Mussels drive polychlorinated biphenyl (PCB) biomagnification in a coastal food web

Despite international regulation, polychlorinated biphenyls (PCBs) are routinely detected at levels threatening human and environmental health. While previous research has emphasized trophic transfer as the principle pathway for PCB accumulation, our study reveals the critical role that non-trophic interactions can play in controlling PCB bioavailability and biomagnification. In a 5-month field experiment manipulating saltmarsh macro-invertebrates, we show that suspension-feeding mussels increase concentrations of total PCBs and toxic dioxin-like coplanars by 11- and 7.5-fold in sediment and 10.5- and 9-fold in cordgrass-grazing crabs relative to no-mussel controls, but do not affect PCB bioaccumulation in algae-grazing crabs. PCB homolog composition and corroborative dietary analyses demonstrate that mussels, as ecosystem engineers, amplify sediment contamination and PCB exposure for this burrowing marsh crab through non-trophic mechanisms. We conclude that these ecosystem engineering activities and other non-trophic interactions may have cascading effects on trophic biomagnification pathways, and therefore exert strong bottom-up control on PCB biomagnification up this coastal food web.


Study site
We conducted the field experiment in a cordgrass dominated saltmarsh platform located on Blythe Island, Georgia (31°10¢47.8²N, 81°32¢02.1²W; Supplementary Fig. S1B). This site experiences diurnal tides and is inundated for 0-5 hours per tide. Porewater salinity ranged from 20‰ to 47‰ (35 ± 2 ‰) and average air temperatures ranged from 21°C in April to an average of 27°C in August 2017. Blythe Island is adjacent to Turtle River, where chemical contaminant data has been well documented [1][2][3] . PCBs in the area are generally attributed to Aroclor 1268.
Aroclor 1268 is a highly chlorinated mixture of PCBs that was previously used by LCP from 1955 -1994 to lubricate high voltage equipment during the manufacturing of chloralkali. After 39 years of discharging hazardous waste into holding pits adjacent to the saltmarsh and directly into Purvis Creek, a tributary of the Turtle River, the U.S. EPA designated the LCP facility and 813 acres of impacted tidal marshlands and uplands a Superfund site in 1996 2 .
Remediation via burying and dredging sediment was conducted at the Superfund site between 1994 to 1999 on 136 of the 813 identified impacted acres 4 . Despite these efforts, PCBs continue to be documented in the sediment 2 as well as in benthic 5 and pelagic 3,6-8 biota throughout the Turtle-Brunswick River estuary and adjacent barrier islands 3,9 that are intensively commercially and recreationally fished despite fishing advisories 3,10 . Due to the available research in the area, Blythe Island served as an ideal location to assess how common benthic salt marsh macro-invertebrates, and suspension-feeding bivalves in particular, may influence PCB assimilation into coastal food webs.
Stable isotope sample preparation Similar to Nifong (2016) 11 , stable isotope values (δ 13 C and δ 15 N) are expressed in standard per mil notation (‰): where X is the element of interest and R is the ratio of heavy to light isotopes ( 13 C/ 12 C or 15 N/ 14 N) of the sample and standard (Vienna Pee Dee Belemnite used for δ 13 C and Atmospheric Nitrogen-AIR for δ 15 N). Instrument accuracy was measured and corrected for each sample run using five measures of in-lab standard USGS-40 (l-glutamic acid) with δ 13 C = -26.39 and δ 15 N = -4.52.
Analytical instrument error for USGS-40 was 0.08 for δ 15 N and 0.10 for δ 13 C across all runs. Post-GPC extracts were solvent exchanged under nitrogen (TurboVap®) from DCM to hexane (final volume of 0.5mL). Lastly, we used alumina solid phase extraction (SPE) (~1.1 -1.15g of 5% water-deactivated alumina) as a final cleanup step, resulting in a final volume of ~0.5mL. A recovery standard (δ-hexachlorocyclohexane) was then added prior to instrumental analysis to evaluate internal standard recoveries.

Instrumental analysis and data quality
Samples were analyzed for 100 PCB congeners (Supplementary Table S1) using an Agilent 6890 gas chromatograph (GC) equipped with a 5973 Mass Selective Detector operated using electron impact ionization. Data was acquired in selected ion monitoring mode. The GC included a DB-XLB column (J&W; 30 m × 0.25 mm diameter × 0.25 μm) and a programmable temperature vaporization inlet. Congener concentrations above our analyte specific method detection limit 12 were summed for a total PCBT concentration in each sample. Figure S1: Cordgrass density. Stem density was recorded prior to the experimental setup and every 4-6 weeks over the 5-month experiment. The data are reported as the mean ± standard error of 6 replicate plots per date. Figure S2: Baseline PCB concentrations for crabs sampled from Sapelo Island, GA. PCBs (ng/g ww) are lipid normalized and shown as mean ± standard error of 3 extractions taken from a homogenized composite of whole marsh crabs (n=143) and a homogenized composite of whole fiddler crabs (n=143). The large standard error for marsh crabs is due to the fact that that composite of crabs were sampled from both on and off mussel mounds. Table S1. PCB congeners sought in various matrices during this study. For each sample, those detected above the analyte specific method detection limit were summed and reported as a total PCB concentration for that sample. Co-planars analyzed are in bold and congeners that represent PCB1268 are italicized.  Supplementary Table S2. Effects of mussels on dioxin-like toxicity. Experimental treatment effects on the concentration of five coplanar PCBs pg/g ww (dw in parentheses for sediment and pseudofeces). ND = not detected above method detection limit. If a coplanar was detected in multiple treatment enclosures concentrations are shown as mean ± standard error. TEF values per coplanar PCB is provided in the first row. The average total coplanar concentration per treatment and their toxic equivalency values from coplanar PCBs in sediment, fiddler crabs and marsh crabs are also shown as the mean of 3 replicate enclosures ± standard error. The last column contains TEQ values reported as mean; min -max pg/g ww (dw in parentheses for sediment and pseudofeces) for dioxin-like PCBs from this study as well as from other invertebrates previously reported on in the literature.  Supplementary Table S3. Summary of response variables measured in each treatment. A "Y" signifies that the response variable was measured for the specified treatment. Cordgrass and benthic algae variables, porewater salinity, and soil temperature were measured in every plot per treatment (n=6 replicate plots per treatment), invertebrate densities were measured in every enclosure with invertebrates, stable isotopes were measured in every plot of identified treatments with the exception of some plots used for PCB analyses due to lack of enough available material. For all PCB related response variables, half of the replicate enclosures per treatment (n=3) were used for these analyses (Supplementary Fig. S1C). Fiddler crabs and mussels Marsh crabs and mussels Fiddler crabs, marsh crabs, and mussels