Appendices

Appendix A: Protocol and Rationale for setting Provincial Sediment Quality Guidelines

Protocol for setting Sediment Quality Guidelines

Rationale for setting Sediment Quality Guidelines

In developing guidelines to provide adequate protection for biological resources, the Ministry has attempted to ensure that the methods employed consider the full range of natural processes governing the fate and distribution of contaminants in the natural environment. Since benthic organisms respond to a variety of stress-inducing factors they are, in essence, integrators of all the physical, chemical and biological phenomena being experienced in their environment and these organisms should form the basis of any method used in setting sediment guidelines.

Because individual species may respond differently to stress-inducing factors it is very difficult to study a specific organism (e.g., a sensitive species) with the hope of developing guidelines that will protect the rest of the community. Sensitivity to chemical contaminants has not been fully evaluated for different benthic organisms and most sediment bioassay work has been concerned mainly with a few selected species (e.g., the mayfly Hexagenia). While the mayfly has traditionally been used as a "sensitive" indicator organism for factors such as low dissolved oxygen, its sensitivity relative to other benthic organisms has not been clearly established for chemical contaminants. Therefore, in developing PSQGs, the Ministry has not relied on single-species data.

Similarly, a method that relies heavily on those species that are known to be extremely tolerant of contaminants in sediment cannot result in guidelines that will adequately protect less tolerant members of the aquatic community. It has been demonstrated that some populations can adapt to varying levels of environmental contamination with increasing tolerance to these contaminants occurring in succeeding generations. This can present difficulty in laboratory studies of reared populations since these may lack the genetic diversity found in natural populations and responses may not be consistent with those observable under field conditions. There is also concern over placing heavy reliance on laboratory data as in most situations contaminants in sediments exist as mixtures of various substances. Laboratory tests generally tend to examine the effects of single substances, and laboratory data can be difficult to apply to field situations.

In developing the protocol for setting sediment quality guidelines, the Ministry considered a number of different approaches developed by state and federal agencies in North America that employed various degrees of biological assessment. The various suggestions for the development of sediment quality guidelines can be summarized in five approaches as possible means of setting sediment quality guidelines. At present, no single approach can adequately account for all the factors that operate in natural sediments and each of the five approaches has positive attributes as well as limitations with regard to the development of biologically based guidelines. The rationale used in setting sediment quality guidelines includes a number of considerations which are detailed below. These considerations provided the basis for selecting the best method or combination of methods for the development of Provincial Sediment Quality Guidelines (PSQGs).

1. Sediment quality guidelines should consider a range of contaminant concentrations that is wide enough to determine the level at which ecotoxicological effects become noticeable. This can be achieved most effectively by looking at a large number of organisms under the widest possible range of contaminant exposure. Only then can the appropriate ecotoxicological level be adequately determined. A restricted range may result in the setting of guidelines that are not reflective of actual ecotoxicological effects on organisms and as such may be overprotective. This is especially important where the range of effects used may not cover the entire tolerance range of the species in question.
2. PSQGs should be based on cause-effect relationships between a specific contaminant and benthic organisms since it is necessary to demonstrate that at a certain concentration a contaminant results in adverse effects on benthic organisms.
3. PSQGs should account for contaminant effects in a multi-contaminant medium. Since contaminated sediments usually consist of mixtures of substances, the presence of a number of different contaminants, any or all of which may affect the response of the organisms to the contaminant being investigated must be considered. Since exposure to combinations of contaminants may result in different responses than exposure to a single contaminant (through either synergistic or antagonistic effects), these effects must also be considered. A PSQG method must incorporate this feature into the derivation of a number for specific contaminants.
4. PSQGs should consider chronic effects of contaminants on aquatic biota since these can affect the long term viability of aquatic organism populations. Methods that consider only acute effects do not offer adequate protection, since sediment concentrations reflect long-term conditions and are not subject to the extreme temporal variability of water column contaminant concentrations.
5. The PSQGs should be capable of incorporating and accounting for the range of environmental factors that could have a bearing on the presence or absence of organisms in a given area. Contaminant behaviour and the organisms’ well-being are governed by a variety of natural physical, chemical and biological processes. If these processes are not accounted for in a PSQG method then the resulting guidelines will be unrealistic. For example, organisms may be absent from a given area not because of the level of contaminants but because of unsuitable habitat, low dissolved oxygen, or interspecific competition. In formulating a guideline it is essential that these factors be considered along with the chemical data. If they are not considered, the numerical value obtained would not necessarily be protective of aquatic species. This will also reduce the need for site-specific guidelines, since a full range of environmental conditions will have been covered.

Approaches to developing Sediment Quality Guidelines

As part of the sediment guideline development process, the Ministry has carried out an extensive literature review of possible approaches to the development of sediment guidelines. This effort has resulted in the selection of five potential approaches for this purpose. These are:

1. Sediment Background Approach
2. Equilibrium Partitioning Approach (Water-Sediment and Biota-Water-Sediment Partitioning)
3. Apparent Effects Threshold Approach
4. Screening Level Concentration Approach
5. Spiked Bioassay Approach

The five approaches are discussed below and additional details can be found in the pertinent literature cited for each method.

Sediment Background Approach

In the Background Approach, sediment contaminant concentrations are compared to concentrations from reference background sites where contaminant levels are deemed to be acceptable (OMOE 1987, 1988). Using the Background Approach, levels are set according to a "suitable" reference site or "acceptable" level of contamination. A suitable reference site may be one where sediments are considered to be relatively unaffected by anthropogenic inputs. Alternatively a suitable reference site may be derived through sediment profiles. In the latter, the pre-industrial sediment horizon, as determined through techniques such as palaeontology, could be used to determine background levels. The basis of the Background Approach is the implicit assumption that concentrations above these background values have an adverse effect on aquatic organisms.

For the purposes of PSQG development a "pre-industrial" standard could be adopted only for metals. The strictly anthropogenic (man-made) organic contaminants, for which background levels should theoretically be zero, would require adoption of a contemporary surficial sediment standard, based on a suitable reference site.

Advantages

The data requirements of the Background Approach are minimal in that the method requires only the measurement of the chemical concentrations of contaminants in sediments. As such it can be used with the existing data, thus minimizing the need for additional data collection. The method does not require quantitative toxicological data and avoids the need to seek mechanistic chemical explanations for contaminant behaviour or biological effects.

Background limits have advantages from an enforcement perspective since the Background Approach does provide an indication of the chemical concentration for metals that is expected to occur naturally. While it is possible that biological effects may occur in some species at metal concentrations indistinguishable from non-anthropogenic background, it is difficult to justify enforcement of a standard that has never been realized in nature. Thus background levels for metals can provide a practical lower limit for management decisions. For organic contaminants, which are largely anthropogenic, background should theoretically be zero. In most areas, however, contaminants have found their way into sediment and a contemporary benchmark based on current average concentrations for a suitable reference area may provide the practical lower limit for enforcement.

Limitations

Since the Background Approach relies only on the chemical concentration of contaminants in sediments it has no biological basis. Because biological effects data are not considered, cause-effect relationships between sediment contaminant levels and sediment-dwelling organisms cannot be determined. The exclusive use of chemical data implies that sediment characteristics have no influence on the resultant biological effects, but rather that chemical concentrations alone are responsible for the observed effects. However, sediment characteristics (i.e., grain size, organic content, dissolved oxygen levels) have been shown to be major factors affecting benthic community composition (Gooday et al 1990; Death 1995; Remple et al 2000).

Implicit in the method is the assumption that the chemicals present are in their biologically available forms. The method therefore, makes no allowance for the occurrence of different chemical species with differing biological availability and toxicity.

A further limitation of this approach is that background levels tend to be highly site- specific. They therefore require the designation of a reference site, which itself is likely to be highly subjective.

Equilibrium Partitioning Approaches

Phase partitioning of organic compounds has been used to describe the distribution of certain organic compounds in aquatic compartments. Partitioning, like adsorption, is one of the processes by which organic compounds can be sorbed to sediments. A major difference however, is that partitioning is solubility dependent and therefore, reversible (i.e. equilibrium) partitioning of non-polar organic compounds is a function of their solubility in water. The very insoluble compounds, as a result, partition strongly to sediment with only very minor amounts in water. These compounds tend to have high partition coefficients, as measured by the octanol-water partition coefficient, K_ow. The K_ow is the ratio of the amount of the compound that is soluble in an organic solvent, such as octanol, relative to the amount soluble in water.

The partitioning approaches have been extensively investigated by the U.S. EPA (Pavlou & Weston 1984, Di Toro et al, 1991). A basic assumption of this approach is that the distribution of contaminants among different compartments in sediment is controlled in a predictable manner by a continuous equilibrium exchange among sediment solids and the interstitial water. Partitioning to these two phases can therefore be calculated by the quantity of sorbent in the sediment (for which organic carbon is the primary sorbent) and the partition coefficient K_oc. K_oc values, which can be estimated from K_ow, are normalized to sediment organic content.

The Equilibrium Partitioning approaches also assume that interstitial water is the primary route of organism exposure to contaminants in sediments. Therefore, this approach assumes that only the amount of contaminant partitioning to the water is of interest, the amounts partitioning to the sediments being considered as unavailable.

Using this approach, contaminant-specific partition coefficients are determined (generally expressed in terms of organic carbon content of sediment) and used to predict the distribution of the contaminant between sediment and interstitial water. This approach, however, can only be used for contaminants that partition between environmental phases. Contaminants that do not partition appreciably into sediment organic matter, and those whose chemical behaviour is highly unpredictable (such as metals), cannot be considered using this approach.

Under the Equilibrium Partitioning approach, a generic (i.e. equally applicable to all sites) organic carbon-normalized partition coefficient K_oc is developed and is then multiplied by an existing water quality objectives/guidelines to derive a sediment guideline. In essence, the distribution coefficients for the non-polar organics are used to establish the chemical concentration in the sediments that, at equilibrium, will not exceed water quality objectives/guidelines in the interstitial water. Sediment Quality Guidelines based on the equilibrium partitioning of organics can be calculated in a number of ways, depending on the type of data available:

1. Water-Sediment Equilibrium Partitioning Approach:

   The Water-Sediment Partitioning Approach is a generic partitioning method which derives a sediment quality guideline from the partitioning of a chemical to the water and the sediment solid phases. There is sufficient evidence to show that sediment organic carbon is the primary environmental factor influencing partitioning (Di Toro et al. 1985 in OMOE 1988). The partition coefficient, K_sed, is normalized for organic content and an organic carbon-normalized sediment-water partition coefficient is derived (K_oc). This can either be derived empirically, or calculated from the octanol-water partition coefficient. The partition coefficient is then multiplied by a water quality criterion (such as a water quality objective) to derive a sediment quality guideline.

2. Biota-Water-Sediment Equilibrium Partitioning Approach

   The Biota-Water-Sediment Partitioning Approach is a generic partitioning method which derives a sediment guideline from an existing tissue residue criterion. It is a two step approach utilizing a generic water-biota bioconcentration factor (BCF) to relate the tissue criterion to a corresponding water concentration. For bioaccumulative substances this relationship determines the tissue-water concentration level (TWCL). The TWCL is the value that must not be exceeded in water in order to prevent exceedance of the tissue residue criteria from which the TWCL was derived. The TWCL, therefore, is equivalent to a water-quality criterion. Following this step the approach is similar to that described for the water-sediment approach with the TWCL used in place of the water quality criterion.

Advantages

Generic Partitioning Approaches are biologically based to the extent that existing water or tissue criteria are biologically based and, therefore, provide more defensible guidelines than the Background Approach. Since they make use of the virtual no-effect levels determined from existing Provincial Water Quality Objectives and Guidelines (PWQO/Gs) the sediment guidelines derived through generic partitioning approaches can be considered no-effect levels for the protection of those end-uses the water quality guidelines were designed to achieve.

The partitioning approach relies on an existing toxicological rationale which has been established during the development of the water quality criterion being used. Thus, a new toxicological evaluation is not required provided that the water quality criterion has been derived to protect those benthic organisms which are exposed to the interstitial water. However, a corresponding limitation to the approach is its applicability only to chemicals which have water quality criteria. Moreover, if the water and sediment criteria are meant to protect different organisms then an assumption is made that the two sets of organisms are of equal sensitivity to given levels of contaminants.

Limitations

The basic assumption that availability of an organic compound to aquatic organisms is controlled by the amounts partitioning to the water ignores both the sediments and food chain effects as potential sources. It has not yet been proven that the interstitial water is the only significant route of exposure and for the highly hydrophobic compounds (those with high K_ow); all of these sources may be significant routes of exposure.

Tissue residue criteria are generally based on human health considerations and human food consumption patterns (MacDonald 1994). Therefore, the tissue residue criteria apply to human food organisms such as fish, rather than benthic organisms. Similarly, the BCF applies to fish, and the water concentration (TWCL) thus derived applies to the water column in which the fish lives. This approach is limited by the substantial gap that exists between the water column compartment and the interstitial water compartment that is assumed to be in equilibrium with the sediments. The reduction in contaminant concentration from the interstitial water to the water column compartment is likely to be highly site-specific depending on local-circulation.

Current use of the Partitioning Approach is limited to those contaminants that exhibit predictable partitioning behaviour. Since the partitioning of metals in sediments is highly unpredictable (e.g., sediment-water partition coefficients for metals can span a wide range of values differing by orders of magnitude depending on such factors as redox potential, pH, dissolved oxygen and organic matter content of the sediment) and polar organics generally do not partition into sediment, the partitioning approaches are considered applicable only to non-polar organic compounds.

The scientific validity of a sediment guideline obtained through the partitioning approaches relies heavily on the accuracy of the partitioning coefficients (K_oc) used. The published values for partition coefficients obtained by different authors can differ by an order of magnitude. This presents great difficulty in choosing a representative value for use in guideline development work and unless a standard approach is used it will be difficult to obtain consistent or compatible guidelines using the EP approach.

At present the Equilibrium Partitioning Approach cannot account for all the forms a contaminant can exist in and all the possible sediment constituents it can partition to. This is currently a drawback to the Equilibrium Partitioning Approach since the various forms of a contaminant have their own toxicity and partitioning characteristics. Several species of a contaminant may be bioavailable and toxic, but often their concentrations are more or less linearly dependent on the concentration of a single species. While it has been possible to establish that one species correlates with the observed toxic effects for the non-polar organics, this has not been possible for the metals or the polar organics. The partitioning approach does not work for metals or polar organics due to the multiplicity of adsorption mechanisms these undergo. It is not even clear which sediment components are controlling partitioning.

Apparent Effects Threshold (AET) Approach

The Apparent Effects Threshold (AET) Approach, as developed by Tetra Tech (1986) is a statistically based approach that attempts to establish quantitative relationships between individual sediment contaminants and observed biological effects. The biological effects can be both field measured effects such as changes in benthic community structure and laboratory measured effects through the use of sediment bioassays. The basis of this technique is to find the sediment concentration of a contaminant above which significant biological effects are always observed. These effects can be any or all of a number of different types, such as chronic or acute toxicity, changes in community composition, and bioaccumulation and are considered in conjunction with the measured sediment contaminant levels. Inherent in the approach is the assumption that observed effects above this level of contamination are specifically related to the contaminant of interest, while below this level any effects observed could be due to other contaminants.

Advantages

The AET approach is effects based and therefore more defensible than the partitioning approaches in relation to the protection of benthic organisms. The method assumes a direct cause-effect relationship between sediment concentrations of a contaminant and the occurrence of significant biological effects.

Unlike the partitioning approach the AET approach makes no assumptions regarding contaminant availability from the various environmental compartments. Therefore the effects on biota can be due to contaminants available through both adsorption from sediments and interstitial water and through absorption from ingested matter.

Limitations

The method is unable to separate the biological effects that may be due to a combination of contaminants. While assuming a cause-effect relationship, the method cannot clearly demonstrate a cause-effect relationship for any single contaminant. Thus, while definite ecotoxicological effects can be established, these cannot be attributed to any one chemical contaminant.

In using the AET approach care must be exercised in selecting the species of organism to be used and the particular type of effects (endpoints) to be considered. If the data used consist of mixed species and endpoints, the least sensitive of these will always predominate and the guidelines derived may not protect other more sensitive species. For example, if the data base for a particular contaminant contains data on acute toxicity to tubificid oligochaetes, then the AET will be designed to protect against acute toxicity to tubificids. It will not protect species that are more sensitive nor will it provide protection against chronic effects.

For most practical purposes this method requires chronic toxicity data since results from the existing database indicate guidelines tend to be higher than those calculated by other means, in some cases by an order of magnitude. This is usually due to the use of acute toxicity data which needs a correction factor to adjust to chronic toxicity. The development of a chronic toxicity database (i.e., one based on reproductive effects and effects on the most sensitive life stages) itself requires a very extensive set of information which at present does not exist in a standardized form. In order to obtain such information, considerable laboratory testing will have to be carried out. In addition, for data from different investigators to be useful, consistency in procedures and definition of endpoints will be necessary.

In practice, guidelines generated by the AET approach are likely to be underprotective since this method determines the contaminant level above which biological effects are always expected. Biological effects, however, can be and are observed at chemical concentrations lower than these values, though these effects may not occur in all samples.

The AET approach is applicable for all types of contaminants, making use of both laboratory tests on sediments (spiked sediments) and field data. In laboratory tests of field-collected sediments it may not be possible to separate the effects of mixtures of chemicals. If spiked sediments are used, only single contaminant or known (specific) mixtures can be used and therefore this method suffers from some of the same limitations as the Spiked Bioassay method (discussed below). In using field collected sediments in conjunction with other field data (e.g. community composition), it is not possible to separate the effects of mixtures of contaminants and this method suffers from the limitations affecting the Screening Level Concentration Approach.

The Screening Level Concentration (SLC) Approach

The Screening Level Concentration (SLC) Approach, like the AET Approach, is an effects based approach applicable mainly to benthic organisms. The SLC approach uses field data on the co-occurrence in sediments of benthic infaunal species and different concentrations of contaminants. The SLC is an estimate of the highest concentration of a contaminant that can be tolerated by a specific proportion of benthic species. In its original derivation and application, the 95^th percentile was used (Neff et al 1986).

The SLC, as developed by Neff et al (1986), is calculated through a two step process. First, for a large number of species (at least ten for each chemical) a species SLC (SSLC) is calculated by plotting the frequency distribution of the contaminant concentrations over all sites (at least ten) where the species is present. The 90^th percentile of this distribution is then taken as the SSLC for that species (Figure 1a). The 90^th percentile was chosen to provide a more conservative estimate of the SSLC. Extreme sediment concentrations may be an aspect of specific sediment characteristics resulting in low biological availability relative to the sediment concentration. By choosing the 90^th percentile, these values are excluded. In the second step, the SSLCs for each species are plotted as a frequency distribution and the 5^th percentile is interpolated from this distribution (Figure A1b). This is the SLC and represents the concentration which 95% of the species can tolerate.

Figure A1. Screening Level Concentration Calculation

Enlarge this image

Figure 1a shows the calculation of the Species Screening Level Concentration (SSLC) for contaminant 'X' for species 'A'. Sites where species 'A' is present is plotted on the x-axis (no units) and the concentration for contaminant 'X' (in µg/g organic carbon) is plotted on the y-axis. The sites where species 'A' is found is plotted from lowest to highest as a function of the concentration. The resulting frequency distribution plot shows the contaminant concentration measured at all sites where the species is present and is represented by a series of points that increase from low to high concentrations. The figure shows a dotted line from the y-axis that represents the 90^th percentile concentration of the data. This value is identified as the Species Screening Level Concentration for Species 'A'.

Figure 1b shows the calculation of the Screening Level Concentration (SLC) for all species for contaminant 'X'. The Species Screening Level Concentration (SSLC) for all species is plotted on the x-axis (no units) and the Species Screening Level Concentration for contaminant 'X' (in µg/g organic carbon) is plotted on the y-axis. Different species are referred to on the x-axis (e.g., species 'D', 'Q', 'P', 'I', etc., including species 'X' from Figure 1a). The different species are plotted from lowest to highest as a function of their calculated Species Screening Level Concentration (i.e., the 90^th percentile calculated for each species from the equivalent Figure 1a for each species). An arrow from Figure 1a shows how the Species Screening Level Concentration for species 'A relates to the data point for the Screening Level Concentration for species 'A' and contaminant 'X' in Figure 1b.

A basic assumption in the method is that the data cover the full tolerance range of each species. This assumption requires that a large range of chemical concentrations be sampled in each case (at least two orders of magnitude) since an SLC will be generated whether or not this assumption is true. This is important though sometimes difficult to verify. The difficulty lies in the fact that the full tolerance range of most species is not known.

Sediment contaminant concentrations for the non-polar organics are normalized to TOC content of the sediments. Since these compounds generally partition strongly to organic matter, the normalized concentration should more closely represent contaminant availability to benthic organisms. For metals and polar organics, bulk sediment concentrations are used since the best normalization procedures for representation of metal availability are as yet unresolved.

Advantages

Since the SLC Approach does not make any assumptions about the absence of a species and considers only those species present, the SLC approach does not require a priori assumptions concerning cause-effect relationships between sediment contaminant concentrations and the presence or absence of benthic species. As no relationship is assumed it is not necessary to take into account the wide variety of environmental factors that affect benthic communities, such as substrate type, temperature and depth.

However, valid inferences can be drawn from this type of analysis regarding the range of sediment contaminant concentrations that can be tolerated by the sediment infauna since field data on the co-occurrence of benthic infaunal species and sediment contaminant concentrations are used.

Since the SLC Approach uses field data on the co-occurrence in the field of contaminants and benthic species, the environmental factors acting on the species distribution are already integrated into the data-set and the response determined is a measure of both the environmental factors and the contaminant levels. It also integrates changes in chronic responses such as reproduction/fecundity and sensitive life-stages, since it is a cumulative measure of effects. In addition, it integrates into the biological response any synergistic or additive effects from multiple contaminants as they would occur in natural sediments. Because of this, the SLC Approach overcomes the difficulties of applying bioassay data to field situations, and the lack of uncertainty associated with partition coefficients.

While it was originally developed primarily for use with non-polar organics (using TOC normalization) it is also appropriate for metals and polar organics as well since it can be used with or without TOC normalization.

At present the size of the database has determined that the SLC level be set at the 5^th percentile of the SLC frequency distribution. However, as the database continues to expand it should be possible to reliably calculate the 1^st percentile (i.e. the level of a contaminant that 99% of the species present can tolerate). The precision of the SLC is directly related to the size of the database and the range of variability of the various factors within the database. Therefore great care must be taken to include data taken over the full range of conditions since a database skewed to either lightly or heavily contaminated areas will yield guidelines that are either too conservative (overprotective) or do not provide adequate protection for aquatic life (under protective).

Limitations

The major limitation of the SLC Approach is the difficulty in determining a direct cause-effect relationship between any one contaminant and the benthic biota, since very rarely is a single contaminant present in natural situations. Therefore, the effects observed are related to the entire mixture of chemicals.

The range and distribution of contaminant concentrations and the particular species used to generate them can significantly affect the calculation of the SLC value. The use of only low values of contaminant concentration may not encompass the entire tolerance range of the species and the concentration would be below the level that would adversely affect the distribution of that species. In such situations, an SLC would still be generated but the value would be conservative and unrealistic. This can be overcome by ensuring that the database includes values from heavily contaminated areas.

The SLC is also sensitive to the species used in the database. Unlike the Partitioning Approach, the SLC Approach does not make any assumptions regarding the possible routes of effect from aquatic contaminants, all possible modes of exposure are taken into account. Since contaminant availability from the sediments may differ in relation to the feeding habits of the organisms used, the proportion of species from each of the feeding groups will determine the shape of the SLC curve. This can also be overcome by limiting the database to those organisms living in or feeding on the sediment.

Spiked Bioassay Approach

In this approach, dose-response relationships are determined by exposing test organisms, under controlled laboratory conditions, to sediments that have been spiked with known amounts of contaminants (OMOE 1987, 1988). Sediment quality guideline values can then be determined using the sediment bioassay data in a manner similar to that in which aqueous bioassays are used to establish water quality criteria. Where chronic toxicity data are not available, an approximation can be obtained by using acute toxicity endpoints that have been adjusted downwards by a factor of ten to obtain a chronic protection level and then applying a suitable safety factor.

Advantages

The major advantage of this approach is that a direct cause-effect relationship can be determined, at least under laboratory conditions, for a specific chemical or combination of chemicals for any species of organism.

Limitations

Despite this advantage, limitations exist that, at present, preclude the use of this method for setting guidelines. Techniques have not been standardized for spiking sediments and differences in methods/techniques can strongly influence the results. In addition, laboratory bioassays performed under controlled conditions may not be directly applicable to field situations where conditions may vary considerably from those encountered in the laboratory. In order to derive realistic guidelines from the Bioassay Approach efforts will have to be made to test different sediments with various chemical mixtures in differing proportions and using different organisms, as would exist in field situations.

Summary Evaluations of the Various Approaches to PSQG Development

The major objectives in the development of sediment quality guidelines are to provide protection to aquatic organisms and ensure water quality protection, as well as guidance in decision-making related to abatement efforts and remedial action. As such, they are intended to be both proactive and reactive in application. The primary basis for such decisions is the protection of biological resources against the lethal and sublethal effects of contaminated sediment.

The biological resources that could potentially be impacted by contaminants in sediment span a wide range. These include organisms that could be impacted directly, namely the benthic species that live in or feed on the sediment, and water column organisms that could sorb contaminants released from the sediment to water and/or through the consumption of benthic organisms; and those impacted indirectly such as non-aquatic consumers (humans and wildlife) of top aquatic predators such as fish.

In reviewing the five approaches to setting sediment guidelines, it is apparent that each approach has certain merits as well as limitations. The Background Approach while lacking a biological basis, does provide a good indication of the levels at which metals are expected to occur naturally and thus provides a realistic lower limit for guideline development.

The Partitioning Approaches to sediment guideline development use existing criteria such as a water quality or tissue residue criteria which can be considered as virtual no-effect values. The resulting sediment guidelines can therefore also be considered as virtual no-effect values for the protection of water column organisms from sediment-bound contaminants. The Partitioning Approach is attractive because it is capable of providing a measure of contaminant availability from sediments with a minimum of data. Due to the incorporation of various safety factors in the generation of PWQOs, this approach is able to provide an estimate of the no-effect level of a contaminant in sediments. How protective this value may be depends on the sediment organisms, the size of the safety factor, and the type of sediment. The approach is limited by its assumption of a single route of exposure for aquatic organisms and its restriction to the non-polar organics.

The AET Approach appears best suited to discriminating between contaminated and uncontaminated areas within a site, since the data used tend to be highly site specific. As a result, any guidelines derived will also be site-specific. The major limitation lies in the assumption of a cause-effect relationship that the methods prove unable to demonstrate. There is also a lack of chronic effects data suitable for AET applications, particularly if consistency in level of protection (i.e. single species and endpoint) is desired. Therefore, the AET Approach is judged less acceptable than the other effects-based approaches.

The SLC Approach has an advantage in that no cause-effect relationships are assumed and therefore, it does not need to account for all of the natural environmental factors that can affect organisms. The effects of these are already integrated into the data. The effects of multi-contaminant interactions are also factored into the data set used in the calculations and, with a sufficiently large database, the effects of other contaminants can be minimized. The SLC Approach would be less defensible on a theoretical basis than the Spiked Bioassay Approach if the data bases for the two approaches were comparable. It has been found, however, that relevant information from bioassays is considerably lacking, especially in relation to the impacts of chemical mixtures on benthic populations. Due to the scarcity of Spiked Bioassay data, it is difficult to achieve consistency in the level of protection (i.e. a variety of species and endpoints must be considered). The problem could be rectified with further chronic data acquisition, particularly if standard spiking techniques were adopted. In practice, the methodology has not been standardized and variations in experimental protocol can greatly influence the results. The ability to transpose laboratory derived results to natural situations is also questionable.

Since there is presently a significant lack of adequate data for use in the development of sediment quality guidelines using the Spiked Bioassay Approach, the SLC Approach offers the best means of developing sediment quality guidelines for the protection of the benthic community. This is especially true since there a good database for the Great Lakes Region already exists.

In accordance with the merits and limitations of the various approaches to sediment guideline development, their use can be summarized as follows:

• Partitioning Approaches have been used to develop virtual no-effect levels for the protection of water quality and uses, and health risks associated with humans and wildlife through the consumption of fish. These can be used to set sediment contaminant levels that are also protective of these same uses.
• The Effects-Based Approaches (AET, SLC and Bioassay) are being used to develop guidelines for the protection of benthic organisms. Based on the existing information base, only the SLC approach is of immediate use in the development of sediment quality guidelines.
• The Background Approach has been used to establish levels where adequate data do not exist for application of any of the other methods or where the methods used are inappropriate for the type of compound. In addition, background levels provide a practical lower limit for management decisions.

As sediment bioassay techniques are refined and standardized it may be necessary to revise the protocol to accommodate these techniques as well, though it is unlikely that these will ever supplant field based approaches such as the SLC, since some field verification of laboratory results will always be necessary.

Data Requirements

A PWQO is required for setting levels according to the Partitioning Approach. In order to maintain consistency between sediment and water quality guidelines, levels set by other agencies will not be used.

At least three estimates of partitioning coefficients would be required to set a guideline using the partitioning approach. Guidelines based on fewer than the minimum number of estimates would be regarded as tentative.

The range of contaminant concentrations for the SLC calculations should span at least two orders of magnitude and include data from both heavily contaminated areas and relatively clean areas. Data from clean areas are needed to ensure that sensitive species are included in the SLC calculation, while heavily contaminated areas are needed to ensure that the full tolerance range of all the species is covered.

The database for the SLC calculations should be based on primarily benthic infaunal species and should minimize the reliance on epibenthic species. A minimum of 75% benthic infaunal species would be required to ensure that the observed effects are from sediment associated contaminants and not from water column effects.

Consistency in the species data used has to be ensured. This requires checking the data for synonymies, unusual species distributions, and level of identification. The minimum acceptable taxonomic level would be the genus, provided that species level identifications were also included in the data set from which the information was derived. Data using only generic level identifications could not be used.

The SLC database must include a large range of areas sampled in order to minimize the effects of unmeasured but co-varying contaminants. Since these are unlikely to occur in the same relation at all other areas, the effects of other contaminants can be reduced or excluded if a sufficiently large number of different areas are included.

A minimum of 10 observations are required to calculate an SSLC. A minimum of 20 SSLCs are required to calculate an SLC. This low number has been chosen so as not to exclude the less common or more sensitive species that may not be present at more highly contaminated sites.

Calculation of the Provincial Sediment Quality Guidelines

No Effect Level

Since this is intended as the level at which contaminants in sediments do not present a threat to water quality and uses, benthic biota, wildlife or human health, the parameter values used in deriving the No Effect Levels (NEL) must be the most stringent criteria.

The NEL is principally designed to protect against biomagnification through the food chain. Since these effects are most often observed with the nonpolar organics, this guideline level is not applicable to most of the trace metals. The partitioning approaches are used to set these guidelines since, with appropriate safety factors the PWQOs are designed to protect against biomagnification of contaminants through the food chain, as well as all water quality uses and organisms.

At present, reliable partition coefficients can only be derived for the nonpolar organics, since only these compounds undergo predictable partitioning behaviour in sediments. NEL Guidelines cannot be calculated for metals and polar organics.

Non-Polar Organics

The NEL for non-polar organics is obtained through a chemical equilibrium partitioning approach using PWQOs.

The calculations for each criterion are as follows:

The PWQO value is multiplied by an organic carbon-normalized sediment-water partition coefficient, K_oc. Normalization was recommended by Pavlou and Weston (1984) and OMOE (1988), since sediment organic carbon has been found to be the primary environmental factor influencing partitioning.

A PSQG is then derived through the equation:

SQG = K_oc × PWQO⁄G

where PSQG is the sediment quality guideline normalized to the sediment organic carbon content (TOC). This is to a bulk sediment basis by assuming a 1% TOC concentration. A 1% level for sediment organic carbon is used for converting to a bulk sediment basis, since calculations using the SLC Approach have shown that this is the lowest effect level of organic carbon in the sediment. A bulk sediment calculation based on the actual organic carbon content of the sediment has been avoided for this reason.

The organic carbon-normalized partition coefficient is calculated from either an experimentally derived sediment-water partition coefficient:

K_sed = ([X]_sed ⁄ O.C.) ⁄ [X]_iw

where

[X]_sed

is the concentration of compound X in the sediment (as mass of X⁄mass of organic carbon)

[X]_iw

is the concentration of the compound in the interstitial water (as gms/L) (Pavlou 1987) or it can be reasonably accurately derived from the octanol-water partition coefficient according to the formula developed by Di Toro et al (1985; in OMOE 1988).

log₁₀ K_oc = 0.00028 + 0.983 log₁₀ ( K_ow )

The K_oc value used is derived by taking the geometric mean of the available K_ow values.

Both measured and calculated K_ow values can be used to derive a K_oc and a number of values are required to estimate the K_oc used.

K_oc values should be calculated from laboratory derived sediment-water partition coefficients whenever possible, rather than from values derived from the octanol-water partition coefficient (K_ow).

Since the NEL criteria make use of the PWQOs, which employ safety factors to ensure conservative levels, it is anticipated that the sediment guidelines derived from these will be conservative as well. While the distribution of non-polar organics in the pre-colonial sediment horizon should technically be zero, it is recognized that a certain amount of sediment contamination has occurred from remote sources through atmospheric inputs. Since guidelines set below these background levels would be impractical, the background levels must form the lower limits of any sediment quality guidelines. To this end, background levels for the non-polar organics are provided in this document for comparative purposes. These are based on the average of the upper Great Lakes, deep basin surficial (top 5 cm) sediment concentrations, or in some cases, on concentrations in bluff materials. It is expected that where the NEL criteria derived by the partitioning method fall below these background levels, the background levels will provide the practical lower limit for management purposes.

The deep basin surficial sediment concentrations from the upper Great Lakes can be considered as representative of atmospheric inputs of the persistent (generally nonpolar) organics. Table 4 gives the background levels for those compounds for which upper Great Lakes level have been calculated, and these can be considered as normal background levels for management purposes.

Lowest Effect Level

The Lowest Effect Level (LEL) is the level at which actual ecotoxicological effects become apparent. It is derived using field-based data on the co-occurrence of sediment concentrations and benthic species. The Screening Level Concentration Approach described in the previous section is used for all types of contaminants.

The calculation of the SLC is a two step process and is calculated separately for each parameter. In the first step, for each parameter the individual SLCs (termed Species SLCs) are calculated for each of the benthic species. The sediment concentrations at all locations at which that species was present are plotted in order of increasing concentration (Figure A1a). From this plot, the 90^th percentile of this concentration distribution is determined. The 90^th percentile was chosen to provide a conservative estimate of the tolerance range for that species. This would serve to eliminate extremes in concentrations that may be due to specific and unusual sediment characteristics. The 90^th percentile is that locus below which 90% of the sediment concentrations fall.

In the second step, the 90^th percentiles for all of the species present are plotted, also in order of increasing concentration (Figure A1b). From this plot, the 5^th percentile and the 95^th percentile are calculated. These represent the concentrations below which 5% and 95% of the concentrations fall.

Metals, Nutrients, and Polar Organics

Calculate the 5^th percentile of the SLC based on bulk-chemistry sediment data. Since the guidelines are derived for province-wide application, the locations used should span a wide range of geographical areas within Ontario of varying sediment concentrations of the contaminant. It is important to ensure that both high sediment concentrations as well as low concentrations are used in the data set to ensure the result is not biased towards one end or the other, since this could bias the resulting SSLC. A minimum of 10 observations would be required to calculate a SSLC for any one species. This relatively low minimum has been chosen so as not to exclude less common species, or more importantly, the more sensitive species that may not be present at the more contaminated sites and thus may not be represented at the majority of sites. A minimum of 20 SSLCs (i.e. 20 species) would be required for calculation of an SLC.

Non-polar Organics

Calculate the SLC as above, but using contaminant concentrations normalized to the organic carbon content of the sediments (i.e. mass of contaminant/mass of organic carbon as expressed by TOC).

The organic carbon normalized sediment contaminant concentrations are converted back to a bulk sediment concentration assuming a 1% TOC. A limit of 1% TOC has been imposed on the calculation since calculations using the SLC approach have shown that this is the lowest effect level of organic carbon in the sediment.

The Ministry also recognizes that certain parameters addressed in these guidelines, such as the trace metals, occur naturally in aquatic environments. In an area as geologically diverse as Ontario, natural sediment levels can vary considerably from one region of the province to another as a result of differences in local geology. Therefore, the Ministry realizes that certain sites will naturally exceed the LEL. In such cases, the local background levels, based on the pre-colonial sediment horizon, will form the practical lower limit for management decisions as described in the Implementation Section of this document.

Calculation of Site-Specific Background

Site-specific background is calculated as either:

1. the mean of 5 surficial (top 5 cm) sediment samples taken from an area contiguous to the area under investigation, but unaffected by any current or historical point source inputs; or,
2. the mean of 5 samples taken by a sediment core from the pre-colonial sediment horizon. The pre-colonial horizon is generally determined as the sediment below the Ambrosia sediment horizon. Except in areas of high sedimentation, such as river mouths, this can be estimated as that sediment lying below the 10 cm sediment depth.

Severe Effect Level (SEL)

This level represents contaminant levels in sediments that could potentially eliminate most of the benthic organisms. It is obtained by calculating the 95^th percentile of the SLC (the level below which 95% of all SSLCs fall).

Metals, Nutrients, and Polar Organics

Calculate the 95^th percentile of all SSLCs using the bulk chemistry values.

Non-polar Organics

Calculate the SLC as for the metals, but normalizing the data to the organic carbon content (TOC) of the sediments. The TOC-normalized SLC is then converted to a bulk sediment value at the time of application to a specific site, based on the actual TOC concentration of the sediments at that site (to a maximum of 10%, the 95% SLC guideline for TOC).