The interplay of environmental toxicology and chemistry in the development of sediment quality criteria


  • See Table S1 for the number of citations and rank of all the “Top 100” papers, which in this essay are references [ Karickhoff et al., Pignatello et al., Cornelissen et al. 8–10,Swartz et al. 15,Di Toro et al. 21,Berry et al. 24,Ankley et al. 27].

The search for rational methods to develop criteria for various media—water, sediment, and tissue—and various receptors—aquatic, terrestrial, and human—has been a consistent and continuous enterprise from the beginning of environmental toxicology. Many of the 100 most frequently cited papers in Environmental Toxicology and Chemistry directly or indirectly address topics relevant to criteria. Subsets of these papers are concerned with the interaction between environmental toxicology and environmental chemistry.

The first aquatic criteria were based on what today would be called best professional judgment applied to the toxicity experiments available at that time. Very little, if any, environmental chemistry was considered. The transition came with the development of the technical guidelines for water quality criteria 1, which provided an explicit methodology for the toxicological component of the criteria: a required minimum data set (tests using eight species from specific groups of organisms), a method of analysis to produce the acute criteria (the 5th percentile from the species sensitivity distribution), and an acute to chronic ratio derived from paired experiments. The criteria developed in the early 1980s, and those that continue to be developed, employ either this framework or closely related frameworks.

The criteria recognized that, in addition to the concentration of the toxicant, there are environmental chemical factors that need to be considered. Examples include water hardness (Ca + Mg concentrations) affecting the toxicity of certain divalent cations and pH affecting ammonia toxicity. Both of these were included in the criteria. An example of a chemical factor that was recognized but not included is dissolved organic matter concentration. Nevertheless, these water quality criteria marked the transition from judgment-based criteria to scientifically based criteria—compare the pre technical guidelines “Blue Book” 2 and “Red Book” 3 to the post technical guidelines “Gold Book” 4.

The development of sediment quality criteria followed a different path. The methods that have been, and are being, developed are either mechanistic criteria or empirical criteria that consider, or do not consider, the chemical factors in sediments that affect toxicity 5. For the most part, the water quality criteria that excluded the effects of other water quality parameters were reasonably successful, particularly for organic chemicals, where the effects due to other chemical constituents in the water were not large. By contrast, the toxicity of chemicals in sediments can vary by factors of 10- to 100-fold depending on sediment characteristics. This problem can be thought of as a variation in the bioavailability of the total quantity of the chemical in sediments—that is, that fraction of the total concentration that can interact with the target that causes the toxicity.

In view of this dramatic variability, it was clear that understanding the reason(s) for the variability in bioavailability was necessary to develop successful mechanistic sediment quality criteria. The key experiment was reported by Adams et al.6, who tested the toxicity of kepone in four quite different sediments. Their insight was to relate the toxicity to the pore water concentration rather than to the sediment concentration. Equally important was their noting that the pore water median lethal and effective concentration (LC/EC50) and the LC/EC50 derived from exposures to water only were essentially equal. They rationalized this observation by hypothesizing that the route of exposure was exclusively via the pore water.

This rationale was consistent with their findings, but it suggested that the use of pore water concentration to understand the toxicity variation across sediments would apply only to organisms that were primarily exposed via the pore water. Another rationale, which implied a much wider applicability of their important findings, was suggested as part of the equilibrium partitioning (EqP) model proposed by Di Toro et al. 7. It was based on the idea that the pore water and sediment solids were in partitioning equilibrium and that the toxicity of the chemical was determined by the chemical activity or fugacity of the equilibrium mixture (hence the name equilibrium partitioning). For kepone, the mortality and growth dose-response curves for the different sediments overlapped to within a factor of approximately 2 using either pore water concentration (CPW) or organic carbon normalized sediment concentration (CSed/fOC) as the dose metric, where CSed is the sediment concentration per unit dry weight and fOC is the weight fraction of organic carbon 7. This would be expected if the partitioning equilibrium between pore water and sediment follows the partitioning model

equation image

where KOC is the organic carbon/water partition coefficient.

The EqP model reduced the problem of predicting the bioavailability of chemicals in sediments to the problem of predicting the partitioning of chemicals between the particulate phases that made up the sediment and the pore water. It also suggested that the EqP model was applicable to sediment-ingesting and water-ingesting animals because the chemical activity was the same in each phase.

A number of issues needed to be investigated to validate and clarify the applicability of the EqP model. Whether sediments and pore water are in equilibrium was being investigated prior to the development of the EqP model. It was found by Karickhoff and Morris 8 that the time to reach equilibrium can be quite long for hydrophobic chemicals. Also, as found by Pignatello 9 and Cornelissen et al. 10, a fraction of the chemical sorbed to particles becomes fixed in the particle and resists desorption. This problem would not be apparent in laboratory spiked sediments but would be evident in field-collected sediments, where contact time between contaminants and sediments is much longer. Finally, the partitioning model may not be entirely appropriate for the particular sediment in question. The sorption to black carbon has been found to be important 11. These mechanisms make the original EqP criteria for organic chemicals conservative because they increase the extent of partitioning. Such problems do not invalidate the chemical activity basis for EqP criteria but signal a need for direct measurement of chemical activity in sediments 12. The EqP model is the basis for the U.S. Environmental Protection Agency (U.S. EPA) sediment benchmark for certain pesticides 13, 14, for a model that deals with mixtures of polycyclic aromatic hydrocarbons (PAHs) by Swartz et al. 15, and for the U.S. EPA benchmark for mixtures of PAHs 16, 17.

The application of the EqP model to potentially toxic metals required that their partitioning in sediments be understood. Metal speciation in sediments involves a number of distinct phases, as demonstrated by sequential extractions 18. In terms of EqP, the problem was predicting the chemical activity of the toxic form of the metal in pore water. The free ion activity model 19, based on the work of Sunda 20, suggested that the concentration of the divalent cation (e.g., Cd2+) was the species of interest.

It was known that very insoluble solid phase metal sulfides (e.g., CdS) were present in sediments. What had not been realized is that most sediments, including freshwater sediments, contain a pool of reactive iron monosulfides (FeS), quantified by the acid volatile sulfide (AVS) extraction, that could react rapidly to precipitate the metals (Cd, Cu, Ni, Pb, Zn, Ag) that form insoluble metal sulfides, thereby reducing the concentration of the cation to very low concentrations, well below toxic levels. This observation was derived from cadmium toxicity and subsequent chemical experiments by Di Toro et al. 21. The accompanying idea—that the proper extraction to determine if excess AVS was present relative to metal in a sediment—was to measure the metal concentration in the same extraction used to measure AVS 22. The simultaneously extracted metal (SEM) was compared to the AVS to determine if metals in sediments were bioavailable. An excess of AVS served as a sink for the metal, precipitating it as an insoluble metal sulfide and rendering it unavailable to the biota. The necessary analytical methods were evaluated by Allen et al. 23, and extensive testing was conducted by Berry et al. 24 of laboratory-spiked sediments and by Hansen et al. 25 of field-collected sediments.

The original partitioning model applied only to the situation where AVS exceeded SEM. When the situation was reversed, however (e.g., in oxic sediments), the SEM–AVS model made no prediction. It had been found subsequently that normalizing SEM–AVS by the organic carbon fraction in the sediment extended the limits of no observed toxicity to the region of positive SEM–AVS. A sediment biotic ligand model which considered both the AVS and organic carbon partitioning supported this empirical approach 26. The EqP and SEM–AVS models were the basis for a proposed sediment quality criteria for metals by Ankley et al. 27 and for the subsequent U.S. EPA sediment benchmark for metals and metal mixtures 28.

The two approaches—the mechanistic criteria derived from EqP and the empirical approaches typified by the Long and Morgan effects range–low (ER-L) and effects range–medium (ER-M) 29 and subsequent threshold effect level (TEL), probable effect level (PEL), and consensus criteria 30—have different objectives. The EqP-based criteria aim to establish a direct relationship between a chemical or mixture of chemicals and sediment toxicity. The empirical criteria aim to predict the probability that a sediment is toxic based on the total chemical concentrations in that sediment. It is a remarkable and disturbing observation that for most of the toxic sediments in large existing field-collected data sets, neither method can identify the chemical(s) causing the toxicity, nor can they predict that the sediment is toxic 31. Thus, there is much remaining work to be done in understanding the causes and extent of sediment toxicity. In particular, within the EqP framework, partitioning models are necessary for many classes of compounds, such as oxyanions and charged organic chemicals.

The EqP models are examples of the interplay of environmental chemistry and environmental toxicology. The need to consider both in the risk analysis and the solution of environmental problems is more apparent now than in previous decades. A good example is the use of additives to sediments to decrease the toxicity of contaminated sediments 32. The success of these methods depends on their ability to decrease the chemical activity of the toxicants by increasing the sorption to the added particles. The total concentration of toxicant does not change, only the chemical speciation is changed, with the resulting change in toxicity. As the problems become more complex from both the toxicological and chemical sides, it is critical that both disciplines are employed in their solution. Our Society of Environmental Toxicology and Chemistry (SETAC) is appropriately named.


Table S1.