ABSTRACT

Capping of subaqueous dredged-material disposal areas is becoming a common method for isolating contaminated sediments from the environment. This paper presents two methods for determining how thick a sediment cap must be to isolate contaminants from the sediment reworking activities of organisms. Both methods use data on local benthic communities to make quantitative predictions of the minimum cap thickness necessary to isolate contaminated dredged material. For Boston Harbor, this thickness is about 50 cm.

INTRODUCTION

Capping of contaminated sediments is becoming a common method of disposing of and confining pollutants in dredged material. As part of the Boston Harbor Navigation Improvement Project (BHNIP), contaminated sediments dredged from channels in Bostonís Inner Harbor will be disposed of in confined aquatic disposal (CAD) cells to be created in the northern portion Bostonís Inner Harbor. The contaminated sediments will be capped with clean sand. Because capping of sediments is costly, it is important to develop efficient capping strategies which minimize costs while maintaining the integrity of the cap. One important question to be answered is: How thick should the sediment cap be in order to contain contaminated sediment?

To isolate contaminants in dredged sediments from the water column and from organisms, sediment caps must be thick enough to bury contaminants below the surface mixed-layer of the sediment. There are a variety of processes that contribute to the depth of the mixed layer and to the rate of transport within the sediment. These processes include resuspension by waves and currents (Dyer 1986), advection and diffusion of porewater components (Berner 1980), advection of fine particles into coarser materials (Huettel et al. 1996), and sediment reworking by animals, termed bioturbation (Berner 1980). Because some burrowing organisms have the ability to penetrate caps as thick as 50 cm (Thayer 1983, Brannon et al. 1985, 1987, Gunnison et al. 1987) the depth of bioturbation is perhaps the most important factor in determining the minimum thickness of sediment caps.

The purpose of this paper is to present models for predicting the sediment-mixing depth within capped sediments at a particular location given knowledge of the benthic community structure of the region or a list of species which could potentially colonize a sediment cap. For sediment tracers or contaminants which are tightly bound to sediment particles, the mixing depth will be equal to the minimum cap thickness required to isolate these contaminants. If the tracers or contaminants possess an aqueous phase, then the minimum cap thickness must also account for diffusion or advection of porewater into the cap (e.g., Murray et al. 1993). Two modeling approaches for determining sediment cap thickness will be developed which depend upon these two types of benthic data (community structure or a species list). Mathematical models of particle mixing will be developed which directly incorporate data on benthic community structure. These models will then be used to predict the minimum cap thickness necessary to isolate dredged material generated by the BHNIP from burrowing organisms in Boston Harbor.

This theoretical approach has several advantages over past attempts to determine an optimal cap thickness. Benthic community structure varies widely from region to region (e.g., Sanders 1956, Lie and Kelly 1970). Thus, results of studies which have examined the transport of contaminants in the presence of a few species of burrowing infauna in laboratory microcosms (e.g., Brannon et al. 1985, 1987, Gunnison et al. 1987) may or may not be applicable to all locations with dredging operations. This theoretical approach will predict different cap thickness for different areas and thus may be superior to an approach that sets one standard cap thickness for all regions.

METHODS

The main assumption of the models developed in this paper is that the movement of particles in the sediment is controlled primarily by the activities of benthic infaunal organisms. Many theoretical and empirical studies support this assumption (Thayer 1983, Rice 1986, Wheatcroft et al. 1990, Matisoff 1995). These studies indicate that organisms are responsible for the mixing processes that could potentially excavate deep sedimentary deposits such as contaminated dredged material underlying a sediment cap. For the purposes of this investigation, the porewater phase will be neglected. Instead, the focus will be on transport of the particles that comprise the cap and dredged material. However, extensions of the model that include tracers or contaminants possessing a porewater phase will be discussed.

The second assumption of the model is that all species available to colonize a sediment cap are found in the regional species pool. The regional species pool, also called the "realized species pool" (Osman and Dean 1987), consists of all species which have been collected in an area. If this assumption is correct, then if a comprehensive list of benthic species is available for a large enough region surrounding a proposed sediment capping site, then all species which will likely colonize a sediment cap will be found on that list. Since most large-scale dredging operations require environmental impact statements that include an assessment of the benthic community, it is likely that benthic species lists for specific regions will be available prior to capping. Benthic communities in Boston Harbor, for example, have been extensively surveyed by the Massachusetts Water Resources Authority (MWRA) and other agencies.

The most difficult part of predicting the effects of benthic organisms on sediment caps is determining which species will colonize the cap. Because no quantitative benthic-ecological theories exist which could make such a prediction, the best one can do is to consider a range of possible communities which could develop on a sediment cap. One end-member of this range will be termed the "worst-case-scenario community". It is composed of the deepest-dwelling organisms found in the regional species pool that could potentially excavate dredged material from underneath a sediment cap. The other community to be considered in this report will be termed the "ambient" community. This is composed of those species in the region which can be found in sediments similar to sediments which comprise the cap. Adopting the "worst-case-scenario community" approach, (to be described below) will yield the most conservative prediction of the minimum sediment-cap thickness (thickest cap). It is the approach to be used if only a species list is available. The "ambient" community approach can be taken if data on the abundance and spatial distribution of benthic species is available. A separate theory for each of these approaches will be developed next.

"Worst case-scenario" method

Biology

The first step in this method is to obtain a comprehensive list of species in the region that defines the species pool. The region must be large enough to contain all species that are likely to colonize sediment at the dredged-material-disposal site. For the dredged-material-disposal site in Boston's Inner Harbor, for example, the regional species pool was considered to be all species which had been collected anywhere within Boston Harbor from 1991 until 1994 (Kropp and Diaz 1995). It was thus implicitly assumed that any species that had not appeared in the harbor over these years would not likely appear in the future.

The next step is to identify all deep burrowing organisms found in the data set. This step requires knowledge of lifestyles of benthic organisms. References that aided in accomplishing this task for Boston Harbor taxa are cited in Tables 1 and 2. From this list of deep-burrowing organisms a benthic community is constructed which contains the maximum densities of the deepest dwelling organisms which could potentially rework deep sediments. For Boston Harbor, literature values for sediment reworking rates, burrowing depths and densities were used (Table 1).

Table 1."Worst-case-scenario" benthic community from Boston Harbor. Species list is from Kropp and Diaz (1995).

Table 2. Ten most abundant taxa, densities and feeding modes of organisms collected in Dorchester Bay. Abundances represent mean numbers with standard deviation shown in parentheses. Organism dry weights were used to estimate feeding rates from Cammen"s (1980) empirical ingestion rate model. Interface feeders were considered to move particles at the sediment surface. Burrowers were considered to move particles one body length from the point of ingestion. Conveyor-belt feeders were considered to move sediment from their feeding depth to the sediment surface. Subductive feeders were considered to move particles from the surface to the bottom of their tubes.

Once a "worst-case-scenario community" is constructed, the effects of this community on caps of different thickness must be determined. This was accomplished by assuming that all species have the ability to move sediment from their burrowing depth to the sediment-water interface. Although many deep-burrowing species such as the heart urchin (order Spatangoida) or some burrowing shrimp (superfamily Thalassinoida) might not move sediments encountered at depth to the surface (Thayer 1983, Nickell and Atkinson 1995), this assumption was used as a worst-case scenario. Then, following the arguments of Rice (1986), the deposition of material at the sediment-surface can be treated using an advective model.

Mathematical model

The rate of deposition of sediment at the sediment-water interface can be considered to be the sum of the external (allochthonous) sedimentation rate, and the rate of biodeposition by burrowing deposit feeders. If B is the biomass of deposit feeders, r is the feeding rate (units found in Table 1), n and r represent sediment porosity and solids density, and b represents the depth distribution of deposit feeders, then the rate of deposition of dredged material at the sediment-water interface for different cap thickness, L, will be determined by the integrated biomass of burrowers found below the sediment cap multiplied by the biomass-specific biodeposition rate.

(1)

Assuming the depth distribution of organisms is Gaussian with a mean of m and standard deviation s, then,

(2)

where erfc is the complimentary error function. Values of erfc are tabulated in most statistical handbooks and are can be calculated by many mathematical packages (e.g., MATLAB, 1993).

Using equation 2, the rate of movement of contaminated dredged material to the sediment surface can be calculated for caps of various thickness which have been colonized by the "worst-case-scenario" benthic community. If organisms breach the cap and begin reworking contaminated sediments into the cap material, changes in concentration of particulate contaminants of concern within the sediment cap can be modeled using w_(x,L) as the advective term in the diagenetic equation:

(3)

where C(x,t) is the concentration of a conservative tracer of contaminated dredge material (Rice, 1986). Although it is beyond the scope of this study, equation 3 can be expanded to account for other processes affecting the distribution of tracers or contaminants of concern such as sorption and desorption, porewater diffusion and advection. The effects of contaminant input to the top of the cap due to deposition of surrounding ambient sediment, which could potentially occur at high rates within the BHNIP CAD cells, can also be explored. Methods of expanding this equation to account for these kinds of processes have been reviewed by many authors (e.g., Berner 1980, Schwarzenbach et al. 1993, Boudreau 1997).

"Ambient-community" method

Biology

In many urban harbors, extensive data on the abundance and distribution of benthic species are available. In addition, past experiments might indicate that the community of organisms colonizing the capped disposal sites will be similar to nearby ambient benthic communities. In this case, data from the ambient community can be used to determine the minimum cap thickness. In addition, these data can be used to model changes in the concentration of contaminants within the sediment cap given caps of different thickness and particle-reworking rates.

Sediment caps used in the BHNIP are composed of sandy sediments. Thus, the colonization of these caps could produce a benthic community similar to communities inhabiting nearby sandy sediment. One such community is located in Dorchester Bay (71^o 01' W, 42^o 19.5' N, 5 meters depth, Figure 1). To determine the benthic community structure of this site, samples were collected from this area using 4.7-cm diameter and 16-cm diameter cores. Smaller cores were vertically sectioned at 1-cm intervals, preserved in formalin, and later transferred into ethanol using a 250-mm sieve. Larger cores were sieved whole over a 1-mm mesh sieve and preserved in the same manner. Organisms were identified to species, body lengths were measured, and dry weights were determined after oven drying at 70^oC.

Figure 1. Map of Boston Harbor showing the location of the first CAD cell created for the BHNIP and the location of the site sampled to collect data for the "ambient community" method for determining the minimum cap thickness.

Mathematical model

Benthic community structure data of this type can be readily entered into a model which can be used to determine minimum sediment-cap thickness. Particle-mixing rates are again assumed to be due primarily to the deposit-feeding activities of benthic organisms. Given data on the sizes and depth distributions of benthic species, pathways of particle movement are assumed to begin at the point of sediment ingestion, and end at the point of sediment egestion. The rate of particle movement is determined by organism ingestion rates. Ingestion rates can be determined from literature values or can be calculated using Cammenís (1980) empirical ingestion rate model.

The sediment can be envisioned as a series of boxes (Figure 2). The rate of particle movement from box to box is calculated using a transition matrix. By converting the particle movement rates to transition probabilities, particle mixing can be analyzed using Markov chain theory. Changes in the distribution of dredge material can be determined by the formula: N_t+1 = N_t^. P, where N_t is the distribution of dredge material at time t and P is the transition matrix.

Figure 2. Diagrammatic representation of sediment mixing within a CAD cell. Each box represents a sediment depth interval. The bottom box represents the depth below the mixing zone. Arrows represent particle trajectories from one depth interval to another.

RESULTS

"Worst-case scenario"

Table 1 lists the deepest-dwelling organisms which are capable of reworking sediments collected by the MWRA sampling program from 1991 -1994 (Kropp and Diaz 1995). Included in this list are species known to occupy both fine-grain and coarse-grain sediments. Of the species listed in Table 1, Clymenella torquata was chosen as the "worst-case scenario" organism. This species lives as deeply as any other deposit-feeding species found in Boston Harbor and can transport sediment at a rapid rate. Because this species can inhabit sandy sediment (Rhoads and Stanley 1966), it is possible that it could colonize the sandy capping material used in the BHNIP. To construct a "worst-case-scenario community", the maximum published densities of C. torquata were used (600 m^-2, Sanders et al. 1962). The burrowing depth was also determined by averaging values from the published literature. It was assumed that the range of published burrowing depths for this species represented that variance in depth distribution (Figure 3). The maximum feeding rate found in the literature was used in the model (4g dry sediment worm^-1day^-1, Fuller 1994).

Figure 3. Modeled depth distribution of Clymenella torquata biomass.

Figure 4. Predicted rate of deposition of dredged material at the sediment surface for sediment caps of different thickness.

The results of the model (Equation 2) are shown in Figure 4. These results indicate that a 25-cm thick sediment cap would isolate dredged material from the deepest-dwelling deposit feeders in Boston Harbor. A cap which would also contain contaminants diffusing upward would need to be somewhat thicker. Using an estimate of 30 cm to chemically seal contaminated sediments (Palermo et al. 1989) results in a cap thickness of 55 cm, very close to the 50-cm standard.

"Ambient" community

Benthic community structure data from the sandy community sampled in for this study are displayed in Table 2 and Figure 5. The vertical distribution of organisms are shown in Figure 5. The vertical distribution of C. torquata was calculated as three times its body length (Magnum 1964). The feeding rates of the deposit-feeders was calculated using Cammenís (1980) ingestion rate model.

Figure 5. Depth distribution of species listed in Table 2.

Figure 6. Profiles showing the change in the depth distribution of dredged material over time. The initial cap thickness was 19 cm. The slight deepening of the cap visible after 10 years (line with circles) is due to the modeled sedimentation rate of 1 mm per year.

The model was run to determine changes in the distribution of dredged material and cap material assuming the cap were colonized by this community. A number of caps of different thickness were used in these runs. Results for 19-cm and 10-cm caps are shown in Figures 6, 7 and 8 . Results indicate that a 19-cm thick cap would effectively isolate dredged material from this community (Fig. 6). If a 10-cm cap were used, organisms would initially deposit dredged material on the sediment surface (Fig. 7). Over time, however, the concentration of dredged-material tracers or contaminants at the sediment surface would drop due to mixing with cleaner cap material (Fig. 8).

Figure 7. Profiles showing the change in the depth distribution of dredged material over time with an initial cap thickness was 10 cm. The solid line indicates the distribution after 50 days. The broken line indicates the distribution after 1 year. The line with the plus marks indicates the depth distribution after 2 years. The line with circles indicates the distribution after 10 years.

Figure 8. Predicted changes in the fraction of dredged material in surface sediments over twenty years.

DISCUSSION

In past capping operations, it has typically been assumed that the minimum cap thickness required to isolate contaminated sediments from the water column and biotic community was 50 cm (SAIC 1995). This thickness was based on the results of a series of laboratory studies that examined changes in contaminants in capped sediments (Brannon et al. 1985, 1987, Gunnison et al. 1987). These studies concluded that a 50-cm cap effectively isolated contaminants from surface-dwelling and water-column organisms. One limitation of these studies was that only two cap thicknesses were examined: 5 cm and 50 cm. Thus, the actual minimum cap thickness required to isolate contaminants from the water column and from biota may fall somewhere in between these depths. In fact, the required cap thickness may vary from region to region depending on a variety of factors including which species are available to colonize the cap.

Another limitation of these laboratory studies is that benthic organisms used in the laboratory might not be representative of the organisms occurring near a capping site. One study that used these methods but also considered the local benthic community was conducted by Palermo et al (1989). In this study, the initial thickness estimate of 50 cm was increased to 80 cm to include the burrowing depth of the geoduck (Panope abrupta), a local clam species. In this case, the known burrowing depth of a local benthic organism was much more important in determining the sediment-cap thickness than the results of laboratory studies which included burrowing infauna. Thus, an approach to determining sediment-cap thickness, which uses local biological data to make quantitative predictions of sediment-mixing depths, may be superior to laboratory studies using a few species of infauna.

By utilizing biological data collected at proposed dredged-material-disposal sites, the methods developed in this paper can be used to make site-specific determinations of cap thickness. For some areas, the required cap thickness will be much less than 50 cm. In areas possessing very deep-dwelling infauna, such as burrowing shrimp, these methods might determine that the minimum cap thickness is greater than 50 cm. This highlights another advantage of these methods. Because these methods require an examination of benthic community structure, they may identify areas which are not suitable for the dumping and capping of dredged material due to the presence of deep-burrowing species.

These methods also possess disadvantages. They require the existence of an extensive benthic data set. Even with an extensive set of data, rare but important species could still be missed. Also, it may be difficult in some areas to identify the geographical extent of the species pool. Finally, it requires specific knowledge of organism characteristics. Despite these disadvantages, the ability to make quantitative determinations of the sediment-cap thickness required to isolate contaminated sediments makes this a potentially valuable technique.

Implications for the BHNIP

The "worst-case-scenario" approach predicted that a 25-cm thick cap would isolate dredged material from burrowing benthos. The "ambient-community" approach predicted that a 19-cm thick cap would be sufficient. The discrepancy is due to the more conservative nature of the "worst-case-scenario" approach. Because the fauna comprising the "ambient community" included the "worst-case-scenario" species, Clymenella torquata, the "ambient community" estimate of 19-cm can be accepted without running the risk of neglecting the deepest burrowing species in the region. If it is assumed that a 30-cm cap is required to contain contaminants diffusing upward (Palermo et al. 1989), then the sum of these cap-thickness estimates is very close to the 50-cm estimate commonly used in capping operations.

Future Directions

This modeling approach can be expanded to account for other processes that may transport contaminants. Molecular diffusion or porewater irrigation can be incorporated into the diagenetic model (Equation 3) and into the matrix model to predict changes in distributions of contaminants possessing a porewater phase. This approach would likely produce a more accurate estimate of minimum-cap thickness than the common practice of summing the thickness required to chemically seal contaminants and that required to account for bioturbation.

Because these models predict the rate at which contaminated sediments are brought to the surface of the sediment for caps of different thickness, the results could be included as part of a cost-benefit or risk-analysis approach to determining cap thickness. A cost-benefit or risk analysis could be used to determine an acceptable rate of contaminant transfer to the sediment surface or water column. The acceptable rate of transfer would likely be a function of the types of contaminants found in the dredge material and the risks that those contaminants pose to the environment. The models could then be used to determine the cap thickness necessary to limit contaminant transfer to the acceptable rate.

REFERENCES