The Effect of Microplastic on the Uptake of Chemicals by the Lugworm Arenicola marina (L.) under Environmentally Relevant Exposure Conditions

Abstract

It has been hypothesized that ingestion of microplastic increases exposure of aquatic organisms to hydrophobic contaminants. To date, most laboratory studies investigated chemical transfer from ingested microplastic without taking other exposure pathways into account. Therefore, we studied the effect of polyethylene (PE) microplastic in sediment on PCB uptake by Arenicola marina as a model species, quantifying uptake fluxes from all natural exposure pathways. PCB concentrations in sediment, biota lipids (C_lip) and porewater measured with passive samplers were used to derive lipid-normalized bioaccumulation metrics C_lip, Biota sediment accumulation factor (BSAF), Bioaccumulation factor (BAF) and the Biota plastic accumulation factor (BPAF). Small effects of PE addition were detected suggesting slightly increased or decreased bioaccumulation. However, the differences decreased in magnitude dependent on the metric used to assess bioaccumulation, in the order: C_lip > BSAF > BPAF > BAF, and were nonsignificant for BAF. The fact that BAF, that is, normalization of C_lip on porewater concentration, largely removed all effects of PE, shows that PE did not act as a measurable vector of PCBs. Biodynamic model analysis confirmed that PE ingestion contributed marginally to bioaccumulation. This work confirmed model-based predictions on the limited relevance of microplastic for bioaccumulation under environmentally realistic conditions, and illustrated the importance of assessing exposure through all media in microplastic bioaccumulation studies.

Introduction

It has been hypothesized that ingestion of microplastic increases exposure of aquatic organisms to hazardous contaminants.¹⁻³ This increased chemical exposure is often perceived as a major concern. Because chemical sorption to polymers is reversible, the transport of chemicals via plastic is possible in two directions: both transporting chemicals from plastic into organisms, and transporting chemicals from the organisms lipids into the plastic (“cleaning”).^4,5 Some laboratory studies have shown an elevating effect of plastic in food on the uptake of chemicals,⁶⁻¹² whereas others found no effect, for all or some of the chemicals studied.^7,12−15 These studies often used an experimental design where parallel uptake from water or food/sediment was not explicitly considered, rendering them less ecologically relevant for conditions in nature, where these parallel uptake pathways do occur.^16,17 Instead, often plastic loaded with persistent organic pollutants (POPs) was considered as the only route for chemical uptake. Furthermore, usually (relatively) clean organisms were used, which forces chemical transfer from the plastic to the organism. Modeling studies have attempted to asses more environmentally relevant conditions systematically, by (a) including the environmental uptake pathways water, sediment and food, and (b) accounting for the fact that organisms in the environment already would be chemically contaminated, which reduces the fugacity gradient driving chemical transfer. These studies indicated that the uptake of plastic-associated chemicals in organisms would be a minor contribution to total bioaccumulation in more environmentally realistic scenarios.^{4,5,13,16−20} Therewith, there tends to be growing consensus among recent studies by various groups that plastic in the environment will have minor effects on bioaccumulation in organisms.^{4,5,7,13−20} Although first-principles on chemical partitioning and kinetics suggest this, there is a lack of empirical environmentally realistic studies confirming these modeling outcomes. The few experimental studies that included uptake routes other than plastic so far indicated no measurable vector effect of plastic on bioaccumulation.^7,14,15,21 Quantification of chemical uptake from all environmental pathways jointly is however still lacking, as well as normalization of the bioaccumulation on lipid content of the organisms. Additionally, in the relatively small volume of the bioassays in previous studies the addition of an extra absorbing pool, namely the plastic, diluted porewater POP concentrations.^4,7 This hampered the comparability of outcomes with environmentally relevant settings, where the excess availability of POPs from the surrounding media is virtually infinite. Therefore, empirically, the lack of effect of plastic on bioaccumulation under environmentally realistic exposure conditions with all uptake pathways quantified and accounted for yet has to be proven.²²

The aim of the current study was therefore to assess the role of microplastic as a vector of plastic associated chemicals for Arenicola marina (L.) (lugworms) under environmentally relevant exposure conditions and with full quantification of all exposure pathways. Sediment, porewater and two realistic plastic doses were PCB spiked and equilibrated for 6 weeks after which PCB concentrations were assessed in sediment and in the porewater with the aid of polyoxymethylene (POM) passive samplers.^23,24 This enabled determination of the in situ partitioning of PCBs among biota lipids, plastic, porewater and sediment. There were two exposure scenarios. Since the addition of clean plastic was expected to slightly dilute the chemical concentrations, the first scenario was referred to as the “chemical dilution” (CD) scenario. Another set of PCB congeners was used to represent a second scenario. The spiked quantity of these PCB congeners was slightly increased to roughly a priori compensate for the anticipated PCB dilution by plastic addition. This second scenario aimed to represent open seafloor conditions where PCBs can be considered present from a virtually infinite source,⁴ which is why the second scenario is referred to as the “Infinite Source” (IS) scenario. Lugworms were chronically exposed for 28 days with end points survival, feeding activity, growth, lipid content and PCB bioaccumulation. As such, we focused on relevant scenarios where (a) as in nature, chemicals spread out over the environmental compartments and parallel uptake pathways exist, (b) environmentally relevant low plastic and chemical doses were used. We used the polymer high density polyethylene, because of its relatively high affinity for POPs and large global production. A mixture of environmentally relevant particle sizes in the smaller microplastic size range was used (10–180 μm).^25,26 Ten PCB congeners were used as a proxy for environmentally sorbing and native plastic associated chemicals, spanning a wide hydrophobicity range. The ∑PCB concentrations in the sediment were a factor >24 lower than the reported NOEC, and thus were too low to cause any toxic effect.²⁷ In bioaccumulation assessment for benthic invertebrates, several bioaccumulation metrics are usually applied. Biota sediment accumulation factors (BSAFs) correct for variations in lipid content (f_lip) and organic matter content (f_OM) of organisms and sediment, respectively.²⁸ Hence, for bioassays with microplastic, this metric can reveal whether significant differences in bioaccumulation exist between organisms that were or were not exposed to plastic, as they are supposed to eliminate the above-mentioned differences between organism and sediment characteristics. However, BSAF is composed of four measured variables ((C_organism/f_lip)/(C_sediment/f_OM)), which makes this type of metric inherently sensitive to error propagation and may limit its rigor in detecting subtle differences in bioaccumulation.²⁸ Instead, Bioaccumulation factors (BAFs) correct observed bioaccumulation for porewater concentrations ((C_organism/f_lip)/(C_PW)) and thus more directly eliminate differences in bioaccumulation caused by differences in chemical concentrations in the porewater of the sediment. Here we report for the first time effects of microplastic on lipid normalized bioaccumulation in lugworms, lipid and organic matter (OM) normalized BSAFs and BAFs, and use and evaluate these metrics with respect to the vector effect of plastic on bioaccumulation. Furthermore, we introduce and evaluate the Biota Plastic Accumulation Factors (BPAF = (C_organism/f_lip)/C_plastic) as a new metric relevant for the assessment of bioaccumulation from microplastic. Furthermore, the relative importance of PCB uptake pathways for the various scenarios was assessed using a plastic-inclusive biodynamic model.^4,18,20

Materials and Methods

Materials

Polyethylene (PE, green fluorescent UVPMS-BG, spherical, diameter 10–180 μm, density 0.94 kg/L)²⁴ was used in the bioassay. PE polymer identity was confirmed by FTIR (ThermoFisher, iN10 MX). For microscope images and particle size distributions of the PE the reader is referred to the publication by Velzeboer et al.²⁴ Polyoxymethylene sheets (POM, 76 μm thickness) were employed as passive samplers.^23,24,29 The selected PCB congeners were 28, 31, 44, 52, 101, 118, 138, 153, 170, and 180. Further details are provided in the Supporting Information (SI).

Sediment Sampling and Pretreatment

The sediment was sampled from the Eastern Scheldt (The Netherlands).^7,30 PE was added to the sediment, accomplishing plastic concentrations of 0, 0.05 and 0.5% DW, which are within and above the range found in the marine environment, respectively.^7,31−33 Subsequently, the sediment-plastic mixture was spiked with the PCB congeners and mixed for 6 weeks. During the last 4 weeks of mixing, three POM passive samplers (≈0.3 g each)^23,24,29,34 were added to each PE-sediment mixture for determination of porewater PCB concentrations. Six and 4 weeks have been shown sufficient to reach chemical equilibrium between sediment porewater, and 10–180 μm PE particles and POM passive samplers, respectively,^{4,23,24,34−42} for PE also is confirmed by the linearity of the logK_PE – logK_OW plot (SI Figure S7). Further details are provided as SI.

Experimental Set Up

Glass aquaria with dimensions 16 × 16 × 16 cm were filled with 4 kg wet PE-containing PCB equilibrated sediment (3.2 kg DW, ± 8.7 cm thick layer) and covered with stainless steel gauze with a mesh size of 2 mm to prevent exchange of lugworms. These aquaria were placed per five replicates in large (80 × 40 × 40 cm) aquaria, following previously published procedures.⁷ Subsequently, ± 90 L of seawater from the Eastern Scheldt (The Netherlands) was added. After a two week stabilization period, the bioassay was started by adding pools of 5 individuals of A. marina to each small aquarium. Following our previous bioassays, no extra food source was provided.^7,43 Three times a week, water quality was measured, and about 30 L of overlying water was refreshed. Further details on the maintenance of the systems and the test organisms can be found in the SI.

Treatments

In this study we aimed at scenarios with and without a diluting effect of PCBs in the environment by plastic, referred to as the chemical dilution (CD) scenario and the infinite source (IS) scenario. These two chemical exposure scenarios were combined within the same experimental units to eliminate any influence of biological variability in the comparison of the two exposure scenarios. We achieved this by spiking the sediment plastic mixtures with pairs of chemically comparable PCB congeners. PCB pairs were 28 and 31*, 52 and 44*, 101 and 118*, 153 and 138*, and 180 and 170*. Within each of these pairs of chemically comparable PCB congeners, one of the congeners was spiked equally among all treatments (CD scenario, ∑PCBs ≈ 5 μg/kg DW sediment mixture). The other congeners per pair, the IS scenario PCB congeners (congeners marked *), were spiked in higher quantity in the treatments with PE to compensate for the anticipated dilution effect of the added PE (0.05% PE treatment: ∑PCBs* ≈ 7 μg/kg DW sediment mixture; 0.5% PE treatment: ∑PCBs* ≈ 23 μg/kg DW sediment mixture). Two plastic free treatments were included. The first one, referred to as ‘0% PE A’, consisted of the CD scenario PCB congeners plus an equal spike of IS scenario PCB congeners (∑PCBs* ≈ 5 μg/kg DW sediment mixture). This 0% PE A treatment did not receive an enhanced spike of the IS scenario PCBs, because no dilution by added PE would occur at 0% PE. The second plastic free treatment (‘0% PE B’) contained the same CD scenario PCB congener spike concentrations as the ‘0% PE A’, but now with the IS scenario PCBs at the higher spiked quantity of the 0.5% PE treatment (∑PCBs* ≈ 23 μg/kg DW) (SI Figure S1). The extra spiked quantities of the IS scenario PCB congeners were designed in such a way that the porewater concentrations in the sediment were expected to be similar in the PE free and PE containing treatments. The latter was based on a priori estimates of the organic matter-water and PE-water partitioning coefficients K_OM(44) and K_PE,⁴⁵ respectively (SI Table S2). In this experimental design, the effect of PE on bioaccumulation in the CD scenario can be seen from the difference between concentrations of the CD scenario PCB congeners among all four treatments. The effect of PE on bioaccumulation in the IS scenario can be seen from the difference between concentrations of the IS scenario PCB congeners in the 0% PE B and the 0.5% PE treatment. Comparison of the IS scenario PCB congeners in the 0% PE A, 0.05% PE, and 0.5% PE treatment, is a comparison of systems with a designed similar porewater concentration. Note that as K_OM and K_PE in the design phase were estimated using literature values,^44,45 an exact compensation for the dilution effect leading to identical porewater concentration was not anticipated. PCB concentrations added to the sediment mixtures are listed in SI Table S1.

End Points

During the exposure assay, mortality was assessed daily and dead lugworms were removed. Feeding activity was assessed following previous procedures, as the number of faeces heaps produced per organism per day,^7,46−48 and additionally as mass of faeces heaps produced per organism per day. The latter was done by flattening all sediment surfaces with a spatula at the 27th day, and subsequently collecting all faeces heaps at the 28th day. Of these faeces, wet weight (WW), dry weight (DW, 60 °C during 24 h) and ash free dry weight (AFDW, 600 °C during 2 h) were determined and corrected for the number of surviving organisms to calculate faeces weight produced per organism per day. Results were corrected for the initial polyethylene fraction of the sediment (f_PE) to estimate the OM content of the faeces. Thereby the assumption was made that all PE burned during the AFDW determination. After the exposure period of 28 days, the lugworms were transferred to clean seawater to clear their guts overnight.^7,43 Lugworms were rinsed with demineralized water, air-dried on tissue paper for 15 min and subsequently stored at −18 °C. After defrosting and homogenization of the tissue, WW, DW (≈1 g tissue per pool), AFDW, lipid content⁴⁹ and PCB concentrations were analyzed.

PCB Analysis and QA

PCB concentrations of initial (i.e., t = 0 days) porewater, initial and final (i.e., t = 28 days) sediment and initial and final lugworm tissue were determined. Sample preparation and PCB analysis followed previously published procedures.^7,50,51 For details the reader is referred to Besseling et al. 2013.⁷ Porewater concentrations (C_PW) were determined by analyzing the POM passive samplers.³⁴ After the 4 weeks equilibration, POM strips were rinsed with demineralized water, air-dried for 15 min on tissue paper and stored at 7 °C until analysis. Recoveries of PCB congeners were determined in triplicate and averaged for the individual congeners 83.6 ± SE 1.5%. PCB concentrations were corrected for procedural blanks. The concentration of PCB congener 52 was only incidentally identified in initial lipids as well as in the sediment at the start and end of the bioassay, and therefore left out of further analysis of the these metrics.

Data Analysis

Normality of the data and equality of variances were tested with a Shapiro-Wilk Normality test and Levene’s test, respectively. Linear regression (LM), ANCOVA, Kruskal–Wallis, Tukey HSD and Nemenyi–Damico–Wolfe–Dunn (NDWD) tests were performed with R statistical software (R Development Core Team), with a significance level of α = 0.05. Unless stated otherwise, results are reported ± standard error. C_PW,t=0 was derived from the passive sampler data using measured PCB partitioning coefficients to 76 μm POM provided by Hawthorne et al.²³ (K_POM, SI Table S2, eq S1). Calculation methods for C_PW, t=28, K_OM, K_PE, C_PE, C_OM, BSAF, BPAF, and BAF are provided in the SI (eq S2–S9). We included a figure that illustrates, theoretically, the various scenarios associated with each of the bioaccumulation metrics BSAF, BPAF, and BAF (SI Figure S2).

Bioaccumulation Modeling

Bioaccumulation was modeled according to Koelmans et al.⁴ For details, the reader is referred to the SI.

Results and discussion

Lugworm Survival, Feeding Activity, Weight, and Lipid Content

Overall lugworm survival was 81%, with no significant differences between the treatments (SI Figure S3A). This implies that the PCBs nor the PE additions had physical or chemical effect on survival. Feeding activity started in all treatments within the first week of exposure. The feeding activity expressed as number of heaps per individual worm was significantly highest in the treatment with 0.05% PE and lowest in the treatment with 0.5% PE (SI Figure S3B, ANOVA, p-value = 0.010, Tukey HSD, p-value = 0.006). However, the size of the faeces heaps was observed to be rather variable. Furthermore, individual heap count seemed to be affected by the number of alive lugworms (linear regression, treatment and survival both explanatory for activity (heaps/individual/day), R² = 0.91). This can be explained by a high feeding activity causing the heaps to be less well distinguished from one another. Heap production in mass might therefore be a better indicator of lugworm condition than number of heaps and is more relevant as a relative measure of egestion rate. In the 0% PE treatments the heap mass production per individual varied from 9.4–19.9 g DW, whereas in the treatments with PE this was significantly lower by a factor of 2, that is, only 4.1–11.6 g DW (SI Figure S3C, ANOVA, p-value = 7.20 × 10^–3, Tukey HSD, p-values ≤0.049). After correction by f_PE, the only significant difference in f_OM in the heaps was the f_OM of the 0% PE A treatment being higher than that of the 0% PE B treatment (SI Figure S3D, black markers, Kruskal–Wallis, p-value = 0.012, NDWD, p-value = 0.022). The concurrence of lowest feeding activity with highest PE concentration confirms the negative effect of plastic on feeding activity that was previously observed for 7.4% PS⁷ and 5% UPVC.⁵² Weight loss was calculated as average weight of the surviving lugworms divided by average initial weight in that pool. The initial pooled average weight varied between 3.0 and 4.1 g WW/individual. After 28 days of exposure this was for the surviving lugworms reduced to 2.5–3.5 g. This weight loss of 3.7–28.8% did not significantly differ among the treatments (SI Figure S3E). On average the lipid fraction was 1.8 ± SE 0.03%. No significant difference in lipid fraction of the lugworms exposed to the different treatments was found (SI Figure S3F).

PCB Concentrations in Porewater and Sediment

PCB concentrations in porewater at the start of the bioassay (C_PW,t=0, Figure 1) can be used to check whether PE indeed diluted concentrations of CD scenario PCB congeners in the porewater and whether concentrations of IS scenario PCB congeners in the porewater remained more similar among treatments. Overall, C_PW values ranged between 0.001 and 10 ng/L, therewith being environmentally relevant.^53,54 PE treatment, chemical exposure scenario and K_OW were significant explanatory variables of C_PW (ANCOVA, R² adj = 0.91, p-values ≤ 0.021). The negative trend of C_PW with logK_OW in Figure 1 was explained by spiking in similar quantities (SI Table S1), which resulted in lower PCB concentrations in the porewater with increasing hydrophobicity of the congeners. The C_PW of the IS scenario PCB congeners was somewhat higher and showed a higher variability among treatments than that of the CD scenario PCB congeners, which was explained by the higher spiking concentrations of the IS scenario PCB congeners (SI Table S1). As such, the extra spiking of the IS scenario PCB congeners turned out to be an overcompensation for the sorption by PE. This difference from exact compensation could be expected, since literature K_PE values were used for the experimental design,⁴⁵ which were in line with, though not exactly equal to, measured K_PE values (SI Figure S7).²⁴ Per scenario, the order of C_PW among the different treatments was generally equal for all congeners (Figure 1). Although designed to have a similar C_PW, C_PW of the CD scenario PCB congeners was a factor 1.9 lower in the 0% PE A than in the 0% PE B sediment at the start of the experiment (0.19–0.41 log unit, SI Figure S6, ANCOVA, R² adj = 0.97, p-values ≤ 3.35 × 10^–4). IS scenario PCB congeners had a relatively low C_PW in the 0% PE A sediment too. We explain the lower than expected C_PW of all congeners in the 0% PE A sediment by sediment heterogeneity during the preparation phase or random variability. For instance, spiking the CD scenario PCB congeners caused already an average factor 1.1 lower concentration of PCBs in the 0% PE A sediment compared to the 0.05% PE sediment. After the preparation phase of 6 weeks sediment mixing, porewater PCB concentrations were a factor 1.1–1.8 lower in the 0% PE A sediment compared to the 0.05% PE sediment, even before exposure started. A plausible explanation for these lower C_PW values is a higher f_OM, or a different OM quality, in the initial 0% PE A sediment compared to the other treatments, which also is consistent with the aforementioned faeces heaps OM content. In the sediment containing 0.05% PE, C_PW of the CD scenario PCB congeners was higher compared to the 0% PE (A) sediment (a factor 1.8 increase, range 1.2–2.6, Tukey HSD, significant for 3 out of 5 congeners, p-values ≤ 0.015), and statistically identical compared to the 0% PE (B) sediment (factor 0.94, Tukey HSD, p-values ≥ 0.198). At the ten times higher PE dose of 0.5% PE, C_PW of the CD scenario PCB congeners was insignificantly different compared to the 0% PE A sediment (Tukey HSD, p-values ≥ 0.057) and was reduced compared to the 0% PE B sediment by a factor 3.5 (1.9–5.6, Tukey HSD, p-values ≤0.011). The latter phenomenon has previously been referred to as the ‘dilution effect of plastic’.^4,7C_PW of the IS scenario PCB congeners were in the 0.05% PE sediment equal (Tukey HSD, p-values ≥ 0.231) and in the 0.5% PE sediment elevated compared to the 0% PE A sediment (Figure 1, dark gray color, factor 1.4–2.3 increase, Tukey HSD, significant for 3 out of 5 congeners, p-values ≤0.022). This was caused by the extra spike but also implied that the PCBs sorbed less to PE and OM than a priori assumed. After all, the extra spike of the IS scenario PCB congeners was designed to keep the C_PW in the treatments with PE more or less constant compared to the 0% PE A treatment. The difference between these sediments was also increased by the random variability that lowered the C_PW of all PCB congeners in the 0% PE A treatment. Nevertheless, a substantial part of the extra spike was indeed sorbed to the PE, as one can see by comparison with the 0% PE B treatment. In the latter sediment, as expected, the extra spike of IS scenario PCB congeners in absence of PE resulted in higher porewater concentrations (factor 8.2–22.5). The extra spike of IS scenario PCB congeners thus compensated for the dilution effect as planned, although at the same time it was not fully representative of an open seafloor scenario where porewater concentrations would have been the same among environments with and without PE (0% PE A, 0.05% PE, and 0.5% PE). This, however, does not interfere with interpretation of treatment effects, as will be discussed later on.

Figure 1.

Average C_PW values ± SE 0.011–0.075 (not shown) at t = 0 days measured with passive samplers. CD scenario PCB congeners: 28, 52, 101, 153, 180. IS scenario PCB congeners: 31, 44, 118, 138, 170. LogK_OW values from Van Noort et al.,⁶²SI Table S2.

Concentrations in OM and PE at the start of the experiment (t = 0 days), and in porewater, OM and PE after exposure at t = 28 days were calculated (SI Figures S4, S7), using measured total (sediment plus PE) concentrations in sediment at t = 0 and 28 days, porewater concentrations at t = 0 and assuming OM-porewater and PE-porewater partition coefficients to remain constant during 28 days of exposure (SI, eqs S2–S6). PCB concentrations in porewater did not differ substantially between 0 and 28 days (SI Figure S5), hence constant exposure was concluded and C_PW,t=28 was used to derive the accumulation factors discussed hereafter.

OM-Porewater and PE-Porewater Partitioning Coefficients

Partitioning coefficients to OM (K_OM) were calculated from C_PW,t=0 and the PCB concentrations in the sediment (C_SED,t=0) and are close to literature values⁴⁴ (SI Figure S6). Because of the measured, as designed, statistically identical porewater concentrations of the CD scenario PCB congeners in the 0% PE B and 0.05% PE sediment, K_OM values of the 0% PE B sediment were used to calculate K_PE values in the sediments with PE. The following formula was used:

1

with K_p-total being the PCB partitioning to the total mixture of sediment including OM and PE, and f_OM and f_PE being the fraction OM and polyethylene in that sediment mixture, respectively. This follows the same procedure as Rakowska et al.,^55,56 who derived activated carbon partitioning coefficients in mixtures of activated or black carbon and sediment. LogK_PE showed a linear increase with logK_OW (SI Figure S7, LM, p-value = 2.17 × 10^–12). LogK_PE values did not differ significantly between the treatment with 0.05 and 0.5% PE. Furthermore, they were in line with previous findings by Lohmann et al.⁴⁵ and Velzeboer et al.²⁴ This conformance in K_PE among PE treatments and literature studies that used pure PE, confirms the reliability of the presently used procedures. The PCB concentrations in PE (C_PE) were calculated from K_PE and C_PW according to SI eq S5 and ranged up to about one μg/g (SI Figure S8). The above analysis provided a clear view on the chemical concentrations in all relevant media, water, sediment and PE, and how PE additions in the different treatments changed these concentrations. This facilitates the interpretation of the data on the effects of PE additions on bioaccumulation.

Effects of PE on Bioaccumulation

Effects of PE treatments were assessed first by evaluating lipid normalized PCB concentrations (C_lip) in the lugworms, representing a direct measure of bioaccumulation (Figure 2, Figure 3A). After 28 days of exposure to 0.05% PE, concentrations of CD scenario PCB congeners in lugworm lipids were increased by on average a factor 2 (1.6–2.5, significant for four out of five congeners (not for PCB 180) compared to the 0% PE A treatment (Tukey HSD, p-values ≤ 4.33 × 10^–4). This factor difference is similar to that found in earlier studies that assessed parallel pathways,^7,12 and complies with the general conclusion by Koelmans et al.¹⁷ that a factor two increase or decrease may occur due to complex counteracting mechanisms affecting accumulation. However, compared to the 0% PE B treatment, C_lip remained equal or was lower after exposure to 0.05% PE by on average a factor 1.3 (1.1–1.7, Figure 2, Figure 3A, SI Figure S9, significant for one out of five congeners, Tukey HSD, p-value = 0.047). Exposure to sediment containing 0.5% PE caused bioaccumulation of the CD scenario PCB congeners to be on average a factor 1.6 (1.1–2.2) lower for all five congeners compared to the 0% PE A treatment, and even a factor 4.2 (3.2–6.1) compared to the 0% PE B treatment. This suggests that the factor two increase in apparent bioaccumulation observed at the 0.05% compared to the 0% PE A treatment is more than compensated for by chemical dilution at a PE dose of 0.5%. One could explain the increased bioaccumulation after exposure to the 0.05% PE compared to the 0% PE A treatment by release of PCBs from the ingested PE (vector effect). However, the lack of this increase when comparing the 0.05% PE to the 0% PE B treatment, in which C_PW was similar, makes it more likely that the low C_PW observed in the 0% PE A treatment (Figure 1) explains the low C_lip in that treatment. For the five IS scenario PCB* congeners, bioaccumulation was strongly elevated after exposure to the 0% PE B treatment compared to the 0% PE A treatment (Figure 2, Figure 3A, SI Figure S9, ANOVA, p-values ≤1.70 × 10^–5, Tukey HSD, p-values ≤ 3.74 × 10^–5), which is explained by the higher spike quantity used. For the IS scenario PCB* congeners, the presence of 0.05 or 0.5% PE resulted as expected in decreased bioaccumulation compared to the 0% PE B treatment (Tukey HSD, p-values ≤1.80 × 10^–4) and no significant differences with the 0% PE A treatment. The elevation of bioaccumulation in the 0% PE B treatment was up to a factor 10 compared to the treatments where these PCBs were added in the presence of plastic (Figure 2). As there was extra spiking of PCBs in the PE treatments too, the lack of this elevated bioaccumulation in the treatments with PE compared to the IS control thus can be attributed to sorption to PE.

Figure 2.

Average PCB concentrations ± SE (lipid normalized) in lugworms after exposure to the different treatments and their background PCB concentrations before start of the exposure assay for the representative PCB congeners 153 and 138. Left panel: PCB congener 153 spiked equally in all treatments representing the CD (chemical dilution) scenario. Right panel: PCB congener 138 extra spiked in the treatments with PE and the 0% PE B to correct for the dilution mechanism representing the IS (infinite source) scenario. Where error bars are invisible they are small and thus lie behind the markers. The (similar) results for eight more PCB congeners can be found in SI Figure S9 and the results of all congeners combined in Figure 3A.

Figure 3.

Average bioaccumulation and biota to sediment, PE and porewater accumulation factors per PCB congener. Panel A: Lipid normalized PCB concentrations ± SE in lugworms after exposure to the different treatments. Concentrations of 10 congeners as a function of their hydrophobicity (logK_OW). Panel B: BSAFs normalized on concentrations of PCBs in lipids in lugworm tissue and OM in sediment. Panel C: BPAFs normalized on concentrations of PCBs in lipids in lugworm tissue and PE in sediment. Panel D: Bioaccumulation factors (logBAFs) normalized on concentrations of PCBs in lipids in lugworm tissue (±SE 0.011–0.198, not shown). Linear regression line with cut off at logK_OW > 6.8: logBAF = 1.44 × logK_OW – 1.53, R² = 0.95.

In summary, a small increase in bioaccumulation was detected for the CD scenario PCB congeners after exposure to the 0.05% PE compared to the 0% PE A treatment. However, when comparing to reference systems with statistically identical porewater concentration (0% PE B) as would occur in nature, no difference in bioaccumulation was detected. Dilution due to PE addition was detected at higher PE dose and for the IS scenario PCB congeners. Dosing extra PCBs to compensate for dilution in the IS scenario confirmed but did not provide a clear additional mechanistic view on these processes when looking directly at bioaccumulation in lipids of lugworms. Hereafter, we tease out dilution effects due to presence of plastic by normalizing bioaccumulation to concentrations in respectively sediment organic matter (BSAF), PE (BPAF), or porewater (BAF) below.

Biota to Sediment Accumulation Factors (BSAFs)

Biota to sediment accumulation factors (BSAFs) were calculated from PCB concentrations in the lipids (C_lip) and OM (C_OM) (SI eq S7, Figure 3B). Overall, for BSAF, variability among treatments was less than when C_lip was used as a metric for bioaccumulation (compare Figure 3A and B). In case increase of bioaccumulation would occur due to extra spiking or chemical dilution effects would occur due to addition of PE, these would respectively increase or decrease C_lip as well as C_OM. Accordingly, BSAFs can be expected to provide a clearer view on the role of PCB uptake from ingested PE than C_lip. Average BSAFs ranged 30–272. This implies that bioaccumulation did not comply to Equilibrium Partitioning Theory (EPT), which would suggest normalized BSAF values of 1–2.²⁸ Similarly high BSAFs for PCBs were reported for lugworms by Diepens et al.⁵⁷ and Besseling et al.,⁷ and for several other species and chemicals.^58,59 One reason for higher BSAFs is that binding to Oesterput OM was relatively low, that is, K_OM values were an order of magnitude lower than K_OW values (SI Figure S6), where K_OW is taken as a proxy for the binding affinity to lipids. BSAFs can also be enhanced due to biomagnification from ingested sediment OM. This would imply a higher BSAF with higher logK_OW, which however was not observed. This in turn can be explained by not having reached steady state for the more hydrophobic congeners in lipids.⁵⁷ For the CD scenario PCB congeners, BSAFs were elevated after exposure to 0.05% PE compared to the 0% PE A treatment by a factor 4.4 (1.9–10.6), which was significant for one congener (PCB 28). After exposure to 0.5% PE, BSAFs were elevated by a factor 3.6 (2.5–4.6) which was significant for three congeners (PCB 28, 153, 180, Figure 3A, NDWD, p-values ≤0.030). However, compared to the 0% PE B treatment, the treatment with similar porewater concentration, BSAFs were not elevated after exposure to PE. Also for the IS scenario PCB congeners compared to the 0% PE B treatment, BSAFs did not increase after exposure to PE (factor 1.2 (0.9–1.7), statistically insignificant). For both the CD and IS scenario PCB congeners, differences in BSAFs after exposure to either 0.05 or 0.5% PE were not significant (average ratio BSAF_0.5%PE/BSAF_0.05%PE = 1.1, range 0.4–2.0). This implies that a vector effect of PE is not likely, because a 10 times higher PE dose then would have resulted in a higher BSAF. The latter corresponds with the lack of difference in BSAFs of the PE treatments compared to the 0% PE B treatment as mentioned above. We conclude that by using BSAF as a metric for bioaccumulation, magnitude and statistical significance of differences among treatments were smaller compared to C_lip, as expected.

Biota-to-Plastic Accumulation Factors (BPAFs)

Biota to plastic accumulation factors (BPAFs) were calculated from PCB concentrations in the lipids (C_lip) and PE (C_PE) (SI eq S9, Figure 3C). Overall, for BPAF, variability among the 0.05% and 0.5% PE treatments was even less than when BSAF was used as a metric for bioaccumulation (Figure 3B,C). BPAFs ranged 7.6–44.8, suggesting a contribution of OM bound PCBs to C_lip. A decreasing trend of BPAF with increasing K_OW was observed, which might be explained by (a) slow kinetics toward the worm lipids of PCBs with a higher hydrophobicity,⁵⁷ and (b) sorption to PE increasing more than proportionally with logK_OW (SI Figure S7). BPAFs were lower than BSAFs, as K_PE values were higher than K_OM values. There were no significant differences (average ratio BPAF_0.5%PE/BPAF_0.05%PE = 1.0, range 0.5–1.2) between these plastic normalized BPAFs after exposure to 0.05 and 0.5% PE for all tested PCB congeners (Kruskal–Wallis, p-values ≥0.22). This again implies that no vector effect of PE was found. After all, when PE would act as a vector for PCBs, a 10 times higher PE dose would have resulted in a higher BPAF. The lower variability among PE doses when using BPAF instead of BSAF indicates that normalizing on PE is a useful, more representative way to interpret bioaccumulation in sediments containing plastic.

Bioaccumulation Factors

Lipid normalized bioaccumulation factors (BAF, SI eq S9) were calculated as the ratio of C_lip and C_PW (SI eq S5, Figure 1). Overall, for BAF, the variability among treatments and among treatment replicates was far less than that for all other metrics. LogBAF increased linearly with increasing hydrophobicity of the PCB congeners (Figure 3D), with a hydrophobicity cutoff visible at logK_OW > 6.8.⁶⁰ Interestingly, BAF showed no significant differences among treatments in both the CD and IS scenario, except for the least hydrophobic PCB congener. Only for PCB 28, the BAF at 0.05% and 0.5% PE was enhanced by a factor of 5.3 and 2.3 respectively, compared to the 0% PE A treatment (NDWD, p-values ≤0.045). This also implies that the BAF normalization removed the aforementioned deviation for the 0% PE A treatment for all but one PCB. Differences in BAF between the 0.05 and 0.5% PE treatments were not significant either, and the ratio between them again was one, equaling that of the aforementioned BSAF and BPAF.

In summary, differences among treatments appeared to decrease in terms of magnitude and statistical significance in the order: C_lip > BSAF > BPAF > BAF, to become nonsignificant when BAF was used as a metric for treatment effect. This implies that the main driver of treatment effects was the difference in porewater concentration (Figure 1), which in turn was driven by partitioning phenomena among worm lipids, PE, water and sediment OM. Because C_lip closely followed the concentrations in the porewater for each of the PCBs, we conclude that bioaccumulation was not affected by extra exposure due to PE ingestion.

Model Supported Assessment of Relative Importance of Uptake Pathways

We modeled bioaccumulation of the PCBs (dC_B,t/dt; μg × g^–1d^–1) as a mass balance of uptake and loss processes:^{4,13,17,19,20}

2

A detailed description of the modeling is provided as SI. The first term in eq 2 quantifies uptake from the porewater. The second and third term quantify uptake from ingested sediment and ingested PE, respectively. The fourth term quantifies loss due to elimination and egestion. For the CD scenario PCB congeners, bioaccumulation after 28 days was modeled using SI eq S17, after which the relative shares of the uptake pathways on accumulation were assessed. A tiered parameter estimation was applied. First, parameters were set at default values except the sediment ingestion rate (IR), which was fitted using the bioaccumulation data of the 0% PE B treatment. This resulted in IR = 9.98 g/g DW × d^–1, a value which is close to the value estimated following the equation provided by Cammen⁶¹ of 7.1 g/g DW × d^–1. For the 0.05% and 0.5% PE scenario calculations, IR was set at 55% and 33% of this IR value in the control, derived from the measured heap mass production, here taken as a relative measure of ingestion. Second, the uptake rate constant in the gut (plastic-gut fluid exchange coefficient k₁)⁴ was optimized, which resulted in values of 0.080 and 0.27 d^–1 for the 0.05% and 0.5% PE treatments, respectively. Using this parametrization, the terms in eq 2 were evaluated for both PE scenarios (SI Table S3). It appears that at 0.05%, PE contributed less than 3% to PCB uptake and also the loss rate changed marginally (not shown), indicating no substantial effect of PE at this environmentally realistic dose. At the high concentration of 0.5%, PE contributed more to uptake, but still to a minor extent for most PCBs with up to 62% of uptake for the most hydrophobic congener. This percentage should not be interpreted as extra bioaccumulation but as the contribution of plastic to the total uptake term in SI eq S1, meaning that the overall uptake term does not necessarily increase. For the latter PCBs, loss rates are predicted however to increase, leading to an overall predicted factor two decrease in bioaccumulation, as was suggested before.^4,5,16,18 These flux estimates, based on measured concentrations in all exposure media confirm the lack of a vector effect deduced from the bioaccumulation metrics discussed in the previous sections, thus mechanistically explaining the empirical observations.

Our results further illustrate that measuring bioavailability using passive samplers is crucial to understand exposure at the extremely low aqueous phase concentrations typical for hydrophobic chemicals in environmentally realistic exposure studies. To date, most studies that tested the effects of plastic ingestion on bioaccumulation did not assess aqueous phase concentrations with the detection limits that can be achieved with passive samplers, and they usually neglected the possibility of aqueous exposure. With the aid of metrics that normalize bioaccumulation to concentrations in the various exposure media and biodynamic modeling, we showed that these extremely low pg/L concentrations in porewater still drive exposure and can explain bioaccumulation, a phenomenon that is widely recognized in the bioaccumulation literature. This work confirms model-based predictions on the limited relevance of microplastic for bioaccumulation under environmentally realistic exposure conditions, and illustrates the importance of assessing exposure through all media in microplastic bioaccumulation studies.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.7b02286.

• Tables on PCB concentrations; partitioning coefficients and model outcomes on relative importance of PCB uptake pathways. Schematic representation of treatment design. Figures on physical end points, C_PW, logK_OM, logK_PE, C_PE, and C_lip. Details on materials and methods, calculation of all used metrics, and bioaccumulation modeling including equations and model input. (PDF)

The authors declare no competing financial interest.