Environ. Sci. Technol. 1993, 27, 875-882
Assimilation of Hydrophobic Chlorinated Organics in Freshwater Wetlands: Sorption and Sediment-Water Exchange John H. Pardue,. Patrick H. Masscheleyn, Ronald D. DeLaune, and William H. Patrick, Jr.
Wetland Biogeochemistry Institute, Louisiana State University, Baton Rouge, Louisiana 70803-7511 Donald D. Adrlan
Department of Civil Engineering, Louisiana State University, Baton Rouge, Louisiana 70803-751 1 In laboratory studies, sorption of hexachlorobenzene (HCB) in wetland soil and floodwater was strongly influenced by high levels of naturally-occurring dissolved organic carbon (DOC). Organic matter-dominated wetlands (e.g., marshes) containing high concentrations of DOC demonstrated no advantage over mineral-dominated wetlands (e.g., bottomland hardwood forests) for sorption of HCB due to partitioning in this DOC phase. Sorption of HCB was described adequately using a three-phase model that included the DOC phase. However, partition coefficients in the bottomland hardwood soil were timedependent. This time-dependent phenomena was not observed in the freshwater marsh soil. Sorptive assimilation of organic compounds in wetlands appears to be dependent on the unique hydrological conditions that promote sediment-water exchange and accretion rather than any enhanced sorptive ability of wetland soils themselves. Introduction
these highly productive systems. Wetlands generally have the highest concentrations of DOC of any aquatic system. Organics can partition into this phase, increasing their apparent solubility. Increased concentrations of DOC result in less partitioning of the organic compound to settleable solids (4-7) and may result in increased transport of the organic out of the system. The second process, rate-limited or nonequilibrium sorption, has been observed in numerous studies in soil, sediments, and subsurface materials (8-10). This effect is thought to be due to the diffusion-limited partitioning of organic compounds into soil structures such as soil organic matter or pore spaces. The end result of this process is that desorption of organic compounds from the soil is hindered, due to the increased fraction of slowly reversible, resistant sorbate. In this case, observed partitioning of the organic compound between particles and water increases over time. We investigated the sorption and sediment-water exchange of hexachlorobenzene (HCB) in two wetland types, a mineral-dominated bottomland hardwood forest and an organic matter-dominated freshwater marsh, to determine the potential for these aquatic systems to assimilate hydrophobic organic pollutants by sorption. The major objectives of this study were (1)to characterize the effect of DOC on sorption of HCB in these two wetland soils and floodwaters, (2) to determine the importance of rate-limited sorption of HCB in these wetland soils, and (3) to quantify rates of sediment-water exchange in these systems. No previous studies have examined the dynamics of these sorption processes in wetland systems. HCB was chosen as a test compound because it is a hydrophobic chlorinated priority pollutant commonly found in hazardous waste sites. In addition, a large estuarine wetland area in coastal Louisiana has been contaminated with HCB, and its assimilation in the wetland environment is of interest (1I).
Wetlands, transitional areas between terrestrial and aquatic ecosystems, perform important functions in the environment including wildlife habitat, aquifer recharge, and flood control (1). In addition, wetland environments have been hypothesized to serve as a buffer zone, as filters, and as sinks for a wide variety of anthropogenic pollutants including toxic organic compounds. The use of natural and constructed wetlands is a growing alternate technology for treatment of various types of wastewater. There are common features in many wetland environments that suggest enhanced assimilation of toxic organics. These include the high organic content of wetland substrates, a large diversity of detrital microorganisms, a wide range of redox zones (from highly oxidized to highly reduced), accretion of the wetland soil which results in burial of the pollutants, and hydrologic conditions (reduced water velocity, shallow water depth) which enhance sedimentation and sediment-water contact. In contrast, there are other common features of wetland environments that may inhibit assimilation of organic contaminants. Large quantities of dissolved organic carbon (DOC) are found in wetlands which can sorb organics, effectively enhancing their solubility and preventing their sorption onto settleable particles (2,3). Also, organics can be sequestered in redox zones where microbial degradation cannot occur. At present, little published work has been performed on the overall fate of toxic organics in wetlands. Sorption of toxic organicsin wetlands is likely influenced by two phenomena with opposite results. The first process is the enhanced solubility of organic compounds in the presence of DOC. In wetlands, the source this DOC is the leaching and incomplete degradation of plant material in
Study Sites. Bulk soil, water, and intact soil cores were sampled from two wetland sites; a forested bottomland hardwood wetland (Spring Bayou Wildlife Management Center, Avoyelles Parish, LA) and a freshwater Panicum hemitomonmarsh (Lake des Allemands, St.John the Baptist Parish, LA). The bottomland hardwood site is intermittently flooded under the influence of the Mississippi and Red Rivers. The soil is a mineraldominated Sharkey clay (Vertic Haplaquept, pH = 6.5, organic carbon = 28 g of C/kg of soil, clay = 6896, silt = 27 % ,sand = 5 % ) which supports bald cypress, oaks, and honey locusts with little understory or ground cover. The study site has been previously characterized (12) and is typical of the extensive bottomland hardwood wetlands in the Mississippi alluvial valley. The freshwater marsh
0 1993 American Chemical Society
Environ. Sci. Technol., Vol. 27, No. 5, l9Q3 875
0013-936X/93/0927-0875$04.00/0
Experimental Section
site is a continuouslyflooded, fringing marsh in the upper end of the Barataria Basin, LA. The soil is an organic matter-dominated peat soil (pH = 6.3, organic carbon = 230 g of C/kg of soil) which supports large stands of Panicum hemitomon (maiden cane). HCB Sorption in Wetland Soil and Floodwater. Suspensions (18d water:soil ratio) were constructed using both bottomland hardwood and freshwater marsh soil. Suspensions were incubated in microcosms as described by Patrick et al. (13) with some modifications. Two suspensions were used for each soil; one maintained under aerobic conditions (+500 mV), the other under anaerobic conditions (-200 mV), encompassing the range of redox conditions present in these wetland soils. Samples incubated under various redox conditions were expected to contain a wide range of concentrations of DOC due to differences in microbial processes occurring under these conditions (14). Suspensionswere equilibratedfor 10days before the addition of HCB. One day before HCB was added, 1g of NaN3was added to inhibit microbial activity and the loss of HCB via biodegradation. HCB (equivalent to 10 mg/kg of soil) was added to the suspensions in a microliter quantity of methanol. Gas purging of the microcosms was terminated after addition of the NaN3 to minimizevolatilizationlossesfromthe system. Thestirred suspensions were then sealed with a Teflon-coatedrubber stopper. Nonequilibrium sorption was studied by sampling the suspensions over time (1day, 2 weeks, 16 weeks, and 24 weeks) and by calculating the observed partition coefficient for the original HCB added. Soil from these suspensions was also used to construct sorption isotherms at different DOC concentrations. Contact time for these isotherms between HCB and the wetland soil was arbitrarily chosen as 24 h. Sorption isotherms were generated using subsamples of the suspensions incubated for different times (1 day, 2 weeks, and 16 weeks) which contained different DOC concentrations. DOC in the stirred suspensions increased over time due to leaching and mechanical breakdown of particulate plant organic matter (14).coupled with the inhibition of DOC oxidation by NaN3. Subsamples of the suspensions were removed with a glass syringe and needle and added to Teflon FEP centrifugetubes. Samples were amended with various amounts of HCB (C-45pgig of soil) and shaken for 24 h. Samples were centrifuged (17OOg for 1b) in a refrigerated centrifuge (Dupont-SorvallRCB-5). The supernatant was carefully decanted to a glass centrifuge tube containing hexane:acetone (1:l)and extracted by shaking for several hours. The dissolved fraction (true dissolved plus DOC-associated) is operationally defined as the HCB remaining in solution after this separation step. DOC in the supernatant was determined by the method of Moore (15). HCB remaining on the soil was determined by extractingseveral aliquotsat each sampling time using bexane:acetone (l:l,shaking for 48-72 h, extraction efficiency, 6C-70% ). HCB was measured by high-resolution gas chromatograpby (HRGC) using a DB-5 capillary column (30 m). Chromatographic parameters were as follows: 2 cL splitlessinjection; injector, 275 "C; temperature program, 32 O C initial temperature heated to 200 OC at 30 OCimin; and electron capture detector, 375 O C . Response factors were calibrated daily using standards. The minimum detection limit was 0.02 ng. 2,4,5,6-Tetrachloro-m-xylene was used as an internal standard. 876
Envlron. Scl. Technol.. Vol. 27, No. 5. 1993
Nz gas out
N2 gas in
$4
Syringe
Flgure 1. Gas purging apparatus used for determination of Kmc.
The partition coefficient for HCB between DOC and water, Kdm, was determined by using the gas purging approach described by Hassett and Milicic (16). Samples of both wetland soils were incubated in microcosms as described earlier. After 2 months of incubation, the bulk samples were removed and centrifuged under the same conditions used to separate dissolved and particulate phases in the isotherm experiments. Aliquots of the bulk supernatant (180 mL) containing DOC were placed in a gas purging apparatus consisting of a 250-mL gas washing bottle containing a small Teflon-coated stir bar. A schematic of the apparatus is presented in Figure 1. A length of Teflon tubing was placed through the outlet arm of the gas washing bottle to periodically remove samples of the solution. An aliquot of 14C-labeledbexachlorobenzene ([WIHCB, 10.8 mCi/mmol, Sigma Radiochemicals) was injected into the solution, and the mixture was equilibrated for 1b while stirring. The entire apparatus was maintained in a water bath at 25 "C. After equilibration, purging of the solution was begun using Nz gas with a low-pressure needle valve (500 mL/min). At short intervals, samples of the solution (2 mL) were removed through the Teflon tubing using a glass syringe. Samples were injected immediatelyinto 20 mL of liquid scintillation cocktail and counted on a liquid scintillation counter. Radiolabeled HCB was used without further purification; therefore, selected samples were extracted and analyzed using HRGC to confirm HCB concentrations. The experiment was duplicated using deionized water (DOC = 0) to establish a baseline HCB volatilization rate according to the method in ref 16. The partition coefficient, Kdoo wasdeterminedbynonliiearregression (SAS,NLIN, Cary, NC) of the general solution to the algorithm described by Hassett and Milicic (16). Partition coefficients were also determined for HCB in wetland floodwater. Water was collected from the study sitesandcentrifuged (1700gfor 1h) toremoveallsettleable particles before use. DOC concentrations were adjusted to the same level (22 mg/L) with deionized H20. Aliquots of the water were placed in Teflon FEP centrifuge tubes. [l4C1HCB,equivalent to 0.6, 1.0, 3.0, and 6.0 pgiL, was added to replicate tubes and shaken overnight. Solids, equivalent to 0, 0.1, 0.3,0.6, and 1.0 g/L, were added to the centrifuge tubes and shaken until an equilibrium was reached (24-48 h). Samples were recentrifuged, and
aliquots were taken for liquid scintillation counting. Random samples were extracted and analyzed for HCB using HRGC to check agreement with 14Ccounts. Sediment-Water Exchange. Diffusive flux between the water column and the wetland soil was determined by using intact cores taken from the field sites. Soil cores (15-cmdiameter) were incubated in the laboratory in PVC containers lined with a Teflon-coated film. A floodwater depth of 6 cm was maintained over the cores. An aliquot of [l4C]HCB was added to the floodwater and mixed, and the activity was monitored over time. Cores were capped to minimize volatilization. Molecular diffusion coefficients, D,, were determined using 3Hz0in duplicate soil cores using the reservoir method described by Van Rees et al. (17). Accretion Rates. Short soil cores (15-cm diameter X 40-cm depth) were removed from each wetland site. Cores were sectioned into 3-cm horizontal sections, air-dried, and ground in a mortar and pestle. Accretion rates were determined by measuring 137Csactivity in core sections using a Ge-Li detectorimultichannel analyzer. The core section with maximum 137Csactivity was assigned a date of 1963, corresponding to the date of maximum l37Cs atmospheric fallout (18).
between two samples will result in changes in the observed partitioning by a factor of
Theoretical Section
(KdocCdoc/KpCpart)l (9) where Cpart is the concentration of particles (mg/L) and % true dissolved % doc-assoc + % particulate = 100. Sediment-Water Exchange. Sediment-water exchange can be most simply expressed as a simple Fickian diffusion process as follows:
Sorption. In general, sorption of organic compounds is described most simply as a linear partitioning process: S = KpC, (1) where S is the sorbed concentration (mg/kg), Kp is the “true” partition coefficient (L/kg), and C, is the “true” dissolved concentration (mg/L). Because of the strong dependence of sorption on the organic carbon content of the soil, the partition coefficient Kp is often normalized as KO,= Kp/foc where KO, is the organic carbon normalized partition coefficient (L/kg) and focis the fraction of organic carbon (dimensionless). In the presence of a third phase, such as DOC, the sorption equation is modified to
Kp-obsl/Kp-obsZ= (1-t ( ~ ~ o c ~ ~ , c ~( ~ / ~~ ~o 6c ~ ~) /o(6) (c ~~ / ~ 0 6 ~ where 1 and 2 denote samples with two different DOC concentrations. Ratios of a series of observed partition coefficients (Kp-obsl/Kp-obsl) measured from samples with different DOC concentrations can be plotted versus DOC using eq 6 and will have a theoretical slope of (Kdoc/(106+ Kdoccdocd). Speciation. Equations describing the speciation of organic compounds in these three phases can be derived using the equilibrium approach above. The percentage of compound in true dissolved, DOC-associated, and particulate form are given by eqs 7-9 (3, 4): % true dissolved = 100/{1+ (KPCp,,/1O6)
+
(KdocCdoc/lo6)] (7)
+
+
% doc-assoc = 100/{1 (106/K~,cc~oc)
(KpCpart/KdocCdoc)) 70 particulate = 100/(1 + (106/KpCpart) +
+
6cfW/6t= ( D , / R ) ( ~ ~ c , / ~ x ~ ) (10) where Cf, is the concentration of HCB in the floodwater (mg/L), C, is the concentration of HCB in the sediment pore water (mg/L), R is the retardation factor (dimensionless), x is the distance (cm), D , is the molecular diffusion coefficient (cm2/day),and t is the time (day). Appropriate initial and boundary conditions are given by: C,(X,O) = 0
(loa)
s = Kp.,b,(Cw + Ld,,)
(3) where KPobsis the “observed” partition coefficient (L/kg) and Ldoc is the concentration of organic associated with DOC on a volume basis (mg/L). At present, it is difficult to separate organics associated with all three phases (true dissolved, DOC associated, and particulate) in a single measurement. Therefore, it is necessary to determine the partition coefficient for the organic compound between DOC and water, alone, as
When two-phase equilibrium sorption is the only mechanism of solute retardation in the soil, R can be expressed as:
R = 1 + pKp/B
(11) where p is the sediment bulk density (kg/L = g/cm3), 0 is the volumetric water content (cm3/cm3),and Kp is the sediment-water partition coefficient (L/kg).
(4) where Mdocis the concentration of organic associated with DOC on a weight basis (mg/kg) and Kdocis the partition coefficient (L/kg). The relationship between Kpand Kp. obs can then be expressed as:
Results and Discussion
Kp-,bs = Kp/(l + (KdocCdoc/106)) (5) where Cdoc is the concentration of DOC (mg/L). This equation defines the effect of increasing concentrations of DOC on the observed partitioning of the organic compound. Increases or decreases in the DOC concentration
Sorption of Hexachlorobenzene in Wetland Soils. Microcosm studies consistently demonstrated lower sorption of HCB on the freshwater marsh soil, despite the higher fraction of organic carbon (0.23 vs 0.028 for the bottomland hardwood soil). Isotherms were typically linear (over the range of added HCB) for the freshwater Environ. Sci. Technol., Vol. 27, No. 5 , l S S 3
877
;::
10’
,
-
3c
.
I
0 .
Hours
(Ei
EL Hardwood EL Hardwood MARSH (*E)
31
0
z
e
I
I
6
7
d
9
1 o4
15
12
HCB in solution (pg/L)
50,
,
/
I
I
0 1
lo3
1
000
001
002
003
D0C-l
005
006
007
(rng/L)-’
Flgure 3. Determination of partltion coefflcients between HCB and wetland DOC. (Inset) Example of data from gas purglng apparatus in the presence and absence of DOC.
D
HCB in solution (pg/L)
Flgure 2. Sorption of HCB by freshwater marsh and bottomland hardwood wetland soil (A) and effect of increasing concentrations of DOC on HCB sorption in freshwater marsh soil (B).
marsh soil, while soluble concentrations of HCB in the bottomland hardwood soil appeared to be controlled by the water solubility of HCB itself [6 pgIL (19)l (Figure 2). Concentrations of DOC ranged from approximately 10 to 250 mg/L in the microcosm suspensions, with the DOC of the freshwater marsh soil 2-10 times higher than the bottomland hardwood soil. Freshwater marsh isotherms constructed with soil that had been incubated 2 weeks and 16 weeks had progressively lower Kp-obs’S (lower sorption) than those constructed after 1day of incubation (Figure 2). A corresponding 5-fold increase in DOC in the freshwater marsh soil over this time frame was also noted. These results suggest that DOC influences sorption of HCB, particularly in the freshwater marsh soil. The inverse relationship between DOC and the partition coefficient has been demonstrated previously for PCBs in estuarine sediment (6)and in laboratory studies using lake sediments ( 4 ) . Since wetlands generally have the highest DOC concentrations of any aquatic system (141, quantifying this effect is most important in understanding toxic organic partitioning. The positive correlation between the sediment-water partition coefficient, Kp, and f o c is well-established but was not observed in these wetland soils. In the freshwater marsh soil, high concentrations of DOC appear to increase the solubility of HCB, canceling any sorptive “advantage” the organic soil has over the mineral soil. Predicting the magnitude of this DOC effect should be possible given the knowledge of the partition coefficient between HCB and natural DOC. Partition coefficients between wetland DOC and HCB (Kdoc) were determined and ranged from 104.5to 105.2using the gas purging approach. Partition coefficients were inversely related to 878
0.04
Envlron. Sci. Technol., Vol. 27, No. 5, 1993
DOC concentration. This effect has been noted previously (16) and was attributed to adsorption of the organic compounds on the glass purging apparatus. Therefore, measured partition coefficients were plotted against the inverse of the DOC concentration to eliminate any glass adsorption effects (16) (Figure 3). The intercept of the regression line (the point corresponding to infinite DOC where no HCB would be bound to the glass appsrqtw) was taken to be the best estimate of the DOC-water partition coefficient (104.38).This assumes, of course, that Kdoc is the same for DOC at both wetland sites. The strong linear relationship (r2 = 0.99) in Figure 3 suggests this; however, previous work has shown differences in partition coefficients between DOC from different sources (2).The Kdoc vahe is comparable to that determined for 2,2’,5,5’tetrachlorobiphenyl using this same technique (10485104.87)(16). The Kdocalso falls within the wide range of values previously determined for HCB with dextran (103.08),groundwater DOC (105.65), and humic acid (105.g8) (7). The partition coefficient for HCB determined in the present study for wetland DOC was st least 1 order of magnitude lower than for commercial humic acid and for natural DOC isolated from groundwater. Differences in the molecular composition of DOC from various sources probably account for this wide range of values, as noted previously (2). Again, the definition of DOC is highly operational in nature. In the present study, centrifugation was used to isolate this fraction. Different separation techniques would likely result in different partition coefficients. Knowledge of the partition coefficient of HCB between DOC and water allowsthe use of a three-phase partitioning model (eqs 5-9) to describe sorption in these wetland substrates. This three-phase model was able to predict differences in the observed partition coefficients in the sorption isotherms. Using the data in Figure 2 as an example: increasing the DOC in the freshwater marsh from 19.9 to 51.8 mg/L should decrease the Kp.obsby a factor of 1.5 using eq 6. The measured decrease in the observed partition coefficient was 3294/1880 = 1.75. Therefore, for this example, knowledge of DOC and Kdoc can give some predictive ability for changes in the observed
Table I. Observed Partition Coefficients of HCB in Freshwater Marsh Soil under Various DOC Concentrations after 24-h Contact
samplea 1 (AE-1) 2 (AE-14) 3 (AE-112)
4 (AN-1) 5 (AN-14) 6 (AN-112) slope of K p u b s l / Kpubsxversus DOC theoretical slope = Kduc/(1o6
-k
Kp-,,bs(L/kg) (fsE)
5956 f 337 3296 f 71 1748 i 117 4386 f 268 3612 f 182 1219 f 95 0.013 i 0.005
15000
I
I
I /
12000
-
3000
-
I
'
DOC (mg/L) 10.5
19.9 51.8 31.0 121.0 246.0 r2 = 0.64
6C00 r
0.019
0 BL,H,Kp-abn=135
KdocCdocl)
AE = incubated under aerobic conditions; AN = incubatedunder anaerobic conditions; numbers indicate days of incubation of the soil Drior to addition of HCB. (I
n
0 FM,Kp-o,s=268
Ldi 0
'0
20
3C
40
50
6C
,
70
dprn/mL
Figure 4. Partitioning of HCB in wetland floodwater.
partition coefficient. An overall statistical evaluation of model fit can be obtained plotting a series of partition coefficients versus different concentrations of DOC using eq 6. Theoretically, a plot of Kp-obsl/Kp-obsxversus DOC should yield a slope of ( ~ ~ o c / (-t l oKdocCdocl)), s where Kp. and Cdocl represent observed partition coefficients and DOC concentrations for some initial sample. Partition coefficients from the freshwater marsh suspensions are presented in Table I. For the freshwater marsh data in Table I, the theoretical slope = 0.019. The actual slope from the experimental measurements was 0.013 f 0.005 (SE),which is not statistically different from the theoretical slope at the 5 % level of confidence. Therefore, it appeared that a three-phase model was applicable for predicting changes in the partitioning in wetland substrates. HCB Sorption in Wetland Floodwater. Observed partition coefficients were also determined for HCB in the wetland floodwater. Partitioning of HCB in the floodwater can be much different because the ratio of settleable particles to colloidal phases (e.g., DOC) is much lower than in the wetland soil. For example, particulate concentrations in the water column might vary from loo to lo2 mg/L while concentrations in the soil are 104-105 mg/L. DOC concentrations in the water column and sediment pore water can vary from lo1 to lo2 (20). In many wetland systems, DOC concentrationsgreatly exceed suspended solids concentrations (14). In bulk water samples taken at different times of the year, suspended solids concentration ranged from 10 to 156 mg/L in the bottomland hardwood site but were
