Environ. Sci. Technol. 2008, 42, 3615–3621
Organic Carbon-Water Concentration Quotients (Πsocs and πpocs): Measuring Apparent Chemical Disequilibria and Exploring the Impact of Black Carbon in Lake Michigan LAWRENCE P. BURKHARD,* PHILIP M. COOK, AND MARTA T. LUKASEWYCZ Mid-Continent Ecology Division, National Health and Environmental Effects Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, 6201 Congdon Boulevard, Duluth, Minnesota 55804
Received October 26, 2007. Revised manuscript received February 20, 2008. Accepted March 7, 2008.
Introduction In aquatic ecosystems, bioaccumulation of hydrophobic nonionic organic chemicals is dependent upon distribution of the chemicals between surface sediments, suspended particles, and the freely dissolved state throughout the water column (1, 2). Net bioaccumulation by a fish depends on these chemical distributions because the sediment and water column are the bases for the benthic and pelagic portions of the food web, respectively. The ratios of concentrations of freely dissolved chemical in water and in carbon associated with sediments or suspended particles are called concentration quotients rather than partition coefficients because active fate and transport processes, above and beyond partitioning processes, e.g., sediment resuspension, particle deposition, organic carbon diagenesis, and chemical loading rates to the ecosystem, control the distribution of the chemical between the two separated phases which are rarely at equilibrium (2). Sediment organic carbon–water column concentration quotients (Πsocs) are defined as the ratio of the concentration of the chemical in the sediment on a total organic carbon (TOC) basis (Csoc) to the concentration of the chemical in the water column on a freely dissolved basis (Cwfd) (1, 2): fd Πsoc ) Csoc/C w
Chemical concentration quotients measured between water and total organic carbon (TOC) in sediment (Πsoc) or suspended particulates (πpoc) in southern Lake Michigan reveal up to 2 orders of magnitude differences for polychlorinated biphenyl (PCB), dibenzo-p-dioxin (PCDD), dibenzofuran (PCDF), and polycyclic aromatic hydrocarbon (PAH) compounds with similar octanol–water partition coefficients (Kows). Apparent disequilibria for PAHs, PCDDs, and PCDFs, determined as measured Πsocs or πpocs divided by their organic carbon equilibrium partitioning values, are significantly greater than disequilibria of PCBs with similar Kows. Apparent disequilibria, when adjusted for black carbon content by using published black carbon nonlinear partition coefficients (Kf,bcs) and a Freundlich exponent (nf) value ) 0.7, still exceed equilibrium predictions for the PAHs, PCBs, and PCDDs but with the PCDF disequilibria uniquely below equilibrium. While Monte Carlo analysis of all the variables associated with the black carbon adjusted disequilibria provides wide confidence intervals for individual chemicals, the large class disequilibria differences between PAHs and PCDFs with respect to the PCBs and PCDDs are highly significant. Use of the PCDD Kf,bcs for calculating both the PCDF and PCDD disequilibria eliminates their extreme divergence. On the basis of the complexity of carbonaceous geosorbent effects and the apparent variable degrees of chemical sequestration in particles, the disequilibria can be adjusted by chemical class to meet expected near equilibrium conditions between suspended particles and water in the hypolimnion. Although these adjustments to the disequilibria calculations produce consistent and plausible values, the complexities of variable carbonaceous geosorbent affinities for these chemicals in Lake Michigan presently favor use of measured, rather than a priori modeled, steady-state total organic carbon-water concentration quotients indexed to TOC as biogenic organic carbon.
* Corresponding author phone: (218)-529-5164; fax: (218)-5295003; e-mail:
[email protected]. 10.1021/es702652b
Not subject to U.S. Copyright. Publ. 2008 Am. Chem. Soc.
Published on Web 04/18/2008
(1)
Similarly, suspended particulate organic carbon–water column concentration quotients (πpocs) are defined as the ratio of the concentration of the chemical in the particulate TOC divided by Cwfd: fd πpoc ) Cpoc/C w
(2)
When all the organic carbon is assumed to be biogenic, the relationships between Csoc and Cwfd are described by the TOC equilibrium partition coefficients (Ksoc or Kpoc) which have been approximated by the octanol–water partition coefficient (Kow) (3, 4) or some fraction ( f ) thereof (5). Thus, when field measured TOC-water concentration quotients, Πsoc and πpoc, are divided by the chemical’s estimated equilibrium partition coefficients (Ksoc and Kpoc, respectively), the apparent disequilibria, or approximate fugacity ratios (6), are determined as follows: disequilibrium ) Πsoc/Ksoc ) Πsoc/( fKow)
(3)
disequilibrium ) πpoc/Kpoc ) πpoc/(fKow)
(4)
When not stipulated otherwise in this report, f ) 1.0 so that Ksoc and Kpoc are nominally set as equal to Kow. For a measured freely dissolved chemical concentration in water, an apparent disequilibrium greater than 1 indicates that the measured chemical concentration in the organic carbon fraction of sediment or suspended particulates exceeds the expected equilibrium concentration, and vice-versa when the disequilibrium is less than 1. For many hydrophobic organic chemicals which have recently reduced loadings into aquatic ecosystems, the sediment-water disequilibria are consequently expected to be greater than 1 due to the slow rates for transport of the chemicals from the sediments to overlying water (through diffusion from pore water and sediment resuspension) coupled with slow rates of desorption. In addition to sediment-water disequilibria that tend to be greater than 1 due to significantly reduced chemical loadings from previous years, data from Green Bay of Lake Michigan, Hudson River, Lake Ontario, Lake Erie, and Lake Saint Clair reveal an inverse relationship between polychlorinated biphenyl (PCB) disequilibria and their Kows (4, 6). An VOL. 42, NO. 10, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3615
organic carbon diagenesis model (6) has been proposed as an explanation for why measured Πsocs and πpocs can exceed expected equilibrium conditions and why the disequilibria (Πsoc/Kow and πpoc/Kow) appear to decrease with increasing Kow in aquatic ecosystems. Further, the diagenesis model suggests that organic carbon normalized concentrations of hydrophobic nonionic organic contaminants should be lower in plankton, larger in suspended particulates, and largest in sediments due to the diagenesis process. PCB field data from deep water systems are generally consistent with this trend; however, there is a need to consider alternative explanations such as interchemical differences in partitioning to black carbon which is not incorporated into the disequilibria defined by eqs 3 and 4. When black carbon is present in sediments and suspended solids, equilibrium concentrations in water with respect to TOC are expected to be significantly reduced because of the exceptional affinity of some chemicals for the black carbon phase (7). The primary objective of this study was to determine the extent to which trace amounts of black carbon in suspended particulates and surface sediments impact apparent disequilibria as calculated by eqs 3 and 4. This objective required measurement of high quality Πsocws and πpocs for PCBs, dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs), and polycyclic aromatic hydrocarbons (PAHs) in southern Lake Michigan for 1000 L water samples collected from both the epilimnion and hypolimnion. Derivation of Equilibrium Organic-Carbon–Water Concentration Quotients for Measuring Disequilibria Affected by Presence of Black Carbon. When black carbon (bc) and biogenic organic carbon (bioc) phases are present in TOC of sediments or suspended particulates, both forms of carbon act additively to sorb organic chemicals, but the bc phase has more sorption capacity per unit mass. In this work, the TOC sorption capacity is related to the bioc concentration plus bc as a bioc equivalent concentration. Under this approach, the fraction of bc in sediment (fsbc) must be adjusted to reflect the bc equivalent fraction bioc, (fsbioc)bc. The black carbon equilibrium partition coefficient (Kbc) is commonly characterized (8) by a Freundlich isotherm, as Kf,bcCwfd,n-1, to reflect nonlinear partitioning behavior with respect to concentration of the freely dissolved chemical in fd,n-1 by Ksoc provides the normalwater (Cfd w ). Dividing Kf,bcCw ization factor for relating fsbc to (fsbioc)bc. Thus, if Ksoc ) fKow, the TOC fraction equated to bioc in sediment having bc, (fsbioceq)bc, is as follows: fd,n-1
(fsbioceq)bc ) fsbioc + (fsbioc)bc ) fsbioc + (fsbc)(Kf,bcCw
/( fKow)) (5)
It follows that the measured concentration of chemical in the sediment normalized to TOC with black carbon set equivalent to bioc, (Csoc)bc as µg chemical/kg dry weight bioc, can be calculated as /(fKow))]-1
fd,n-1
(Csoc)bc ) Cs[fsbioc + fsbc(Kf,bcCw
(6)
and the measured sediment-water concentration quotient adjusted for bc equivalence to bioc is determined by dividing eq 6 by Cwfd to give eq 7: (Πsoc)bc )
(Csoc)bc Cwfd
)
Cs[ fsbioc + fsbc(K f,bcCwfd,n-1/(fKow))]-1 fd Cw
(7) Because the measured (Πsoc)bc is normalized to TOC as bioc and the reference equilibrium relationship is defined as (Πsoc)bc ) fKow, the disequilibrium expression (analogous to eq 3) that incorporates the influence of black carbon in sediment is determined as (Πsoc)bc/(fKow). 3616
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 10, 2008
Values for (πpoc)bc may be calculated, parallel to (Πsoc)bc in eq 7, as shown in eq 8 in which fpbioc is mass fraction of particulates that is biogenic organic carbon and fpbc is the mass fraction of particulates that is black carbon. Note also that the reference equilibrium relationship, (πpoc)bc ) fKow, is specifically between freely dissolved and suspended particulate organic carbon associated chemical in the water column and does not require equilibrium with sediments or pore water. (πpoc)bc )
(Cpoc)bc fd Cw
)
fd,n-1 Cp[fpbioc + fpbc(Kf,bcCw /(fKow))]-1 fd Cw
(8) It is important to remember that comparisons of values of (Πsoc)bc or (πpoc)bc to f · Kow to judge disequilibria are subject to uncertainties associated with assuming Kbioc ) fKow and the applicability of the Freundlich equation for each chemical for any particular data set.
Methods Samples. Five sediment samples (top 1 or 2 cm) were collected in 1994 and 1995 from the eastern near-shore and midlake depositional areas in southern Lake Michigan by the use of a box corer (Figure S1, Supporting Information). Details on the sediment samples, their preparation, the sampling locations, and time required for the deposition of the surficial sediment layer have been reported previously (9). Six 1000 L filtrate (dissolved) and nine 1000 L total suspended solids (TSS) water samples (epilimnion samples in duplicate) were collected in October 1994 at sites 180 (44°40′59″ N, 87°13′29″ W) at depths of 1.4 and 40 m, 280 (43°21′16″ N, 87°14′50″ W) at depths of 12.4 and 50 m, and 380 (42°41′05″ N, 86°27′27″ W) at depths of 11 and 50 m on Lake Michigan (Figure S1). Samples were filtered using a stainless steel Pentaplate filtration device (Micro Filtration Systems, model #303300) containing five ashed (at 450 °C for 4 h) 293 mm Whatman GF/F grade (0.7 µm nominal pore size) filters, and the filtrate was extracted using two Goulden Large Sample Extractors (GLSE) in parallel (10). The Pentaplate and GLSE devices were used as described in U.S. Environmental Protection Agency, Great Lakes National Program Office (11, 12), and Anderson et al. (13). Concurrently with the filtration and extraction of the water samples, TSS (for carbon analysis), dissolved organic carbon (DOC), and particulate organic carbon (POC) samples were collected. Additional information is provided in Supporting Information. Sample Analysis. Sample preparation, carbon analysis, and GC/MS analysis procedures are provided in the Supporting Information. Data Analysis. For the procedural blanks, chemical concentrations were calculated by using one-half the minimum detection limit (1/2MDL) for analytes not detected or not quantified (not meeting ion abundance ratio criteria of (30% for the PCBs, PCDDs, and PCDFs), and for analytes with measurable amounts in the blanks, the amounts quantified were used, i.e., no censoring if amounts fell below the MDL. For the sediment and water samples, measured concentrations were blank corrected, and if the blank corrected value was less than zero, the concentration was set to 1/2MDL. No censoring of the blank corrected chemical concentrations was performed if the concentration fell below the MDL. A total of 15 blank analyses were performed, and the geometric mean of the blank chemical concentrations was used in performing the blank corrections. MDLs were estimated using three times the noise, found using Finnigan MAT95S signal-to-noise algorithm. Πsocs and πsocs were only calculated when none of the sediment and water values were derived using the 1/2 MDL rule.
FIGURE 1. Precision and reproducibility of water concentration data for the southern Lake Michigan ecosystem. The blue circles represent PCB congeners, except for red circles that represent nonortho chlorinated congeners 77, 81, 126, and 169. The yellow squares, orange diamonds, and green triangles are the PCDDs, PCDFs, and PAHs, respectively. Freely dissolved concentrations of the individual chemicals in the filtrates were determined using a two phase partitioning model (14): fd ) Cfiltrate/(1 + KdocDOC) Cw
(9)
where Cfiltrate is the concentration of the chemical in the water passing the filter, Kdoc is the dissolved organic carbon–water partition coefficient, and DOC is the concentration of dissolved organic carbon in the water (kg/L). Kdocs were estimated using the geometric mean regression relationship: Kdoc ) 0.08Kow (14). Log Kows were taken from Hawker and Connell (15) for the PCBs, Govers and Krop (16) for the PCDDs, and PCDFs, and deMaagd et al. (17) and Mackay et al. (18) for the PAHs. The Kow for indeno[1,2,3-cd]pyrene was estimated using the ClogP algorithm due to the lack of a measured value. Monte Carlo Analyses for Uncertainty of Disequilibria Calculations. Monte Carlo analyses were performed using a Visual Fortran (Version 6.6C) program with Visual Numerics ISML Fortran random number subroutines. The probability distributions, variability, and uncertainties for the analyses are provided in the Supporting Information (Table S1), and the analyses were run with 100000 iterations for eqs 3, 4, 7, and 8. The analyses were performed for each parameter separately (varying only one parameter while having the other variables fixed) and with all parameters simultaneously. The analyses with the ftoc and fsbc parameters were performed simultaneously because fsbioc was determined by difference (i.e., ftoc minus fsbc). In cases where fsbioc was equal to or less than zero, new random numbers were drawn until the difference was greater than zero.
Results The concentrations of PCBs, PAHs, PCDDs, and PCDFs, and organic carbon content data for the surficial sediment samples and dissolved and particulate water samples are provided, in detail, in the Supporting Information (Tables S2-S6). The mean total organic and black carbon contents in the sediments were 3.14% and 0.028%, respectively (19). Concentrations of PCBs, PCDDs, and PCDFs in the sediment samples have been reported and evaluated previously (9) while the concentrations of PAHs are reported here. The PAH values compare favorably with measurements of Simcik et al. (20) from southern Lake Michigan; i.e., our average values are not significantly different, R ) 1%, for 12 of 15 PAHs from sites 18 and 19 in the Simick et al. (20) report (Table S7, Supporting Information). For the water samples, concentrations of dissolved and particulate PCBs compare favorably to measurements by Anderson et al. (13) and Trowbridge and Swackhamer (21) (Tables S8 and S9, Supporting Information). Data for PAHs in Lake Michigan waters are limited to near shore surface
waters off Chicago, IL (22), and these concentrations were substantially higher than those determined in this investigation from the midlake sampling location (Table S10, Supporting Information). The TSS, POC, and DOC values are in reasonable agreement, given the difference in sampling times (fall vs spring), with the measurements performed by Anderson et al. (13) (Table S8, Supporting Information). The total organic and black carbon contents in the suspended particulates at station 380 were 22.18% and 0.0343% for the epilimnion, and 10.84% and 0.0117% for the hypolimnion. Despite these differences in carbon contents of solids, dissolved chemical concentrations in hypolimnion or epilimnion water samples were similar. The precision of replicate water column particulate samples and agreement between epilimnion and hypolimnion water samples across the three water sampling stations are excellent (Figure 1, Figure S2, Supporting Information). These comparisons combined with the comparability of the concentration data with other studies strongly suggest that the data for these water samples are of high quality for all four chemical classes. The computed Πsocs, πpocs, and their coefficients of variation (cv) are provided, in detail, in the Supporting Information, Tables S11 and S12. The epilimnion and hypolimnion Πsocs were determined using the average concentrations in sediment and the epilimnion and hypolimnion freely dissolved concentration data, respectively. Because the Πsocs and πpocs are based upon single hypolimnion and epilimnion water samples, the variances for the water samples were estimated by calculating the variance of the six water samples collected in this study across stations 180, 280, and 380. Because two of the water sampling stations are located in other regions of the lake, the cv values probably under-represent the actual precision attainable for station 380. The average cv values for the Πsocs (and πpocs) were 44% (38%), 33% (38%), and 65% (62%) for the individual PCBs, PAHs, PCDDs, and PCDFs, respectively. On the basis of duplicate analysis of the sediment samples, we estimate that approximately a quarter of the uncertainty is due to analytical measurement variability.
Discussion Measured Disequilibria. The Πsocs, πpocs, and associated disequilibria (Πsoc/Kow and πpoc/Kow) for PCBs, PCDDs, PCDFs, and PAHs are shown in Figure 2A-H as a function of their Kows for the southern Lake Michigan ecosystem. Because most chemicals’ Cfd w s measured in epilimnion and hypolimnion water samples are very similar (Figure 1), differences in each chemical’s Πsoc/Kow and πpoc/Kow are largely related to differences in Csoc and Cpoc. There are large differences in disequilibria, for both the sediment and suspended particulates in epilimnion and VOL. 42, NO. 10, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3617
FIGURE 2. Measured values in southern Lake Michigan for (1) log Πsoc and log πpoc for epilimnion (A, C) and hypolimnion (B, D) water samples; (2) total organic carbon based disequilibria, Πsoc/Kow (E, F) and πpoc/Kow (G, H) with representative chemical 95% confidence intervals based on Monte Carlo analysis; (3) disequilibria incorporating literature reported Kf,bcs and Freundlich exponent nf ) 0.7, (Πsoc)bc/Kow (I, J) and (πpoc)bc/Kow, (K, L); (4) disequilibria (M, N, O, P) resulting from use of PCDD Kf,bcs for PCDFs and then adjusting mean hypolimnion (πpoc)bc/Kow values for each chemical class to 1.0 to account for the maximum amount of unavailable chemical measured in TOC. See Figure 1 for the chemical symbol key. hypolimnion, between the chemical classes. These differences are likely to be related to differences in chemical affinities for TOC. Although Monte Carlo analysis of all the variables associated with the TOC based disequilibria provide wide 95% confidence intervals for individual chemicals, analysis of variance confirms the significance of the apparent differences in the class disequilibria between PAHs, PCBs, and PCDFs/PCDFs. The PAHs, PCDDs, and PCDFs have larger Πsoc/Kow and πsoc/Kow values than those for PCBs of similar hydrophobicity, suggesting that PAHs, PCDDs, and PCDFs behave differently than PCBs. Only the PCBs’ disequilibria exhibit a log Kow dependence (negative slope), especially for congeners with log Kow < 6. This may be associated with exceptional volatilization losses of the less hydrophobic, more volatile PCB congeners from the water column over time, as well as effects of biogenic organic carbon diagenesis reported previously (6). The PAH disequilibria are consistently much greater than the PCB, PCDD, and PCDF disequilibria. The disequilibria for PCDDs and PCDFs are consistent with disequilibria derived from the data of Suarez et al. (23) (Figure S3, Supporting Information), i.e., the sediment and particulate disequilibria have little dependence upon Kow. In contrast, disequilibria for PCDDs and PCDFs, derived from the data of Persson et al. (24) (Figure S4, Supporting Information), increase with increasing Kow. 3618
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 10, 2008
If the equilibrium approximations of Ksocs and Kpocs are based on fKow 1.0. This indicates that the observed exceptional departures from equilibrium are primarily due to factors other than uncertainty for Koc values. Numerous investigators (7, 8, 25–27) have concluded that black carbon is often responsible for a majority of the total sorption of PAHs, PCDDs, PCDFs, and planar PCBs in sediments because of the higher affinities of planar aromatic compounds for black carbon in comparison to biogenic carbon, derived from the diagenesis of plant and algae materials. The preferential partitioning of planar aromatic compounds to black carbon could, consequently, explain the differences in disequilibria among the four classes of chemicals measured in this study. This hypothesis is supported by Persson et al. (24) who concluded that PCDD/F behavior in the water column and sediments of the Grenlandsfjord, Norway, fjord (which is highly contaminated with black carbon from a magnesium production plant) was controlled by the black carbon phase in this ecosystem. Assessing Impacts of Trace Amounts of Black Carbon on Disequilibria. To evaluate the influence of black carbon on disequilibria in southern Lake Michigan, (Πsoc)bcs and (πpoc)bcs were calculated on the basis of eqs 7 and 8, respectively, with fKow ) Kow. Values for Kf,bc were estimated using a Freundlich exponent value of nf ) 0.7 with the
equations of Koelmans et al. (28) for linear regressions between log Kf,bc and log Kow for PAHs and PCBs. Kf,bcs for PCDDs and PCDFs were estimated using the regression equations from Barring et al. (29) and nf ) 0.7. The measured (Πsoc)bc and (πpoc)bc values were divided by Kow to provide the disequilibria (Figure 2I-L). As expected, the black carbon adjusted disequilibria for sediment-water (Figure 2I,J) and suspended solids-water (Figure 2K,L) are numerically smaller than the corresponding disequilibria values for all chemicals based on TOC without consideration of black carbon (Figure 2E-H). The same is observed for the data of Persson et al. (24) (Figure S4, Supporting Information). The black carbon adjusted disequilibria are smaller because the adjusted TOC for the sediment and suspended solids includes the black carbon fraction after conversion to its biogenic carbon equivalent. If the black carbon adjusted disequilibria accurately account for differences in each chemical’s partitioning, the values within and among the chemical classes would be expected to become similar. Comparison of the patterns between the disequilibria, with and without the black carbon adjustment for the PAHs and PCBs, reveals that the disequilibria for the two classes do not become more similar after the black carbon adjustment. While PCBs and PCDDs have similar disequilibria as judged by class averages, PCBs with greater log Kows have consistently lower (closer to equilibrium) values of (Πsoc)bcs and (πpoc)bcs. The PCDF (Πsoc)bcs and (πpoc)bcs are significantly less than 1.0 and thus very divergent from PCBs and PCDDs. The greater than 10-fold less than equilibrium values obtained for the black carbon adjusted PCDF disequilibria strongly indicate that the estimated Kf,bc values used are proportionally too large. From the analysis of Hauck et al. (30), it appears that the value of nf probably contributes more variance in the water based disequilibria adjusted for black carbon than the value of Kf,bc. This is confirmed in our Monte Carlo analysis of all the variables affecting the disequilibria calculations (Supporting Information, Table S13). However, rather than adjust nf, it seems more appropriate to hypothesize that PCDFs and PCDDs should have very similar Kf,bcs. This is consistent with the molecular structure similarities and the well recognized common carbon column elution behavior (31). When the Kf,bc/Kow regression equation for PCDDs is used for the PCDFs, the PCDF disequilibria are greater than 1.0 and similar to those of the PCDDs (data not reported). While the chemical concentrations in lake water are expected to be less than equilibrium concentrations with respect to sediments, the (πpoc)bc/Kow values plotted in Figure 2L for the hypolimnion should all be close to 1.0 since near equilibrium partitioning conditions are expected between suspended particulates and water (32). There are a number of factors which could explain or contribute to the observation of disequilibria greater than 1.0. The equilibrium approximation Kpoc ) fKow would require f values ranging from 2 to 600, with PAHs in the f ) 500 range, in order to let (πpoc)bc/(fKow) ) 1.0, and having Kpocs substantially greater than the chemical’s Kow is not supported in the literature (3–5). A more likely explanation for the observed hypolimnion (πpoc)bc/Kows > 1.0 is overestimation of (πpoc)bc and specifically (Cpoc)bc. The black carbon adjusted disequilibria are dependent upon the relative amount of black carbon in the samples, and in this study, the measured fpbcs are quite small (Supporting Information Tables S2, S3). If fpbc is underestimated, (Cpoc)bc will be overestimated and thus hypolimnion (πpoc)bc/Kows overestimated. The potential existence of complex mixtures of multiple carbonaceous geosorbents (CGs) (27), in the nonbiogenic component of TOC treated as black carbon (BC) only here, can result in underestimation of the
amount of chemical associated with BC and other exceptional CGs. An additional explanation for overestimated (Cpoc)bcs, and thus overestimated (πpoc)bc/Kows, is the assumption that all the chemical measured on/in the solids is available for exchange with water. As has been pointed out (27), these chemicals, and particularly the PAHs, may be essentially sequestered in particles and thereby unavailable for exchange with water. During the ponar dredge collection of surface sediment samples, we observed chunks of coal. It is possible that coal dust is a component of both the surface sediments and suspended particulates analyzed for this study. This may have resulted in measurement of total PAH concentrations much greater than that of the available PAH concentrations. Unfortunately, the available data do not include a quantitative characterization of the various CG components nor the degree to which the chemicals were sequestered in particles. Therefore, after using the PCDD Kf,bcs from Barring et al. (29) to recalculate the hypolimnion (πpoc)bc/Kows for PCDFs, the average hypolimnion (πpoc)bc/Kows were determined for each of the four chemical classes. These averages when divided into 1.0 provide an estimate of the fraction of total chemical not sequestered: i.e., 0.0039 for PAHs; 0.338 for PCBs with log Kow g 7.0; 0.165 for PCDDs; and 0.206 for PCDFs. In order to have average hypolimnion (πpoc)bc/Kows equal 1.0 for PAHs, PCBs with log Kow g 7.0, PCDDs, and PCDFs, these fractions were then used to correct each chemical’s hypolimnion (πpoc)bc/Kow value for the combined effects of chemical sequestration and enhanced CG absorption. The resulting values are plotted in Figure 2P. Each chemical class is thereby clustered about the theoretical equilibrium partitioning line, although the PCBs, having a persistent and reproducible log Kow related negative slope, have obvious systematic deviations for congeners with log Kow < 7.0 due to the (πpoc)bc/Kow correction based only on congeners with log Kow g 7.0. While the hypolimnion (πpoc)bc/Kow values plotted in Figure 2P are mainly illustrative of the potential impact of quite possible, but not verifiable, systematic errors unavoidably incorporated into the disequilibria calculations, the corrections can be independently used to calculate the disequilibria plotted in Figure 2M-O. The high concordance of the (Πsoc)bc/ Kow values plotted in Figure 2M,N with the (πpoc)bc/Kow values (Figure 2P) is striking (r ) 0.935 and 0.954, respectively). Although the PAHs are noticeably shifted below 1.0 for the epilimnion (πpoc)bc/Kow values, the r ) 0.935. Environmental Significance. Assessing the impact of black carbon on the partitioning of nonionic organic chemicals in Lake Michigan on the basis of the chemical’s distributions between surface sediments and lake water is complicated by effects of fate and transport processes; e.g., sediment resuspension, particle diagenesis, deposition, and volatilization. However, the inclusion of associated data for the chemical distributions between suspended particulates and water associated with them provides a closer look at TOC and BC sorption effects on freely dissolved chemical concentrations in water (Cfd w s). Chemical mass balance models (32) have assumed πpoc/(fKow) ≈ 1.0, especially for the hypolimnion. While there is some uncertainty for the effect of DOC on Cwfds, this alone can not explain the observation of most values of πpoc/(fKow) being significantly greater than 1.0 (Figure 2E-H). The most plausible way for measured BC adjusted disequilibria, (πpoc)bc/(fKow), to approximate 1.0 for PAHs, PCBs, PCDDs, and PCDFs in the Lake Michigan hypolimnion is to incorporate sequestration in CGs which include BC (27). This is suggestive of the importance of GCs in the Lake Michigan ecosystem. Reducing uncertainties, associated with the presently available Freundlich isotherm values for Kf,bcs and nfs and their applications, is complicated by the following: potential variability associated with black carbon composition; chemiVOL. 42, NO. 10, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3619
cal sequestration in particles; complexity of site-specific mixtures of CGs; particle size and particle surface conditions; coal particles as a source of PAHs; encapsulation in organic flocs; etc (6, 27). All these challenges are in addition to the need to apply the CG equilibrium partitioning relationships to ultra trace concentrations of PCDDs and PCDFs in water and suspended particles with very small, but significant, concentrations of GCs. Thus, direct a priori prediction of the sediment and particulate-water column disequilibria for hydrophobic organic chemicals in ecosystems having even trace amounts of black carbon presently seems problematical at this time. This study points to a need for investigations of the application of Kf,bcs and nfs, under the Freundlich model, or alternative nonlinear models (33), for hydrophobic organic chemicals in additional aquatic ecosystems. These should probably include measurement and consideration of partition coefficients for different CGs (27) as well as determination of fractions of chemicals which are essentially sequestered in the CGs. It should be appreciated that, despite the complexities and uncertainties for assessing partitioning of these chemicals in association with the CG fraction of total organic carbon, relative degrees of equilibrium-disequilibrium can be inferred through use of the traditional TOC equals biogenic carbon approach. The recent and developing research on black carbon and other CGs provides, at a minimum, an important advancement in understanding why apparent disequilibria can exceed biogenic carbon based expectations.
Acknowledgments The authors thank Damian Shea and David Mount for their reviews of this report, Brian Butterworth for MAT95 support, Elizabeth Durhan and Barbara Sheedy for contributing sample collection and preparation support, and Glen Warren for sampling coordination and ship time within the Lake Michigan Mass Balance Study effort. The information in this document has been funded wholly by the U.S. Environmental Protection Agency. This paper has been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents reflect the views of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
(4)
(5) (6)
(7)
(8) (9)
(10)
(11)
(12)
(13)
(14) (15) (16)
Supporting Information Available Concentrations in the sediment; epilimnion and hypolimnion suspended solids (particulates); epilimnion and hypolimnion water (dissolved, i.e., passing filter) column samples; their ancillary data (POC, DOC, and BC); and the calculated Πsocs, πpocs, and disequilibria are provided for PAHs, PCBs, PCDDs, and PCDFs from Lake Michigan. Tables comparing the Lake Michigan data to the data of Anderson et al. (13), Simcik et al. (20), Trowbridge and Swackhamer (21), and Offenberg and Baker (22) are provided. Figures with the sampling locations and comparing field replicate and hypoliminion vs epilimnion data, Suarez et al. (23) data, and Persson et al. (24) data are provided. This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Thomann, R. V.; Connolly, J. P.; Parkerton, T. F. An equilibrium model of organic chemical accumulation in aquatic food webs with sediment interaction. Environ. Toxicol. Chem. 1992, 11 (5), 615–629. (2) Burkhard, L. P.; Cook, P. M.; Mount, D. R. The relationship of bioaccumulative chemicals in water and sediment to residues in fish: A visualization approach. Environ. Toxicol. Chem. 2003, 22 (11), 2822–2830. (3) DiToro, D. M.; Zarba, C. S.; Hansen, D. J.; Berry, W. J.; Swartz, R. C.; Cowan, C.E.; Pavlon, S. P.; Allen, H. E.; Thomas, N. A.; Paquin, P. R. Technical basis for establishing sediment quality 3620
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 10, 2008
(17)
(18)
(19)
(20)
(21)
(22) (23)
criteria for nonionic organic chemicals using equilibrium partitioning. Environ. Toxicol. Chem. 1991, 10 (12), 1541–1583. U.S. Environmental Protection Agency. Methodology for Deriving Ambient Water Quality Criteria for the Protection of Human Health (2000); Technical Support Document Volume 2: Development of National Bioaccumulation Factors; EPA-822-B03-030; Office of Water: Washington, DC, 2003. Seth, R.; Mackay, D.; Muncke, J. Estimating the organic carbon partition coefficient and its variability for hydrophobic chemicals. Environ. Sci. Technol. 1999, 33 (14), 2390–2394. Gobas, F. A. P. C.; MacLean, L. G. Sediment-water distribution of organic contaminants in aquatic ecosystems: The role of organic carbon mineralization. Environ. Sci. Technol. 2003, 37 (4), 735–741. Gustaffson, O.; Haghseta, F.; Chan, C.; MacFarlane, J.; Gschwend, P. M. Quantification of the dilute sedimentary soot phase: Implications of PAH speciation and bioavailability. Environ. Sci. Technol. 1997, 31 (1), 203–209. Accardi-Dey, A.; Gschwend, P. M. Assessing the combined roles of natural organic matter and black carbon as sorbents in sediments. Environ. Sci. Technol. 2002, 36 (1), 21–29. Burkhard, L. P.; Cook, P. M.; Lukasewycz, M. T. Biota-sediment accumulation factors for polychlorinated biphenyls, dibenzop-dioxins, and dibenzofurans in southern Lake Michigan lake trout (Salvelinus namaycush). Environ. Sci. Technol. 2004, 38 (20), 5297–5305. Goulden, P. D.; Anthony, D. H. J. Design of a large sample extractor for the determination of organics in water; National Water Research Institute, Canada Centre for Inland Waters, Contribution Number 85-121; Burlington, Ontario, Canada, 1985. U.S. Environmental Protection Agency. Standard operating procedure for the sampling and filtration of hydrophobic organic contaminants in Great Lakes waters; Great Lakes National Program Office: Chicago, IL, 1994. U.S. Environmental Protection Agency. Standard operating procedure for the extraction of dissolved-phase hydrophobic organic contaminants in Great Lakes waters using the Goulden extractor; Great Lakes National Program Office: Chicago, IL, 1994. Anderson, D. J.; Bloem, T. B.; Blankenbaker, R. K.; Stanko, T. A. Concentrations of polychlorinated biphenyls in the water column of the Laurentian Great Lakes: Spring 1993. J. Great Lakes Res. 1999, 25 (1), 160–170. Burkhard, L. P. Estimating dissolved organic carbon partition coefficients for nonionic organic chemicals. Environ. Sci. Technol. 2000, 34 (22), 4663–4668. Hawker, D. W.; Connell, D. W. Octanol-water partition coefficients of polychlorinated biphenyls. Environ. Sci. Technol. 1988, 22 (4), 382–387. Govers, H. A. J.; Krop, H. B. Partition constants of chlorinated dibenzofurans and dibenzo-p-dioxins. Chemosphere 1998, 37 (9–12), 2139–2152. de Maagd, P. G. J.; ten Hulscher, D. Th.E. M.; van den Heuvel, H.; Opperhuizen, A.; Sijm, D. T. H. M. Physiochemical properties of polycyclic aromatic hydrocarbons: aqueous solutilitynoctanol/water partition coefficients, and Henry’s law constants. Environ. Toxicol. Chem. 1998, 17 (2), 251–257. Mackay, D.; Shiu, W. Y.; Ma, K. C. Physical-chemical properties and environmental fate handbook, CRCnetBASE 2000; Chapman & Hall CRCnetBASE, CRC Press LLC.: Boca Raton, FL, 2000; CD-ROM. Lukasewycz, M.T T.; Burkhard, L. P. Complete elimination of carbonates: A critical step in the accurate measurement of organic and black carbon in sediments. Environ. Toxicol. Chem. 2005, 249, 2218–2221. Simcik, M. F.; Eisenreich, S. J.; Golden, K. A.; Liu, S. P.; Lipiatou, E.; Swackhamer, D. L.; Long, D. T. Atmospheric loading of polycyclic aromatic hydrocarbons to Lake Michigan as recorded in the sediments. Environ. Sci. Technol. 1996, 3010, 3039–3046. Trowbridge, A. G.; Swackhamer, D. L. Preferential biomagnification of aryl hydrocarbon hydroxylase-inducing polychlorinated biphenyl congeners in the Lake Michigan, USA, lower food web. Environ. Toxicol. Chem. 2002, 21 (2), 334–341. Offenberg, J. H.; Baker, J. E. PCBs and PAHs in Southern Lake Michigan in 1994 and 1995: urban atmospheric influences and long-term declines. J. Great Lakes Res. 2000, 262, 196–208. Suarez, M. P.; Rifai, H. S.; Palachek, R.; Dean, K.; Koenig, L. Distribution of polychlorinated dibenzo-p-dioxins and dibenzofurans in suspended sediments, dissolved phase and bottom sediment in the Houston Ship Channel. Chemosphere 2006, 62, 417–429.
(24) Persson, N. J.; Gustafsson, O.; Bucheli, T. D.; Ishaq, R.; Nœs, K.; Broman, D. Soot-carbon influenced distribution of PCDD/Fs in the marine environment of the Grenlandsfjord, Norway. Environ. Sci. Technol. 2002, 36 (23), 4968–4974. (25) Cornelissen, G.; Gustafsson, O. Sorption of phenanthrene to environmental black carbon in sediment with and without organic matter and native sorbates. Environ. Sci. Technol. 2004, 38 (1), 148–155. (26) Cornelissen, G.; Gustafsson, O.; Bucheli, T. D.; Jonker, T. O.; Koelmans, A. A.; Van Noort, P. C. M. Extensive sorption of organic compounds to black carbon, coal, and kerogen in sediments and soils: mechanisms and consequences for distribution, bioaccumulation, and biodegradation. Environ. Sci. Technol. 2005, 39 (18), 6881–6895. (27) Cornelisson, G.; Breedveld, G. D.; Kalaitzidis, S.; Christanis, K.; Kibsgaard, A.; Oen, A. M. P. Strong sorption of native PAHs to pyrogenic and unburned carbonaceous geosorbents in sediments. Environ. Sci. Technol. 2006, 40 (4), 1197–1203. (28) Koelmans, A. A.; Jonker, M. T. O.; Cornelissen, G.; Bucheli, T. D.; Van Noort, P. C. M.; Gustafsson, O. Black carbon: The reverse of its dark side. Chemosphere 2006, 63, 365–377. (29) BarringH.; Bucheli, T. D.; Broman, D.;.; Gustafsson, Ö. Sootwater distribution coefficients for polychlorinated dibenzo-p-
(30)
(31)
(32)
(33)
dioxins, polychlorinated dibenzofurans and polybrominated diphenylethers determined with the soot cosolvency-column method. Chemosphere. 2002, 49 (6), 515–523. Hauck, M.; Huijbregts, M. A. J.; Koelmans, A. A.; Moermond, C. T. A.; Van Den Heuvel Greve, M. J.; Veltman, K.; Hendriks, A. J.; Vethaak, A. D. Including sorption to black carbon in modeling bioaccumulation of polycyclic aromatic hydrocarbons: uncertainty analysis and comparison to field data. Environ. Sci. Technol. 2007, 41, 2738–2744. Smith, L. M.; Stalling, D. L.; Johnson, J. L. Determination of part-per-trillion. Levels of polychlorinated dibenzofurans and dioxins in environmental samples. Anal. Chem. 1984, 53 (11), 1830–1842. Endicott, D. D.; Richardson, W. L.; Kandt, D. J. MICHTOX: A mass balance and bioaccumulation model for toxic chemicals in Lake Michigan; U.S. EPA report EPA/600/R-05/158; 2005. Allen-King, R. M.; Grathwohl, P.; Ball, W. P. New modeling paradigms for the sorption of hydrophobic organic chemicals to heterogenous carbonaceous matter in soils, sediments, and rocks. Adv. Water Resour. 2002, 25, 985–1016.
ES702652B
VOL. 42, NO. 10, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3621
