Contaminant AMPLIFICATION. in the Environment

Contaminant AMPLIFICATION in the Environment

A new approach suggests that phenomena, such as bioconcentration, biomagnification, and bioaccumulation, result from two fundamental processes. R OBIE MACDONALD, DON MACKAY, AND BRENDAN HICKIE As we view the behavior of contaminants in the environment, the question comes to mind: Why are certain contaminant concentrations amplified to unusually high levels in certain environmental media, such as lipids and aerosol particles, and in certain locations, such as high- altitude lakes? This is obviously of concern because high concentrations can lead to greater exposures and amplified effects on organisms. Effective management of commercial chemicals and those produced inadvertently, especially those that are persistent, bioaccumulative, and toxic (PBT), requires that we fully understand the fundamental mechanisms causing concentration amplification. This understanding can lead to predictive and then preventive strategies. Wania discussed aspects of this issue in a pioneering study and suggested three fundamental causes: equilibrium partitioning, kinetic effects, and changes in phase composition, volume, or temperature (1). Here, we take a somewhat different approach and suggest that there are only two basic and quite distinct processes or operations, which we term solvent switching and solvent depletion. These, in turn, can cause two effects: concentration amplification and fugacity amplification. In the environment, these processes can occur singly or in combination. Discriminating between the two effects clarifies the fundamental causes of high contaminant concentrations, contributes to a fuller understanding of the fate of chemicals in the environment, and may help to anticipate how global change operates on contaminants. We also suggest that under certain transient or dynamic conditions, a filtering effect may cause high concentrations. In this feature, we introduce and illustrate these concepts using an example drawn from analytical chemistry and then discuss a number of natural environmental phenomena. BEVERLY DOYLE Solvent switching and solvent depletion An environmental chemist can quantify the presence of a hydrophobic organic contaminant at a 1 ng/l concentration in water. If a total mass of 10 ng of analyte is required, the obvious strategy is to obtain an adequate sample of the water, say 50 L, and increase the concentration by extracting it with an organic solvent, such as dichloromethane (DCM). If the extraction is completed with a total vol- 2002 American Chemical Society DECEMBER 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 457 A

ume of 0.5 L of DCM, possibly split into several sequential batches, the concentration in the extract will be amplified 100 times to 100 ng/l. Solvent switching is responsible for this concentration amplification, that is, the analyte has become more concentrated by exploiting its high partition coefficient between DCM and water. There is no increase in fugacity (a thermodynamic parameter closely related to partial pressure). Instead, the equilibrium criterion of fugacity will probably fall in value. To further increase the concentration, the extract can be evaporated down to a volume of 5 ml. Both concentration and fugacity now increase by another factor of 100 because of solvent depletion. The net result is an overall concentration increase by a factor of 10,000 due to two effects: solvent switching, which causes a 100-fold increase, and solvent depletion, which contributes another factor of 100 increase. A 1-mL injection of this concentrated extract into the instrument supplies the required quantity of analyte. FIGURE 1 Contaminant concentration fugacity plot An organic solute is extracted from water with dichloromethane (DCM) under equilibrium conditions (solvent switching, ) followed by DCM evaporation (solvent depletion, SD). Similar processes contribute to bioconcentration, bioaccumulation, and biomagnification. Molar mass of the solute is 250 g/mol; Z, which is defined in the text, for water is 0.0008 mol/m 3. Pa; initial concentration in water is 1 ng/l or 4 nmol/m 3 ; initial fugacity is 5 µpa. Concentration 100 ng/l 1 ng/l Extraction 5 µpa 10,000 ng/l SD Evaporation Fugacity 500 µpa Figure 1, a schematic plot of concentration versus fugacity, depicts and discriminates between the two contributing causes of this amplification. Solvent switching merely increases the concentration by exploiting the partition coefficient of the analyte between DCM and water, but it does not increase the analyte s fugacity. Solvent depletion increases both the concentration and fugacity because the concentration increase occurs in the single medium of DCM. Thermodynamically, the first process proceeds spontaneously because there is no increase in free energy indeed, there will be a thermodynamically favored decrease. The second process involves a free energy increase (or an entropy decrease) that can only be achieved by energetic intervention in this case, by evaporating the solvent, which induces a greater free energy decrease elsewhere in the system. We suggest that there is a third process. Filtering can be applied under dynamic or unsteady-state conditions and viewed as causing contaminant amplification, but it is really only mechanical in nature. Again, an example is drawn from analytical chemistry. If the water contains 0.5 ng/l of a truly dissolved chemical and an additional 0.5 ng/l sorbed onto particles, then filtration at a rate of 100 L/h will accumulate 50 ng/h on the filter. The water and particle phases have separated, resulting in a high concentration on the filter. There is no increase in fugacity or concentration. Instead, there is merely preferential mechanical accumulation of a solvent phase particles that has a high concentration because of their high partition coefficient with respect to water. This filtering process can appear to cause high concentrations in an environmental situation in which there is mechanical accumulation of a high concentration phase. Examples are filter-feeding organisms or accumulation of aerosol particles on foliage. A fourth possibility exists. Biological systems have evolved the capability to regulate internal concentrations by sequestering, excreting, or pumping desired or undesired elements, such as metals and nutrients. This may also result in amplification, but to our knowledge these specialized mechanisms do not apply to PBT chemicals. Solvent switching and solvent depletion processes also occur in the environment. We suggest that when probing the science underlying contaminant fate in the environment, it is useful to determine how the overall concentration increase is separately attributable to fugacity and concentration amplification. We present the following environmental examples. Bioconcentration, bioaccumulation, and biomagnification In many respects, these three phenomena in fish are analogous to DCM extraction and evaporation. Bioconcentration is like solvent switching from water to lipid phases in fish, whereas biomagnification is caused by solvent depletion as lipids are digested. Bioaccumulation is the sum of both processes. As shown in Figure 2, if a fish is exposed to water containing a nonmetabolizable hydrophobic contaminant, such as a PCB, at a concentration of 1 ng/l or an equivalent fugacity of 5 micropascals (µpa), the ultimate concentration of PCB in fish could be 10 5 ng/l or ng/kg, or 100 ng/g. Thus, the fish water concentration ratio or bioconcentration factor is 10 5 L/kg. This may be caused by the lipid-to-water partition coefficient (e.g., K LW of 10 6 ) and the presence of 10% lipid in the fish. K LW is usually equated to the octanol water partition coefficient, K ow, in models of the bioconcentration phenomenon. The 10 5 concentration amplification is entirely attributable to solvent switching from water to lipid. The fugacity of the PCB in the fish approaches that of the water, 5 µpa, 458 A ENVIRONMENTAL SCIENCE & TECHNOLOGY / DECEMBER 1, 2002

and there is no fugacity amplification. As a result, the bioconcentration line in Figure 2 is vertical. If, however, under bioaccumulation conditions, the fish is also fed a diet containing 10% lipid, which is in equilibrium with the water and contains a chemical at a concentration of 100 ng/g of food, the animal can biomagnify that concentration and end up with an internal concentration of 300 ng/g. This represents a fugacity amplification or biomagnification factor of 3, going from 5 to 15 µpa. A bird or mammal fed the same diet can reach much higher concentrations with a fugacity amplification of about 30. The overall bioaccumulation process thus represents a concentration amplification by a factor of 3 10 6 but a fugacity amplification of only a factor of 3. Biomagnification is a fascinating phenomenon that involves a sequence of solvent depletion and solvent switching steps. It can be described by tracking the fate of the contaminants as solutes in the neutral lipid solvent, notably triglycerides, present in the ingested food. In the gut lumen, lipids are hydrolyzed by digestive enzymes to monoglycerides and free fatty acids, causing a loss of solvent and a modest increase in fugacity in the gut contents as shown in studies with fish by Gobas et al. (2, 3). As digestion proceeds, the contaminant is forced to redistribute from lipid to other organic matter, which causes a higher fugacity because of the lower partition coefficient into this other solvent. Digestion also reduces the amount of both solvents. Solute concentration, C (mol/m 3 ), is related to fugacity, f (Pa), with a proportionality constant, Z (mol/m 3 Pa), which expresses the affinity of the solute for the solvent in question. Therefore, C is Zf. In this case, the Z value of the lipid is much higher than that of the other organic matter by a factor such as 30; thus, there is simultaneous solvent depletion (of lipids) and switching into a less-favored solvent. The net result is a fugacity increase by about a factor of 3. Mackay and Fraser have given a numerical example of this phenomenon (4). While the fugacity increases in the gut, the products of lipid hydrolysis diffuse into the brush border cells lining the intestine, where triglycerides are resynthesized and form packets referred to as chylomicrons. This newly FIGURE 2 formed solvent pool causes a localized increase in Z and fugacity reduction within the cell, which provides a fugacity gradient that allows the chemical to diffuse into the cell (a solvent switching process). This gradient is maintained by the ongoing production and release of chylomicrons into the bloodstream by a process of cellular excretion called exocytosis and the subsequent transport of these lipid packets to the liver and other tissues. This process has been referred to as the co-assimilation or the fat-flush hypothesis (5, 6). Because the assimilation efficiencies of the chemical and lipid are similar in magnitude, the final critical solvent depletion step in biomagnification takes place in the tissues where virtually all the assimilated lipid is metabolized for energy. Taken together, these steps allow chemicals to be accumulated from food against an apparent fugacity gradient and reach biomagnification factors or fugacity amplifications of 30 or more. A digression It is interesting to consider the relative importance of these two amplification processes. On an arithmetic basis, the concentration increase from water to food is almost 100 ng/g, and from food to fish is a further 200 300 ng/g. Biomagnification or fugacity amplification is thus apparently twice as important as bioconcentration. On a logarithmic scale, which is the most relevant thermodynamically because free energy is related to the log of the concentration, the relative contributions are 5 decades (factor of 10 5 ) by bioconcentration and 0.5 decades (factor of 3) for biomagnification. Here, bioconcentration is more important than biomagnification by a factor of 10. The reader is left to adjudicate this controversy! Concentration fugacity plots of bioconcentration and bioaccumulation Bioconcentration (left) involves only solvent switching (); bioaccumulation (right) involves from water to food and solvent depletion (SD), which leads to a corresponding fugacity increase in the biomagnification stage. Concentration 1 ng/l Water 5 µpa Fugacity 1 ng/l Water 300 ng/g Biomagnification 5 µpa Fish 15 µpa Food and SD Fish 100 ng/g 100 ng/g Bioconcentration Wet deposition Aerosols. Rainfall often contains concentrations of contaminants, such as PAHs, PCBs, or metals, which are factors of 10 4 to 10 6 times their concentrations in air. The rain-to-air concentration ratio is termed a washout ratio and can represent a considerable amplification. Two processes have occurred. First, and most important, the contaminant has partitioned from the gaseous to an aerosol phase, with possibly 50% in the gaseous form and 50% sorbed to particles. Thus, there DECEMBER 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 459 A

FIGURE 3 Solvent switching and solvent depletion in the environment Natural processes involving solvent switching () result in a concentration increase but no fugacity increase, whereas solvent depletion (SD) increases fugacity and possibly concentration. Processes illustrated are (a) wet deposition, (b) fog water, (c) deposition in snow and subsequent melting, (d) sedimentation, and (e) lipid dynamics. Concentration (a) Aerosol (d) Air Rain Suspended sediment (b) Coalesced Fog fog water SD Fugacity is a much higher concentration in the aerosol. The rain has scavenged the aerosol particles to the extent that each liter of rainwater has removed all the aerosol from possibly 200,000 L of air. It has therefore removed half the contaminant from this volume of air, and the concentration in the rain is now 100,000 times that of the original total air concentration. In addition, there may be equilibrium partitioning between the gas phase and the water with a ratio of perhaps 1000 in favor of the water (i.e., the air water partition coefficient, K AW, is 0.001). Thus, the total concentration in the rain is 101,000 times that of the total concentration in air. Drinking 1 L of this rainwater will result in an exposure approximately equivalent to inhaling 101,000 L of air, a process that would normally require about a week. The bulk concentration in the rain is much higher than that of the air, but it is lower than that in the aerosol phase. This concentration (and exposure) amplification is entirely the result of solvent switching. There is no fugacity amplification. In Figure 3a, both lines are vertical. The key solvent switching process is from the gas to aerosol phase, the aerosol phase being greatly preferred by the contaminant. The corresponding partition coefficient (aerosol air) is of the magnitude of 10 11 and depends strongly on the contaminant vapor pressure or K ow. Fog water. A similar phenomenon can occur with fog, in which the air water interface can sorb substantial quantities of contaminant, resulting in bulk concentrations in fog water that greatly exceed the Air Bottom sediment and SD (e) (c) and SD Initial Snow Meltwater SD Air Final Lipid dynamics during migration or starvation concentration of water in equilibrium with the air (7). When the fog droplets coalesce, the solvent of interfacial area is lost and there is a fugacity increase as shown in Figure 3b. Snow and snowmelt. Snowfall is an important mechanism of contaminant deposition, with some snowfalls possibly contributing in excess of 10% of total annual inputs to small, remote lakes (8). In a recent review of the mechanisms involved in the deposition of hydrophobic contaminants in snow, it was suggested that the two processes occur sequentially (9). First, snow like rain can scavenge aerosols from the atmosphere. This is again simple solvent switching between gas and aerosol. Second, the contaminant can sorb to the air ice interface in appreciable quantities. Franz and Eisenreich reported corresponding washout ratios as large as 10 6 (10). Hoff et al. measured specific areas of snow of ~0.1 m 2 /g (11). A typical sorption partition coefficient is 2(g/m 2 )/(g/m 3 ); that is, the ratio of concentration on the interface (C i, g/m 2 ) to concentration in air (C A, g/m 3 ). It follows that 10 6 g of snow could contain 0.1 10 6 2 or 200,000 C A g of contaminant. The fugacities of the contaminant in the air and snow are likely to be equal, but when this snow melts to form 1 m 3 of water and the surface vanishes, the concentration in the meltwater is likely to exceed the equilibrium water air concentration namely, C A /K AW. The K AW is typically 0.01, so the meltwater concentration corresponds to a fugacity of ~20,000 times that of the air. As has been pointed out, this could have immense implications for the behavior of chemicals in melting snowpacks (9, 12). It is likely that during melting there will be appreciable evaporation of any chemical with a significant vapor pressure, including semivolatile compounds. Fugacity amplifications by factors of thousands seem feasible with local chemical concentrations in air greatly exceeding those in the atmosphere from which the snow originated. It is even possible that during melting there could be precipitation of a pure chemical phase, if the fugacity exceeds the vapor pressure of the pure substance. This is essentially solvent depletion with the air ice interface acting as the solvent as depicted in Figure 3c. A further consideration is that the meltwater from this snow is likely to be very contaminated, analogous to the acid pulses associated with early spring snowmelt events in acid-sensitive systems (13). In the arctic marine environment, meltwater accumulates as a buoyant layer under the melting ice because of its low salinity. This occurs at a time when there is rapid growth of epontic (under-ice) algae and subsequent incorporation of this biomass into marine food webs with the corresponding bioaccumulation. This phe- 460 A ENVIRONMENTAL SCIENCE & TECHNOLOGY / DECEMBER 1, 2002

nomenon may well explain why the individual sums of PCB and DDT concentrations and fugacities in some ice algae samples can be about six- to eight-fold higher than in other samples of ice algae and pelagic (oceanic) algae (14). Sedimentation of organic particles in lakes The production and fate of particulate and dissolved organic matter are important in determining the bioavailability and fate of hydrophobic contaminants in aquatic systems. The (autochthonous) organic matter produced primarily by algae within lakes includes lipids, proteins, and polysaccharides, and is thus a complex solvent pool in which contaminants may accumulate (15). As algal cells die and sink, some of the more labile organic matter is consumed by bacteria, releasing CO 2 and dissolved organic carbon, which is further used by benthic bacteria and deposit-feeding organisms. For example, the organic carbon fraction of particles settling in the center of Lake Michigan can decrease by a factor of about 10 as they drop from the lake surface to the sediment (16). Organic carbon content also declines severalfold with greater depth in sediment cores and becomes increasingly dominated by higher-molecular-weight recalcitrant compounds (17, 18). Moreover, Axelman et al. reported a nearly 10-fold increase in carbon-normalized PCB concentrations in settling particles in the Baltic Sea as the organic carbon degrades (19). It is thus likely that there are both concentration and fugacity increases from solvent depletion as shown in Figure 3d. Benthic organisms may then dwell in and approach a fugacity higher than their pelagic companions and become more contaminated on a lipid-normalized basis. Lipid dynamics in organisms Many species rely on fat reserves to meet their energy requirements during periods of low food intake, such as hibernation, migration, molting, or scarce food availability. Increases in tissue concentrations of some contaminants that are nearly proportional to the extent of fat reserve depletion have been documented in a number of species, including migrating fish, marine mammals, and hibernating polar bears (20 23). In the study by Ewald et al., the lipid content in sockeye salmon muscle declined by ~60% during their spawning migration, whereas the concentrations of PCBs and DDTs in muscle lipid increased by factors of 3.7 and 5.5, respectively (20). There is a corresponding fugacity increase resulting from this solvent depletion, as shown in Figure 3e, which can be expected to cause increases in contaminant concentration in blood and other organs, including the gonads. There may be a corresponding increase in toxic impact, raising the prospect that migration can induce a toxic stress at a time when fish are already stressed by an arduous journey that must be completed if the population is to remain viable. There may be enhanced transfer of lipid-soluble contaminants to eggs, disrupting endocrine processes. Increases in mixed function oxidase enzyme activity and tissue concentrations of persistent organic pollutants associated with the depletion of fat reserves have been observed in captive fish and starving beluga whales (22, 24). Earlier perspectives on amplification In a pioneering exploration of this issue, Wania described these effects but suggested three causes rather than our two (1). His equilibrium partitioning effect and rapid phase change effect are essentially our respective solvent switching and solvent depletion. Indeed, Wania also used biomagnification as an example. He states, The fugacity increase is caused by a rapid decrease in phase volume. We do not agree that a rapid decrease is necessary. If the decrease is slow, there may be time for diffusion to reduce the fugacity, but this is, in our view, a separate issue of equilibration kinetics. FIGURE 4 Mass balance diagram for a single compartment At steady state, a fugacity (f ) increase may result if the transport parameter value in (D in ) exceeds that for total loss, including degradation (D out ). N in and N out are the transport rates in and out, respectively. At steady state, N in = N out and f out /f in = D in /D out. N in = D in f in f out N out = D out f out Wania s third dynamic or kinetic effects are depicted in Figure 4, in which there is fast transport into a phase, but only slow transport out. The rates of these transport processes, N in and N out (mol/h), can be expressed as Df, where D is a transport parameter (determined by flow or diffusion rate and Z value) and f is fugacity. In Figure 4, D in exceeds D out. Because at steady state N in must equal N out, f out must exceed f in and the rate f out / f in will approach D in /D out. This is a similar concept to the original explanation of bioconcentration, in which the rate constant for contaminant transport in, k 1, is much greater than the contaminant rate constant out, k 2,, and the contaminant concentration ratio tends toward C 2 /C 1. Situations with D in exceeding D out do occur, but the fundamental reason for the inequality of these values under steady-state conditions is solvent depletion. In an entirely reversible diffusive situation, D in equals D out. For D in to exceed D out implies the presence of a one way or nondiffusive process that transports the chemical into the phase in association with matter that is not expelled either because the matter is being reduced or degraded or because it is being retained and builds up in quantity. Examples of the latter include filtering or settling of particles from water and migration of fish into but not out of lakes. These cases are inherently unsteady-state in nature because there is continuing buildup of a contaminated phase, but only at the input fugacity. A steady- DECEMBER 1, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 461 A

state situation can only apply if the contaminated phase or solvent is depleted. If a contaminant is subject to degradation, this has the effect of introducing another removal process and thus increasing the total D out, and the tendency to amplify by solvent depletion will be reduced. Similarly, degradation in a phase tends to reduce the fugacity in that phase below that of the surrounding media or phases. Amplification will thus be most pronounced for persistent contaminants, such as certain organohalogens. We thus maintain that fundamentally, solvent switching and solvent depletion are the only two causes of concentration amplification in the environment, but we also recognize that under unsteady-state conditions there can be a local buildup of quantity in the contaminated phase by selective filtering. This issue is, however, somewhat esoteric, and the more important task is to understand and predict the causes of concentration amplification. As Wania pointed out, viewing chemical fate in this light helps to focus the search for locations of amplification in the following three situations. First, there are phases with high solvent capacity or high Z values to which chemicals, especially persistent ones, have solvent switched because of selective chemical affinities or low temperatures these are often ideal phases for monitoring. Second, there are situations in which solvents or sorbing media, such as surfaces, lose capacity because of solvent depletion due to phenomena such as lipid hydrolysis or metabolism, organic carbon degradation, or loss of sorbing surfaces arising from melting or coalescence. Finally, a compartment can experience a large D in but small D out either because of solvent depletion or unsteady-state filtration of material with a high capacity for the chemical. Onto global change We routinely expect to observe concentration changes by solvent switching at approximately equal fugacities, which is essentially an environmental manifestation of Nernst s distributive law, where the ratio of concentrations is a constant partition or distribution coefficient. But we should also be aware that when a contaminant resides in a solvent or at an interface that is depleted by digestion, degradation, coalescence, or other phenomena that reduces the dissolving power of the solvent, a fugacity amplification is expected that can also contribute to concentration amplification with corresponding environmental effects. Appreciating these concepts can help clarify the fundamental nature of the many transport and transformation processes that contaminants experience in the environment. Returning to the earlier digression, we might ask how these two processes would be affected by global change. In principle, the effect on solvent switching by a 5 C temperature rise, for example, can be predicted by temperature coefficients for phase partitioning. Modeling suggests that temperature changes of this order are likely to lead to only small changes in risk, and probably reduced risk as increased temperature drives contaminants out of the water or off surfaces (25). In contrast, solvent depletion processes are sensitive to the projected changes and probably offer greater opportunity for surprises. For example, changes in temperature and hydrology can alter the nature of scavenging (snow, rain), the trophic structure (population size distribution, length of food chain), and individual lipid dynamics, each of which can easily be imagined to vary contaminant concentrations by a factor of 2 10. Robie Macdonald is a chemical oceanographer with the Institute of Ocean Sciences of the Federal Department of Fisheries & Oceans, located in Sidney on Vancouver Island, British Columbia (Canada). His primary research interest is the fate and transport of contaminants in the Arctic marine environment. Don Mackay is director of the Canadian Environmental Modeling Centre at Trent University. His interests are the properties, fate, transport, and modeling of organic chemicals in the environment. Brendan Hickie is a conjunct professor in Environmental and Resource Studies at Trent University (Canada). His research focuses on contaminant behavior in the aquatic environment including bioaccumulation. Address correspondence to Mackay at Canadian Environmental Modeling Centre, Trent University, Peterborough, ON K9J 7B8, Canada. References (1) Wania, F. Environ. Sci. Pollut. Res. 1999, 6, 11 19. (2) Gobas, F. A. P. C.; Zhang, X.; Wells, R. Environ. Sci. Technol. 1993, 27, 2855 2863. (3) Gobas, F. A. P. C.; Wilcockson, J. B.; Russell, R. W.; Haffner, G. D. Environ. Sci. Technol. 1999, 33, 133 141. (4) Mackay, D.; Fraser, A. Environ. Pollut. 2000, 110, 375 391. (5) Nichols, J. W.; et al. Drug Metab. Dispos. 2001, 29, 1013 1022. (6) Schlummer, M.; Moser, G. A.; McLaughlin, M. S. Toxicol. Appl. Pharmacol. 1998, 152, 128 137. (7) Hoff, J. T.; Mackay, D.; Gillman, R.; Shiu, W. Y. Environ. Sci. Technol. 1993, 27, 2174 2180. (8) Welch, H. E.; Muir, D. C.; Billeck, B. N.; Lockhart, W. L.; Brunskill, G. J. Environ. Sci. Technol. 1991, 25, 280 286. (9) Wania, F.; Hoff, J. T.; Jia, C. Q.; Mackay, D. Environ. Pollut. 1998, 102, 25 41. (10) Franz, T. P.; Eisenreich, S. J. Environ. Sci. Technol. 1998, 32, 1771 1778. (11) Hoff, J. T.; Gregor, D.; Mackay, D.; Wania, F.; Jia, C. Q. Environ. Sci. Technol. 1998, 32, 58 62. (12) Hoff, J. T.; Wania, F.; Mackay, D.; Gillham, R. Environ. Sci. Technol. 1995, 29, 1982 1989. (13) Gubala, C. P.; Driscoll, C. T.; Newton, R. M.; Schofield, C. L. Environ. Sci. Technol. 1992, 25, 2024 2030. (14) Hargrave, B. T.; et al. Environ. Sci. Technol. 2000, 34, 980 987. (15) Skoglund, R. S.; Swackhamer, D. L. Environ. Sci. Technol. 1999, 33, 1516 1519. (16) Meyers, P. A.; Eadie, B. J. Org. Geochem. 1993, 20, 47 56. (17) Johnson, T. C.; Evans, J. E.; Eisenreich, S. J. Limnol. Oceanogr. 1982, 27, 481 491. (18) Kemp, A. L. W.; Johnson, L. M. J. Great Lakes Res. 1979,5, 1 10. (19) Axelman, J.; Bromon, D.; Naf, C. Ambio 2000, 29, 210 216. (20) Ewald, G.; Larsson, P.; Linge, H.; Okla, L.; Szarzi, N. Arctic 1998, 51, 40 47. (21) Cameron, M. E.; Metcalfe, T. L.; Metcalfe, C. D.; Macdonald, C. R. Mar. Environ. Res. 1997, 43, 99 116. (22) White, R. D.; Hahn, M. E.; Lockhart, W. L.; Stegeman, J. J. Toxicol. Appl. Pharmacol. 1994, 126, 45 57. (23) Polischuk, S. C.; Letcher, R. J.; Norstrom, R. J.; Ramsay, M. A. Sci. Total Environ. 1995, 160/161, 465 472. (24) Jorgensen, E. H.; Bye, B. E.; Jobling, M. Aquat. Toxicol. 1999, 44, 233 244. (25) McKone, T. E.; Daniels, J. I.; Goodman, M. Risk Analysis 1996, 16, 377 393. 462 A ENVIRONMENTAL SCIENCE & TECHNOLOGY / DECEMBER 1, 2002