Persistent organic pollutant accumulation in Pacific abyssal plain sediments and biota: Implications for sources, transport, and deep-sea mining

Despite its vast size and ecological and economic importance, the deep-sea (here defined as >200 m depth) is one of the least understood ecosystems on Earth (Ramirez-Llodra et al., 2010; Rogers, 2015; Danovaro et al., 2020). Characterized by 434 million km² of seafloor with an average depth of 4.2 km (Rogers, 2015), this vast deep-sea environment supports global health through carbon export and sequestration, nutrient cycling and regeneration, climate regulation, waste detoxification, and provisioning economically important fish stocks (Irigoien et al., 2014; Boyd et al., 2019; Da Ros et al., 2019; Drazen et al., 2020). While much remains to be discovered (Rogers, 2015), researchers have established that the deep ocean is facing numerous anthropogenic stressors, including chemical pollution (Kallenborn, 2006; Ramirez-Llodra et al., 2011) in water, sediments, and biota (Lohmann et al., 2006; Domingo and Bocio, 2007; Letcher et al., 2010; Dasgupta et al., 2018). Persistent organic pollutants (POPs), defined by their persistence in the environment, bioaccumulation, and toxicity, are continually discharged and transported into the deep sea despite protections under global and regional conventions (Arctic Monitoring and Assessment Programme [AMAP], 2004; Barbo et al., 2023; Sobek et al., 2023; United Nations Environment Programme [UNEP], 2023a, 2023b).

Sources of organic pollutants to the global ocean are varied and include inputs from coastal regions, atmospheric deposition, shipping discharges, waste disposal, and deep-sea fishing and mining (Tornero and Hanke, 2016; Sanganyado et al., 2020). The profile of POPs (type and congener-specific concentrations) within environmental media can provide insight into the industrial sources (e.g., Aroclor mixtures, flame retardants, dyes/pigments) and transport (e.g., atmospheric deposition, sinking ocean currents) of these contaminants (Hu and Hornbuckle, 2010; Anh et al., 2021; Mastin et al., 2022; Barbo et al., 2023). For example, polybrominated diphenyl ethers (PBDEs) 209 (a flame retardant used in plastics) and semivolatile POPs such as less chlorinated polychlorinated biphenyls (PCBs) are often transported to remote ocean locations by long-range atmospheric transport (Wania and Mackay, 1996; Jurado and Dachs, 2008; Sobek et al., 2023). Undertaking a comprehensive congener-specific analysis enables the differentiation between POP sources as well as routes and rates of transport to ocean habitats (Wang et al., 2007; Domínguez et al., 2010; Yeo et al., 2020; Anh et al., 2021; Mastin et al., 2022).

The comparatively few studies that have investigated POP transport processes to the deep ocean have identified deep-water formation, proximity to continental shelves, steep geomorphologies of submarine canyons, and long-range atmospheric transport as contributors to POP concentrations in different regions (Froescheis et al., 2000; Axelman and Gustafsson, 2002; O’Sullivan and Sandau, 2013; Jia et al., 2019; Wagner et al., 2019; Sanganyado et al., 2021; Sobek et al., 2023). Pollutant sources from deep-water formation occur when surface water contaminants sink along with the water masses that form deep ocean currents. Deep-water formation sources were found to be an important driver of perfluorinated compounds in younger waters (<10 years since formation) compared to deep-ocean currents formed >100 years ago (Sanganyado et al., 2021). Deep-ocean currents formed more than 1000 years ago, such as those in the abyssal plain of the Pacific Ocean, are thus unlikely to be sources of anthropogenic POPs produced only within the past 100 years. Indeed, scientists previously assumed that deep abyssal plain ecosystems were far too removed to be affected by anthropogenic pollutants as transport to the seafloor through diffusion and advection could take millennia (Froescheis et al., 2000; Turner, 2015; Sanganyado et al., 2021). Sinking ocean particles (e.g., marine snow including phytoplankton, zooplankton, detritus, feces, particulate organic matter (POM), and plastics) that accumulate POPs as they descend through the water column are another vector for POP transport to the deep sea. This pathway is dependent on particle size and density, with sinking rate slowed significantly as increasing pressure to abyssal depths produces consequent increases in seawater density (Giering et al., 2019). However, uptake of particles into vertically migrating organisms (e.g., zooplankton) that release fecal pellets with POPs at depth can accelerate transport of POPs to deeper waters (Turner, 2015; Yeo et al., 2020). Additionally, a potentially significant contaminant transport process is the flux of carrion to the seafloor (sometimes referred to as deadfall; e.g., Houde et al., 2005; Weijs et al., 2013; Jepson et al., 2016; Desforges et al., 2018; Jeong et al., 2020). While research has identified carrion as an important food resource to the benthos, little has been done to understand the transfer of contaminants in animal carcasses to abyssal plain ecosystems (Barry and Drazen, 2007; Drazen et al., 2008; Drazen et al., 2012; Bryant et al., 2022).

Despite limited data on POP transport and accumulation in the deep sea, modeled estimates suggest up to 60% of historically released POPs, such as PCBs, have been removed from oceanic pelagic ecosystems through burial in seafloor sediments (Wagner et al., 2019; Sobek et al., 2023). Our lack of empirical data to support these estimates, however, hampers our ability to effectively regulate and mitigate anthropogenic impacts to the ecosystem structures and functions provided by this vast environment (Danovaro et al., 2020; Drazen et al., 2020; Levin et al., 2020; Sobek et al., 2023). Knowledge of anthropogenic organic contamination in the Pacific abyssal plain is also necessary to estimate the potential risk of deep-sea mining re-releasing largely banned POPs back into mid- and deep-ocean foodwebs through sediment disturbance and mining effluents (Ginzky et al., 2020; Sanganyado et al., 2021).

Here, we provide data on POPs in the deep abyssal plain, specifically in a region targeted for deep-sea mining and an adjacent region designated as a protected reference zone (PRZ). We targeted three categories of human-made organic contaminants with varying durations of production (i.e., PCBs from 1920s to 1980s, per- and polyfluoroalkyl substances (PFAS) since the 1940s; PBDEs since the 1970s) with the goal of garnering information on concentrations in relation to other ecosystems and toxicity benchmarks, sources, transport, accumulation, and relative contributions of legacy versus emerging POPs. To that end, all 209 PCB congeners, 46 PBDE congeners, and multiple PFAS were measured in abyssal plain sediment and in muscle and liver from rattail fish, Coryphaenoides sp., a deep-ocean predator and scavenger.

Ten sediment samples, five for PCB and PBDE analyses collected together, and five for PFAS analyses, were taken from five sample sites in the eastern Clarion-Clipperton Zone (CCZ) of the Pacific abyssal plain in May of 2021 (Figure 1; Table 1, which includes individual sample results). Four sample sites were from an area permitted for deep-sea mining exploration (test-mining area) and one sample site was from the area designated as a PRZ where mining will be excluded so that impacts can be assessed in the future by comparing mined areas to protected areas. Samples were collected using a piston-driven tubular multi-core sediment sampler lowered to the seafloor. The depth of sample collection ranged from 4234 m to 4297 m. Collection methods were adapted from EPA Method 537 and 537.1 for targeted organic contaminants sampling (United States Environmental Protection Agency [U.S. EPA], 2020).

Briefly, after deployment of each multi-core device, sediment from the top 2 cm of 2–3 of the cores were collected using fresh nitrile gloves and a cleaned stainless-steel scoop and pooled to achieve ≥200 g designated for PCB and PBDE analysis and ≥100 g designated for PFAS analysis. Sediment in contact with the wall of the multi-core tubes was avoided during sediment collection. Sediment for PCB and PBDE analysis was placed in certified clean 250 ml amber glass jars, sealed with tape, and further contained in Whirl-Pak^® bags. Sediment for PFAS analysis was placed in certified clean 500 ml Nalgene^® high density polyethylene bottles and sealed with tape. All sediment samples were stored at –20°C before shipment for analysis. All tools used for sediment handling were cleaned with laboratory grade soap (i.e., Liquinox) and warm water, rinsed with milliQ water, and rinsed with an organic solvent (i.e., methanol) before use on another sample. One field blank for PCB and PBDE testing and three field blanks for PFAS testing were also collected to account for potential contamination during sampling. Field blanks were prepared by introducing milliQ water to multi-core tubes for the same duration as sediment samples and handled in the same manner as samples for each analyte.

A total of 24 rattail fish (Coryphaenoides sp.) were collected for muscle and liver tissue analyses (two fish analyzed for PCBs, two for PBDEs, and eight for PFAS from each of the two sampling areas; Table 1) at depths of 4167 m to 4291 m, from May to June and November to December 2021 (Figure 1; Table 1). Rattail fish, also known as grenadiers, are benthopelagic, inhabiting the waters immediately above the seafloor in bathyal and abyssal zones of the deep ocean (Dunlop et al., 2019). They are slow-growing, reaching lengths of 1 m and ages of approximately 75 years (Black et al., 2021). They are predators that use barbels to detect and consume benthic invertebrates buried in the sediment but are also characterized as scavengers that can detect carrion from a significant distance (Barry and Drazen, 2007; Drazen et al., 2008; Drazen et al., 2012; Dunlop et al., 2019). Fish were collected using an autonomous trap designed by A.K. Sweetman and equipped with a KUMQuat acoustic release. For a description of the trap, see Hartl et al. (2024). The trap was baited with Pacific mackerel (Scomber japonicus) and deployed to the seafloor for periods of 24–72 hours. All fish were dead upon recovery with evidence of barotrauma, though organs were intact during sampling. Samples were immediately placed on ice and transferred to a cold laboratory (4°C–7°C) where fish were measured (total length). Dorsal muscle (N = 24 fish) and liver (N = 11 fish) were dissected and stored in sterile plastic bags at –80°C until shipment on dry ice for analysis (Table 1).

Sediment samples, aqueous field blanks and fish tissue samples were shipped on dry ice to SGS AXYS Analytical Services Ltd. (Sidney, BC, Canada) for POP analyses. Additionally, sediments and aqueous field blanks were sent to the Water Quality Centre at Trent University (Peterborough, Ontario, Canada) for PFAS analysis. Detailed analytical methods are provided in Supplemental materials (Text S1).

Briefly, samples were processed and analyzed by SGS AXYS for all 209 PCB congeners using method MLA-010 Rev. 12 Ver. 09 (Modified EPA method 1668) and for 46 PBDE congeners using method MLA-033 Rev. 06 Ver. 13 (Modified EPA Method 1614). Samples for PCB analysis were spiked with ¹³C-labeled surrogate standards prior to extraction and analysis using a HP 6890 SPB-Octyl capillary gas chromatography column coupled to a Micromass Ultima high-resolution mass spectrometer equipped with a CTC autosampler. For quality assurance, samples were analyzed with a procedural blank, a spiked matrix sample, and a sample duplicate (if sufficient sample volume allowed) every 7 to 20 samples. Only PCB congeners that met quality assurance/quality control acceptance criteria were reported. Samples for PBDE analysis were spiked with isotopically labeled BDE surrogate standards, solvent-extracted, cleaned on a series of chromatographic columns and spiked with isotopically labeled internal recovery standards prior to instrumental analysis using the same capillary gas chromatography column coupled to the Micromass Ultima high-resolution mass spectrometer. Calibration was achieved using a series of BDE solutions covering the working response range of the instrument along with labeled surrogate, cleanup and recovery standards.

Fish tissue samples were processed and analyzed for numerous PFAS (Table S1) using method MLA-110 Rev. 02 Ver. 12 (Modified EPA Method 1633) by SGS AXYS Analytical Services Ltd (Table S1). Briefly, after addition of isotopically labeled surrogate standards the sample was extracted with methanolic potassium hydroxide solution, with acetonitrile, and finally with methanolic potassium hydroxide solution, each time collecting the supernatants. The supernatants were combined, treated with ultra-pure carbon powder and evaporated to remove methanol. The resulting solution was diluted with water and cleaned up by solid phase extraction on a weak anion exchange sorbent. The eluate was spiked with recovery standards and analyzed using a Waters Acquity ultrahigh performance liquid chromatography coupled with a TQ-S Xevo tandem mass spectrometry (UPLC-MS/MS). Analysis of the sample extract was performed on the UPLC-MS/MS reversed phase C18 column using a solvent gradient. The column was coupled to a triple quadrupole mass spectrometer run at unit mass resolution in the Multiple Reaction Monitoring mode, using negative electrospray ionization. Initial calibration of the UPLC-MS/MS instrument was performed by the analysis of five or more calibration solutions. A mid-level calibration standard was analyzed to verify the initial calibration and injected after every 10 samples or every 12 hours, whichever occurred first, and at the end of the instrumental run sequence. Sample-specific detection limits were determined by converting the area equivalent to 3.0 times the estimated chromatographic noise height to a concentration in the same manner that target peak responses were converted to final concentrations. These limits accounted for any effect of matrix on the detection system and for recovery achieved through the analytical work-up. Reported data met established quality control acceptance criteria. Concentrations of PCBs, PBDEs, and PFAS in aqueous field blanks were two or more orders of magnitude below sediment values, demonstrating measured concentrations were representative of concentrations in the sediment and not from the sampling process.

Perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) were also measured in sediment samples and aqueous field blanks following EPA Method 8327 and included ultrasonic-assisted basified methanolic extraction and analysis using an Agilent 1100 LC and SCIEX Triple Quad 5500 QTrap MS with a liquid chromatography tandem mass spectrometer by the Water Quality Centre at Trent University. Quantitation was completed using isotopically labeled standards and matrix-matched calibration standards. All blanks were below levels of quantification (sediment, 0.04 ng/g wet weight; water, 0.4 ng/L), and spiked recoveries were all >80%.

Total concentrations of PCBs, PBDEs, and PFAS for each sediment, muscle tissue, and liver tissue sample from the PRZ and the test-mining area are reported in Table 1. All samples for each organic contaminant measured in sediment, muscle tissue, and liver tissue were summarized using the mean and range to characterize overall contamination in this region. We also conducted general linear regression analyses to better understand the relationship between fish tissue POP concentrations and fish size. A one-way analysis of variance (ANOVA) was conducted to determine significant differences in POP concentrations and contaminant distributions between sample sites where samples sizes allowed (where N ≥ k + 1). These same summary statistics and statistical comparisons were conducted for dioxin-like and non-dioxin-like PCBs. While this dataset is small, normal quantile plots were evaluated and showed that data were not significantly outside the bounds of normality. Despite this measure of normality, a non-parametric Kruskal Wallis comparison also accompanied one-way ANOVA tests. For all statistical tests that were possible, a 90th percentile significance level (α = 0.1) was used because true differences are harder to detect at lower samples sizes. As such, we reduced the bright-line used to identify differences to 90% confidence to guard against small sample sizes limiting the detection of potential true differences.

We estimated the proportion of PCBs in sediment and tissue samples that likely originated from historical industrial sources (Aroclors) and those that originated from emerging unintentionally released PCBs (non-Aroclors) to determine the contribution of Aroclor (legacy) PCBs versus non-Aroclor PCBs to the abyssal plain ecosystem. For this comparison Aroclor PCB congeners were identified as those that made up ≥1% of at least one of the following Aroclor formulations: 1016, 1242, 1248, 1254, 1260 (Frame et al., 1996; Agency for Toxic Substances and Disease Registry [ATSDR], 2000). We defined non-Aroclor PCBs as those that made up between 0.05% and 0% of the same Aroclor formulations. Those PCB congeners that made up between 0.05% and 1% were defined as low Aroclor PCB congeners. The proportional contribution of indicator PCBs (28, 52, 101, 118, 138, 153, and 180) to the summed total PCBs in sediment, muscle tissue, and liver tissue in both sampling areas were also determined. Last, the concentration profiles of all tested PCB congeners, PBDE congeners, and PFAS chemical species were compared graphically among sample areas and media to provide insight into transport mechanisms to, and accumulation in, this region of the abyssal plain. A principal component analysis (PCA) with sampling area (i.e., test-mining area and PRZ) as a supplementary variable was also conducted for sediment PCB and PBDE congener concentrations to provide insight into regional differences in PCBs and PBDEs transported to deep-sea sediments.

Toxic equivalence (TEQ), or the weighted total of dioxin-like PCBs relative to the most toxic form of dioxin, was also calculated for each sample. These values are calculated by assigning a toxic equivalent factor (TEF) established through international agreements to dioxin-like PCBs (Schecter et al., 2006). The TEQ is then calculated by multiplying the actual weight of each dioxin-like compound by its corresponding TEF and summing the results. These values were in pg/g wet weight and assuming that nondetectable concentrations were zero, half the reporting limit, and equal to the reporting limit.

Sediment concentrations of total PCBs in the area designated for test-mining ranged from 10.9 to 39.6 pg/g dry weight, while in the PRZ the concentration from the one multicorer sample was 35.5 pg/g dry weight (Table 1; Figure 2). For PBDEs, sediment concentrations ranged from 9.1 to 297.3 pg/g dry weight in the test-mining area and was 5.1 pg/g dry weight in the PRZ. Only two types of PFAS were analyzed in sediment samples: PFOS and PFOA (Table S1), and only PFOA was detected in the PRZ sample with a concentration of 1500 pg/g wet weight (Table 1).

Fish tissue concentrations of total PCBs and total PBDEs were several orders of magnitude higher than sediment concentrations, while concentrations of PFAS between the one sediment sample with PFOA and fish tissues were within an order of magnitude. Total PCBs in liver tissue (range of 57.7–122 ng/g wet weight; Table 1) were significantly higher (one-way ANOVA and Kruskal Wallis P < 0.01) than in muscle tissue (range of 0.67–2.46 ng/g wet weight; Figure 3). Total PBDE concentrations in fish muscle tissue (0.30–5.05 ng/g wet weight) and liver tissue (1.33–4.37 ng/g wet weight; Table 1) were similar (one-way ANOVA P = 0.40; Kruskal Wallis P = 0.38). A significance test could not be conducted to compare tissue PCB or PBDE by sample area due to small sample size when samples were broken down to sample area (Figure 3). Total PFAS in fish muscle (range of non-detectable to 16.6 ng/g wet weight) may have been greater (Kruskal Wallis P = 0.01; Table 1) than in liver tissue (range of non-detectable to 0.69 ng/g wet weight), though this relationship was not statistically significant (α = 0.10) for the ANOVA comparison (one-way ANOVA P = 0.12). Total concentration of PFAS in muscle tissue (one-way ANOVA P = 0.08) may have been greater in the test-mining area versus the PRZ based on ANOVA results; however, nonparametric results were not statistically significant (Kruskal Wallis P = 0.16; Figure 3).

Significant declines in POP contamination concentrations were seen with increased fish total length but whether these results were driven by spatial factors such as habitat where fish were collected or allometric factors relating to fish size and contaminant accumulation is not clear. For instance, only the smallest tested individual had detectable concentrations of PFAS in liver tissue resulting in a significant decline in liver PFAS with size when fish from the two sampling areas were combined (N = 5, P = 0.08, R² = 0.57). However, this result could be spatially driven as the one sample with detectable liver PFAS was from the PRZ while all others that did not have detectable liver PFAS were from the test-mining area. A decline in muscle tissue PCBs with an increase in fish total length was also evident when fish from both sample areas were combined (N = 4, P = 0.03, R² = 0.94). However, the two lowest total PCB concentrations were from the two largest fish collected in the PRZ, while the two highest total PCB concentrations were from the two smallest fish collected from the test-mining area (Table 1). Total PFOS concentrations in muscle tissue also showed a significant decline with fish total length in the PRZ (N = 7, P = 0.05, R² = 0.57). While this relationship was driven largely by the two smallest individuals (Table 1), both were collected from two different locations in the PRZ and therefore was not likely to be the result of where the fish were collected (Figure 1). No relationship between muscle tissue PFAS and fish total length was evident for individuals collected from the test-mining area (N = 8, P = 0.21), or between muscle tissue PBDEs and fish total length (N = 4, P = 0.14). PBDE in liver tissue, however, did demonstrate a significant increase in concentration with fish total length when samples from both sample areas were combined (N = 3, P = 0.046, R² = 0.99). This result was less likely to be driven by collection site given that the lowest and highest PBDE liver concentrations came from the same location (test-mining area; Table 1).

Relative PCB congener concentrations in sediment, fish muscle tissue, and fish liver tissue were similar in both sample areas in the eastern CCZ, meaning that those congeners that were generally higher in concentration than others were higher in both regions and those lower, lower in both regions (Figure 4). Indeed, a PCA using sampling area as a supplementary variable demonstrated the correlation in PCB congener concentrations between sampling areas (Figure 5). PCB congener profiles shifted from higher concentrations of less chlorinated PCB congeners to higher concentrations of more chlorinated PCB congeners from sediment to muscle tissue and from muscle tissue to liver tissue. For instance, PCBs in muscle tissue consisted largely of mono-, di-, tri-, tetra-, penta-, hepta-, and hexa-PCBs, while total concentrations of PCBs in liver tissue consisted largely of penta-, hepta-, hexa-, and octa-PCBs (Figure 4). However, for both sediment and muscle tissue in both sampling areas, PCB 11 and PCB 44 had higher concentrations than nearly all other PCB congeners. This result was not seen in liver tissue, where PCB 129, 153, and 180 were the dominant congeners.

To better understand the sources of PCBs to sediment and biota, congeners were depicted by their proportional contributions to the summed total, and those congeners associated with historical Aroclor mixtures were identified to understand their proportional contribution to PCBs in sediment, muscle tissue, and liver tissue (Figure 6). In sediment, Aroclor PCBs accounted for 49% of total PCBs in the PRZ and 51% in the test-mining area, while non-Aroclor PCBs made up 31% of total PCBs in the PRZ and 30% in the test-mining area. The low-Aroclor category could be assigned to either of the other two categories (Aroclor or non-Aroclor) given their extremely low concentrations (0.5%—1%) in historical Aroclor mixtures. In muscle tissue, Aroclor PCBs accounted for 56% of total PCBs in the PRZ and the test-mining area, while non-Aroclor PCBs made up 20% of total PCBs in the PRZ and 18% in the test-mining area. In liver tissue, Aroclor PCBs accounted for 70% of total PCBs in the PRZ and 73% in the test-mining area, while non-Aroclor PCBs made up only 3% of total PCBs in the PRZ and test-mining area. However, PCB 11 was the single PCB congener that had the highest proportion of total PCBs in sediment (18% in both sample areas) and one of the highest in muscle tissue (7% in test-mining and 8% in the PRZ; Figure 6). PCB congeners that also made up higher proportions of total summed PCBs in muscle tissue were PCB 44, 129, 153, and 180. The latter three of which were the highest relative PCB congeners reflected in liver tissue, with PCB 153 and 180 being commonly associated with Aroclor mixtures.

Previously, the common indicators PCB 28, 52, 101, 118, 138, 153, and 180 have been suggested to comprise the majority of total PCB in environmental samples because they make up the greatest proportion of historical Aroclor mixtures. However, for sediment and tissue samples from the abyssal plain in the eastern CCZ, indicator PCBs were not representative of total PCBs. Indeed, indicator PCBs comprised only 10% (test-mining area) and 11% (PRZ) of total summed PCBs in sediment, 15% (test-mining area) and 14% (PRZ) in muscle tissue, and 45% (test-mining area and PRZ) in liver tissue. The remaining proportion of PCBs, making up the majority in both sediment and fish tissues, were non-indicator PCBs.

Another subset of PCBs commonly tested are the World Health Organization 12 (WHO12) dioxin-like PCBs (77, 81, 105, 114, 118, 123, 126, 156, 157, 167, 169, 189) commonly measured to assess health risks (U.S. EPA, 2003; Fernandes et al., 2008; Domingo, 2017). Here, concentrations of dioxin-like PCBs in muscle and liver tissue were 0.11–0.16 ng/g wet weight and 3.33–5.07 ng/g wet weight in the test-mining area and 0.06–0.09 ng/g wet weight and 7.57 ng/g wet weight in the PRZ. The TEQs, assuming that undetectable concentrations of dioxin-like PCBs were zero, half the reporting limit, and equivalent to the reporting limit for sediment, muscle tissue, and liver tissue, are reported in Table 2.

PBDE congener concentrations varied by sample area, with the fully brominated PBDE congener (209; Figure 7) dominating total PBDEs in sediment and muscle tissue in the test-mining area. Sediment in the test-mining area also had lower concentrations of several other highly brominated congeners, while PRZ sediment had comparatively little and less brominated PBDE congeners. Muscle tissue PBDE congener concentrations were several orders of magnitude higher than sediment concentrations, though congener profiles were similar between sediment and muscle tissue in the test-mining area. Fish from the PRZ also had a greater concentration of PBDE 209 compared to other congeners, though the relative difference was smaller for the PRZ compared to the test-mining area. PBDE profiles of fish liver tissue also varied by sample area (Figure 7), with PBDE 37 having the highest concentration in the test-mining area and PBDE 100 and 154 having the greatest concentrations in liver tissue from fish collected in the PRZ.

All sediment PFOS and PFOA levels were below detection except for the sediment sample from the PRZ, which had a PFOA concentration of 1500 pg/g wet weight (1.5 ng/g wet weight). Conversely, fish tissue samples were analyzed for 40 different chemical species of PFAS (Table S1). Unlike sediment, PFOA was not detected in muscle or liver tissue. Instead, PFOS comprised all measured PFAS present in muscle tissue (range in the test-mining area of 0.83–16.6 ng/g wet weight; range in the PRZ of 0.49–3.92 ng/g wet weight). Liver tissue samples did not have detectable PFAS concentrations except for one sample from the PRZ. For the one liver tissue sample with detectable PFAS, PFUnA (0.396 ng/g wet weight) and PFTrDA (0.298 ng/g wet weight) were the two chemical species measured.

The long-term persistence, wide-ranging transport, bioaccumulative nature, and toxicity of POPs have resulted in global effects on aquatic ecosystems and human health (Breivik et al., 2002; AMAP, 2004; Breivik et al., 2007; UNEP, 2023a, 2023b). Here, we provide the first evidence that these contaminants have reached the remote ecosystem of the Pacific abyssal plain, with PCBs, PBDEs, and PFAS being present in sediment and accumulating in deep-sea biota associated with the seafloor. While there is no evidence of PCB-induced toxicity in any fish species at the total PCB concentrations measured here (Henry, 2015), there have been no investigations of PCB toxicity in deep-sea fishes, which may have greater sensitivity. Tolerance limits and screening level thresholds for human ingestion of fish have been set by regulatory bodies to limit health effects from PCB exposure through fish consumption (Simpson et al., 2023). Here, concentrations of total PCBs in abyssal tissue samples did not exceed the U.S. Food and Drug Administration’s (FDA) regulatory consumption limit of 2.0 µg/g (United States Food and Drug Administration, 2023). Additionally, while muscle tissue samples did not exceed the U.S. Environmental Protection Agency’s (EPA) consumptive screening level for human cancer risks of 12 ng/g, all liver tissue samples far exceeded this threshold (U.S. EPA, 2000). Toxic equivalence (TEQ) values provide the relative potential for human and wildlife consumptive health risks for each sample from dioxin-like PCBs (U.S. EPA, 2003; Fernandes et al., 2008; Giesy and Kannan, 2008; Domingo, 2017; Simpson et al., 2023). Here, the TEQ risk-based threshold of 0.15 pg/g wet weight (U.S. EPA, 2000) was exceeded in 3 of 4 samples of deep-sea fish muscle tissue when the concentrations below detection were assumed to be half of the detection limit. However, deep-sea organisms may have different sensitivities to dioxin-like PCBs than the organisms used to develop TEQs which may not be representative of deep-sea organism consumptive risks. Additionally, abyssal rattail fish are generally not harvested for human consumption.

PBDE production started in the 1970s, much later than PCBs (1920s). Despite the later start to production, the extensive usage of PBDEs along with their persistence have resulted in ubiquitous distributions in the environment (Wang et al., 2007; Besis and Samara, 2012). PBDEs are brominated flame retardants mixed with polymers without chemical bonding. As such PBDEs can easily leach from plastics into the surrounding environments (Guerra et al., 2010). While associated with numerous forms of human toxicity (Turyk et al., 2009; Bruchajzer, 2011; Wikoff and Birnbaum, 2011), total PBDE concentrations in fish muscle and liver tissue here were several orders of magnitude below current human consumption levels of concern (U.S. EPA, 2000; Integrated Risk Information System, 2008; Virginia Department of Health, 2010). For PFAS, Barbo et al. (2023) estimated that a single serving of fish per year, with the median level of PFAS found across the nation (11.8 ng/g wet weight), would result in a significant increase of human blood serum PFOS levels. Here, several abyssal seafloor-associated fish in this study had tissue concentrations close to or higher than this median concentration, suggesting that if rattail collected from abyssal depths were to be consumed, they could contribute to human PFAS exposure (von Stackelberg et al., 2017; Barbo et al., 2023).

The relatively large difference between sediment and tissue POP concentrations compared to freshwater ecosystems also seemed to suggest an additional and possibly more direct source of these POPs to rattail fish other than benthic prey (e.g., carrion), though to varying degrees. For instance, Pacific abyssal sediment PCB and PBDE concentrations were one to several orders of magnitude lower than those found in freshwater and estuarine ecosystems (Santschi et al., 2001; Elskus et al., 2020; Yeo et al., 2020; Yin et al., 2020; Wagner et al., 2022; National Centers for Coastal Ocean Science, 2023). Conversely, total PCBs in fish muscle and liver tissue were similar to those measured in freshwater fish muscle and whole body concentrations in bottom-dwelling fish collected across the U.S. (Stahl et al., 2009). Total PBDEs measured in Pacific abyssal fish muscle and liver were also similar, and in some cases greater than (Crane, 2017) total PBDEs in freshwater fish muscle and whole body concentrations for bottom-dwelling fish (Stahl et al., 2013) collected across the U.S. Additionally, despite sample collections from the same sites as PCBs and PBDEs, PFAS levels were undetectable in all but one sediment sample. However, that one sample had a concentration of PFAS considered high for freshwater soils and sediments across the U.S. and Canada (Awad et al., 2011; Vedagiri et al., 2018). This result suggests high spatial variability in PFAS concentrations across this region of the Pacific abyssal plain. However, more representative sampling of Pacific abyssal sediments will be needed in future studies to understand the extent of PFAS variability and identify areas of concern. PFAS measured in abyssal seafloor-associated fish muscle tissue were also comparable and even exceeded the median PFAS concentrations measured in fish from the Great Lakes, which were determined to have the highest measured fish tissue PFAS concentrations across the U.S. (Barbo et al., 2023). These high tissue concentrations were found even where sediment PFAS concentrations were undetectable.

Atmospheric deposition is considered a primary input source of POPs to the global ocean, particularly regions far from land (Wagner et al., 2019). The marine biological pump has also been recognized as a significant vertical transport mechanism for POP movement from surface waters to deep ocean habitats and ultimately benthic deposition (Dachs et al., 2002; Galbán-Malagón et al., 2012). Of biological pump export mechanisms, the slower transport process of smaller sinking particles (e.g., POM) would be expected to result in more spatially uniform distributions at the seafloor given that smaller particles are subject to more advective transport while sinking. In contrast, more rapid transport of larger biogenic particles and organisms (e.g., sinking carrion) would be expected to produce more spatially disparate POP distributions at the seafloor (Martí et al., 2001; González-Gaya et al., 2019).

Despite sediment collections from the same sample locations across the eastern CCZ, PCBs, PBDEs, and PFAS showed different relative spatial distributions. For instance, abyssal sediment PCB congener concentrations were relatively uniform compared to sediment PBDE congener concentrations, which were more variable based on collection site. Concentrations of abyssal sediment PFAS in the eastern CCZ were even more spatially variable than PBDEs, with levels below detection for all but one relatively high concentration of PFOA. While data were limited, the uniformity of PCB concentrations and congener distributions in abyssal sediments across the eastern CCZ suggests the primary driver of POP transport to the Pacific abyssal plain is likely a regional and consistent process, such as sinking particles (Ma et al., 2018). Studies have demonstrated that PCBs sorb to natural particles such as POM and zooplankton, as well as to plastics, all of which have been consumed by aquatic organisms (Mato et al., 2001; Karapanagioti et al., 2011; Bakir et al., 2012).

PBDEs also sorb to these same natural and anthropogenic particles, though research has suggested that, despite somewhat similar octanol-water partition coefficients (log K_ow range of 5.7–8.3 for PCBs, ATSDR, 2000; log K_ow range of 6.3–7.0 for PBDEs, U.S. EPA, 2017), PBDEs sorb with a somewhat lesser affinity compared to PCBs (Yeo et al., 2020). As such, the transport of PCBs to the seafloor through sinking particles may be more efficient than for PBDEs. Indeed, similar low concentrations of PBDE 47 and 99 were detected in sediment in both the test-mining area and PRZ. These results suggest that more highly brominated PBDEs may have degraded to less-brominated forms in the photic zone or that these lower brominated congeners, which have a higher vapor pressure, were transported atmospherically to this remote Pacific region before being transported to the seafloor through sinking particles that, at least partially, contributed to PBDEs in abyssal sediments in the eastern CCZ (Wang et al., 2007; Domínguez et al., 2010; Yeo et al., 2020). Conversely, PBDE 206, 207, 208, and particularly 209 were highest in sediment from the test-mining area, while in the PRZ none of these congeners was detected in sediment. These differences suggest that PBDE transport was also partially driven by a spatially limited and episodic process, such as sinking carrion and particulate waste from vertically migrating species.

The sporadic distribution and highly variable concentration of PFAS in Pacific abyssal sediment also suggests that PFAS were transported to the abyssal seafloor primarily via vertically migrating species and organisms that accumulate PFAS in shallower waters and sink to the seafloor after dying, where high concentrations of PFAS from carrion are incorporated into sediment and regional food webs through scavenging. Previous research has estimated that for emitted PFAS to circulate globally through deep ocean current formation would take at least 4500 years (Yamashita et al., 2008). Also, while research has identified that PFAS adsorb to zooplankton fecal matter in the Indian, Pacific, and Atlantic oceans, resulting in the vertical transport of PFAS to deeper waters (González-Gaya et al., 2019), the less hydrophobic nature of many chemical species of PFAS, compared to PCBs and PBDEs, may result in lower vertical transport efficiencies through sinking particles (Ng and Hungerbühler, 2014; Casal et al., 2017; Kadlec et al., 2024).

In freshwater and coastal ecosystems, sediment congener profiles are often reflective of congener profiles for bottomfish muscle tissue, particularly when food resources are primarily benthic (Castro-Jiménez et al., 2013; Sanganyado et al., 2021). Here, the profiles for abyssal sediment PCB and PBDE congeners and PFAS chemical species were only partially reflective of muscle tissue PCB congener profiles and were not reflective of PBDE congener profiles in the PRZ (though were in the test-mining area) or of PFAS chemical species. While sample size was limited, the differences in sediment and muscle tissue POP concentrations, and in congener and chemical species profiles between sediment and fish tissue, suggest a different and relatively highly concentrated source of these POPs to seafloor-associated fish other than benthic prey species (Castro-Jiménez et al., 2013). Thus, these differences provide additional support that carrion-fall may be an important and direct vector of POPs into deep-sea food webs, particularly for scavenging biota. Further, while results indicated a relationship between POP concentration and fish total length, there were not enough data to identify whether this relationship was driven by collection site or an allometric relation between fish size and contaminant accumulation. However, past studies have suggested that rattail fish have high mobility (e.g., Priede et al., 2003; Drazen et al., 2012) and have shown ontogenetic changes in diet as a function of body size (Pearcy and Ambler, 1974; Drazen et al., 2008; Stowasser et al., 2009; Harbour et al., 2021).

Congener profiles in environmental media can provide insight into the anthropogenic sources of POPs. Unfortunately, due to expense, many studies evaluating POP contamination only test for a select subset of congeners. For example, the most common PCB congeners targeted for analysis are described as indicator PCBs (28, 52, 101, 118, 138, 153, and 180) because they make up the greatest proportion of industrial technical mixtures (e.g., Aroclor mixtures; Melymuk et al., 2022) intentionally manufactured from the 1920s until production was banned globally in the 1990s. Previously, indicator PCBs have been found to have the highest concentrations in environmental samples given their long-term and widespread use (Melymuk et al., 2022). However, improved analytical techniques capable of quantifying all 209 PCB congeners indicate that non-Aroclor PCBs are an increasingly significant component of total PCB profiles, suggesting newer sources of PCBs are of increasing concern (Vorkamp, 2016; Bartlett et al., 2019; Liu and Mullin, 2019; Megson et al., 2019; Hermanson et al., 2020; Rodenburg et al., 2020). For example, PCB 11 (a non-Aroclor PCB) is a component of azo dyes and pigments (among the most important synthetic organic colorants in manufacturing) and is found increasingly in atmospheric, sediment, and tissue samples (Hu and Hornbuckle, 2010; Vorkamp, 2016; Anh et al., 2021; Mastin et al., 2022).

In the present study, the disproportionately high concentrations of non-Aroclor PCBs in sediment and muscle tissue demonstrate that unregulated and unintentionally released novel PCBs are accumulating in deep-sea sediments and biota (Mastin et al., 2022; Megson et al., 2022; Sobek et al., 2023). Indeed, the particularly high concentrations of PCB 11 suggest that current pigment and dye manufacturing processes are contributing substantially to PCB accumulation in the deep ocean (Mastin et al., 2022; UNEP, 2023a). Given that PCBs are entirely anthropogenic, and deliberate uses have been severely restricted for decades, the increasing prominence of novel unintentional PCBs in environmental media is concerning. A global monitoring plan (GMP) was enacted by the Stockholm Convention to evaluate the effectiveness of efforts to eliminate POPs from environmental media. However, the GMP does not require that all PCB congeners be measured and reported, including PCB 11. Consequently, these findings further demonstrate the need to include non-Aroclor PCBs, particularly PCB 11, in the GMP to track and regulate novel PCB accumulation in environmental media. The predominance of Aroclor PCBs in abyssal sediment and fish muscle and liver tissue also showed historical PCBs contributed to abyssal eastern CCZ ecosystems.

DecaBDE is a technical mixture used as a flame retardant in polymers (e.g., plastics), where the main component is PBDE 209. The predominance of this fully brominated PBDE in abyssal sediment and fish muscle tissue suggests that flame retardants incorporated into plastics were a primary source of PBDEs to the deep abyssal plain in the test-mining area of the eastern CCZ (Yeo et al., 2020). Given that PBDEs degrade to less-brominated PBDEs in environmental media, often through photodegradation, the high concentration of PBDE 209 in abyssal sediment suggests a recent, and relatively rapid, transport process that was protected from photochemical degradation (Domínguez et al., 2010; Eljarrat et al., 2011). Altogether, our results show that POP transport to one of the most remote locations on the planet may be relatively rapid and that other emerging contaminants should not be assumed to be absent from deep abyssal seabed ecosystems.

Another relevant consideration for POP accumulation in seafloor sediment in the eastern CCZ is the potential for disturbance through deep-sea mining efforts that would introduce sediment plumes at the seafloor and in midwater ecosystems through the release of mining effluent (Drazen et al., 2020; Levin et al., 2020). Plume and midwater effluent releases (at approximately 1200 m depth; Hauton et al., 2017) would include sediment-associated contaminants previously removed from shallow and mid-water circulation and fisheries through burial in the deep sea. Also, because these POPs are bioaccumulative, there is concern that resuspended POPs could enter midwater and deep-sea food webs. The lack of information on the ecological consequences of deep-sea mining makes it difficult for regulators, such as the International Seabed Authority, to develop toxicity thresholds that are adequate for minimizing the adverse effects caused by deep-sea mining (Kaikkonen et al., 2018). Ginzky et al. (2020) and Sanganyado et al. (2021) acknowledged that baseline data on the biological and chemical conditions of the deep seabed are required to ensure that future environmental monitoring programs for organic pollutants can adequately determine the effects of mining activities before impacts become widespread. One approach to monitoring deep-sea mining impacts is to preserve a portion of the seafloor in the eastern CCZ from mining (i.e., the PRZ) as a reference area to determine the level of impact in the mining area. However, for this approach to work the mining area and PRZ would need to be similar prior to mining so that any differences between the PRZ and mining area can be attributed to changes from mining and not natural differences between the two areas. Here, we evaluated the differences in POP concentrations and congeners profiles between the test-mining area and PRZ to determine if these areas were similar enough for this approach to work. Our results suggest that significant POP transport and concentration differences may exist between the two areas. However, more data are needed to fully characterize POP concentrations in the PRZ and test-mining area before any differences or similarities can be attributed to mining or lack thereof.

Data are freely available in supplemental material file (Data S1 and Data S2).

The supplemental files for this article can be found as follows:

Text S1. Analytical methods—organics analyses

Table S1. Per- and polyfluoroalkyl substances (PFAS) tested for using method MLA-110 Rev. 02 Ver. 12 (Modified EPA Method 1633) by SGS AXYS Analytical Services Ltd.

Data S1. POP data in sediment.

Data S2. POP data in fish muscle and liver tissue.

We are grateful to The Metals Company for providing the funds and resources to complete this work, particularly Dr. Leigh Marsh, Dr. Michael Clarke, and Claire Dalgleish, as well as the captain and crew of the Maersk Launcher for their professional support during participating campaigns of the NORI-D baseline project. We are also so very thankful to Dr. Jeffery Drazen from the University of Hawaii for providing support, leadership, and organization to the larger effort of characterizing baseline environmental conditions in the eastern CCZ. We also thank Sidney (Suma) Thomas from SGS AXYS, Inc. for their patience and assistance with the completion of analytical work. A very special thanks also goes to Rosalind Pinkard and Stephanie Swann for their steadfast support and help in managing this project.

The Metals Company, Inc. provided the funds for this project under contract with the University of Maryland to meet the requirements of an Environmental Impact Assessment under the International Seabed Authority.

The authors declare no competing interests. While the funds that paid for the work conducted herein was from a mining company, the contract between The Metals Company and UMD was in service of an independent environmental impact assessment required by the International Seabed Authority. Additionally, the contract between The Metals Company and UMD did not exclude the authors from publishing all data and interpretations of those data without influence from The Metals Company.

Conceptualization: DKS, LY.

Methodology: All authors.

Data and analysis: All authors.

Visualization: All authors.

Project administration and funding acquisition: DKS, TH, AKS, LY.

Writing and revising: All authors.

Approved submission of the manuscript: All authors.

How to cite this article: Sackett, DK, Anderson, D, Henry, T, Sweetman, AK, Yonkos, L. 2024. Persistent organic pollutant accumulation in Pacific abyssal plain sediments and biota: Implications for sources, transport, and deep-sea mining. Elementa: Science of the Anthropocene 12(1). DOI: https://doi.org/10.1525/elementa.2024.00042

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Guest Editor: Jeff Drazen, University of Hawaii at Manoa, Honolulu, HI, USA

Knowledge Domain: Ocean Science

Part of an Elementa Special Feature: Deep-Sea Mining of Polymetallic Nodules: Environmental Baselines and Mining Impacts from the Surface to the Seafloor