Environ. Sci. Technol. 2009, 43, 3757–3763
PAH Bioavailability in Field Sediments: Comparing Different Methods for Predicting in Situ Bioaccumulation STEPHAN A. VAN DER HEIJDEN* AND MICHIEL T. O. JONKER Institute for Risk Assessment Sciences, Utrecht University, P.O. Box 80177, 3508 TD Utrecht, The Netherlands
Received November 24, 2008. Revised manuscript received March 11, 2009. Accepted March 27, 2009.
In situ exposure concentrations of chemicals in sediments and their depending risks are determined by site-specific parameters(e.g.,sedimentorganiccarboncomposition),controlling bioavailability. Over the years, several analytical methods have been developed to assess bioavailable concentrations or fractions. Some of these methods have been successful in the laboratory, but few attempts have been made to test their potential for predicting actual in situ bioavailability. In this study, solid-phase microextraction (SPME)-fibers and aquatic worms (Lumbriculus variegatus) were exposed in situ at three locations. In addition, laboratory-based methods, i.e., methods with which sampling of the bioavailable fraction/concentration took place in the laboratory, being SPME, polyoxymethylene solidphase extraction (POM-SPE), hydroxypropyl-β-cyclodextrin(HPCD), and 6 h-Tenax extraction were applied to sediments collected at the locations. Using equilibrium partitioning-based calculations, biotic PAH levels were calculated from the concentrations or fractions extracted by the used methods. In general, method-predicted concentrations were within a factor of 10 of those measured in field-exposed oligochaetes, with in situ SPME and laboratory-based POM-SPE yielding the best results. As a reference, the currently used generic risk assessment approach overpredicted biotic concentrations by a factor of 10-100, which corresponded to in situ SPME-derived sediment-water distribution coefficients and biota-tosediment accumulation factors being up to 2 orders of magnitude higher and lower, respectively, than generic values. These observations advocate site-specific risk assessment for PAHs, for which potential tools were evaluated in this study.
Introduction Considering the widespread contamination of sediments with polycyclic aromatic hydrocarbons (PAHs), efficient and reliable risk assessment is of vital importance. Within the current risk assessment paradigm, a comparison is made between total contaminant concentrations in the solid phase and fixed risk limits (1), rationalizing any remediation practice. However, risks may often be overestimated as limit values are intrinsically based on generic sorption parameters. These parameters only consider sorption of contaminants to natural organic matter, whereas recent studies propagate * Corresponding author phone: +31 30 2535018; fax: +31 30 2535077; e-mail:
[email protected]. 10.1021/es803329p CCC: $40.75
Published on Web 04/14/2009
2009 American Chemical Society
the notion of site-specific compositions of sorption domains within the solid phase, having differential binding strengths (2, 3). Though unfeasible, site-specific analysis of sorption by quantification of all sorbing fractions (e.g., natural organic carbon, black carbon, oil, coal) and their distinct sorption characteristics would provide a sound basis for the ultimate definition of exposure concentrations. A more realistic approach would be a direct measurement of exposure concentrations or the bioavailable fraction of a contaminant (i.e., the fraction of the solid-bound concentration being available for uptake by organisms), as these already comprise solid phase, chemical, and biotic characteristics (4). Over the past decade, several analytical tools have been developed that aim at measuring these concentrations or fractions, including 6 h-Tenax extraction (5), hydroxypropyl-β-cyclodextrin (HPCD) extraction (6), supercritical fluid extraction (SFE) (7), solid-phase microextraction (SPME) (8), polyoxymethylene solid-phase extraction (POM-SPE) (9), semipermeable membrane device (SPMD) extraction (10), mild solvent extraction (11), and persulfate oxidation (12). Several studies have investigated the ability of one or two of these bioavailability-assessing methods to predict actual bioaccumulation of PAHs or other hydrophobic contaminants in environmental samples to which test organisms were exposed under laboratory conditions. Hitherto, promising results have been obtained with SFE (13), 6 h-Tenax (14, 15), SPME (16–18), SPMD (19), and POM-SPE (20). However, laboratory conditions do not necessarily competently reflect the situation in which organisms are normally exposed in the environment. As a result, a bioavailability-assessing method should only be considered successful when it is capable of predicting bioaccumulation occurring under field conditions. Still, a limited number of studies have addressed this issue; only SPMD-predicted bioaccumulation and Tenax-extractable fractions have been compared to bioaccumulation measured in field-exposed organisms (14, 21–24). Hence, it is not yet possible to conclude which method is best suited for use in actual site-specific risk assessment strategies. The objective of the present study was to compare the ability of several of the above-mentioned methods to predict PAH bioaccumulation in aquatic worms occurring under true field conditions. To this end, Lumbriculus variegatus as well as SPME fibers were exposed in situ at three different field locations in The Netherlands. In addition, sediment samples were taken for the assessment of bioavailability in the laboratory, using SPME, POM-SPE, and extraction by HPCD and Tenax. Although SPME has been applied in the field before for measuring contaminant concentrations in porewater (25), to our knowledge, this is the first study relating SPME-predicted concentrations to actual concentrations in field-exposed organisms. Additionally, as SPME yielded in situ freely dissolved PAH concentrations, the measurements enabled determination of true in situ organic carbonnormalized sediment-water distribution coefficients (Kocs) and bioconcentration factors (BCFs). While the first may be compared to generic sorption parameters currently used in risk assessment, the last yields insight in actual bioaccumulation in the field.
Materials and Methods Chemicals and Extraction Media. Chemicals used were acetone, acetonitrile, methanol, hexane (HPLC grade; LabScan, Dublin, Ireland), atropine (>99%; Sigma Aldrich, Steinheim, Germany), calcium chloride, hydrochloric acid, sodium azide (Merck, Darmstadt, Germany), aluminum oxide (Super I; ICN Biomedicals, Eschwege, Germany), and 2-meVOL. 43, NO. 10, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3757
thylchrysene (99.2%; Community Bureau of Reference, Geel, Belgium). Poly(dimethylsiloxane) (PDMS)-coated disposable SPME fiber (glass core diameter 110 µm, 28.5 µm thick coating) was supplied by Poly Micro Technologies, Phoenix, AZ. Prior to use, the fiber was cut into pieces of 3 or 5 cm, which were thermally desorbed at 275 °C for 16 h under a constant flow of helium. POM was obtained from Vink Kunststoffen BV, Didam, The Netherlands. Before use, strips of appropriate dimensions were cut and cold-extracted with hexane (30 min) and methanol (3 × 30 min), after which they were air-dried. HPCD (>98%; CAVASOL W7 HP Pharma) was obtained from Wacker-Chemie GmbH, Burghausen, Germany. Tenax-TA (60-80 mesh; Chrompack, Middelburg, The Netherlands) was pre-extracted with hexane (1.5 h) and acetone (1.5 h) and dried overnight at 80 °C. Organisms. Lumbriculus variegatus was selected as test organism, because (i) it is a commonly used test organism in ecotoxicological studies, (ii) it has a limited PAH biotransformation capability (26), (iii) the worms are resistant to multiple stress factors (e.g., anoxia, temperature shifts, high contaminant levels), and (iv) their size and surface/volume ratio enable relatively fast equilibration with environmental media. The worms were laboratory-raised at a temperature of 23 ( 1 °C in aerated 45 L flow-through aquaria, containing cellulose substrate. The organisms were fed with flake fish food (King British, UK) once a week. Before testing, organisms were gut-purged during a 16 h period under mildly running tap water. In Situ Exposures. Worms and SPME fibers were exposed at three field locations in The Netherlands: (i) a rural canal where PAHs likely originate from creosote-treated sheetpiling (location 1), (ii) a branch of the river Hollandsche IJssel, mainly receiving PAHs from a large diesel-powered pumping station 100 m upstream (location 2), and (iii) a ditch with PAH contamination probably also originating from creosotetreated sheetpiling (location 3). The oligochaetes were exposed in bottomless stainless steel enclosures (16 × 16 × 25 cm (L × W × H); see Supporting Information Figure S1), which were inserted into the sediment without disturbing the stratified structure. The enclosure walls projected 8 cm from the sediment surface and reached a depth of 17 cm, effectively keeping the organisms (∼13 g wet weight/enclosure) from escaping. Enclosures were covered with 1 mm-mesh stainless steel gauze to prevent predation by fish. To enable easy insertion and collection of the fragile SPME fibers, SPME in situ devices (SiDs) were designed. The SiDs consisted of stainless steel gauze soldered to a 1 mm thick stainless steel body, creating a freely permeable envelope in which four 3 cm SPME fibers were placed (see SI Figure S1). During exposure, the fibers were positioned in the top 3.5 cm of the sediment, mimicking the spatial preference of L. variegatus (27). At each location, four SiDs were fixed to a frame with stainless steel pins, thus yielding four sublocations, rather than replicates. The frame was anchored into the sediment by a stainless steel rod. At each location, also four enclosures were inserted within a defined cluster between the SiD positions (SI Figure S2). Special care was taken to prevent disturbance of the sediment within the cluster boundaries. SPME fibers and L. variegatus were field-exposed for 30 days at approximately 18 °C. Collection and Processing of in Situ-Exposed Materials. Upon finishing the in situ exposures, a sediment layer of 10-15 cm was sampled from the enclosures and stored in clean plastic containers for worm recovery in the laboratory. In addition, sediment samples were taken at exposure depth (i.e., the top 3.5 cm of the sediment layer), at a minimal distance from each SiD. Directly upon extraction of the SiD-frame from the sediment, SPME fibers were collected from the SiDs with tweezers, wiped with a wet 3758
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009
tissue, their length was measured, and they were transferred to brown HPLC vials, containing an insert filled with 180 µL of acetonitrile. The vials were stored in a coolbox and back in the laboratory, internal standard was added. Worms were manually collected from the sediments. Recovered worms were gut-purged for 16 h, frozen, freeze-dried, and homogenized in a mortar prior to Soxhlet extraction. Sediments were also frozen, freeze-dried, and homogenized before laboratory-based SPME, POM-SPE, and Soxhlet, HPCD, and 6 h-Tenax extractions. A pilot test demonstrated that these sediment manipulations did not consistently affect PAH bioavailability (see SI Table S1). Laboratory-Based Methods for Predicting in Situ Bioaccumulation. Concentrations of PAHs in sediment porewater were measured in the laboratory using SPME according to the procedure described by Jonker et al. (28). Ambercolored 7 mL vials were filled with 2-4 g of sediment, 6.5-7 mL of Millipore water, containing 0.01 M calcium chloride and 25 mg/L sodium azide, and two 5 cm SPME fibers. Systems were incubated for either 3, 7, 14, 28, or 56 days to assess equilibration time. POM-SPE and POM cleanup was performed exactly as described by Jonker and Koelmans (9), using 3 g of sediment, 250 mL of the above-mentioned solution, and 1 g of POM. HPCD extraction was carried out according to a slightly modified version of the procedure described by Reid et al. (6). Weighed amounts (0.75 g) of sediment were transferred to 20 mL glass vials to which 15 mL of aqueous solution was added, containing 50 mM of HPCD and 25 mg/L of sodium azide in Millipore water. The vials were horizontally shaken (150 rpm) for 24 h and centrifuged (3000 rpm) for 20 min. Subsequently, 10 mL of supernatant was transferred to clean vials and extracted with hexane (2 × 3 mL). Finally, hexane extracts were concentrated and exchanged to acetonitrile, after which internal standard was added. We determined 6 h-Tenax-extractable fractions according to the method described by Moermond et al. (14). Modifications were the amount of sediment extracted (0.5 g), the composition of the aqueous solution (25 mg/L sodium azide and 0.01 M calcium chloride), and hexane was finally exchanged to acetonitrile. All determinations were performed in quadruplicate (SPME and POM-SPE) or triplicate (HPCD and 6 h-Tenax extractions). PAH Analyses. Sediments and worms were Soxhletextracted with a 3:1 hexane-acetone mixture for 16 h. The extracts were concentrated, cleaned-up through aluminum oxide columns, concentrated again, and exchanged to acetonitrile. Finally, internal standard was added, and the extracts were transferred to brown HPLC vials. All determinations were adjusted for procedural blanks and recoveries. Extracts were analyzed for 13 PAHs (phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[e]pyrene, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, benzo[g,h,i]perylene, dibenz[a,h]anthracene, and indeno[1,2,3-c,d]pyrene) by using HPLC, as described previously (28). Lipid and Organic Carbon Analyses. Lipid contents of freeze-dried worms were determined gravimetrically by Soxhlet extraction with a 3:1 hexane-acetone mixture, and evaporation to dryness. Organic carbon fractions (foc) of sediments were determined as described in ref 29, including removal of inorganic carbon prior to analysis by the addition of hydrochloric acid. Atropine was used as standard and for carbon quantification, a Carlo Erba NA 1100 elemental analyzer was used. All results were adjusted for blanks.
Results and Discussion Field Samples. In SI Table S2, sediment organic carbon fractions, worm lipid percentages, and averaged PAH concentrations in sediments (Cs), in situ exposed SPME fibers (Cf), and L. variegatus (Cb) are presented for the three field
locations. The table shows ΣPAH concentrations in sediments ranging from 548 mg/kg OC for location 2, to 170 and 5 mg/ kg OC for locations 1 and 3, respectively, and organic carbon fractions varying between 0.03 and 0.24. These data clearly indicate three distinct testing environments. All in situ-exposed SPME fibers were successfully retrieved from the sediments. As shown in SI Table S2, some PAHs were not determined in the fibers, despite the chemicals’ presence in the biotic and sediment phases. In these cases, the volume of the extraction phase was too small, causing the absorbed PAH concentrations to be below detection limits. Though no kinetic experiment was conducted in the field to assess the equilibration time, in situ SPME (isSPME)derived logKoc values (discussed and derived below) correlated linearly with hydrophobicity, as shown in SI Figure S3. Such a linear relationship strongly suggests the occurrence of equilibrium (30). Moreover, isSPME-derived logKoc values appeared not significantly different (t-test) from logKoc values determined in the laboratory under highly dynamic conditions using POM-SPE (see SI Table S3). Therefore, isSPMEderived data (Koc, BCF, and predicted Cb values) can be assumed to reflect equilibrium conditions. Likewise, worms are also supposed at equilibrium with their surroundings, as Leppa¨nen and Kukkonen (26) demonstrated that equilibrium was reached within a week for pyrene and benzo[a]pyrene in L. variegatus exposed in sediment. Note that the exposure duration in the present study was 30 days. Oligochaetes were recovered from locations 1 and 2. At location 3, no test organisms were found in the sampled sediment. Although prior to the in situ work worm survival was tested and approved in the laboratory using sediment samples taken from all locations, the actual exposure point at location 3 was about 2 m away from the one where the samples for worm survival were taken. These results indicate considerable spatial heterogeneity and stress the necessity of a detailed prescreening. Worms recovered from locations 1 and 2 showed no visible alterations of physical constitution (e.g., size, mobility, color) compared to unexposed individuals. Biotic PAH concentrations were low enough (see SI Table S2) to exclude toxic effects, which could have influenced uptake (31). Also, lipid fractions were 14.9% ((1.4; n ) 4) and 10.1% ((0.76; n ) 4) of dry weight for worms from locations 1 and 2, respectively, whereas that of nonexposed worms was 10.4% ((0.16; n ) 3). Therefore, starvation of worms due to nutrient depletion most probably did not occur. This deduction is supported by the observation that nitrogen and organic carbon fractions were not significantly lower in sediment collected outside of the enclosures upon conclusion of the field experiment (see SI Table S4). Having measured the rapidly desorbing PAH fractions in the sediments using Tenax (discussed below), conservative estimates of the depletion of the bioavailable sediment-associated PAH pool by the worms were derived. Depletion appeared negligible at both locations (1.7 and 0.3% averaged for all PAHs at location 1 and 2, respectively; see SI Table S5). In Situ Sediment-Water and Biota-Sediment Distribution. By applying previously determined SPME fiber-water partition coefficients (Kf) (28, 32), freely dissolved PAH concentrations in porewater (Cw) were calculated from concentrations in in situ-exposed SPME fibers (Cf), using the equation Cw ) Cf/Kf. These Cw values subsequently allowed calculation of actual BCF and linear in situ Koc values, thus enabling comparison with generic values used in current risk assessment. In situ BCF values have been presented elsewhere (32); Koc values will be discussed below as well as biota-to-sediment accumulation factors (BSAFs), also determined in this study. Koc values were calculated according to Koc ) ((Cs/foc)/ Cw), using Cs values and organic carbon fractions from SI Table S2. In SI Table S3, the resulting in situ logKoc values
for the three locations are presented, along with generic logKoc values, which were derived from logKows by using a PAHspecific one-parameter linear free energy relationship from ref 33; logKoc ) 1.14logKow-1.02. In SI Figure S3, a clear difference can be observed between generic logKocs and measured values for locations 1 and 2. In most cases, the latter ones are 1 to 2 orders of magnitude higher than the generic values. Many previous studies have meanwhile presented similar results (3, 28, 29, 34, 35) and have argued that the main cause of variation in sediment and soil Koc values of (particularly) PAHs is related to the quality of the carbon fraction (3, 28, 29, 34–36). The merit of the present results is a verification of the enhanced sorption phenomenon and the illustration of the limited usefulness of a generic sorption concept under true in situ conditions. Although no attempt was made to measure soot or black carbon in the present sediments, the enhanced sorption at location 2 is most probably attributable to the presence of diesel soot (36) and/or diesel oil (37), emitted by the upstream pumping station. Oil was detected in this sediment at a concentration of around 2000 mg/kg (unpublished results). The high logKoc values at location 1 may similarly be related to the likely PAH source, i.e., creosote tar, which supposedly sorbs PAHs to a comparable degree as oil does (37). Interestingly, note that the in situ Koc values from locations 1 and 2 are not significantly different (t-test). As illustrated by SI Figure S3, the logKoc values measured at location 3 agree more closely with generic values, though the presumed PAH source at this location is creosote tar as well. This observation may be explained however by the very high organic carbon fraction of the sediment (0.24). At such carbon levels, any effect of enhanced sorption to for instance black carbon or tar will be obscured (3). BSAFs were calculated using the equation BSAF ) Cb/ (Cs/foc). In SI Figure S4a, resulting values are presented in relation to their hydrophobicity. The figure shows that BSAF values decrease for PAHs with a logKow > 5.5. This result is similar to that obtained in other studies (24, 38), where the effect was either explained by nonequilibrium conditions in organisms, due to sterically retarded membrane passage of larger PAHs, or to enhanced sorption to sediment of the more hydrophobic PAHs. Interestingly, the fiber-to-sediment accumulation factors (FSAF; FSAF ) Cf/(Cs/foc)) presented in SI Figure S4b display an even more pronounced decrease with increasing logKow as compared to the BSAF values. Because equilibrium for all tested compounds in in situ exposed SPME fibers is highly plausible (see above), nonequilibrium conditions do not seem to be a valid explanation for this observation. Rather, sorption to sediment seems to become stronger relative to sorption into the biotic and PDMS phase as hydrophobicity increases. Note that sorption to sediment organic carbon is stronger than to worms in any case (i.e., all BSAF values
