prepared by adding an equal amount of polymer mixture in the cups with clean filters, without any matrix.

The recovery (%) of each polymer was calculated by dividing the indicator compound response in the spiked samples by the control samples (spiked after extraction) and expressed as a percentage (spiked samples/ control samples × 100). The matrix effect (%) for each indicator compound was calculated based on the ratio of the average relative peak area of the analyte with the matrix to the average relative peak area without the matrix ((control samples/ reference samples − 1) × 100). If more than one indicator compound is applicable, the quantification of the target polymers was done based on the respective analyte indicator compound with the least matrix effect.

A procedural blank was processed with the same reagents without the fish matrix with each batch of samples to check for possible background contamination. Fourteen procedural blank samples were treated and analyzed with the same protocol as the samples. The limit of detection (LOD) and limit of quantification (LOQ) were calculated as 3 and 10 times the standard deviation of the response of the quantifier compound in the blanks, respectively. The average concentrations of the target polymers in the procedural blanks ranged from < LOD to 0.31 µg. All samples were blank-corrected for the individual polymer. The LOD, LOQ, average in blanks, recovery (%), and matrix effect (%) of the quantifier compounds are listed in Table S6 in the supplementary material for the list of target compounds.

Statistical analysis

Statistical analysis was conducted using R Statistical Software version 4.2.3 to determine the significance of differences in concentration levels between female and male tilapia samples. The Shapiro–Wilk test was employed to assess the normality of the data. Due to the non-normal distribution, the Wilcoxon rank-sum test was used for group comparison.

Results and discussion

Method validation and performance

Sample analysis using the pyrolysis system in this study employs sample cups with an inner diameter of 3.8 mm and a height of 8 mm. When analyzing biota samples, it is required to concentrate the samples to fit into these small cups by minimizing the matrix interferences with the indicator ions of the target polymers. Residues such as lipids or proteins present in the sample can introduce matrix interferences, potentially leading to overestimations of the target polymers [29, 31, 37, 42, 43]. For the analysis of fish fillet samples with Py-GC–MS, we evaluated PLE and chemical digestion with KOH for sample preparation.

To evaluate the potential matrix interferences and recoveries using PLE, three different extracts were prepared. First, as a reference standard, a mixture of polymers was extracted in a pre-calcinated sea sand and an external calibration curve was prepared with this extract in increasing volume from 1 to 60 µL. Second, 1 g (dw) fish fillet was extracted without any polymers (unspiked fish extract). Another calibration curve was obtained by adding the same amount of the reference standard and unspiked fish extract in the cups with increasing volumes (external calibration with the fish matrix). Third, 1 g (dw) of fish fillet was extracted with the mixture of polymers and a matrix-matched calibration curve was obtained by analyzing 1–60 µL of this spiked fish extract.

The comparison of the external calibration curves with and without the fish matrix revealed the influence of the matrix's presence on polymer responses during pyrolysis. Here, we observed a signal enhancement of 85% for the indicator markers of PE when pyrolyzed with the matrix and a signal suppression for styrene trimer (31%), the marker of PS used in this study (Fig. S2—supplement). Dierkes et al. [29] evaluated the matrix interferences of PLE extracts of different matrices including wood, leaf, fish fillet, etc., on the indicator compounds of PE, PP, and PS. Similar to our findings, no significant interferences were observed for PP and PS for fish fillets, when 2,4-dimethyl-1-heptene and styrene were used for the identification of these polymers, respectively. However, the release of short-chained alkanes and alkenes from the fatty acids in the fish matrix showed interferences with the thermal degradation products of PE.

In addition, the comparison of external and matrix-matched calibration curves provided insights into the signal interferences when polymers are co-extracted and co-pyrolyzed with the matrix. Results showed a significant signal enhancement for PE, PP, and PMMA (Figs. S2, S3—supplement), with calculated recoveries above 300%. As the signal enhancement issue by co-extraction and co-pyrolysis has not been previously reported for PP and PMMA, Rauert et al. [31] highlighted the potential of overestimation of PE concentrations due to high matrix interferences when extracted from high-fat content matrices using PLE. They optimized a method tailored for PE analysis that included additional enzymatic and chemical digestion steps before PLE to ensure proper matrix removal. From our results, we concluded that both co-extraction and co-pyrolysis of polymers with the fish matrix can enhance the matrix interferences for several compounds of our target list. This requires additional sample treatment steps for the removal of the matrix, compromising the practicality of the automated PLE procedure. Additionally, only a small portion of the final extract can be transferred to the pyrolysis cup, introducing an additional dilution factor to the results.

Chemical digestion with 10% KOH was assessed for the treatment of fish fillets for Py-GC–MS analysis. Initially, the filterability of a minimum of 1 g of digested fish fillet (corresponding to ≥ 200 mg dry weight) was evaluated on a 0.7-µm microfiber glass filter with an 8-mm diameter surface, then the recovery experiments were done with the most optimal protocol and the least matrix interferences.

Digestion with 10% KOH is widely applied for the treatment of fish muscle tissues (Table S1-supplement) [27]. First, wet and freeze-dried fillet samples were digested with 20 mL 10% KOH with an overnight incubation at 50 °C. In both cases we encountered incomplete digestion, leading to filter clogging. In addition, foaming was observed during vacuum filtration of the wet samples, which was potentially generated by the emulsion of oil and water [44]. Therefore, the method was refined further using a freeze-dried fish fillet. Next, the combination of KOH and NaClO was tested as a digestion solvent. Süssmann et al. [41] previously tested this combination for alkaline oxidative digestion for fish tissues, resulting in filterable suspensions using a 1-µm Ø 47-mm filter, with minimal steps and no impact on the analyzed polymers' integrity. In our study, this combination improved the digestion of fish fillets but still led to clogging of the 0.7-µm Ø 8-mm microfiber glass filter. The discrepancy in filterability of the digested tissues may be attributed to the use of a smaller filter mesh size and the smaller surface area in our filtration set-up. To improve the filterability, 20 mL of isopropanol was added to the digested samples for the dissolution of the remaining matrix causing clogging. The resulting solution proved to be clear and readily filterable through the targeted surface area.

The method performance was evaluated based on the recoveries and matrix interferences for each pyrolysis product of the targeted polymers. The efficiency of the applied chemical digestion and filtration steps was measured by comparing the responses of indicator compounds in spiked fish samples to those in control samples, which contained the same amount of polymers spiked into the cup on unspiked fish extracts. The method yielded acceptable recoveries for 5 targeted polymers (PMMA, PE, PS, PP, and PVC) ranging from 56 to 89% with a relative standard deviation of (RSD) < 20%. PET, however, could only be recovered at 10.9 ± 6.5% (Fig. 2). Similarly, previous studies reported low recoveries of PET when KOH digestion was applied at 50 °C–60 °C [40, 45]. This can be attributed to the hydrolysis of PET in alkaline solutions at temperatures higher than 30 °C [41]. However, PET was quantified by several studies applying alkaline digestion at elevated temperatures (Table S1-supplement), emphasizing the importance of accurately determining the extraction efficiency of the polymers prior to analysis of real samples. Due to the low recoveries obtained, PET was not quantified in the tilapia fillet samples in this study.

Fig. 2

Recovery and average matrix effect of polymers with the optimized methodology using KOH digestion followed by dilution with isopropanol and filtration (n = 3)

The mean matrix interferences for the indicator compounds of PMMA, PP, and PS ranged from − 12 to 26%. However, we observed a strong matrix effect for some indicator compounds of PE and PVC (Table S7—supplement). Interferences from the organic matrix for the short-chained alkane and alkene pyrolysis products of PE have been highlighted by some studies [29, 31, 37, 39]. Similar to these observations, we encountered high matrix interferences for the short-chained alkene indicator compounds of PE, where the matrix effect of, e.g. 1-hexadecene was calculated as 553%. However, the interfering signals decrease with the increasing number of carbons, thus the use of long-chain alkanes/alkenes above C16 was recommended as suitable markers for PE analysis [28]. In our study, the 1-triacontene (C₃₀H₆₀) was utilized for the quantification of PE that showed a matrix effect of 15% and was recovered with 87 ± 10%. Similar to PE, the organic-rich matrices can potentially generate the same pyrolysis products with PVC, which consist of aromatic molecules [33, 42]. We encountered high matrix effects for these compounds ranging from 71 to 193%, with the exception of 1,2-dihydronaphthalene (Table S7-supplement). However, during validation studies, we observed a significant response variation for 1,2-dihydronaphthalene between two sets of unspiked tilapia samples, before and after replacing the same GC column with a new one (Fig. S5- supplement). Therefore, PVC was excluded from the quantification of the tilapia samples due to concerns regarding precision.

MPs in tilapia samples

The validated Py-GC–MS method was used to identify and quantify PE, PP, PS, and PMMA in 24 tilapia samples collected from RAS. Fillets of 12 male and 12 female tilapia were analyzed in triplicate (72 subsamples) and the average concentration per polymer per sample was reported. The four targeted polymers were detected in 42% of the fish analyzed (> LOD) and the total concentrations ranged from < LOD to 1.48 µg/g tissue with an average of 0.14 ± 0.32 µg/g, as 3 samples had concentrations above LOQ (Fig. 3). The highest concentrations were determined for PE in 2 samples (M2, F10), at levels of 0.18 and 1.25 µg/g tissue, respectively. PP was detected in 2 samples (M5, F5; 0.11 µg each), PS in 8 samples (M5, M6, M10, M12, F5, F6, F10, F12) within a concentration range of 0.09–0.36 µg/g, and PMMA was detected in 2 samples (M1 and M6, 0.01 and 0.02 µg/g tissue) (Fig. 3, Table S9—supplement). The statistical analysis revealed no significant difference in terms of total MP concentrations between female and male samples (P = 0.606). High variations within the replicates from the same fillet were observed (RSD = 173%) (Table S8—supplement), as commonly reported in the literature [36, 40]. This can be attributed to the non-homogenous distribution of the polymers within the tissues. Increasing the subsample size, ideally to the entire fillet, would enhance the result robustness. However, aiming to analyze smaller particle sizes often requires a trade-off of the initial sample intake size.

Fig. 3

The mass concentrations detected in 24 adult tilapia samples from RAS: a concentrations above LOD, b concentrations above LOQ. M: male samples, F: female samples

MPs in the edible fish muscles were reported in various fish species from different environments [27]. Most of these studies were conducted with spectroscopic techniques such as µ-FTIR and µ-Raman, and only a few studies used mass spectrometry for MP analysis. Figure 4 summarizes the reported MP concentrations and the size ranges of the detected polymers.

Fig. 4

Summary of MP concentration in the fish fillet analyzed with spectroscopic techniques (purple, unit = MPs/g), and with mass spectrometry (green, unit = µg/g). Arrows were used if no upper or lower size limit of the detected particles was specified. Concentrations in particle numbers (MPs/g) and mass (µg/g) indicate the average MP concentrations in the analyzed species and the individual concentrations of analyzed species, respectively

With spectroscopic analysis, plastic particles were detected from 0.45 to > 5000 µm in the edible fish fillet. A wide variation in reported concentrations was observed among studies with mean concentrations ranging from 0 to 212,000 MPs/g. The majority of the studies reported mean values below 1 MP per gram of tissue (Fig. 4, Table S1—supplement), whereas some studies reported no detected particles in the analyzed muscle tissues [46,47,48]. Matias et al. [24] analyzed MPs in the edible muscle of 55 juvenile seabass samples collected from a RAS using µ-FTIR. MPs were detected in 82% of the muscle samples with a mean concentration of 0.36 ± 0.32 MPs/g. The particle sizes ranged from 30 to 2181 µm.

Some studies have conducted spectroscopic analysis of particles smaller than 10 µm in fish muscle tissues. Zitouni et al. [49] analyzed particles larger than 0.45 µm in benthopelagic fish samples and mean concentrations of particles have been found ranging from 1.78 to 6.03 MPs/g. Makhdoumi et al. [23] reported mean concentrations up to 1.6 MPs/g for 7 riverine fish species with an average concentration of 0.85 ± 0.38 MPs/g. The MP analysis in 4 species from the Mediterranean Sea by Ferrante et al. [50] revealed concentrations with average values of 9.50 × 10⁴–8.66 × 10⁴ MPs/g for the particle sizes of 1.6–2.8 µm. The concentrations were extrapolated based on the MP analysis of the 0.1 g of subsamples from the pooled fillet samples.

The application of spectrometric techniques for the identification and quantification of MPs in biota samples including fish is limited to a few studies [22]. Some studies utilized Py-GC–MS only for the characterization of polymer composition in the fish organs including the GI tract, liver, and stomach [51, 52]. MPs in fish muscle tissues were previously quantified with Py-GC–MS by Ribeiro et al. [40] Haave et al. [36] (Table 1). Haave et al. [36] conducted Py-GC–MS analysis for MPs ≥ 10 µm in wild animals and reported PS in the muscle tissues of codfish (2/3) with 1 µg/g. Ribeiro et al. [40] examined 5 polymer types (PS, PE, PP, PMMA, and PVC) in seafood including oysters, squid, pawns, oysters, carbs, and sardines. As MPs (≥ 2.7 µm) have been quantified in each species, the highest concentrations were detected in wild sardines from 17 to 2491 µg/g. In sardines, PP, PS, and PE were the most frequently detected polymers and PE showed the highest concentrations up to 2352 µg/g tissue. In the study of Hermabessiere and Rochman (2021), polymers including, PE, PP, PS, PMMA, PVC, and PC were analyzed in mussels using Py-GC–MS. PE and PVC were quantified within the range of 18.5–41.6 µg/g tissue. Another study analyzing MPs (PE, PP, PS, and PVC) in mussels from coastal China using TGA-FTIR-GC/MS reported average values of 0.16–1.71 µg/g [34] and PE was found as the most abundant polymer type.

Table 1 Summary of the studies that applied Py-GC–MS for the identification and quantification of MPs in fish muscle tissues

The mass concentrations reported for wild sardines by Ribeiro et al. [40] are several orders of magnitude higher than our results, which may be due to the methodological differences as well as the inherent variability of the samples. Factors such as species type, diet, growth conditions, and environmental factors affect the MP intake and translocation in fish [53]. Studies investigating fish from different trophic levels revealed a negative relationship between fish body size and the MP concentrations in the muscle tissues [54, 55]. Additionally, the feeding behaviors of the species can affect the MP uptake. Albignac et al. [33] analyzed MPs in the size range of 0.7–500 µm in 6 marine organisms from different feeding modes using Py-GC–MS/MS. Their analysis revealed concentrations from 105 to 7780 µg/g (dw), with deposit feeders exhibiting the highest MP concentrations, followed by filter feeders. The lowest MP concentrations were detected in carnivore species. Tilapia samples analyzed in our study were collected from a more controlled environment compared to wild conditions. Although the presence of MPs in the RAS environment, including in water and commercial feed pellets, has been previously reported [24, 56], the comparatively lower concentrations found in the fillet tissues can be attributed to the lower MP exposure through ingestion.

Sources of contamination and implications for human exposure

In this study, market-ready tilapia samples obtained from RAS were chosen as the target species considering the growing share of aquaculture in meeting global food demands [57]. RAS offers sustainable farming in a controlled environment and optimized feed for fish growth. However, it is noteworthy that RAS environments are not entirely free of MPs due to the widespread use of plastics in the infrastructure. Additionally, the levels of MPs in source water and fish feed are crucial factors influencing fish exposure to these particles. According to Matias et al. [24], MPs were found in both tank water and fish feed, with average particle concentrations of 37.2 MPs/L and 3.2 MPs/g, respectively. Their study identified fibers of man-made cellulose and PET as the most abundant polymers, while polymers associated with RAS components, such as PE and PP, were less prevalent. Conversely, another study focusing on a marine RAS system reported that the infrastructure contributed the most to MP levels, followed by the feed and source water [58]. Comparing our results with these findings is challenging due to differences in methodologies and study scopes. In our study, we detected trace amounts of PE, PP, and PS in the tilapia samples. However, with the current dataset, it is not possible to pinpoint the exact sources of these compounds. Further research involving samples from different system components and feeds would provide a more comprehensive understanding.

The findings of this study provide valuable data for assessing human exposure to MPs. For a comprehensive exposure and risk assessment, it is crucial to consider both the concentrations of MPs in edible fish tissues and the frequency of fish consumption. Although specific data on tilapia consumption are not available, we estimated the dietary intake of MPs using freshwater fish consumption data from the WHO Global Environment Monitoring System—Food Contamination Monitoring and Assessment Programme (GEMS/Food). This data indicates a minimum per capita fish consumption of 0.1 g/day (G17 cluster countries) and a maximum of 27 g/day (G09 cluster countries) [59]. Based on these consumption rates and the mean MP concentration found in our study, the estimated daily MP intake through tilapia consumption ranges from 0.014 to 3.78 µg/day per capita. However, these estimates should be interpreted with caution, as tilapia represents only a small fraction of the overall human diet.

Conclusions

In this study, Py-GC–MS was utilized to identify and quantify MPs in fish fillet samples for particles larger than 0.7 µm. We optimized a sample treatment protocol involving chemical digestion for the analysis of four plastic polymer types (PE, PP, PS, and PMMA) resulting in a filterable extract on a 0.7-µm mesh-sized filter (Ø 8 mm). The developed methodology was applied to 12 male and 12 female adult tilapia samples from a RAS. MPs were identified and quantified above LOQ in 3 out of 24 of the fillet samples and above LOD in 10 out of 24 samples with an average concentration of 0.14 ± 0.32 µg/g.

We believe that our study is a step forward in the application of Py-GC–MS for mass quantification of MPs in fish by acknowledging areas of improvement. We observed a high variability between the subsamples of the same fish fillet, which can possibly be attributed to the heterogeneous distribution of the MPs. Further refinements are needed to increase the subsample size, ideally analyzing the entire fillet. It should be noted that achieving this with filters of nanometer-level mesh sizes poses significant challenges. In our study, we employed 0.7-µm glass microfiber filters, which, to the best of our knowledge, is the smallest filter mesh size for the analysis of MP with Py-GC–MS in fish fillet samples used in the literature. While Py-GC–MS offers advantages in detecting smaller particles, further optimization would be required to use smaller mesh-sized filters for the analysis of nanoplastic particles. Additionally, PET and PVC were excluded from our target list due to low recovery and matrix interferences. The target list of polymers can be expanded by testing different digestion temperatures and additional solvents for further matrix removal.