Managing health risks of per ﬂ uoroalkyl acids in aquatic food from a river-estuary-sea environment a ﬀ ected by ﬂ uorochemical industry

Substantial per ﬂ uoroalkyl acids (PFAAs) production still occurs in China, and the consumption of aquatic products is a critical exposure pathway of PFAAs in humans. In this study, specimens of 16 freshwater and 40 marine species were collected in the river-estuary-sea environment a ﬀ ected by a mega ﬂ uorochemical industry park in China in 2015, and the edible tissues of these organisms were analyzed for PFAA levels. Per ﬂ uorooctanoic acid (PFOA) was the dominating contaminant with an overall contribution of more than 90%, and concentrations as high as 2161 ng/g wet weight (measured in the freshwater winkle). All species with the greatest PFOA levels were benthic. The trophic magni ﬁ cation factor (TMF) of PFOA was 1.10 for freshwater species and 1.28 for marine species, indicating that PFOA was slightly magnifying. Analysis of carbon source indicated that freshwater species were more benthic feeding, while marine species were more pelagic feeding. Aquatic food consumption screening values of PFOA were modi ﬁ ed according to estimated daily intake (EDI) values, which generated recommendations for limited meal categories and the do-not-eat category. Thus, this study provides recommendations for mitigating the health risks of PFAA-contaminated aquatic food, ranging from food selection to consumption frequency and proper food processing.


Introduction
Among per-and polyfluoroalkyl substances (PFASs), perfluoroalkyl acids (PFAAs) have a wide array of industrial applications. They consist of a fully fluorinated carbon chain (C n F 2n+1 -) and a charged functional group, which gives them enhanced properties, such as strong acidity and high surface activity at very low concentrations, compared to nonor partially fluorinated alkyl compounds . Earlier research indicated that the eight-carbon (C8) perfluoroalkyl-derived surfactants have the best activity, making perfluorooctanoic acid (PFOA)-and perfluorooctane sulfonic acid (PFOS)-related compounds among the most applied PFAAs. However, more recently, their applications have been restricted due to concerns over their persistent, bioaccumulative, and toxic properties and their presence in wildlife (Betts, 2007). The C4 compound perfluorobutane sulfonic acid (PFBS) was developed by a major producer to replace PFOS (3M, 2002). This development assisted efforts to control PFOS production and emission at a global scale (Stockholm Convention, 2009). But for PFOA, the restrictions brought more of a shift in production location rather than elimination. For example, the 2010/2015 PFOA Stewardship Program claimed that the goals were met, yet only eight leading companies joined the program (USEPA, 2015); In the EU, it is estimated that the annual import of textiles could contain 1000-10,000 tons of PFOA-related substances (ECHA, 2018). Thus, the production and use of PFOA is expected to continue until effective alternatives are available to the majority of the fluorochemical industries. The occurrence of PFOA remains a major concern that requires global control, and thus, PFOA, its salts, and PFOA-related compounds were listed in Annex A of the Stockholm Convention with specific exemptions (decision SC-9/12) as of May 2019 (Stockholm Convention, 2019).
The widespread presence of PFAAs in the aquatic environment is linked to their water solubility, especially as salt or free acid (USEPA, 2017), and the use of aqueous forms in industrial applications, such as the ammonium salt of PFOA (APFO) in the production of fluoropolymers by emulsion polymerization (European Commission, 2010). Thus, aquatic organisms can accumulate PFAAs, leading to human exposure through consumption of aquatic species. Past estimates of the biomagnification factor (BMF) and trophic magnification factor (TMF) of PFAAs were inconsistent, with values ranging over several orders of magnitude from ≪1 to ≫1 (Franklin, 2015). Apparently, PFOS has a higher bioaccumulation potential than PFOA in aquatic biota, and PFOS displays relatively clear biomagnification along the food chain, especially for top predators like water birds and marine mammals, while there are no clear trends for PFOA among species from different trophic levels (TLs) (Houde et al., 2011;Ahrens and Bundschuh, 2014), which might be due to the faster elimination of PFOA than PFOS. Hence, because TLs or positions in a food chain cannot be used alone as guidance for categorizing aquatic food according to PFAA contamination levels, more factors should be taken into consideration following exposure under natural conditions.
Animal studies indicate that PFAA uptake can lead to many adverse outcomes, including metabolic disorders, endocrine perturbations, and immune and developmental toxicity (Krafft and Riess, 2015). Human health studies also suggest that PFAAs are associated with certain problems like adverse immune outcomes and dyslipidemia, as well as cancer for residents living in manufacturing locations (Sunderland et al., 2019). These concerns have led to stricter regulations of PFAAs for drinking water safety (USEPA, 2016). For food safety, several countries have already published tolerable daily intake (TDI) as healthbased guidance values for PFAAs (EFSA, 2008;Danish EPA, 2015;FSANZ, 2017;USEPA, 2017;ATSDR, 2018). These values are critical not only for assessing the human health risk of PFAAs, but also for providing more practical suggestions to reduce the risk. In this case, the calculated TDI values may be improved by considering individual differences in dietary habits.
Despite environmental and health concerns, PFOA production is still ongoing in China. According to the China Fluorosilicon Organic Materials Industry Association, until 2015, the annual production capacity of PFOA was approximately 200 tons. A majority of the PFOA was used as processing aids to produce fluoropolymers with an annual capacity of up to 130,000 tons, with only minor amounts being exported to foreign companies. We are aware that high amounts of PFAAs are emitted from a mega fluorochemical industry park (FIP) located in the Xiaoqing River catchment of North China (Wang et al., 2014;Wang et al., 2016). Continued PFAA emissions have led to local contamination of the surface and groundwater (Liu et al., 2016), indoor and outdoor dust (Su et al., 2016), and plants and animals produced locally for human consumption (Liu et al., 2017;Su et al., 2017). The PFOA that is discharged as waste from the FIP is transported via the river to the adjacent sea , where an intensive fishery exists. The objectives of this study included investigating the impact of high PFAA emission on the aquatic organisms in the river-estuary-sea system of the Xiaoqing River catchment of North China. We also aimed to analyze factors that influence the bioaccumulation of PFAAs among different aquatic species, and explore strategies to mitigate human health risks via advice on the selection and consumption of aquatic foods.

Research design
This study examined the water transportation pathway of PFAAs, starting from the emission source (the FIP). Freshwater and marine aquatic organism samples were collected along this transport route (Fig. 1). In the minor tributary receiving waste from the FIP, limited aquatic organisms were observed. Sites DY1 and DY2 located upstream and downstream of the FIP, respectively, were selected for monitoring PFAA emission. Sites XQ1-XQ8 were distributed from upstream of the Xiaoqing River to its estuary, and two fish species, along with one crab species, were collected to evaluate the transport of PFAAs through the river-estuary-sea environment. The greatest effort in collecting the widest range of freshwater species was focused on site XQ4 located downstream of the confluence point where the tributary meets the Xiaoqing River. Site XQ-S was located in the Laizhou Bay within an intensive fishery area where the widest possible range of marine organisms was collected. Further description of these sampling sites can be found in Wang et al. (2016).

Sample collection and preparation
The sampling campaign was conducted in October 2015, after the start of the oceanic fishing season (closed from May to September). Fixed fishing nets combined with cast fishing nets were used to catch freshwater organisms, whereas local fishing boats were hired to catch marine organisms (from shore to site XQ-S). We collected common local species at their normal (adult) sizes for analysis. Non-local species that can be bought from local markets were not included. The aquatic organisms were maintained in clean water for a short period (minutes to hours, benthic species took longer time) after being caught to reduce the effect of in-situ water. All aquatic taxa were identified to species level. A few individuals of each species (depending on size and availability) were homogenized to generate the sample material before extraction. To assess the human health risk, only the edible parts were used for PFAA measurements, depending on the dietary habit (muscle, whole body without shell, or whole body, etc.) (Table S1). All aquatic samples were freeze-dried and ground for quantitative extraction. Water samples (XQ1-XQ8, XQ-S, DY1-DY2) were collected in prerinsed 1 L polypropylene (PP) bottles at < 20 cm depth, excluding the surface microlayer. Water parameters, including temperature, pH, dissolved oxygen, and salinity, were measured in situ using an HQd Portable and Benchtop Meter Configurator (HACH Company, USA).

PFAA measurement and QA/QC
A total of twelve linear PFAAs, including nine perfluoroalkyl carboxylic acids (PFCAs) with carbon numbers from C4 to C12 and three perfluoroalkyl sulfonic acids (PFSAs), were quantified in this study (Table S2). The extraction procedures were mostly consistent with previous studies. Briefly, for water samples, 400 mL were extracted using Oasis WAX 6 cc Vac cartridges (Waters Corp., Milford, MA) . For aquatic organisms, approximately 0.5 g of dry samples were extracted by ion-pair extraction, followed by cleanup with Supelclean™ ENVI-Carb™ cartridges (Sigma-Aldrich Co., St. Louis, MO) and Oasis WAX 6 cc Vac cartridges (Loi et al., 2011). Detailed information on the standards, reagents, and extraction is provided in the supplementary material. Concentrated 1 mL extracts were loaded on an Agilent 1290 Infinity HPLC System coupled to an Agilent 6460 Triple Quadrupole LC/MS System (Agilent Technologies, Palo Alto, CA) ( Table S3). Quantification of PFAAs in all samples was conducted using 10-point internal quantification curves of PFAA standards with concentrations ranging from 0.01 to 1000 ng/mL and 10 ng/mL internal standards. The regression coefficients (R 2 ) were > 0.99 in all batches. LOD and LOQ were 0.01-0.10 ng/g and 0.04-0.50 ng/g dry weight (dw) for aquatic organisms, and 0.01-0.06 ng/L and 0.06-0.22 ng/L for water samples, respectively. Matrix spike recoveries (mean value ± standard deviation, n = 4) ranged from 94 ± 4% to 114 ± 6% for aquatic samples and from 78 ± 3% to 125 ± 6% for water samples (Table S2). Extracts with PFAA concentrations above 1000 ng/mL were adjusted in weight or volume and extracted again to ensure that the sample concentrations were within the calibration ranges. Field, transport, procedure and solvent blanks were prepared with every sample batch to monitor potential interference,

Natural stable isotope analysis
The nitrogen and carbon stable isotope ratios of the test organism tissues were analyzed using a DELTA V Advantage Isotope Ratio Mass Spectrometer coupled with a Flash 2000 HT Elemental Analyzer (Thermo Fisher Scientific, Inc., USA). The natural nitrogen and carbon stable isotope contents are denoted as δ 15 N and δ 13 C (‰), expressing the per mil deviation from international standards, the atmospheric 15 N and the 13 C content of Vienna Pee Dee Belemnite (VPDB), respectively (Table S8).
Results of the δ 15 N were used to determine the trophic level (TL) of individual species using the following equation (Eq. (1)): The hypothesis is that the TL of zooplankton is 2 and the enrichment factor constant is 3.8 (Loi et al., 2011). We analyzed the community composition and richness of the phytoplankton and zooplankton in the Xiaoqing River. Forty-one phytoplankton and twenty-seven zooplankton species were identified. But the richness was not sufficient for the stable isotope analysis. Instead, we used the mean value of 6.14 as a constant for δ N zooplankton 15 derived from Wan et al. (2005) that was conducted near this study area.
Then trophic magnification factor (TMF) of PFAAs was calculated using the following equations (Eqs. (2) & (3)) (Tomy et al., 2004): where the unit of PFAA concentration was ng/g, wet weight (ww). Only PFAA concentrations in muscle of the aquatic organisms were used. The value b in Eq. (3) was the slop of Eq. (2). A TMF of > 1 implies that a chemical is biomagnifying.
Results of the δ 13 C were normalized using the carbon-to-nitrogen ratio (C:N) in the following equation (Eq. (4)): In the study by Post et al. (2007), C:N was proved to be sufficient to normalize the untreated δ 13 C for lipid content, which is specifically applicable for aquatic organisms.
The normalized δ 13 C was further used to determine the carbon source of the aquatic species in the following equation (Eq. (5)): The hypothesis is that the zooplankton represent the pelagic source, δ Δ C 13 is the trophic enrichment factor for consumers analyzed using P. Wang, et al. Environment International 138 (2020) 105621 muscle tissues, which is set as a constant of 1. 3‰ (McKinney et al., 2012). The benthic source in this study is presented using the benthic species with the lowest δ C benthic 13 values, which were Eriocheir sinensis (−27.1) for the freshwater species and Zoarces slongatus (−25.0) for the marine species. Carbon source values closer to 0 indicate more benthic feeding and values closer to 1 indicate more pelagic feeding.

Screening values of PFOA
Aquatic food consumption screening values (ACSV) of PFOA (ng/g; ww) were calculated using the following equation (Eq. (6)): The FCSV PFOS (ng/g, ww) is the fish consumption screening value (FCSV) of PFOS determined by the Michigan Department of Health and Human Services (MDHHS) (State of Michigan, 2016), which includes a series of concentration ranges of PFOS that provide guidelines for fish consumption frequency. The R PFOA/PFOS is the PFOA-to-PFOS ratio from published health guidelines that is a measure for co-occurrences of PFOA and PFOS in the same subject. Because the MDHHS guidelines do not provide the FCSV for PFOA, we hypothesized that the R PFOA/PFOS could be used as a measure for the difference between PFOA and PFOS, and thus, the ACSV PFOA could provide guidance for the consumption of aquatic food with PFOA contamination.

Estimated daily intake of PFOA
The estimated daily intake (EDI) of PFOA (ng/kg/day) based on the high end of the ranges for ACSV PFOA was further calculated with the following equation (Eq. (7)): where F is the consumption frequency (converted to a daily basis) of the aquatic food that is derived from the meal category of the FCSV PFOS . R IR/BW (g/kg) is the ratio of the ingestion rate (IR) per meal (g) to the body weight (BW; kg) of consumers. There are differences in IR and BW among different populations. However, the IR and BW values provided by the MDHHS are divided into age groups and in one-to-one correspondence, which generates a constant ratio (0.10 oz/kg converted to 2.84 g/kg).

Statistical and spatial analysis.
The PFAA concentrations in aquatic organisms were converted from the dry weight basis to wet weight basis using the moisture content values that were measured during lyophilization (Table S1). Statistical analysis was performed with Microsoft Excel 2016 (Microsoft Corp., Redmond, WA), SPSS Statistics V20.0 (SPSS Inc. Quarry Bay, HK), and OriginPro 2018 (OriginLab Corp., Northampton, MA). Tests of normality on the concentrations of PFAAs were conducted to ensure that data met the assumptions used for further analysis. Spatial distributions of sampling were illustrated using the Arcmap module in ArcGIS V10.0 software (ESRI, Redland, CA) with the actual recording of the coordinates combined with other detailed geographical information (river, tributary, topography of land and sea, etc.).

PFAAs in aquatic organisms
PFAAs were detected in all freshwater and marine aquatic organisms (Table 1 and Table S4). PFOA was the only compound with a detection ratio of 100%. All long-chain PFAAs had detection ratios over 90%, whereas the short-chain PFAAs had relatively low ratios, except for perfluorobutanoic acid (PFBA) (93%). PFOA concentrations (including mean, median, and maximum values) were several orders of magnitude higher than those of other PFAAs, indicating that PFOA was the dominating PFAA in the aquatic organisms analyzed in this study.
Previous studies on PFAAs in aquatic organisms affected directly by fluorochemical emission provided limited information for a comparison with our data. PFOS has received the most attention in the monitoring of contaminants in aquatic animals (Houde et al., 2011;Ahrens and Bundschuh, 2014;Taylor and Johnson, 2016;Pan et al., 2018;Fair et al., 2019). In this study, the PFOS concentrations (max: 3.55 ng/L) in the Xiaoqing River water were relatively low (Table S4), but in aquatic organisms, the maximum PFOS concentration was 9.73 ng/g (ww). This disparity is due to the high bioaccumulation potential of PFOS (Ahrens and Bundschuh, 2014). The tissues of aquatic animals had low concentrations and detection ratios of short-chain PFAAs, which were likely related to their low bioconcentration factors (BCFs) (Martin et al., 2003) and possibly to depuration during the short period when the freshly caught aquatic animals were maintained in clean water (to reduce the effect of in-situ water) (Cerveny et al., 2018;Zhong et al., 2019). In contrast, freshwater plant species affected by fluorochemical industry emissions accumulated substantially higher amounts of shortchain PFAAs than the aquatic animals analyzed in this study (Wang et al., 2019). According to an earlier study on terrestrial animals collected in the emission impact zone of the same FIP described our report , the PFBA-to-PFOA ratio in home cultured eggs was 0.18, which was much higher than the respective ratio in the aquatic animals with the highest concentrations of ΣPFAAs in this study (4.70e−3). These results may indicate that the consumption of aquatic animals is a more important source of long-chain PFAAs in humans than that of terrestrial animals. Thus, the following sections are mostly focused on PFOA.

PFOA in freshwater species
Concentrations of PFOA in the crucian carp (Carassius auratus) were only 0.33 and 0.38 ng/g (ww) at the upstream sites XQ2 and XQ3, respectively, but 99.0 ng/g (ww) at XQ4 downstream of the FIP. Among sixteen freshwater species monitored at site XQ4, the highest PFOA concentration was found in a mollusk, the winkle (Cipangopaludina chinensis) (2161 ng/g, ww), followed by a fish, the loach (Misgurnus anguillicaudatus) (340 ng/g, ww), and a crustacean, i.e., a crayfish species (Procambarus clarkii) (241 ng/g, ww), which belong to three different phyla. The nine fish species showed relatively high PFOA concentrations, ranging from 7.04 to 340 ng/g (ww). The two amphibians, a toad (Bufo raddei) and a turtle (Trionyx sinensis), had the lowest PFOA concentrations of 0.42 ng/g and 0.41 ng/g (ww), respectively) Note: nd, indicates below LOQ.

PFOA in marine species
Because the seawater in the Laizhou Bay dilutes the river water PFOA levels through the freshwater-saltwater interface, the forty marine aquatic species at the XQ-S site had lower PFOA concentrations than the freshwater species. The highest PFOA concentration was found in a mollusk, the fur clam (Scapharca subcrenata) (642 ng/g, ww), followed by a fish, the fang goby (Odontamblyopus rubicundus) (349 ng/g, ww), and a macroinvertebrate sea worm (Urechis unicinctus) (167 ng/g, ww). Similar to the findings with the freshwater species, the distribution of PFOA concentrations among the species was not associated with their phylum or type. Fourteen out of eighteen fish species had PFOA concentrations of less than 1.00 ng/g (ww) (Fig. 3).

PFOA levels in the aquatic environment
The emission from the FIP was the main source of PFOA in the aquatic environment of the Xiaoqing River (from sites XQ4-XQ8) ( Table  S5). Compared with our previous studies, the water PFOA levels remained very high in recent years, at mean values from 3100 ng/L in September 2011 (Wang et al., 2014), to 17,361 ng/L in June 2013  and 40,368 ng/L in October 2015, measured at sites XQ6 to XQ8, respectively. This trend was corroborated by other studies conducted in the same area in 2014 (Heydebreck et al., 2015;Shi et al., 2015a). Moreover, a study by Chen et al. (2016) on PFASs in the Bohai Sea also suggested broad dissemination of PFOA in the entire Laizhou Bay area, indicating a continued influence of PFOA on local aquatic organisms.

The river-estuary-sea environment
The river-estuary-sea system involves frequent mixing of freshwater and saline water (Table S6). Three common species, including the crucian carp, river crab, and sea bass, were selected to evaluate the influence of the aquatic environment. In the section of the Xiaoqing River with heavy PFOA pollution (sites XQ4-XQ8), the crucian carp had the highest mean PFOA concentration (mean: 90.4 ng/g, ww, range: 33.0-156 ng/g, ww), followed by the river crab (muscle tissue of males, mean: 38.9 ng/g, ww, range: 3.98-150 ng/g, ww) and sea bass (mean: 10.2 ng/g, ww, range: 4.19-15.1 ng/g, ww). Fluctuations of PFOA concentrations were observed throughout the river-estuary-sea environment (Fig. 4a). However, despite the river PFOA concentration decreasing almost 10-fold from site XQ7 (78.0 µg/L) to site XQ8 (8.56 µg/L) (Table S5), PFOA concentrations in the three species collected from estuary (site XQ8) were substantially higher than those from upstream sites, especially in the river crab.
We further analyzed the pairwise correlations between PFOA concentrations in the three species and PFOA concentrations in water, along with four water parameters (temperature, pH, salinity, and dissolved oxygen) (Table S6). There were no strong correlations, except that there was a significant negative correlation (p < 0.05) between PFOA concentrations in sea bass and water pH (Table S7). Typically, the water pH is positively correlated with salinity and cation concentrations (the reaction of cations with dissolved CO 2 produces OH − through hydrolysis and hence increases water pH), but the increasing cation content (e.g., Ca 2+ and Na + ) in water could decrease the BCF of PFOA in aquatic organisms (Xia et al., 2015).

Sex and tissue differences in PFOA levels of crabs
There are sex-specific differences in the accumulation of PFASs within the same species (Cerveny et al., 2018), but these differences may not be comparable with those between species (Babut et al., 2017). In our study, sex was not a critical factor in most aquatic species used for human consumption, except in crabs. Specifically, we measured PFOA concentrations in the edible parts of crabs, i.e., in muscle and fat samples of male crabs and in muscle and roe samples of female crabs (Table S4, Fig S2). In river crabs, PFOA accumulation was generally higher in males than in females. Among male river crabs, PFOA concentrations were slightly higher in muscle samples (mean: 38.9 ng/g, Fig. 2. PFOA concentrations (ng/g, ww) in the freshwater species at site XQ4 (the value for the river crab (Eriocheir sinensis) was obtained from muscle tissue of males). P. Wang, et al. Environment International 138 (2020) 105621 ww) than in fat samples (mean: 31.5 ng/g, ww), even in crabs from site XQ8 with much higher PFOA concentrations. The same trend was also observed in male sea crabs. But the trend was the opposite in female river crabs, in which PFOA concentrations were slightly higher in roe (mean: 8.14 ng/g, ww) than in muscle samples (mean: 6.07 ng/g, ww). A portion of the PFOA burden is likely disseminated into the eggs, which requires further detailed studies.

Influence of TL
The TLs ranged from 2.21 to 5.57 among freshwater species (3.85 ± 0.8) and from 2.35 to 4.83 among marine species (3.45 ± 0.6) ( Table S8). The most contaminated species had relatively low TLs; in freshwater species, the TL was 4.02 for winkle, 2.29 for loach, and 3.62 for crayfish; and in marine species, the TL was 2.38 for fur clam, 3.35 for fang goby, and 2.60 for a sea worm. The TMF of PFOA was 1.10 (p = 0.60) for freshwater species and 1.28 (p = 0.29) for marine species. This indicated that PFOA was biomagnifying among the tested aquatic species, but the trend was not significant, which was a limitation of using field-derived biomagnification factors and trophic magnification factors as indicators of the bioaccumulation potential (Franklin, 2015). Among the three monitored species of the river-estuary-sea environment, the sea bass had the highest TL, followed by the crucian carp and river crab. This trend was similar among the contaminated section of the Xiaoqing River (from site XQ4 to XQ6), whereas their TLs did not differ greatly at the estuary site (XQ8) (Fig. 4b). The freshwater winkle had the highest PFOA level in this study. According to its biology, the winkle grazes on algae and lives on the sediment, perhaps even inside the sediment. PFOA has a high sorption coefficient from water to particles and sediment (Ahrens et al., 2010), and high PFOA levels have been recorded in Xiaoqing River sediments (Shi et al., 2015a;Wang et al., 2016). Furthermore, Robinson et al. (1984) found that the lipids of aquatic organisms can merge with the sediment, generating sediment with an increased nutritive value that is ingested by some aquatic species. This phenomenon could explain the observation that the TL of the winkle (4.02) was comparable to that of some predator fishes.  (ng/g, ww) in marine species collected between the shore and site XQ-S (values for the two sea crabs (Portunus trituberculatus and Charybdis japonica) were obtained from muscle tissue of males). P. Wang, et al. Environment International 138 (2020)

Influence of carbon source
The carbon source of the aquatic organisms is related to their food and habitat (Borgå et al., 2012). In this study, the carbon source was 0.30 ± 0.23 for freshwater species, and 1.11 ± 0.37 for marine species (Table S8). This indicated that freshwater species were more benthic feeding, while marine species were more pelagic feeding. Comparing the TL with carbon source, it seemed that benthic feeding would lead to higher TLs than pelagic feeding. The three species of the river-estuary-sea environment displayed different trends on their TLs and carbon sources (Fig. 4b). The crucian carp showed relatively consistent trends of both TL and carbon source. However, the TL and carbon source of the river crab showed opposite trends, and there were limited sex-specific difference. This might be related to the habitat of the river crab, which lives on the riverbank, ingesting carbon sources that differ from those consumed by species living in the water. The change of carbon sources in the sea bass showed a jump from sea to river, with carbon source altered from pelagic to benthic, and TL increased. The migration of this fish is associated with its physiological adaptations, such as the developmental status, fluid osmotic pressure, etc. How the changing carbon sources or adaptations can affect the bioaccumulation of PFOA in the sea bass, or how PFOA can influence the sea bass migration may require further studies.

Ecological risk evaluation of PFOA
The exposure of PFOA in both environmental medium and organisms would pose potential ecological risk to the aquatic environment. In the study by Valsecchi et al. (2017), annual average environmental quality standards (AA-EQS) for PFOA were derived. In this study, the PFOA concentrations in both freshwater (XQ4 to XQ7) and estuary (XQ8) almost all exceeded the AA-EQS for the protection of pelagic community in freshwater (30 µg/L) and sea water (3 µg/L), respectively. As mentioned above, the PFOA levels remained very high in recent years. This might be a reason why the richness of phytoplankton and zooplankton in the Xiaoqing River was poor.
The PFOA concentrations in water from sites XQ4 to XQ8 and XQ-S all largely exceeded the AA-EQS for the protection of predators (0.1 µg/ L in freshwater and 0.02 µg/L in sea water, respectively). And 72% of the PFOA concentrations in aquatic organisms exceeded the AA-EQS for the protection of predators (0.9 ng/g, ww). Thus, more studies are needed to investigate the influence of PFOA contamination to the local predator species, especially avian in the estuary and wetland environment. And in such monitoring campaign, both whole fish and fillet are necessary to obtain sufficient information on ecological risk evaluation (Mazzoni et al., 2019).

Assessment of consumption screening values of PFOA
Previous studies have demonstrated that aquatic food consumption is a major pathway for human exposure to PFAAs. However, the human health risk of PFOA via consumption of contaminated aquatic food depends not only on PFOA concentrations in the food but also on the dietary habits (Fair et al., 2019). In the Michigan Fish Consumption Advisory Program for PFOS by the MDHHS, the FCSV ranges were established based on the dietary habit, which includes 1 to 16 meals per month and 6 meals per year as limited meal categories, and a do-not-eat meal category (State of Michigan, 2016). The TDI values for PFOA and PFOS, as well as the PFOA-to-PFOS ratio from published health guidelines changed greatly by different countries/regions in recent years (Table 2). Especially, the European Food Safety Authority (EFSA) updated the TDI values in 2018 that were much less than those set in 2008 (EFSA, 2008;, but the new EFSA's TDI is not yet put in force in EU because it is not accepted by all the EU Member States. This indicated that with more epidemiological evidences, the toxicity of PFOA and PFOS were more serious, and the toxicity of PFOA was higher than that of PFOS. To explore risk assessment and management options of dietary frequencies for PFOA in this study, four scenarios were established, with the ACSV calculated using Eq. (6) and EDI calculated using Eq. (7). Then the percentage (P, %) of PFOA concentrations in aquatic organisms measured in this study that fall into the corresponding ACSV ranges were also calculated (Table 3).
Scenario 1: The R PFOA/PFOS is equal to 8 based on the TDI values for PFOA and PFOS set by the FSANZ (2017). Results were presented as ACSV 1 , EDI 1 and P 1 .
Scenario 2: The R PFOA/PFOS is equal to 1 based on the oral non-cancer reference doses (RfDs) for PFOA and PFOS set by the United States Environmental Protection Agency (USEPA) (USEPA, 2017). Results were presented as ACSV 2 , EDI 2 and P 2 .
Scenario 3: The R PFOA/PFOS is equal to 0.44 based on the TDI values P. Wang, et al. Environment International 138 (2020) 105621 for PFOA and PFOS set by the EFSA (EFSA, 2018). Results were presented as ACSV 3 , EDI 3 and P 3 . Scenario 4: The EDI was set as a constant (0.8 ng/kg·bw/day). Results were presented as ACSV 4 and P 4 .
The calculated EDI values were well within the corresponding TDI values in each scenario for scenario 1 to scenario 2, but exceeded the corresponding TDI values in scenario 3. And scenario 4 was designed to adjust scenario 3. From scenario 1 to scenario 4, the human health risk increased via the consumption of the aquatic food with top PFOA concentrations (Fig. 5). More aquatic organisms fell into stricter meal categories. Especially the do-not eat category, from zero in scenario 1, 4.71% in scenario 2, 10.6% in scenario 3, to 24.7% scenario 4. These results brought critical challenges for human health risk assessment and management of PFOA via consumption of aquatic food with high PFOA residue. Stricter guidelines are better for the protection of human health, but might be difficult for sufficient management by local governments. The governments need to choose proper guidelines to protect the health of local residents, and work with the manufacturers to reduce the emission of PFOA. For aquatic food consumers, taking proper options can also mitigate the health risk.

Health risk evaluation of PFAAs
The aquatic organisms marked as 'edible' in Fig. 2 and Fig. 3 are routinely consumed, and the species with top PFOA levels require special attention focused on the dietary habit to mitigate the health risk. The analysis of factors affecting the bioaccumulation of PFOA in aquatic organisms suggested that certain benthic species accumulate higher amounts of PFOA than other species. But if these species appear on the list of food preferences, the ACSV of PFOA can provide detailed suggestions on meal frequency.
It should be noted that the ACSV results are more suitable to describe the limits for a single food source. Thus, even if these values might be within the TDI guidelines, adding other food sources with substantial PFOA exposure might lead to PFOA overconsumption, leading to increased health risks. Considering that other PFAAs examined in this study had much lower levels than PFOA, the ACSV of PFOA could be considered a summary of the 12 PFAAs for health suggestions. However, other PFASs, such as hexafluoropropylene oxide dimer acid (HFPO-DA) and chlorinated polyfluorinated ether sulfonate (F53B), were also detected at substantial levels in the same study area (Heydebreck et al., 2015;Shi et al., 2015b). Although the information on the health effects of these substances is still limited, there are rising concerns and including them in future risk assessments would improve the ASCV results. Furthermore, the trend to lower the values in health guidelines indicates an increasing need for protecting human health from PFAA pollution.

Treatment of aquatic food with PFAA pollution
There are additional measures for mitigating the health risk of PFAAs, including guidelines for the proper treatment of contaminated aquatic food, which we summarized from previous studies. Maintaining contaminated aquatic organisms alive in clean water for a short period before cooking would decrease the PFOA pollution to some extent; especially, it would clean the contaminated in-situ water as it did in this study. An extended cultivation period would benefit from the detoxification processes. In a study on the fate of PFASs in the food web by Cerveny et al. (2018), the authors observed a decreasing of the PFOA concentration in the liver of brown trout after living for six months in water with similar PFOA levels. Moreover, treatment of edible tissues of aquatic products with uncontaminated water could lead to the  Note: The meal categories match the ranges provided by the MDHHS (State of Michigan, 2016); F is the consumption frequency converted from the meal category; ACSV 1 , ACSV 2 , ACSV 3 and ACSV 4 represent the upper limit of the concentration ranges; P is the percentage of PFOA concentrations in aquatic organisms measured in this study that fall into the corresponding ACSV ranges. P. Wang, et al. Environment International 138 (2020) 105621 depuration of PFOA, using treatment periods of hours or days depending on the species (Taylor et al., 2017). Selecting the body parts of the contaminated aquatic animals for consumption is also important. In a study by Fair et al. (2019), PFASs concentrations were two to three times higher in the whole fish than in the fillets, indicating that cooking the whole fish (or entire body parts) would result in more PFOA consumption than cooking the fillet (muscle). However, cooking itself cannot consistently reduce contamination by PFASs. Although the concentrations might change, it is likely due to the treatment, while the amounts of PFASs usually remain unchanged during cooking (Bhavsar et al., 2014;Vassiliadou et al., 2015;Taylor et al., 2019).

Conclusions
In this study, 16 freshwater species and 40 marine species affected by industrial fluorochemical waste were investigated for exposure to 12 PFAAs. PFOA was dominant with an overall contribution of > 90%. Freshwater species had generally higher PFOA concentrations than marine species related to higher PFOA concentrations in the river water. The winkle (Cipangopaludina chinensis) displayed the highest PFOA concentration (2161 ng/g, ww) among freshwater species, whereas the fur clam (Scapharca subcrenata) had the highest PFOA concentration (642 ng/g, ww) among marine species. The species with top PFOA levels were all benthic. The TMF of PFOA was 1.10 for freshwater species and 1.28 for marine species, indicating that PFOA was slightly biomagnifying among the tested aquatic species. The carbon source was 0.30 ± 0.23 for freshwater species, and 1.11 ± 0.37 for marine species, indicating that the freshwater species were more benthic feeding, while the marine species were more pelagic feeding. Comparing the TL with carbon source, it seemed that benthic feeding would lead to higher TLs than pelagic feeding. PFOA concentrations in water showed high ecological risk to pelagic community, while PFOA concentrations in both water and aquatic organism showed very high ecological risk to air-breathing predators. We established the PFOA ACSVs and compared the derived EDI values with TDI guidelines for providing practical suggestions on the consumption frequency of PFOA-contaminated aquatic food based on the 1-to-16 meals-per-month and 6 meals-per-year categories, as well as the do-not-eat category. We also assessed studies on detoxification and depuration of aquatic organisms. Thus, this study provides a comprehensive evaluation of how to mitigate the health risk of PFOA in aquatic food, from the food selection and consumption frequency to proper treatment before cooking.
However, by analyzing so many species simultaneously, details on individual species were limited in this study, and further interdisciplinary research is needed. Specifically, the biology of benthic species should be considered for assessing the large differences in PFOA exposure levels among these species. Moreover, the effect of the river-sea habitat exchange on PFOA bioaccumulation needs to be further examined in the migrating species, and ecological risk of PFOA to pelagic community and predators should be highly concerned.