Environ. Sci. Technol. 2004, 38, 6256-6262
Method for Testing the Aquatic Toxicity of Sediment Extracts for Use in Identifying Organic Toxicants in Sediments LARRY J. HEINIS, TERRY L. HIGHLAND, AND DAVID R. MOUNT* U.S. Environmental Protection Agency, Office of Research and Development, National Health and Environmental Effects Laboratory, Mid-Continent Ecology Division, 6201 Congdon Boulevard, Duluth, Minnesota 55804
Biologically directed fractionation techniques are a fundamental tool for identifying the cause of toxicity in environmental samples, but few are available for studying mixtures of organic chemicals in aquatic sediments. This paper describes a method for extracting organic chemicals from sediments and then re-introducing them into water column toxicity tests in a way that mimics, at least in part, the partitioning processes that govern bioavailability in sediment. This involves transferring solvent extracts of sediment into triolein and then placing the mixture inside low-density polyethylene dialysis tubing in a configuration similar to semipermeable membrane devices (SPMDs) used for environmental monitoring. For four model compounds, SPMDs were shown to effectively maintain water column exposure in static systems for 10-14 d, with partition coefficients similar to KOW. Toxicity tests indicated that the SPMDs were compatible with four of five freshwater organisms tested and could be used to measure both lethal and sublethal end points. An example application showed good correspondence between organism responses in intact sediment and extracts in SPMDs for both field-collected and spiked sediments. The SPMD-based method offers a simple, flexible test design, amenable to several different test organisms, and the ability to work with complex mixtures of contaminants while maintaining partitioning behavior similar to that within intact sediments.
Introduction Biologically directed fractionation methods, often called toxicity identification evaluation (TIE) methods, are an effective means to identify the cause(s) of chemical toxicity in environmental samples (1). Much of the application of TIE has been in the context of evaluating toxicity in effluents or other water samples, in part because of their application in whole effluent toxicity assessment under wastewater permitting programs such as the National Pollution Discharge Elimination System (NPDES) in the United States. Parallel needs exist to identify toxicant(s) in aquatic sediments which, like effluents, can contain extremely complex mixtures of anthropogenic and naturally occurring chemicals. Though not yet as widely applied as water column methods, TIE * Correspondence author telephone: (218)539-5169; fax: (218)529-5003; e-mail:
[email protected]. 6256 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 23, 2004
methods for sediment toxicity have been developed for both isolated interstitial water and whole sediments (2, 3). Nonpolar organic compounds are common contaminants in aquatic sediments and have been implicated as causes of toxicity in many sediments (4-8). For solid-phase sediment TIE, the addition of sorbents such as coconut charcoal and carbonaceous resin have been effective in reducing the bioavailability of nonpolar organic chemicals in sediment and, therefore, their toxicity (8, 9). While these procedures are useful for indicating that sediment toxicity may be attributable to nonpolar organic chemicals, they do not provide any means to isolate, fractionate, or identify the specific chemicals causing toxicity. Broad scan analysis of sediment extracts by GC/MS generally yields an unresolved complex of overlapping peaks that prevents confident identification of individual constituents (10, 11). And even if all constituents could be identified analytically, the task of evaluating the potential toxicity of each one of the hundreds or thousands of organic compounds present in the sediment would be daunting at best. These same difficulties exist in the analysis of organic toxicants in effluents. In that case, reverse-phase chromatography has been used to isolate organic compounds from the effluent matrix and iteratively fractionate the mixture components. For example, in the scheme proposed by the U.S. Environmental Protection Agency (U.S. EPA) (12), effluent is passed over a column of C18 resin, and then the column is eluted with a series of solutions with decreasing polarity. The resulting fractions are diluted in water and tested directly for toxicity, and nontoxic fractions are discarded. Toxic fractions can be further fractionated using HPLC, with subsequent toxicity testing and chemical analysis of the toxic HPLC fractions. The fractionation process reduces the original complex mixture into simpler subsets of the mixture that are much more easily assessed analytically, while toxicity testing focuses the analysis on only those fractions containing the causative toxicant(s). A parallel process has been applied to interstitial waters from sediments (2, 13), wherein interstitial water is isolated from sediment by centrifugation and then fractionated by reverse-phase chromatography. While this method may be effective in some cases, the logistical burden of extracting sufficient volumes of interstitial water (liters) is daunting. Furthermore, this approach relies on the stability of chemical concentrations during extraction and processing of the interstitial water, which can be problematic (e.g., contamination with particles, losses due to sorption). Finally, many common sediment contaminants are hydrophobic and present in interstitial water at very low concentrations, which limits the chemical mass that can be recovered via interstitial water. An alternative is to solvent extract sediments and then conduct toxicity tests on dilutions of the extract and/or fractions thereof (14). However, there are difficulties with this approach. One of the features that makes direct toxicity testing of extracts from water samples feasible is that chemicals can be assumed to be present in extracts at roughly the same relative concentration as they were in the original sample. Thus, the relative toxicological potencies of extracted chemicals should be more or less conserved if the extract is diluted with water and tested for toxicity. For sediments, this is not the case. Bioavailability of organic compounds in sediments is thought to be proportional to chemical activity, which is approximated by the concentration in the interstitial water at equilibrium (15). The concentration of organic compounds in interstitial water is related to that in the solid 10.1021/es049661c Not subject to U.S. copyright. Publ. 2004 Am. Chem.Soc. Published on Web 09/23/2004
TABLE 1. Theoretical Example Demonstrating Changes in Toxic Potency Associated with Direct Testing of Sediment Extractsa parameter log KOW log KOCb water column LC50 (µg/L) sediment LC50 (µg/g of OC) sediment concentration (µg/g of OC) toxic unitsc in sediment concentration in extract (µg/L) concentration in diluted extract (µg/L) toxic unitsd in diluted extract
diazinon 3.30 3.24 10.7 18.8 188 10 18800 3.75 0.35
DDE 6.76 6.65 1.39 6140 614 0.1 61400 12.3 8.84
a Assumes 100 g of sediment at 1% OC solvent extracted into 10 mL volume and then diluted by 5000× with water. b Calculated from KOW per ref 15. c Toxic units ) sediment concentration/sediment LC50. d Toxic units ) concentration in diluted extract/water column LC50.
and organic contaminants are absorbed into the device in proportion to their freely dissolved concentration in water and their partition coefficient (which approximates KOW). In this paper, we present an approach to assessing organic toxicants in the context of sediment TIE by using SPMD in reverse. A solvent extract of sediment is exchanged into triolein, and the resulting mixture is placed within LDPE tubing. The tubing is then suspended in clean water, allowing organic chemicals to partition between the SPMD and the surrounding water. As the SPMD equilibrates with the water, organic contaminants present in the extract/triolein mixture generate a steady-state partitioning. In this manner the SPMD acts as a partitioning-driven delivery device, enabling organic compounds to be dosed into water in a way that mimics partitioning behavior in sediments while compensating for chemical losses due to sorption, volatilization, or degradation.
Experimental Methods phase by the organic carbon partition coefficient (KOC), which is generally proportional to the KOW of the chemical (15). As a result, two chemicals having the same chemical activity in interstitial water may have greatly different total concentrations in sediment if they have different KOC values. It follows, then, that the relative chemical activity of mixture components in a solvent extract of a sediment can be very different from that in the intact sediment or its interstitial water. The potential impact of this issue on sediment TIE is made clear through the following thought experiment (see Table 1). Consider a sediment contaminated with two pesticides, diazinon and DDE. Assume that the primary toxicant in the intact sediment is diazinon, with 100-fold greater toxicity than DDE (sediment toxic units in Table 1). If this sediment were solvent extracted, the situation would change. Because DDE has a higher KOC, the concentration ratio of DDE to diazinon in the sediment extract would be much higher than in the interstitial water. If the sediment extract were diluted in water and tested for toxicity, we would observe toxicity caused primarily by DDE, with 25-fold greater potency than diazinon (extract toxic units in Table 1). Hence, even if a TIE conducted on the sediment extract correctly identified DDE as the primary toxicant in the diluted extract, it would miss the true toxicant in the intact sediment because of the differential partitioning of the two chemicals and the consequent over-estimation of DDE bioavailability. While this thought experiment intentionally represents an extreme example, it makes clear the desirability of an approach that could compensate for this problem. The first alternative explored in our laboratory was to spike extracts from contaminated sediment back into uncontaminated sediment. After equilibration, which re-establishes normal organic carbon partitioning behavior, the relative potency of chemicals in the extract amended sediment should be proportional to that in the original sediment (assuming similar character of organic carbon and chemical distribution within organic carbon in both sediments). By inserting a chemical fractionation step before spiking, the extract could be simplified and the toxic fractions could be tracked and analyzed in a manner analogous to the procedures used for effluents and other aqueous samples. While we were successful in transferring and testing organic contaminants by this method (unpublished data), it had the disadvantage of requiring lengthy periods of time (28-42 d) for equilibration. Because TIEs often involve multiple, iterative experiments, a method with a shorter equilibration time was needed to meet the logistical needs of sediment TIE. Semipermeable membrane devices (SPMDs) were initially developed for use as passive environmental sampling devices (16) and are composed of low-density polyethylene (LDPE) lay-flat tubing containing a purified lipid, triolein. In use as a sampling device, the SPMD is deployed into a water body,
Experimental Overview. A wide variety of experiments are described in this paper that can be broken down into four primary groups and are outlined below. General procedures are delineated in subsequent sections, with unique details included in the presentation of results from individual experiments. Equilibrium Kinetics and Stability. These experiments involved incorporating known quantities of model chemicals into SPMDs and monitoring the rate of equilibration with the surrounding water as well as the stability of chemical concentration in water over a 10-14-d period. Chemicals tested were the polycyclic aromatic hydrocarbon (PAH) phenanthrene alone; a mixture of three PAHs, phenanthrene, fluoranthene, and chrysene; and the pesticide dieldrin alone. Effect of SPMD Configuration on Partitioning. Initial experiments were conducted using SPMDs comprised of 50 cm of LDPE tubing and 0.5 g of triolein, which are standard dimensions commonly used (16). To determine whether smaller designs would perform acceptably, additional experiments were conducted using the mixture of the three PAHs mentioned above, placed into SPMDs of 10 or 25 cm length and containing varying amounts of triolein. Evaluation of Blank Toxicity. To be acceptable for use in sediment TIE, components of the SPMD must not cause artifactual toxicity to the organisms used to evaluate toxicity of sediment extracts. Blank toxicity was evaluated by water equilibrating SPMDs containing only triolein and then adding organisms and monitoring survival (and growth for some organisms) in exposures lasting from 4 to 10 d. Other blank toxicity experiments evaluated SPMDs loaded with extracts from uncontaminated sediment or with excess extraction solvent added to the SPMD. Application to Field-Collected Sediment. The purpose of these experiments was to compare toxicity observed in exposures to SPMDs containing sediment extracts to toxicity observed in exposures to whole sediment conducted according to standard methods (17). The sediments tested were contaminated sediment collected from the East Branch of the Sebasticook River (Corrina, ME) and an uncontaminated sediment spiked with dieldrin. Chemicals and Solvents. Phenanthrene (CAS Registry No. 85-01-8, 99.5% pure), fluoranthene (CAS Registry No. 20644-0, 98% pure), chrysene (CAS Registry No. 218-01-9, 98% pure), and triolein (1,2,3-tri[cis-9-octadeceneoyl]glycerol, CAS Registry No. 122-32-7, 99% pure) were obtained from SigmaAldrich Chemical Co. (St. Louis, MO). Dieldrin (CAS Registry No. 60-57-1, 99.5% pure) was obtained from ChemService Inc. (West Chester, PA). Sodium sulfate (anhydrous), acetone, acetonitrile, dichloromethane, and hexane (HPLC-GC/MS grade) were obtained from Fisher Scientific (Pittsburgh, PA). Sediment Handling, Spiking, and Extraction. Fieldcollected sediments were held in plastic containers with a VOL. 38, NO. 23, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6257
TABLE 2. Organisms and Methods Used in Toxicity Testing with SPMDs common name
scientific name
age (d)
feeding (per d)
Chironomus tentans 7-10 6 mg of Tetrafin as water slurry Hyalella azteca 10-14 1 mL of YCT/algae mixture (17) Daphnia magna 0-2 1 mL of YCT/algae mixture (17) Pimephales promelas 0-2 1 mg dwt (approx) live brine shrimp nauplii Japanese medaka Oryzias latipes 0-2 1 mg dwt (approx) live brine shrimp nauplii
midge larva amphipod water flea fathead minnow
minimum of headspace and stored at 4 °C. Studies requiring a control or uncontaminated sediment used sediment collected from West Bearskin Lake, a remote lake in far northeastern Minnesota (48°3.86′ N, 90°24.61′ W). Dieldrinspiked sediment was prepared by dissolving dieldrin in acetone and then combining this solution with a small amount of clean quartz sand in a 4-L glass jar. This jar was connected to a vacuum pump and rolled continuously until all acetone was removed. West Bearskin Lake sediment was then added to the jar, thoroughly mixed by hand, placed on a roller mill at 2 rpm for at least 2 weeks at 4 °C, and then held without agitation at 4 °C. Sediment was remixed thoroughly by hand before use in experiments. To prepare sediment extracts, sediment was air-dried at 20 °C and ground to a powder by mortar and pestle. A total of 40-100 g of dried sediment was then combined with an equal mass of anhydrous sodium sulfate and extracted three times with 75 mL of hexane:acetone:dichloromethane (60: 20:20, v/v/v) in a sonic bath at 35 °C. The combined extract was dried by passing it through a funnel containing a plug of glass wool covered with a layer of anhydrous sodium sulfate. The extract was concentrated to a final volume of 40 mL in a 50 °C water bath with a stream of dry nitrogen. SPMD Fabrication and Equilibration. Low-density polyethylene (LDPE) tubing, 2.5 cm wide with 100 µm thick walls (CIA Labs, St. Joseph, MO), was cleaned by soaking in a solution of hexane:dichloromethane (80:20, v/v) for 3 h, airdrying, and then drying under vacuum at room temperature for 18 h. A measured mass of triolein was combined with either neat chemical dissolved in acetone or a sediment extract. The extract was then reduced to a constant mass in a 50 °C water bath under a stream of dry nitrogen. The resulting triolein mixture was added to the inside of the LDPE tubing and spread into a thin layer excluding air, and then the ends of the tubing were heat-sealed. Except where noted, experiments were conducted using SPMDs of standard dimensions (50 cm LDPE tubing, 0.5 g triolein; 16). The loaded SPMD was mounted on a support rack in a zigzag configuration to maximize contact with the surrounding water. This support rack was a circular array of stainless steel bolts extending downward from a Plexiglas disk. The SPMD and supporting structure were then placed in a 600-mL glass beaker containing 300 mL of clean Lake Superior water (sand filtered; pH 7.5; hardness 45 mg/L and alkalinity 40 mg/L, both as CaCO3) such that the SPMD was wholly submerged but the Plexiglas disk remained above the water. To speed equilibration, the water was mixed by gentle aeration with a stream of clean air (70 mL/min) supplied through a fritted glass dispersion tube. Each beaker was fitted with a cover and placed into a temperature-controlled water bath at 23 °C. Small volumes of water were added as necessary to compensate for evaporative and sampling losses. Concentrations of chemicals or extracts in SPMDs were expressed as the mass of chemical in the extract or spike divided by the total mass of the SPMD (triolein + LDPE tubing). To calculate expected chemical concentrations in the water phase, we assumed KOW ) KSPMD. Apparent partition coefficients (K′SPMD, in L/kg SPMD) were calculated as the measured concentration in water divided by concentration 6258
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 23, 2004
substrate
end points
1 mm quartz sand layer 1 mm quartz sand layer none none
survival, biomass survival survival survival, biomass
none
survival, biomass
in the SPMD. For sediment extracts containing unknown mixtures of chemicals, we normalized concentrations in sediments and SPMD preparations relative to the total organic carbon in the sediment extracted and the mass of the SPMD into which the extract was placed. As an example, if 100 g of sediment with 3% organic carbon was extracted, that extract represented 3 g of sediment organic carbon. If the entire extract were transferred to an SPMD made from 0.5 g of triolein and 2.5 g of LDPE tubing, we considered that concentration to be 1×. If only half of the extract were placed in the SPMD, the concentration would be 0.5× and so on. Toxicity Testing. Five species of fish and invertebrates were used in toxicity tests (Table 2), and all were obtained from in-house cultures. For toxicity testing, SPMDs were allowed to equilibrate in the test beaker for a minimum of 64 h, after which 10 organisms were added. One to four replicates were tested, and test duration varied from 4 to 10 d depending on the species and test. Exposure vessels were as described under the SPMD Fabrication and Equilibration section. Survival was determined at test termination, along with biomass (where applicable) measured as dry weight (P. promelas and O. latipes) or ash-free dry weight (C. tentans). Nominal test temperature was 23 °C (range 21.1-24.5) and was monitored daily. Test pH was monitored 1-2 times weekly, while conductivity, dissolved oxygen, hardness, and alkalinity were measured at test termination. All parameters remained within normal ranges for Lake Superior water and well within tolerances of the test organisms. Toxicity tests with SPMDs containing sediment extracts also included a control treatment with aeration but no SPMD and a procedural blank consisting of an SPMD containing only triolein. Bulk sediment toxicity tests were conducted according to U.S. EPA guidelines (17), except test vessels were 100-mL beakers containing 30 mL of sediment. Chemical Analysis. Analysis of PAHs was performed using a Hewlett-Packard model 1090 HPLC equipped with a Hewlett-Packard model 1046A fluorescence detector (Agilent Technologies, Wilmington, DE). Water samples were combined with acetonitrile (10% v/v) and separated on a Prosphere PAH 300A 5u column (150 mm × 4.6 mm i.d., Alltech Associates, Wilmington, DE). Program conditions were as follows: 60% acetonitrile/40% water isocratic for 2.5 min, linearly programmed to 90% acetonitrile at the rate of 2.7%/min, and a 1.5 min final hold, all at 1.5 mL/min. Fluorescence detection was programmed as follows: 0 min, excitation wavelength (λex) 250 nm, emission wavelength (λem) 390 nm, pmt gain 15, 10.0 min.; λex 237 nm, λem 420 nm, pmt gain 16, 12.5 min; λex 277 nm, λem 376 nm, pmt gain 16. PAHs were quantitated using external standard solutions ranging from approximately 0.5 to 20 µg/L. Quantitation limits for phenanthrene, fluoranthene, and chrysene were 0.106, 0.129, and 0.125 µg/L, respectively. Analysis precision as measured by the relative percent difference of replicate analyses (absolute difference of the replicate measurements/mean × 100) averaged 9.20% (SD ) 10.9%, n ) 14), 6.88% (5.77%, 7), and 7.42% (3.23%, 7), respectively. Water samples for dieldrin analysis were liquid/liquid extracted into hexane (1× at 2:1 water:solvent). Extracts from water or sediment (see above) were then solvent exchanged
FIGURE 2. Water concentrations (Cw) of dieldrin released from SPMDs loaded with extract from a dieldrin-spiked sediment. Treatments represent 0.94×, 0.36×, and 0.10× concentrations relative to original sediment. Data are plotted as mean ( 1 SD (n ) 2). Regression line is fit as Cw ) 0.732(1 - e-0.306t) (r 2 ) 0.974). Solid symbols represent measured values; open symbols are calculated values if KSPMD ) KOW. FIGURE 1. Water concentrations (Cw) of PAHs released from SPMDs loaded with (a) 52 µg of phenanthrene/g SPMD or (b) 700 µg of phenanthrene, 680 µg of fluoranthene, and 710 µg of chrysene/g SPMD. Data are plotted as mean ( 1 SD (n ) 2). Solid line in panel a is fit as Cw ) 1.29(1 - e-3.39t) (r 2 ) 0.908). Dashed lines represent predicted water concentrations if KSPMD ) KOW. KOW values from ref 36. into 1,1,1-trimethylpentane. Analyses were performed using a Hewlett-Packard model 5890A gas-liquid chromatograph (GLC) equipped with a 63Ni electron capture detector (ECD) (Agilent Technologies, Wilmington, DE). The instrument was operated in the splitless mode using 1-µL injections of 1,1,1trimethylpentane solutions. The injector and detector were operated at 250 and 300 °C, respectively. A 30 m × 0.25 mm i.d. DB-5 column with 0.25 µm film thickness (J&W Scientific, Folsom, CA) was operated at an initial temperature of 80 °C, a 1 min hold, a 20 °C/min ramp to 200 °C, a 1 min hold, a 4 °C/min ramp to 250 °C, and a 4 min hold. Carrier gas (hydrogen) linear velocity was 50 cm/s, and detector makeup gas was nitrogen at a flow rate of 40 mL/min. Dieldrin was quantitated using external standard solutions ranging from 5.00 to 160 µg/L, with a quantitation limit of 0.20 µg/L. The mean recovery efficiency of lab-fortified solutions was 109% (18.9%; 15) and precision averaged 13.3% (19.0%; 8).
Results and Discussion Equilibration Kinetics and Stability. Initial kinetic experiments were conducted using duplicate SPMDs spiked directly with phenanthrene (52 µg of phe/g SPMD) or a mixture of phenanthrene, fluoranthene, and chrysene (700, 680, and 710 µg/g SPMD, respectively). These studies showed that PAH concentrations in water were at or near steady state within 1-2 d and remained consistent throughout the remainder of the experiments (Figure 1a,b). Average PAH concentrations measured in water after 24 h were remarkably close to that predicted based on KOW and, correspondingly, K′SPMD was very close to KOW (data in Figure 1a,b). In a separate experiment, the extract from a dieldrinspiked sediment (59.8 µg of dieldrin/g of OC) was loaded into SPMDs at three different concentrations representing 0.94×, 0.36×, and 0.10× that in whole sediment. The 0.94× treatment was sampled repeatedly and showed a concentration profile similar to that observed for the PAHs, except that the concentrations approached steady state more slowly (Figure 2). However, concentrations in water measured at the end of the experiment (day 13) in all three treatments
TABLE 3. Effect of SPMD Configuration on K′SPMDa avg log K′SPMD LDPE triolein total days 2-10 (mean ( SD) tubing + spike SPMD length (cm) (g) mass (g) phenanthrene fluoranthene chrysene 10 10* 25* 25 25 50*
0.679 0.74 0.427 0.835 1.680 0.845
1.25 0.169 1.73 2.08 2.98 3.08
log KOW (36)
4.76 ( 0.24 4.46 ( 0.09 4.61 ( 0.10 4.97 ( 0.30 5.12 ( 0.18 4.74 ( 0.34
5.37 ( 0.20 5.06 ( 0.05 5.28 ( 0.11 5.27 ( 0.04 5.48 ( 0.11 5.26 ( 0.21
5.99 ( 0.12 5.74 ( 0.04 5.80 ( 0.02 5.87 ( 0.01 5.99 ( 0.13 5.79 ( 0.08
4.57
5.23
5.81
SPMDs tested in duplicate with four sampling periods each (n ) 8 per treatment). Configurations marked with an asterisk (*) are those with LPDE:triolein ratios closest to the standard configuation. a
were within a factor of 2 of predicted values, with log K′SPMD values of 5.13, 5.07, and 5.00 for the 0.94×, 0.36×, and 0.10× treatments, respectively (dieldrin log KOW ) 5.34; 18). Whether the slower kinetics are related to characteristics of dieldrin, the presence of additional materials co-extracted from the sediment, or some other factor is not clear. Effect of SPMD Configuration on Partitioning. Maintaining a 1:1 correspondence of sediment organic carbon mass to SPMD mass requires that substantial amounts of sediment be extracted (e.g., 300 g dwt sediment with 1% OC for one 50 cm SPMD at 1×). This logistic burden can be reduced if SPMD configurations smaller than the 50-cm tubing used in the initial trials yield comparable results. To address this issue, release of a phenanthrene/fluoranthene/ chrysene mixture was evaluated using six different SPMD configurations (in duplicate) which included the original 50 cm length plus three at 25 cm and two at 10 cm (Table 3). The shorter configurations were also tested with larger amounts of triolein relative to LDPE. Stable PAH concentrations in water were achieved over days 2-10 with all configurations (data not shown; SD given in Table 3). Almost all of the resulting K′SPMD values were within 0.5 log units from the literature KOW value, though they were generally biased high. Among the triolein/LDPE ratios tested, those closest to that in the standard 50-cm SPMD showed the closest correspondence to KOW. From this experiment, it appears that a 10-cm SPMD containing 0.1 g of triolein performs effectively in our test system. Using this 10-cm SPMD instead of a 50-cm SPMD enables a 5-fold reduction in the sediment mass that must be extracted for each SPMD. Evaluation of Blank Toxicity. While the experiments described above suggested that SPMDs could be used as VOL. 38, NO. 23, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6259
TABLE 4. Results (mean ( 1 SD) of Toxicity Testing with SPMDs Containing Only Triolein organism
average survival (%) duration trial (d) replicates control blank
TABLE 5. Results (mean ( 1 SD) of 10-d C. tentans Toxicity Tests of SPMDs with Varying Amounts of Extraction Solventa
biomass (% of control)
mean biomass (ash-free dwt/replicate) treatment
n
mean 10-d survival (%)
mg
% of control
control blank (no solvent) 0.05 mL of solvent 0.25 mL of solvent 1.0 mL solvent
2 2 3 3 3
80 ( 14 80 ( 15 97 ( 5 100 ( 0 90 ( 10
5.96 ( 0.74 5.76 ( 0.14 5.90 ( 0.52 6.23 ( 0.25 6.08 ( 1.20
97 ( 2 99 ( 9 105 ( 4 102 ( 20
SPMDs Containing Only Triolein C. tentans
H. azteca D. magna P. promelas O. latipes
1 2 3 4 5 6 1 2 1 2 3 1 1
4 8 10 10 10 10 4 8 4 10 4 7 7
1 4b 2 2 2 2 1 4b 1 1 4 2c 2c
100 90 ( 20 95 ( 7 80 ( 14 90 ( 0 100 ( 0 100 95 ( 7 100 100 100 ( 0 100 100
90 85 ( 10 90 ( 13 81 ( 15 90 ( 20 100 ( 0 100 87.5 ( 10 20 100 68 ( 29 95 ( 7 97.5 ( 4
SPMDs Containing Triolein + Extract from Uncontaminated Sediment C. tentans a
1 2
Not measured.
10 10 b
2 4
95 ( 7 95 ( 7
85 ( 7 95 ( 10
- -a -108 ( 3 97 ( 2 86 ( 5 105 ( 19
a
88 ( 1 85 ( 19
93 ( 8 110 ( 20
n ) 2 in control. c n ) 1 in control.
partitioning-driven delivery devises to establish predictable exposures in water, it was also important to ensure that the SPMDs did not impart artifactual toxicity to test organisms. To evaluate artifactual toxicity, five different species of test organisms were exposed to blank 50-cm SPMDs containing only triolein, along with controls that received only aeration with no SPMD (Table 4). With the exception of D. magna, survival of organisms exposed to blank SPMDs met typical control performance criteria for toxicity tests with these species (generally g80% for a 10-d test; 17, 19), although average survival overall was slightly lower in the blank SPMD treatments. Biomass for C. tentans averaged 99% of control over four trials, while biomass for the two fish species was slightly lower. Survival of D. magna exposed to blank SPMDs was erratic, as high as 100% and as low as 20%. D. magna floating at the surface of the water was common in blank SPMD treatments, a problem that was not resolved. In addition to testing blank SPMDs containing only triolein, we also tested the toxicity of SPMDs containing extracts of a nontoxic control sediment to determine if natural components of a sediment extract would impart toxicity. Both 10-d tests with C. tentans showed survival and biomass comparable to control (Table 4). While limited to a single organism, this experiment did not indicate that extraction of natural components of sediment would impart artifactual toxicity. A final concern regarding artifactual toxicity lies in the potential for residual extraction solvent to remain in the extract after transfer into triolein. Although the triolein + extract mixtures are dried to a constant mass, they are generally heavier than the triolein alone. The extent to which this additional mass represents extracted components of the sediment or extraction solvent resistant to evaporation is unknown. To determine whether residual solvent (if present) would cause artifactual toxicity, 40 mL of the sediment extraction solvent mix was evaporated under nitrogen to a final volume of 4 mL. Triplicate SPMDs (50 cm, 0.5 g of triolein) were prepared containing 0, 0.05, 0.25, or 1.0 mL of this solution. Toxicity testing with C. tentans showed no adverse effect of the residual solvent, with survival and biomass in the solvent treatments comparable to that in the control and blank treatments (Table 5). Taken together, the toxicity tests above suggest that the SPMD-based testing approach is compatible with common test organisms, with the possible exception of D. magna. There was a tendency toward slightly lower performance of 6260
9
Solvent added to 0.5 g of triolein in 50-cm tubing.
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 23, 2004
FIGURE 3. Mean biomass of C. tentans exposed to a contaminated field sediment and a dieldrin-spiked sediment, each tested in whole sediment toxicity tests and as extracts loaded into SPMDs. Dieldrin concentration is expressed as µg/g of OC for sediments and µg/g SPMD for extracts. n ) 4 for field sediment, n ) 3 for dieldrinspiked sediment, and n ) 2 for SPMDs. organisms exposed to blank SPMDs relative to control, but in the context of TIE studies, we do not believe this to be a significant issue. Application to Field-Collected Sediment. Ultimately, the proof of an SPMD-based sediment TIE methodology lies in whether it can be successfully applied to toxic sediments collected from the field. As an example, we evaluated a sediment collected near a former wool processing facility, which was contaminated with several anthropogenic chemicals including chlorobenzenes, PAHs, dieldrin, DDT, and several metals. A 10-d whole sediment toxicity test with C. tentans showed little effect on survival, but biomass was only 20% of control, suggesting substantial sublethal toxicity. This toxicity was greatly reduced or removed by the addition of carbonaceous resin or coconut charcoal, indicating that the causative toxicant was likely organic (data not shown; see refs 8 and 9 for methods). Accordingly, the sediment was extracted and tested at three concentrations using SPMDs. In addition, uncontaminated sediment was spiked with two concentrations of dieldrin, and those were tested for toxicity to C. tentans in whole sediment tests. Finally, one of the dieldrin-spiked sediments was extracted and tested at two concentrations using SPMDs. Results of whole sediment and SPMD toxicity tests with the field sediment and dieldrin-spiked sediment showed a continuous exposure response curve when dieldrin concentration was normalized to organic carbon (for sediments) or SPMD mass (for extracts; Figure 3). While the dieldrin concentration in the field sediment SPMDs did not span as high as the concentration in the intact sediment, the uniformity of the response curve suggests two conclusions. First, that the SPMD-based approach described here can be used to isolate organic toxicants and re-introduce them to
a water column test system that mimics the potency of those materials in the original sediment. Second, that the toxicity of the field sediment to C. tentans is attributable to dieldrin because the responses of C. tentans to whole sediment and SPMD exposures from the field sediment were no greater than from an uncontaminated sediment enriched with dieldrin only, when all data were normalized to dieldrin concentration. Assumptions and Limitations. While the experiments discussed above suggest the utility of this SPMD-based approach for sediment TIE, its application must be in the context of the associated assumptions and potential limitations. One of the key assumptions in this approach is that chemical activity in the sediment extract/SPMD system approximates that in the intact sediment. Absent evidence to the contrary, it is typically assumed that for nonionic organic chemicals in sediments, KOC is comparable to KOW. As indicated here and elsewhere (16, 20), KSPMD is also expected to be comparable to KOW for many of these same compounds, suggesting that partitioning from the sediment extract/SPMD system described here would be similar to that expected from sediments in general. That said, it is clear from the literature that some sediments show measured KOC values greatly different from KOW, such as in the case of PAH partitioning from sediments containing soot or coal (21, 22). Where this occurs, bioavailability in the intact sediment will likely be misrepresented by the SPMD method presented here, with the expected error in rough proportion to the difference between the observed sediment partitioning and KSPMD. Another aspect of sediment partitioning that has received considerable attention is the concept of “fast” and “slow” desorbing pools of organic contaminants in sediments (23). It has been proposed that it is the fast-desorbing pool that controls bioavailability (24). Because organic solvent extraction is expected to extract chemical from both slow- and fast-desorbing pools, it raises the possibility that exposures derived from solvent extracts would over-represent bioavailability relative to the intact sediment. The degree to which desorption kinetics would influence the representativeness of the SPMD method we present is most likely sediment and chemical specific. However, both theoretical argument (15) and experimental evidence (25) indicate that chemical activity in interstitial water is an appropriate measure of chemical bioavailability in sediments. Thus, if the sediment extract/SPMD system recreates that activity, then it is an appropriate representation of bioavailability in sediment, regardless of whether the extracted chemical originated from the slow- or fast-desorbing pool. Once target analytes are known, comparability of chemical activity in the systems can be verified, as indexed by the concentration in water associated with the sediment or SPMD. An alternative approach to the bioavailability issue is to use an extraction method that is based on chemical activity rather than the solvent extraction we have proposed. As an example, SPMDs have been used to derive environmental extracts that were later fractionated and tested for toxicity (26). Other partitioning-based sampling devices such as solidphase microextraction fibers (SPME) (27, 28), thin films (29, 30), or Tenax beads (24) could be used similarly. While these procedures may sample a different subset of sediment contaminants relative to direct solvent extraction of sediment, solvent extracts of the alternative sampling media still retain the KOW-related concentration bias discussed in the Introduction to this paper. For the toxicity of these extracts to be evaluated properly, they must be re-introduced to toxicity tests using a partitioning driven system, such as the SPMD described here. Alternatively, the original sampling device could be dialyzed against water rather than solvent, if that is feasible.
While using partitioning-type samplers to “extract” sediments for toxicity studies is not burdened by an assumption that KOC ≈ KOW, it is subject to other constraints, including differing uptake kinetics for compounds of differing properties, fouling, and long deployment periods required for sample collection. How these factors weigh against the strengths and limitations of the sediment extract/SPMD method we propose here will depend on study-specific circumstances. We feel the comparatively rapid processing and the large chemical mass that can be processed via solvent extraction are significant logistical benefits for sediment TIE. Once causative toxicants are known, alternative sampling procedures focused on bioavailable chemical could be used to confirm that bioavailability in the intact sediment was not misrepresented by the sediment extract/SPMD system. Beyond the SPMD, several other partitioning-based systems have been suggested as delivery devices for use in toxicity testing with organic chemicals (31-33); most of these have been oriented toward the loading of neat chemical onto the partitioning phase (e.g., C18 resin). In the context of sediment TIE, such methods would be more difficult to apply because extracts from contaminated sediments generally contain a broad range of materials that reduce to viscous or oily mixtures that would be difficult to load. The SPMD approach is not compromised by the physical character of sediment extracts, as the resulting mixture can be easily placed in the SPMD even if the extract is oily or dirty. One of the difficulties of designing an appropriate methodology for use in sediment TIE is that the full range of target chemicals is unknown, and thus the compatibility of experimental procedures with all target chemicals cannot be fully quantified. In this paper, we have used three PAHs and dieldrin as model chemicals known to cause toxicity in sediments, but there is a broader array of chemicals potentially in the environment that could cause sediment toxicity. Our model compounds were generally of moderately high KOW and not highly volatile; for compounds with low KOW and/or high loss rates, the stability of concentrations during extended exposures may be compromised. For highly volatile compounds, the aeration is used to speed equilibration of the SPMD may also accelerate loss from the system. We also have not carefully evaluated the influence of a substantial oil phase in sediments on the comparability of partitioning between sediments and SPMDs, although dissolution of PAHs from oily materials using dialysis tubing has been evaluated by others (34). Use of a water column exposure when testing extracts also limits the influence of dietary exposure in determining chemical uptake and resulting toxicity. The absence of dietary exposure could slow the kinetics of chemical uptake by exposed organisms, although the arguments of Di Toro et al. (15) would suggest that this absence may not greatly influence toxicity at steady state. Differences in rate of equilibration between the three spiked PAHs and dieldrin extracted from sediment were observed in our experiments. These differences are not attributable to KOW alone, since the KOW values for the three PAHs bracket that of dieldrin. Where target analytes are known, equilibration kinetics from loaded SPMDs can be directly measured and experimental equilibration times adjusted accordingly. In sediment TIE where the analytes are not known a priori, some standard time will have to be selected, balancing the logistics of extended equilibration times versus to relative likelihood that a steady state concentration will be reached. On the basis of our work thus far, it appears that 4 d may be a reasonable compromise. Although the kinetic experiment with dieldrin indicated longer periods were necessary to reach steady state, concentrations at day 4 were well within a factor of 2 of the apparent steady state. At least in the early stages of sediment TIE studies, this error is probably tolerably small. Future VOL. 38, NO. 23, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6261
research will likely provide better understanding of this issue. Future Directions. The aggregate of the experiments described above indicate that the SPMD-based TIE method described has utility in the study of organic toxicants in sediments and that the approach can help address the challenge of maintaining relative toxicological potency of organic chemicals after they are isolated from the sediment matrix. Because it is a partitioning driven system, the SPMD method can maintain exposure concentrations for 10 d or more (for the chemicals evaluated), with potential to compensate for loss processes such as sorption, evaporation, and uptake by organisms. This allows the conduct of toxicity tests over periods sufficient to measure both lethal and sublethal (biomass) end points. The example of the dieldrin-contaminated field sediment focused directly on dieldrin because of pre-existing knowledge that dieldrin was a contaminant of concern in that sediment. For that reason, fractionation of the sediment extract was not necessary to simplify the mixture of organic chemicals and aid analytical identification of the causative toxicant, as was described in the Introduction. Coupling the SPMD approach with typical chemical fractionation schemes is being pursued in our laboratory and will be a key element of making our approach broadly applicable for sediment TIE. Once sediment toxicants are isolated in extraction solvent, any number of chemical fractionation approaches could be used, such as those based on polarity, hydrophobicity, or other characteristics (e.g., refs 14 and 35). As long as the fractions obtained can be solvent exchanged back into triolein, SPMDs can be used to assess the toxicity of each fraction, focusing chemical analysis on only those fractions demonstrating toxicity. In addition, dilution series can be prepared using multiple SPMDs loaded with different volumes of extract, allowing the quantitation of toxicity and an assessment of recovery. However, the potential for artifactual toxicity should always be kept in mind, and procedural blanks should be included and tested for toxicity as part of any fractionation scheme.
Acknowledgments We thank C. A. Shreve and C. T. Jenson for assistance in the laboratory, D. M. Di Toro for useful discussions on partitioning theory, and C. Rosiu for arranging collection of the contaminated field sediment. R. M. Burgess, E. J. Durhan, M. A. Starus, and three anonymous reviewers provided helpful comments on previous drafts. The information in this document has been funded wholly by the U.S. Environmental Protection Agency. It 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.
Literature Cited (1) Burkhard, L. P.; Ankley, G. T. Environ. Sci. Technol. 1989, 23, 1438-1443. (2) Ankley, G. T.; Schubauer-Berigan, M. K. J. Aquat. Ecosyst. Health 1995, 4, 133-149. (3) Ho, K. T.; Burgess, R. M.; Pelletier, M. C.; Serbst, J. R.; Ryba, S. A.; Cantwell, M. G.; Kuhn, A.; Raczelowski, P. Mar. Pollut. Bull. 2002, 44, 286-293. (4) Swartz, R. C.; Cole, F. A.; Lamberson, J. O.; Ferraro, S. P.; Schultz, D. W.: DeBen, W. A.; Lee, H., II; Ozretich, R. A. Environ. Toxicol. Chem. 1994, 13, 949-962. (5) Swartz, R. C.; Schultz, D. W.: Ozretich, R. A.; Lamberson, J. O.; Cole, F. A.; DeWitt, T. H.; Redmond, M. S.; Ferraro, S. P. Environ. Toxicol. Chem. 1995, 14, 1987-1997. (6) Hoke, R. A.; Ankley, G. T.; Cotter, A. M.; Goldenstein, T. A.; Kosian, P. A.; Phipps, G. L.; VanderMeiden, F. Environ. Toxicol. Chem. 1994, 13, 157-166. 6262
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 23, 2004
(7) Ho, K. T.; McKinney, R. A.; Kuhn, A.; Pelletier, M. C.; Burgess, R. M. Environ. Toxicol. Chem. 1997, 16, 551-558. (8) Ho, K. T.; Burgess, R. M.; Pelletier, M. C.; Serbst, J. R.; Cook, H.; Cantwell, M. G.; Ryba, S. A.; Perron, M. M. Environ. Toxicol. Chem. 2004, 23, 2124-2131. (9) Kosian, P. A.; West, C. W.; Pasha, M. S.; Cox, J. S.; Mount, D. R.; Huggett, R. J.; Ankley, G. T. Environ. Toxicol. Chem. 1999, 18, 201-206. (10) Doskey, P. V. Environ. Sci. Technol. 2001, 35, 247-254. (11) Connell, D. W.; Wu, R. S. S.; Richardson, B. J.; Leung, K.; Lam, P. S. K.; Connell, P. A. Chemosphere 1998, 36, 2019-2030. (12) U.S. Environmental Protection Agency. Methods for Aquatic Toxicity Identification Evaluations: Phase II Toxicity Identification Procedures; EPA/600/3-88/035; U.S. EPA: Duluth, MN, 1989; 21 pp. (13) Kosian, P. A.; Makynen, E. A.; Monson, P. D.; Mount, D. R.; Spacie, A.; Mekenyan, O. G.; Ankley, G. T. Environ. Toxicol. Chem. 1998, 17, 1021-1033. (14) Maltby, L.; Forrow, D. M.; Boxall, A. B. A.; Calow, P.; Betton, C. I. Environ. Toxicol. Chem. 1995, 14, 1093-1101. (15) Di Toro, D. M.; Zarba, C. S.; Hansen, D. J.; Berry, W. J.; Swartz, R. C.; Cowan, C. E.; Pavlou, S. P.; Allen, H. A.; Thomas, N. A.; Paquin, P. R. Environ. Toxicol. Chem. 1991, 10, 1541-1583. (16) Huckins, J. N.; Manuweera, G. K.; Petty, J. D.; Mackay, D.; Lebo, J. A. Environ. Sci. Technol. 1993, 27, 2489-2496. (17) U.S. Environmental Protection Agency. Methods for Measuring the Toxicity and Bioaccumulation of Sediment-Associated Contaminants with Freshwater Invertebrates; EPA/600/R-99/064; U.S. EPA: Washington, DC, 2000; 192 pp. (18) U.S. Environmental Protection Agency. Draft Sediment Quality Criteria for the Protection of Benthic Organisms: Dieldrin; EPA/ 822/R-93/015; U.S. EPA: Washington, DC, 1993. (19) U.S. Environmental Protection Agency. Short-Term Methods For Estimating the Chronic Toxicity of Effluents and Receiving Waters to Freshwater Organisms; EPA/821/R-02/013; U.S. EPA: Washington, DC, 2002; 350 pp. (20) Chiou, C. T. Environ. Sci. Technol. 1985, 19, 57-62. (21) McGroddy, S. E.; Farrington, J. W. Environ. Sci. Technol. 1995, 29, 1542-1550. (22) West, C. W.; Kosian, P. A.; Mount, D. R.; Makynen, E. A.; Pasha, M. S.; Sibley, P. K.; Ankley, G. T. Environ. Toxicol. Chem. 2001, 20, 1104-1111. (23) Karickhoff, S. W. Sorption kinetics of hydrophobic pollutants in natural sediments. In Contaminated Sediments; Baker, R. A., Ed.; Ann Arbor Science: Ann Arbor, MI, 1980; pp 193-205. (24) Kraaij, R.; Seinen, W.; Tolls, J.; Cornelissen, G.; Belfroid, A. C. Environ. Sci. Technol. 2002, 36, 3525-3529. (25) Kraaij, R.; Mayer, P.; Busser, F. J. M.; van Het Bolscher, M.; Seinen, W.; Tolls, J.; Belfroid, A. C. Environ. Sci. Technol. 2003, 37, 268-274. (26) Sabaliunas, D.; Lazutka, J. R.; Sabaliuniene, I. Mar. Pollut. 2000, 109, 251-165. (27) Parkerton, T. F.; Stone, M. A.; Letinski, D. J. Toxicol. Lett. 2000, 112-113, 273-282. (28) Mayer, P.; Vaes, W. H. J.; Wijnker, F.; Legierse, K. C. H. M.; Kraaij, R. H.; Tolls, J.; Hermens, J. L. M. Environ. Sci. Technol. 2000, 34, 5177-5183. (29) Wilcockson, J. B.; Gobas, F. A. P. C. Environ. Sci. Technol. 2001, 35, 1425-1431. (30) Booij, K.; Smedes, F.; van Weerlee, E. M. Chemosphere 2002, 46, 1157-1161. (31) Mayer, P.; Wernsing, J.; Tolls, J.; Gert-Jan de Maagd, P.; Sijm, D. T. H. M. Environ. Sci. Technol. 1999, 33, 2284-2290. (32) Brown, R. S.; Akhtar, P.; Ckerman, J.; Hampel, L.; Kozin, I. S.; Villerius, L. A.; Klamer, H. J. C. Environ. Sci. Technol. 2001, 35, 4097-4102. (33) Pfister, G.; Ju ¨ ttner, I.; Severin, G. F.; Schramm, K. W.; Kettrup, A. Environ. Toxicol. Chem. 2003, 22, 182-188. (34) Woolgar, P. J.; Jones, K. C. Environ. Sci. Technol. 1999, 33, 21182126. (35) Burkhard, L. P.; Sheedy, B. R. Environ. Toxicol. Chem. 1995, 14, 697-711. (36) de Maagd, P. G. J.; ten Hulscher, D. T. E. M.; van den Heuvel, H.; Opperhuizen, A.; Sijm, D. T. H. M. Environ. Toxicol. Chem. 1998, 17, 251-257.
Received for review March 3, 2004. Revised manuscript received July 8, 2004. Accepted July 16, 2004. ES049661C
