Environ. Sci. Technol. 2002, 36, 4968-4974
Soot-Carbon Influenced Distribution of PCDD/Fs in the Marine Environment of the Grenlandsfjords, Norway
both amorphous OC partitioning domains and SC particles as carriers of planar aromatic contaminants if we are to explain the environmental distribution and fate of pollutants such as PCDD/Fs.
N. J. PERSSON,* O ¨ . GUSTAFSSON, T . D . B U C H E L I , †,‡ R . I S H A Q , †,§ K. NÆS,| AND D. BROMAN† Institute of Applied Environmental Research (ITM), Stockholm University, 106 91 Stockholm, Sweden, and Norwegian Institute for Water Research (NIVA), Branch Office South, 4879 Grimstad, Norway
Polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) are persistent, bioaccumulative, and toxic and undergo long-range transport (1, 2). Their distribution between the freely dissolved aqueous state and the various types of solid phases (i.e., their phase distribution) is strongly influencing their uptake in marine organisms and food webs (3, 4). This phase distribution is also believed to be of importance for other PCDD/F fate and transport processes in the marine environment such as degradation (5), sedimentation (6), and volatilization (7). In part because of the significant analytical challenge posed by sampling and quantifying individual PCDD/Fs in different phases in the marine environment (frequently at low aM levels in the dissolved phase), there is a scarcity of data by which laboratory-derived models of PCDD/Fs phase distribution can be verified. Intriguingly, the limited data available on the solid-water distribution (Kd) of PCDD/Fs in the marine and aquatic environment suggest orders of magnitudes higher affinity for the particulate phase than what would be predicted from commonly used bulk organic matter partitioning (OMP) models (8-10). Hence, a primary objective of the current study was to significantly increase the number of available field data on the phase distribution of PCDD/Fs against which hypotheses of PCDD/F partitioning could be tested.
,†
†
The particle associations of polychlorinated dibenzo-pdioxins and polychlorinated dibenzofurans (PCDD/Fs) were studied in both the water column and the surface sediments of a marine fjord system and were found to poorly obey expectations from the organic matter partitioning (OMP) paradigm. The field observations were instead consistent with the presence of a stronger sorbent subdomain such as pyrogenic soot-carbon (SC) playing an important role in affecting the environmental distribution and fate of PCDD/Fs. Solid-water distribution coefficients (Kd) of PCDD/ Fs actually observed in the water column were several orders of magnitude above predictions from a commonly used OMP model. Even when these elevated Kd values were normalized to the particulate organic carbon (POC) content (i.e., KOC), the variability in KOC for individual PCDD/ Fs at different fjord locations and seasons of factors 1001000 suggested that bulk organic matter was not the governing sorbent domain of the suspended particles. Further, POC-normalized particle concentrations of PCDD/ Fs (COC) in a vertical profile (surface water-bottom watersurface sediment) revealed a strong increasing trend with depth. Factors of about 100 higher COC for all PCDD/Fs in the sediment than in the surface water could not be explained by higher fugacity in the surrounding deep water nor with C:N or δ13C indexes of selective aging of the bulk organic matter. Instead this was hypothesized to reflect selective preservation of a more recalcitrant and highly sorbing, but minor, subdomain such as soot. The extent of enhanced PCDD/F sorption, above the OMP predictions, was positively correlated with the SC:POC ratio of the suspended particles in surface and deep waters. Finally, the geographical distribution of sedimentary PCDD/F concentrations were better explained by the SC content than by the bulk OC content of the sediment. Altogether, these field-based findings add to recent laboratory-based sorption studies to suggest that we need to consider * Corresponding author phone: +46 674 7341; fax: +46 674 7638; e-mail:
[email protected]. † Stockholm University. ‡ Present address: Swiss Federal Research Station for Agroecology and Agriculture (FAL), Reckenholzstrasse 191, 8046 Zu ¨ rich, Switzerland. § Present address: Department of Environmental Chemistry (CSIC), Jordi Girona 18, 08034-Barcelona, Spain. | Norwegian Institute for Water Research. 4968
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002
Introduction
Laboratory and field studies on polycyclic aromatic hydrocarbons (PAHs) over the past decade are strongly suggesting that elevated Kd for PAHs, and their spatiotemporal distribution, are stemming from efficient interaction with highly aromatic and planar soot particles (e.g., refs 1119). Such pyrogenic soot-carbon (SC) has recently been found to make up a significant fraction (2-20%) of the total reduced carbon in marine sediments (e.g., refs 20-23). In this work, we test whether the solid-water and spatial distribution of planar PCDD/F in a marine environment, analogous to the PAH case is influenced by SC. This hypothesis is evaluated by probing the distribution of PCDD/Fs, SC, and OC in the marine environment of the Grenlandsfjords, Norway. Because of its relatively high PCDD/F contamination, this system is well-suited as it allows for easier detection and thus generation of a larger set of phase-specific data of individual compounds.
Materials and Methods Study Region. The Grenlandsfjords are five jointed fjords in the south of Norway (59°5′ N, 9°38′ E). The innermost fjord, Frierfjorden, has since the 1950s been substantially polluted by PCDD/F discharges from a magnesium production plant (24). Large amounts of PCDD/Fs, polychlorinated naphthalenes (PCNs), hexachlorobenzene (HCB), and octachlorostyrene (OCS) are formed during the chlorination of magnesium oxide (MgO) to yield water-free magnesium chloride (24, 25). It has been estimated that emissions of PCDD/Fs to the fjord were around 500 g TEQ/yr (TEQ ≡ 2,3,7,8-TCDD toxic equivalents) until 1988 (25). Over the last 10 years, direct emissions have been markedly reduced, but concentrations of PCDD/F are still high in water, sediment, and biota (2628). 10.1021/es020072l CCC: $22.00
2002 American Chemical Society Published on Web 10/23/2002
FIGURE 1. Map showing the location of the sampling sites for surface water, water-column depth profile, and bottom sediments in the Grenlandsfjords, Norway. Sample Collection and Handling. Water samples were taken at various locations in the Grenlandsfjords and the river Skienselva during three excursions on December 1516, 1998; June 29-July 1, 1999; and May 2-5, 2000 (Figure 1). Collection of particulate and dissolved fractions from the water column largely followed previously described protocols (e.g., refs 8 and 16) and will only be outlined briefly here. For each water sample, 600-1400 L was pumped onboard through 12-112-m silicone-coated tubing (i.d. ∼3 cm) at >15 L/min and split to a lower flow rate (∼8 L/min). The sample was led through a 293-mm Whatman GF/F filter (precombusted at 450 °C) held in a stainless steel filter holder for collection of suspended particles. The filtrate was directly passed through a polyurethane foam (PUF) sorbent cylinder of 7 cm diameter and 12 cm length (precleaned by 24 h extraction in toluene and 24 h in acetone), held in a specially
made stainless steel sorbent holder, to collect the dissolved and filter-passing PCDD/F fraction. Samples for determination of particulate organic carbon (POC) were obtained either by subsampling the 293-mm GF/F filters or from 47mm precombusted GF/F filters coupled in parallel to the main sampling line. Samples for analysis of dissolved organic carbon (DOC) was collected from the GF/F filtrate in acidwashed 15-mL plastic test tubes. The pH was adjusted to 2 by adding 200 µL of 2 M hydrochloric acid, and the samples were stored cold until analysis. Filters and sorbents were wrapped in aluminum foil and stored in double layers of plastic bags. Sediment samples from the fjords were retrieved May 2-4, 2000, with a gravity kajak-type corer (10 cm i.d.) described elsewhere (29). Vertically sliced 1-cm sections of the cores were placed in machine-washed (60 °C) polyethylene cans with sealed lids. All water and sediment samples VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4969
were kept at 4 °C for 1-2 days and then at -20 °C until chemical analysis. Chemical Analysis. The content of PCDD/Fs in filter, sorbents, and sediment samples were analyzed largely following previously reported methodology (8, 30, 31). In short, the samples were Soxhlet extracted with toluene (glassdistilled grade, Burdick and Jackson, Fluka Chemie AG) for 24 h with 13C-labeled internal standards added at onset. The internal standard had one to two isomers of each analyzed chlorination degree (i.e., tetra to octa). The cleanup procedure included elution with n-hexane through open multilayer silica gel columns. The silica layers were conditioned with respectively sulfuric acid (40% w/w), potassium hydroxide (36% w/w), and water (10% w/w). Sulfur was removed from the samples by reduction with elemental copper powder (Merck, 70-230 mesh ASTM, 0.063-0.200 mm). This was followed by a two-step preparative HPLC separation of the target compound group (30, 31). Briefly, step one isolated the dicyclic aromatic compounds on an aminopropyl silica column (µBondapak 300 × 7.8 mm, 10 µm). Step two isolated the planar target analytes on a 2-(1-pyrenyl)ethyldimethyl silylated silica column (150 × 4.6 mm, 5 µm). After an additional cleanup through a disposable Pasteur pipet packed with silica gel and eluted with n-hexane, a recovery standard (13C-labeled PeCDD) was added. The samples were analyzed on a high-resolution gas chromatograph (HRGC, HewlettPackard 6890 series) with automated on-column injection to a retention gap (deactivated fused silica precolumn 2 m, 0.53 mm i.d., J&W Scientific, Agilent Technologies). A capillary GC column (30 m, 0.25 mm i.d., 0.25 µm film thickness, SP2331, Supelco Park, Bellafonte, PA) separated the analytes. Helium (>99.99%, AGA AB, Stockholm, Sweden) was used as a carrier gas. Baseline separation for most analytes was enabled in ∼30 min by temperature programming (started at 100 °C for 1 min, raised to 200 °C at 15 °C/min, then raised to 265 °C at 4 °C/min, and held there for 6.5 min). The HRGC was coupled to a high-resolution mass spectrometer (HRMS, Micromass AutoSpec Ultima, Manchester, U.K.). Molecular and qualifier ions of the tetra- to octa-chlorinated DD/Fs were registered in electron impact (EI) mode with 28 eV ionization energy. The mass resolution of the instrument was tuned to 10 000, and the sensitivity was optimized for relevant ions from perfluorokerosene (PFK) before a sample set was analyzed. Simultaneous injection and detection of selected ions from the PFK enabled a check for drift in the magnetic field of the HRMS system during each run. The SC content of the sediment and filter samples was determined with the chemothermal oxidation (CTO) method as described by Gustafsson et al. (14, 23). Particulate organic carbon and nitrogen (POC, PON) and their stable isotope ratios (δ13C, δ15N) were quantified in the filter and sediment samples using isotope ratio monitoring mass spectrometry (irmMS). Prior to analysis, carbonates were removed by in situ acidification in the Ag capsules, following a previously described micro-acidification procedure (14). Isotopic ratios are reported relative to the PeeDee limestone (δ13C) and N2 in air (δ15N). DOC in the water-column filtrate was analyzed with the high-temperature catalytic oxidation (HTCO) technique (Shimadzu TOC-5000).
Results and Discussion Solid-Water Distribution in the Water Column. The measured distribution of PCDD/F between the particulate and the dissolved phases in the Grenlandsfjords water column was compared with predictions from the OMP model (32, 33): OMP
4970
9
Kd ) fOCKOC )
Cs Cw
(1)
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002
where OMPKd is the model-predicted solid-water phase distribution coefficient (Lwater/kgdw solid), fOC is the mass fraction organic carbon in the solid phase (kgOC/kgtotal dw solid), KOC is the organic carbon-normalized solid-water distribution coefficient (Lwater/kgOC), Cs is the particle-sorbed concentration (attomole/kgtotal dw solid), and Cw is the truly dissolved concentration (attomole/Lwater). From our measured PCDD/F concentrations in the filter and PUF sorbent samples, an observed organic carbon-normalized solidwater partition coefficient ((KOC)obs) could be calculated:
COC (KOC)obs ) app Cw
(2)
where COC is the OC-normalized PCDD/F concentration in the filter-collected phase (attomole/kgOC) and appCw is the apparently dissolved concentration measured in the PUF sorbent (attomole/Lwater). The particulate concentration of the most abundant PCDD congener (octachlorodibenzo-pdioxin, OCDD) ranged from 40 aM right outside the fjord mouth to 11 fM near the smelter plant in the base of the fjord, whereas the apparently dissolved (PUF sorbed) concentration ranged correspondingly from 4 to 340 aM. The figures for the most abundant PCDF congener (octachlorodibenzofuran, OCDF) were for the same end-member sites in the particulate fraction 40 aM-110 fM and in the apparently dissolved fraction 2-670 aM. While the resultant (KOC)obs values for the individual PCDD/Fs all were much above the predicted (KOC)OMP, the slope for (KOC)obs as a function of the octanol-water partition coefficient (KOW) showed a positive but decreasing value with increasing hydrophobicity (data not shown). This we interpreted as the well-known artifact termed the particle-concentration effect (PCE; e.g., refs 34 and 35) stemming from inadvertent inclusion of PCDD/Fs sorbed to colloidal organic constituents in the apparently dissolved fraction. That the PUF sorbent collected filter-passing natural organic matter, with its load of sorbed PCDD/Fs, was also apparent from the yellowish appearance that these sorbents and their extract exhibited after sampling. Hence, it was attempted to correct for the PCE by app corr
Cw )
Cw
(1 + DOC × KDOC)
(3)
where DOC is the concentration of DOC in the water (kgDOC/ Lwater) and KDOC is the DOC-water partition coefficient (Lwater/ kgDOC). The DOC was measured in this study (Table 1), and KDOC for hydrophobic organic compounds (HOCs) such as the PCDD/Fs may be estimated from a linear free energy relation (LFER) with KOW proposed by Burkhard (36) based on a very large data set: KDOC ) 0.08KOW (with reported 95% confidence limits of a factor of 20 in either direction). Values for KOW (Lwater/Loctanol) for the PCDD/Fs were taken from Govers and Krop (37). While it is not likely that all filterpassing DOC (with associated PCDD/Fs) was collected on the PUF sorbent, the observation of yellowish color of the PUF sorbent in combination with more linear KOC-KOW relationships after correction for the DOC-associated PCDD/ Fs (e.g., ref 34) suggests that efficiently HOC-sorbing colored dissolved organic matter of coastal waters (35) was partitioned with the “dissolved” phase using this sampling technique. Given the DOC range in the Grenlandsfjords (Table 1) and the published variability in the KDOC estimate (36), the corrCw could be at least a factor eight lower than the PUF-sorbed concentration for the most hydrophobic congeners. However, even in this extreme case, our (KOC)obs would be underpredicted by a factor 9-200 by the OMP model. In fact, the DOC-corrected log(KOC)obs gave a more linear relation with
TABLE 1. Geochemical Characteristics of the Grenlandsfjords Water and Sediment parameter
valuea
nb
rangec
POC (µg/L) PON (µg/L) PSC (µg/L) C:Nd SC:POCe δ13C (‰ PDB) DOC (mg/L)
Fjords Surface Water, 0-2 m 235 ( 127 10 103-563 32 ( 21 8 15-81 15.7 ( 14.9 4 2.4-33.2 9.6 ( 1.2 8 8.1-11.3 0.082 ( 0.097 4 0.014-0.221 -26.7 ( 3.5 8 -29.0 to -18.4 3.3 ( 0.3 10 2.8-3.7
POC (µg/L) PON (µg/L) PSCe (µg/L) C:Nd SC:POCe δ13C (‰ PDB) DOC (mg/L)
River Surface Water, 0-2 m 263 ( 230 6 86-700 22.6 ( 18.0 6 7.4-46.5 8.9 ( 12.4 4 0.9-27 14.5 ( 4.8 6 8.3-21.7 0.019 ( 0.014 4 0.008-0.039 -27.7 ( 1.3 6 -29.0 to -25.2 3.2 ( 0.1 6 3.1-3.4
Fjords Bottom Water, 2-16 m above Sediment POC (µg/L) 47 ( 20 8 30-90 PON (µg/L) 4.9 ( 1.8 6 3.8-8.4 PSCe (µg/L) 5.8 ( 3.2 4 1.4-8.4 C:Nd 9.5 ( 1.5 6 8.2-11.9 SC:POCe 0.15 ( 0.09 4 0.02-0.25 δ13C (‰ PDB) -26.0 ( 2.6 6 -30.2 to -22.5 DOC (mg/L) 2.6 ( 0.7 8 2.0-4.4 TOC (mg/g dw) TON (mg/g dw) SCe (mg/g dw) C:Nd SC:TOC δ13C (‰ PDB) ωf (mm/yr)
Surface Sediment, 0-2 cm 34.2 ( 12.3 17 2.1 ( 0.8 17 3.1 ( 1.7 16 19.3 ( 5.5 17 0.09 ( 0.04 16 -23.4 ( 1.5 11 7.8 ( 2.7 3
2.7-50 0.3-3.2 0.4-7.3 10.1-32.7 0.04-0.16 -26.6 to -8.8 5-13
a Arithmetric average ( standard deviation. b Sample size. c Range (min to max). d Atom ratio. e Year 2000 samples. f Linear sedimentation rate at three stations May 2000.
log KOW (Figure 2), in accordance with HOC partitioning theory (32, 33) as compared to without any DOC correction (r2 was larger for the corrected data set in 27 out of 28 data series investigated). These PCE-corrected (KOC)obs for PCDD/Fs in the Grenlandsfjord system was compared with OMP model predictions of KOC through a LFER to KOW (in Lwater/Loctanol) provided by Seth et al. (38): (KOC)LFER ) 0.33KOW. It is apparent that the (KOC)obs from the real field situation of Grenlandsfjords were considerably higher than the model-predicted (KOC)LFER and consistently so for all the different seasons sampled in the three yearly campaigns (Figure 2). The OMP model underpredicted the (KOC)obs for the PCDDs by a factor of 40960 and for the PCDFs by a factor of 70-3330. It is worth noting the very large scatter observed in these elevated (KOC)obs for any given PCDD/F compound (Figure 2). This variation, frequently of a factor 100-1000, spans a much wider range than expected from the variability of organic matter sorbent quality (i.e., as defined by the 95% confidence interval for (KOC)LFER; Figure 2). Taken together, this large enhancement in particle affinity relative to expectations from partitioning with bulk amorphous organic matter, and the large scatter in these observations calls for explanations. Below we test two hypotheses that potentially may account for these two observations: (a) complications with the sampling method and (b) filter-trapped particles contained sufficiently high fractions of a strong sorbent such as SC. Go´mez-Belincho´n et al. (39) tested different methods of collecting hydrophobic compounds from seawater and found for a 4-fold increase in filtration rate on their PUF column (5.1 cm diameter, 30 cm height) from 490 to 1900 mL/min
FIGURE 2. Observed organic carbon-normalized solid-water partition coefficients (KOC)obs (in Lwater/kgPOC) for samples at various locations and depths in the Grenlandsfjord water-column and river Skienselva. (A) Samples taken in December 1998. (B) Samples taken in June and July 1999. (C) Samples taken in May 2000. Open circles are PCDDs, and filled squares are PCDFs. Dashed line is KOC predicted by the LFER from Seth et al. (38) with their estimated 95% confidence interval. For all three occasions, the observed partitioning to the particles ((KOC)obs) is underpredicted by factors of 40-3330 by the OMP model ((KOC)LFER). Furthermore, the organic carbon-normalized partition coefficients varies by up to a factor 1000 for a given individual PCDD/F compound, suggesting that the OC content is not the key sorbent for the PCDD/Fs. that the measured appCw for fluoranthene decreased only with 30% in their sampling of about 1000 Ls. Since the PCDD/Fs are factors of 50-5000 more hydrophobic and sorptive than fluoranthene to the hydrophobic PUF material, we believe such data support minimum effect of column breakthrough of dissolved PCDD/Fs in our field experiments. The large scatter in the (KOC)obs (Figure 2) also suggests that bulk POC is not a good metric of the sorbent domain for the PCDD/F in marine particles of the Grenlandsfjords. Analogous to the PAH case (8, 11, 13, 14, 16, 17, 19), the consistently high (KOC)obs relative to predictions based on bulk organic matter partitioning point to the presence of a strongly sorbing subphase such as SC within the bulk reduced carbon pool. These water-column data are consistent with a soot-influenced PCDD/F phase distribution in the GrenVOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4971
TABLE 2. Correlation of the Factor Under-Predicted Particle Association (Expressed as (KOC)obs:(KOC)pred) to the Quotient PSC:POC for Filters Sampled in May 2000 in the Grenlandsfjords
a
analyte
r2, pa
n
12378-PnCDD 1234678-HpCDD OCDD 2348/2378-TCDF 12348/12378-PnCDF 23478-PnCDF 123478/123479-HxCDF 123678-HxCDF 234678-HxCDF 123789-HxCDF 1234678-HpCDF 1234789-HpCDF OCDF
0.73* 0.55** 0.48* 0.66** 0.65** 0.77*** 0.60** 0.60** 0.76** 0.96* 0.53* 0.53* 0.51*
6 11 11 12 12 12 10 10 8 4 11 10 11
*, p < 0.05; **, p < 0.01; ***, p < 0.001.
many different locations and seasons in the three different layers are separated by their 95% confidence intervals and can thus be regarded as significantly different at the p < 0.05 level. From the surface to the deep water there is an increase by on average a factor of 12 (min-max, 8-14), and from the deep water into the sediment a further increase by a factor of 8 (5-15) is found. In total, the surface sediment particles carry, on an OC basis, an average factor of 100 (58-140) more of the PCDD/Fs than the particles in the surface waters. This means that either the Cw or the KOC must vary within the vertical profile. The appCw has been measured in the Grenlandsfjords water column in the same vertical profile discussed here for COC. For the tetra- and penta-chlorinated PCDD/Fs, the appCw were about a factor of 5 higher in deep water samples as compared to surface waters. However, for hexa-, hepta-, and octa-chlorinated PCDD/Fs, appCw were more equal or even higher in the surface water. These Cw trends would not be changed by using the corrCw (eq 3), since the DOC concentrations were similar in the surface and the deep water (Table 1). Combining eqs 1 and 2 and solving for appC gives w app
FIGURE 3. Organic carbon-normalized concentration of PCDD/Fs in aquatic particles in the surface water (white bars, n ) 8-10), bottom water (light gray bars, n ) 6), and sediment (dark gray bars, n ) 17). Error bars show 95% confidence intervals. All concentrations increase by roughly a factor 100 from the particles in the surface waters down to the sediment. landsfjords water column. For most of the PCDD/F isomers analyzed, the degree of underpredicted particle association in the water column (i.e., (KOC)obs:(KOC)pred, Figure 2) correlated positively to the quotient PSC:POC on the filters (Table 2). This means that the samples with the most enhanced particle association had the highest PSC:POC ratio, a finding that is in qualitative agreement with the soot-inclusive distribution model. Vertical Trend in PCDD/F Particle Affinity. If bulk OC were the main carrier of PCDD/F molecules in the Grenlandsfjords particles, one would expect similar OC-normalized PCDD/F concentrations (i.e., COC) in samples from within a well-mixed area. Additionally, this also requires the product of Cw and KOC to be invariable. This is analogous to the assumption of constant fugacity and fugacity capacity within well-mixed areas, commonly used in box-type fate models for organic contaminants (e.g., refs 40 and 41). Here, we investigate if the condition of invariable COC (and Cw and KOC) applies in a vertical profile through the water column reaching down into the surface sediment in the Grenlandsfjords. Hence, POC-normalized PCDD/F concentrations for the suspended particles (CPOC) in surface (0-2 m) and deep (0-16 m above sediment, at >48 m water depth) waters as well as TOC-normalized PCDD/F concentrations in the sediment particles (CTOC) were compared (Figure 3). The COC increased markedly with vertical depth. The average COC from 4972
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002
Cw )
COC KOC
(4)
which illustrates that a constant KOC (as predicated by the OMP model) requests a linear relationship between appCw and COC. Unfortunately, the sediment porewater Cw could not be reliably measured in this work, but it is clear that the average 12-fold increased COC between surface and deep water was not obtained in the appCw between these two layers. This analysis suggest that KOC must be correspondingly higher for particles in the deep water. Since temperature influences the PCDD/F solubility in water (42) and probably also to some extent in organic matter, we tested whether this parameter could explain any of the apparent increase in COC and/or KOC observed with water depth. The variation of KOW with temperature has been measured experimentally to estimate the enthalpy of octanol-water phase transfer (∆HOW) for PCNs and PCBs (43) and chlorobenzenes (44) but, unfortunately, not for PCDD/ Fs. The ∆HOW varied from 13 to 32 kJ mol-1 for these related compounds. During our sampling occasions, the temperature was at most 6 °C colder in the deep water. Assuming that the ∆HOW are similar for the PCDD/Fs, this gives only a factor 1.35 higher KOW at the lower deep-water temperatures. Temperature can thus not explain the higher COC in the deep water (Figure 3). Koelmans et al. (45) attributed increasing KOC for PCB, PAH, and pesticides during settling and sedimentation to increasing OC:N ratio of the aquatic particles during the sedimentation process. This is analogous to findings that polysaccharide-rich colloids sorb pyrene a factor of 3 less efficiently than terrestrially derived organic colloids of more lignin-rich humic structures (e.g., ref 35). However, similar OC:N ratios and δ13C signatures in surface and deep water of the Grenlandsfjord (Table 1) do not suggest that such changes in bulk organic matter composition can explain the factor of 12 difference in partitioning. Still, there is a factor of 2 increase in OC:N from bottom water to the sediments. This increase in OC:N along with a simultaneous increase in δ13C (Table 1) are characteristic of bulk organic matter degradation. A dominant portion of the organic matter settling from the surface water is degraded at mid-depth and on the sediment surface as particularly labile compound classes such as proteins and carbohydrates are rapidly remineralized microbially (e.g., ref 46). While local runoff of terrestrial organic matter also may contribute to the bulk organic carbon (47), particularly in the bottom sediments,
FIGURE 4. Linear regression coefficients (r2) of sedimentary PCDD/F congeners (ng/g dw) vs SC (mg/g dw) (black) and TOC (mg/g dw) (gray). Only r2 for regressions where the slopes were significant at p < 0.05 are shown (n ) 13-17). The geographical distribution of sedimentary PCDD/Fs in the fjord sediment is better accounted for by SC than by TOC, and this effect is most pronounced for the PCDFs. the trend in δ13C suggests that it may not be a dominant portion. The increasing COC and/or KOC are consistent with the PCDD/Fs being significantly associated with the SC in the marine particles. SC is much more recalcitrant than biogenic OC (20-23). Unfortunately not resolvable in averaged bulk organic matter properties in this fjord system (Table 1), it is likely that SC, while being only a minor fraction of the bulk reduced carbon, becomes increasingly important as a PCDD/F carrier along any remineralization gradient. This is strongly suggested by the significant correlations between enhanced particle affinity and SC:POC ratio for the water column samples collected in the year 2000 campaign (Table 2), irrespective of sampling depth or distance from source. The impact of SC on affecting the environmental distribution of PCDD/Fs was further evidenced by their sedimentary distribution. Geographical Distribution of PCDD/Fs in Surface Sediments. To further evaluate the relative importance of SC and bulk OC as carriers of PCDD/Fs, these matrix parameters were correlated with the PCDD/F content of surface sediment at 16 geographical locations distributed throughout the Grenlandsfjords area (Figure 1). Regressions of sedimentary PCDD/F with fOC and fSC, respectively (Figure 4), had slopes that were positive and significant. This means that in sediments with higher fSC or fOC the sedimentary PCDD/F was also higher. Importantly, a significantly higher degree of correlation was observed for sedimentary PCDD/F with fSC than with fOC (the regressions with fSC were overall 32% higher than with fOC). Hence, SC describes the geographical distribution of PCDD/Fs in Grenlandsfjords sediments better than bulk organic carbon. This is analogous to a previous study on PAHs, where similar regressions showed that the distribution of compounds such as benzo[a]pyrene in surface sediments of the Gulf of Maine was much better explained by SC than by bulk organic carbon (15). Since both amorphous organic matter and combustion particles could be expected to be largely associated with the “fines” fraction of the sediment, a co-sorting and spurious correlation between fSC and fOC could be expected. Despite this, the stronger coupling of PCDD/F with SC was resolved in the current study. For the PCDD/Fs in the Grenlandsfjords system another potentially complicating co-correlation may be anticipated from the close proximity between the river mouth and the emission pipes from the magnesium plant ( 0.005. We note that the SC:OC ratio in Grenlandsfjords sediments (0.04-0.16; Table 1) is not unusual for anthropogenically influenced coastal sediments where ratios of the order 0.02-0.20 is commonly observed VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4973
(e.g., refs 20 and 22). For instance, in greater Boston Harbor, where elevated Kd for PAHs have been related to soot sorption (13, 14, 19), SC:OC ratios in the range of 0.09-0.18 have been reported (14, 19, 20). The rare data presented in this study on the particle associations of PCDD/Fs in a coastal environment indicate that PCDD/Fs should be considered together with PAHs as compound classes whose solid-water distribution and particle-driven geographical dispersal is importantly affected by their interaction with SC in addition to amorphous organic matter. Beyond helping explain enhanced sorption and phase distribution, soot association may explain obscure observations on other processes affected by sorption such as low bioavailability and inefficient bioremediation. For instance, decreased bioavailability for the PCDD/F isomers that specifically had combustion sources were reported for crabs in Canada (54).
(16) (17) (18) (19) (20) (21) (22) (23) (24) (25)
Acknowledgments This work was financed by a grant from the Norwegian Research Council (139032/720) and Norsk Hydro through the Norwegian Institute for Water Research (NIVA, Project Dioksiner i Grenlandsfjordene). We gratefully acknowledge asssistance during the field program by the Norwegian Coastal Surveillance, the crew on K/O Munin, and Hans Ba¨rring. Yngve Zebu ¨ hr, Kerstin Grunder, and Zofia Kukulska are thanked for skillful analytical support. Brage Rygg and Jarle Håvardstun are thanked for making Figure 1.
Supporting Information Available Dissolved and particulate concentration of PCDD/F, bulk organic carbon composition data, and sampling details. This material is available free of charge via the Internet at http:// pubs.acs.org.
Literature Cited (1) U.S. EPA. Dioxin Reassessment-An SAB Review of the Research and Development’s Reassessment of Dioxin; EPA-SAB-EC-01006; May 2001; www.epa.gov/sab. (2) Van den Berg, M.; Birnbaum, L.; Bosveld, A. T. C.; Brunstrom, B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa, R.; Kennedy, S. W.; Kubiak, T.; Larsen, J. C.; van Leeuwen, F. X. R.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.; Safe, S.; Schrenk, D.; Tillitt, D.; Tysklind, M.; Younes, M.; Waern, F.; Zacharewski, T. Environ. Health Perspect. 1998, 106, 775-792. (3) Hamenlink, J. L., Landrum, P. F., Bergman, H. L., Benson, W. H., Eds. Bioavailability: Physical, Chemical, and Biological Interactions; SETAC Special Publication; CRS Press: Boca Raton, FL, 1994; pp 73-108. (4) Muir, D. C. G.; Lawrence, S.; Holoka, M.; Fairchild, W. L.; Segstro, M. D.; Webster, G. R. B.; Servos, M. R. Chemosphere 1992, 25, 119-124. (5) Friesen, K. J.; Foga, M. M.; Loewen, M. D. Environ. Sci. Technol. 1996, 30, 2504-2510. (6) Broman, D.; Na¨f, C.; Zebu ¨ hr, Y.; Lexe´n, K. Chemosphere 1989, 19, 445-450. (7) Lohmann, R.; Nelson, E.; Eisenreich, S. J.; Jones, K. C. Environ. Sci. Technol. 2000, 34, 3086-3093. (8) Broman, D.; Na¨f, C.; Rolff, C.; Zebu ¨ hr, Y. Environ. Sci. Technol. 1991, 25, 1850-1864. (9) Go¨tz, R.; Enge, P.; Friesel, P.; Roch, K.; Kjeller, L.-O.; Kulp, S. E.; Rappe, C. Chemosphere 1994, 28, 63-74. (10) Bucheli, T. D.; Gustafsson, O ¨ . Environ. Toxicol. Chem. 2001, 20, 1450-1456. (11) Readman, J. W.; Mantoura, R. F. C.; Rhead, M. M. Sci. Total Environ. 1987, 66, 73-94. (12) Broman, D.; Na¨f, C.; Wik, M.; Renberg, I. Chemosphere 1990, 21, 69-77. (13) McGroddy, S. E.; Farrington, J. W. Environ. Sci. Technol. 1995, 29, 1542-1550. (14) Gustafsson, O ¨ .; Haghseta, F.; Chan, C.; MacFarlane, J.; Gschwend, P. M. Environ. Sci. Technol. 1997, 31, 203-209. (15) Gustafsson, O ¨ .; Gschwend, P. M. In Molecular Markers in Environmental Geochemistry; Eganhouse, R. P., Ed.; ACS 4974
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002
(26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) (44) (45) (46) (47) (48) (49) (50) (51) (52) (53) (54)
Symposium Series 671; American Chemical Society: Washington, DC, 1997; pp 365-381. Næs, K.; Axelman, J.; Na¨f, C.; Broman, D. Environ. Sci. Technol. 1998, 32, 1786-1792. Bucheli, T. D.; Gustafsson, O ¨ . Environ. Sci. Technol. 2000, 34, 5144-5151. Jonker, M. T. O.; Koelmans, A. A. Environ. Sci. Technol. 2001, 35, 3742-3748. Accardi-Dey, A.; Gschwend, P. M. Environ. Sci. Technol. 2002, 36, 21-29. Gustafsson, O ¨ .; Gschwend, P. M. Geochim. Cosmochim. Acta 1998, 62, 465-472. Masiello, C. A.; Druffel, E. R. M. Science 1998, 280, 1911-1913. Middelburg, J. J.; Niewenhuize, J.; Van Breugel, P. Mar. Chem. 1999, 65, 245-252. Gustafsson, O ¨ .; Bucheli, T. D.; Kukulska, Z.; Andersson, M.; Largeau, C.; Rouzaud, J.-N.; Reddy, C. M.; Eglinton, T. I. Global Biogeochem. Cycles 2001, 15, 881-890. Oehme, M.; Mano¨, S.; Bjerke, B. Chemosphere 1989, 18, 13791389. Musdalslien, U. I.; Stendal, N. A.; Johansen, J. G.; Oehme, M. Chemosphere 1991, 23, 1097-1108. Schlabach, M.; Knutzen, J.; Bjerkeng, B.; Becher, G. Organohalogen Compd. 1998, 36, 505-508. Oehme, M.; Bartonova, A.; Knutzen, J. Environ. Sci. Technol. 1990, 24, 1836-1841. Næs, K. Norwegian Institute for Water Research (NIVA) Report 765/99. 1999; ISBN 82-577-3671-6 (in Norwegian). Blomqvist, S.; Abrahamsson, B. Schweiz. Z. Hydrol. 1985, 47, 81-84. Zebu ¨ hr, Y.; Na¨f, C.; Bandh, C.; Broman, D.; Ishaq, R.; Pettersen, H. Chemosphere 1993, 27, 1211-1219. Bandh, C.; Ishaq, R.; Broman, D.; Na¨f, C.; Ro¨nquist-Nii, Y.; Zebu ¨ hr, Y. Environ. Sci. Technol. 1996, 30, 214-219. Karickhoff, S. W. Chemosphere 1981, 10, 833-846. Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Environ. Sci. Technol. 1983, 17, 227-231. Gschwend, P. M.; Wu, S.-C. Environ. Sci. Technol. 1985, 19, 90-96. Gustafsson, O ¨ .; Nilsson, N.; Bucheli, T. D. Environ. Sci. Technol. 2001, 35, 4001-4006. Burkhard, L. P. Environ. Sci. Technol. 2000, 34, 4663-4668. Govers, H. A. J.; Krop, H. B. Chemosphere 1998, 37, 2139-2152. Seth, R.; Mackay, D.; Muncke, J. Environ. Sci. Technol. 1999, 33, 2390-2394. Go´mez-Belincho´n, J. I.; Grimalt, J. O.; Albaige´s, J. Environ. Sci. Technol. 1988, 22, 677-685. Lun, R.; Lee, K.; De Marco, L.; Nalewajko, C.; Mackay, D. Environ. Toxicol. Chem. 1998, 17, 333-341. Gobas, F. A. P. C.; Pasternak, J. P.; Lien, K.; Duncan, R. K. Environ. Sci. Technol. 1998, 32, 2442-2449. Shiu, W.-Y.; Wania, F.; Hung, H.; Mackay, D. J. Chem. Eng. Data 1997, 42, 293-297. Lei, Y.-D.; Wania, F.; Shiu, W.-Y.; Boocock, D. G. B. J. Chem. Eng. Data 2000, 45, 738-742. Bahadur, N. P.; Shiu, W.-Y.; Boocock, D. G. B.; Mackay, D. J. Chem. Eng. Data 1997, 42, 685-688. Koelmans, A. A.; Gillissen, F.; Makatita, W.; van den Berg, M. Water Res. 1997, 31, 461-470. Wakeham, S. G.; Hedges, J. I.; Lee, C.; Peterson, M. C.; Hernes, P. J. Deep-Sea Res. Part II 1997, 44, 2131-2162. Næs, K. Norwegian Institute for Water Research (NIVA) Report 464/91. 1991; ISBN 82-577-1977-3 (in Norwegian). Pederstad, K.; Roaldset, E.; Ronningsland, T. M. Mar. Geol. 1993, 111, 245-268. Næs, K.; Oug, E. Chemosphere 1998, 36, 561-576. Næs, K.; Knutzen, J.; Berglind, L. Sci. Total Environ. 1995, 163, 93-106. Reddy, C. M.; Pearson, A.; Xu, L.; McNichol, A. P.; Benner, B. A., Jr.; Wise, S. A.; Klouda, G. A.; Currie, L. A.; Eglinton, T. I. Environ. Sci. Technol. 2002, 36, 1774-1782. Ba¨rring, H.; Bucheli, T. D.; Broman, D.; Gustafsson, O ¨ . Chemosphere 2002, 49, 515-523. Dickhut, R. M.; Miller, K. E.; Andren, A. W. Chemosphere 1994, 29, 283-297. Yunker, M. B.; Cretney, W. J. Environ. Toxicol. Chem. 2000, 19, 2997-3011.
Received for review April 3, 2002. Revised manuscript received September 11, 2002. Accepted September 20, 2002. ES020072L
