Jump to main content.


Assessment and Remediation of Contaminated Sediments (ARCS) Program

Monitoring Links

exit EPA [About PDF]


Table of Contents

Back To Index 

Assessment Guidance Document
Chapter 4

US Environmental Protection Agency. 1994. ARCS Assessment Guidance Document. EPA 905-B94-002. Chicago, Ill.: Great Lakes National Program Office.

Table Of Contents 

SCREENING-LEVEL ANALYSES

4. Screening-Level Analyses

Most of the Great Lakes rivers and harbors that have been designated as AOCs have been industrialized for decades. These AOCs have been subject to a large number of altering and often degrading forces, including discharges of metals, oils, halogenated organic compounds, domestic wastes, and other pollutants; altered sedimentation patterns due to watershed deforestation; and artificial rearrangement of their bedded sediments from dredging, ship traffic, and shoreline construction. One result is that the ecosystems of the Great Lakes AOCs, and sediments in particular, possess a mosaic of chemical and physical characteristics that reflect a multitude of historical anthropogenic alterations.

The chemical and physical characteristics of Great Lakes AOCs are sufficiently complex that conducting even a general inventory is difficult. For many of these AOCs, there are no prior surveys of contaminated sediments except for routine navigational dredging studies performed by the Corps. The available historical studies of contaminated sediments usually included a limited number of chemical and even fewer toxicity tests performed primarily on surficial samples. Detailed chemical and biological tests are expensive, time-consuming, and require relatively large volumes of sediments. Few studies have had the resources to analyze enough samples to create meaningful contour maps of contaminant distributions for significant portions of a river or other water body.

In situations where there were insufficient resources to conduct enough detailed assays to adequately characterize sites, two methods have the potential to fill in data gaps at relatively low cost: indicator (performed during the ARCS Program) and screening-level (the recommended approach) analyses. The analysis approach used in the ARCS Program was designed to test the efficacy of a two-phased sampling design: 1) a set of quick, less expensive assays ("indicator analyses") performed at a large number of stations where sediment cores were collected, and 2) detailed chemical and toxicological assays performed at a limited number of surface sediment stations throughout the study area. Multivariate statistics were then used to explore the potential relationships between indicator analyses and more detailed assays, although the results were often inconclusive (see Indicator Analyses section below).

More recent research has shown that the screening-level analyses discussed in the following section are also quick and less expensive, and may be more comparable from site to site than the indicator analyses tested in the ARCS Program. The screening-level analyses discussed below are appropriate for in-field use, as well as in the laboratory. In the field, screening-level analyses can be very useful for guiding sampling later in the survey, for example, to better delineate an area of high contamination.

Indicator analyses are the field-level analyses performed during the ARCS Program. They are inexpensive surrogates for more expensive parameters. Screening-level analyses are those described in the next section.

Summary of Screening-Level Methods

Screening-level analyses are relatively rapid, inexpensive assays that can be readily applied to sediment assessment. These analyses can be completed in the field or in the laboratory. Due to their relatively low cost, screening-level analyses should be used to focus comprehensive analyses on hot spots where remediation is most likely to occur and on "grey" areas (i.e., areas of intermediate contamination) where the integrated sediment assessment approach is necessary to make a proper evaluation. They can also be used to conduct a preliminary survey of a large area to locate suspected hot spots. Data from these analyses are useful for providing a sufficient number of data points for proper mapping of sediment conditions. Currently used screening-level analyses include fluorometry for PAHs; immunoassays for PCBs, chlorinated pesticides, and PAHs; infrared spectroscopy for petroleum hydrocarbons; thin-layer chromatography (TLC) for semivolatile organic compounds; x-ray fluorescence (XRF) for metals; and rapid toxicity tests. These screening-level analyses are described below.

Total PAHs by Fluorometry

The total PAH assay using fluorometry was developed and tested in surveys along the Buffalo River, and has proven to be fairly rapid (20 samples/day), inexpensive, and strongly correlated with GC/MS and high-pressure liquid chromatography (HPLC) analyses (Friocourt et al. 1985). This assay is particularly sensitive to compounds containing aromatic rings, such as PAHs, which strongly fluoresce at specific wavelengths that can be set on the instrument (e.g., 280-nm for excitation spectra). This sensitivity is enhanced by the specificity of the assay, which reduces the need for complicated separation techniques during sample preparation. Fluorometry does not respond to a wide range of organic compounds found in sediments, such as aliphatic hydrocarbons from oils, fatty acid methyl esters from natural and anthropogenic sources, and phthalate esters, all of which are common interferences in gas chromatographic analyses.

Total PCBs, Chlorinated Pesticides, and Other Organic Chemicals by Enzyme Immunoassay

Enzyme immunoassays are biochemical procedures that rely on the binding of specific chemicals in a sample (plus an enzyme-labeled version of the chemical) to antibodies provided in a test kit; the bound chemicals can then be separated from the rest of the sample and associated interferences by simple washing; then the labeled component is detected by adding a color indicator (Schrynemeeckers 1993). Immunoassay kits are available from several manufacturers. For sediments, a small sample is quickly extracted and purified. The extract is then tested with the immunoassay kit.

Enzyme immunoassays are inherently free of most chemical interferences (Vanderlaan et al. 1991) and are available for several aromatic compounds besides PCBs and pesticides (e.g., petroleum hydrocarbons, PAHs, trinitrotoluene, benzene). These assays are typically used either to indicate the presence/absence at some predetermined concentration for chemical mixtures such as total PCBs, or to provide order-of-magnitude estimates of concentrations of individual chemicals. The detection limit for total PCBs in sediments for various test kits ranges from less than 0.1 ppm to approximately 5 ppm, and the immunoassay results can be strongly correlated with gas chromatography results (Huellmantel et al. 1992). The most confident use of immunoassays is in determining when assay-specific contaminants are below levels of concern because of the low potential for false negative readings.

Enzyme immunoassays can also be performed for PAHs. Production rates and costs are similar to the rates and costs for PCB immunoassays. Other recent field tests of PCB-contaminated soil using immunoassays resulted in an approximate 40-percent decrease in field time and a 44-percent decrease in overall costs compared with more traditional sampling and laboratory analyses (Scallen et al. 1992). Such cost savings will be realized most readily at sites that have a limited number of contaminants of concern, where only a single immunoassay is required. Several different immunoassays would be required to fully characterize sediments containing a wide range of compound classes of concern, because each kit is sensitive to only one of the compounds or classes of compounds described above.

Total Petroleum Hydrocarbons by Infrared Spectroscopy

The infrared assay is intended to be a field version of the extractable residue analysis by USEPA Method 418.1. This analysis is useful for measuring total petroleum hydrocarbons (TPHs) because responses in relatively narrow ranges of the entire infrared spectrum correspond to the presence of specific groups of atoms regardless of the structure of the remaining molecule. For example, a variety of hydrocarbon structures are simultaneously detected in a sample by characteristic changes in carbon-hydrogen or carbon-carbon bonds (e.g., stretching and bending vibrations) that are induced by exposure to infrared radiation. Detection limits are typically in the range of 1-10 ppm.

The infrared method has the advantage of providing a rapid, quantitative determination of TPH concentrations, but also has some limitations that can produce either negative or positive analytical bias (Douglas and Uhler 1993). As a result, this screening method may be less accurate than other techniques for measuring hydrocarbons such as field gas chromatography, which has the disadvantage of a higher cost, or TLC.

Semivolatile Organic Compounds by Thin-Layer Chromatography

The TLC field method, developed by Friedman & Bruya, Inc. (Seattle, Washington) and reported by Newborn and Preston (1990, 1991), can be used for a wide range of semivolatile organic compounds with detection limits of approximately 10 ppm (lower detection limits to approximately 1 ppm are feasible for some compounds). The TLC method was not used in the ARCS Program.

This method involves placing a drop of sample extract near the bottom of a silica gel-coated glass plate. The end of the plate is immersed in an appropriate solvent. As the solvent front moves upward on the plate, the compounds of interest are separated out of the mixture based on their mobility in the solvent-solid phase system, and can then be identified both qualitatively and quantitatively, using ultraviolet light or iodine to visualize the separated chemicals.

Metals by X-ray Fluorescence

Field portable XRF units have been used to analyze soils at Superfund sites (e.g., Fribush 1992; Driscoll et al. 1991) and have been shown to provide rapid (<5 minutes/dried sample) quantification of more than 20 elements at a time. Detection limits for portable units have typically been reported in the 100-1,000 ppm range for most metals, while laboratory-based XRF units have greater resolution and are capable of lower detection limits in the range of 2-25 ppm (Grupp et al. 1989; Fribush 1992). Laboratory XRF units have a somewhat longer analysis time (20 minutes). XRF analyses, unlike other metal analyses that rely on digestion of samples with various acids, do not destroy the sample and require only a small amount of material. XRF results have been found to be correlated strongly with conventional atomic absorption and inductively coupled plasma (ICP) spectroscopy results (Kuharic et al. 1993).

Rapid Toxicity Tests

The Microtox® test is a rapid, sensitive method of toxicity testing based on light emission by the luminescent bacterium Photobacterium phosphoreum in the presence and absence of aqueous toxicants. The emitted light is a product of the bacterial electron transport system and thus directly reflects the metabolic state of the cells. Accordingly, decreased luminescence following exposure to chemical contaminants provides a quantitative measure of toxicity. The test was developed for use in freshwater habitats to assess the toxicity of waterborne pollutants (Bulich et al. 1981). Recently, a solid-phase modification of the Microtox® test was developed. See Chapter 6, Evaluation of Sediment Toxicity, for a more detailed discussion of the Microtox® test.

In addition to Microtox®, other rapid toxicity tests, such as Daphnia IQ ® or the 2-day rotifer toxicity test (Snell and Moffat 1992), can be used for screening-level analyses.

Indicator Analyses

The indicator analyses used in the ARCS Program (Table 4-1) are rapid, low-cost analyses chosen with several considerations in mind, including environmental relevance, probable correlation with conventional analyses, analysis time, analytical cost, and regulatory relevance. The chosen analyses may be divided into two categories: those that produce a direct measure of sediment composition or contamination, and those that produce an indirect measure of sediment quality that may be related to other variables of environmental or regulatory importance.

The metals (cadmium, chromium, copper, iron, lead, nickel, and zinc), total and volatile solids, total organic carbon (TOC), grain size, ammonia, and the Microtox® test (also a screening-level analysis technique) fall into the first category. TOC is an important factor in determining the environmental availability of hydrophobic organic contaminants and metals. Grain size data are useful in determining the geographic extent of sediment deposition zones and layers and perhaps their origin, and provide a quick method to make initial determinations as to the potential contaminant load of the sediments, with finer-grained sediments typically more contaminated than the sandy, coarser-grained fraction. Ammonia is often present in toxic concentrations in harbors and tributaries of the Great Lakes (Ankley et al. 1990). Microtox® has been employed in previous sediment quality investigations as a toxicity screening tool (e.g., Giesy et al. 1988b).

The remaining indicator assays (conductivity, pH, extractable residue, and the organohalogens) may be related to other variables of environmental importance (e.g., bioavailability, organic contamination). Extractable residue is a measure of the solvent-extractable materials (oils and other petroleum-related products) in sediment, some components of which are known toxins (e.g., PAHs), and which have also been shown to influence the availability of many organic contaminants (Neff 1985). The organohalogen analysis is a rapid, inexpensive measure of total halogenated compounds. While the organohalogen analysis is unable to distinguish highly toxic organochlorine compounds (e.g., PCDDs and PCDFs) from other high-concentration, less toxic compounds (e.g., non-planar PCBs), it does give an estimate of the total concentration of these compounds present.

Results for Core Samples Analyzed During the ARCS Program

The following three sample matrices were used in the ARCS Program to assess the indicator analysis approach for core samples collected from the Saginaw River, Buffalo River, and Indiana Harbor AOCs: whole sediment for analysis of grain size, total and volatile solids, metals, solvent extractable residue, organohalogens, and TOC; sediment elutriate for analysis of ammonia and Microtox®; and sediment pore water for analysis of conductivity. The elutriate procedure is designed to mimic the rapid desorption of contaminants from sediments resulting from the open-water disposal of dredged materials (Plumb 1981). (See Chapter 6, Evaluation of Sediment Toxicity, and Burton [1992a] for further discussion of the use of elutriates and pore water.)

The ARCS Program indicator assay data indicate that the three AOCs examined exhibit a wide range of sediment contamination, with Saginaw River the least contaminated and Indiana Harbor the most contaminated (Table 4-2). Only pore water conductivity failed to show this pattern.

Some specific observations drawn from Table 4-2 include:

The core sampling, while fairly costly and time-consuming, yielded information unavailable through traditional surficial grab sampling. Indicator analysis data from many cores showed increasing contamination and toxicity with depth, with highly contaminated and toxic material up to 4 m beneath the surface of the sediment. For example, in the Buffalo River, 50 percent of the cores ended in what was visually characterized as black, oily silt. This material contained elevated concentrations of metals (e.g., maximum lead and copper concentrations of 1,586 ug/g and 951 ug/g dry weight, respectively) and volatile solids (maximum of 19.8 percent dry weight), and was also toxic (50 percent of the deepest samples had an EC50 <50 percent). Sediment cores from the Saginaw River at Station 6, near the wastewater treatment plant discharge, contained an 8- to 20-in. layer of black, oily silt beneath 1 ft or so of cleaner surficial sand. This oily silt layer exhibited metals concentrations 3-15 times higher than those in the sand above (e.g., cadmium = 19 vs. 1.3 ug/g dry weight; chromium = 590 vs. 40 ug/g dry weight; lead = 180 vs. 32 ug/g dry weight). The most highly contaminated sediments in most of the cores from the Saginaw and Buffalo Rivers were found well below the surficial sediments. This distribution will often be the case in Great Lakes AOCs when there has been recent success in controlling contaminants from point sources.

In contrast, Indiana Harbor typically exhibited the most highly contaminated sediments at the surface, probably indicating continuing contaminant inputs.

Correlation Between Indicator and Comprehensive Analyses

As stated earlier, the time and resource requirements of many comprehensive chemical analyses (Chapter 5) and biological analyses (Chapters 6 and 7) limited their application in the ARCS Program to only a few master stations in each demonstration AOC. The quicker and less expensive indicator analyses were performed for a much larger number of stations and samples in order to test whether these indicator analyses could be considered to be both stand-alone measures of sediment contamination and possible correlates with the more detailed master station data. The indicator analysis approach was successful in showing chemical and toxicity concerns at a much lower cost than could be achieved with a comprehensive survey. Therefore, these analyses can be used to focus resources in a follow-up phase of work at a specific site.

Multivariate statistics were also applied to the two data sets to attempt to create predictive, correlative equations for the comprehensive analyses as a function of indicator analysis results. Although significant correlations were often found, the resulting predictive equations were not readily interpretable. For example, in Indiana Harbor, the concentration of total PCBs in surficial sediment was found to be significantly correlated with the concentrations of zinc and cadmium:

PCBs = -58,700 + (11.6xZn) - (2,020 x Cd)
(r2 = 0.9442; P = 0.0031)

where:

PCBs = concentration of total PCBs in surficial sediment, ug/kg dry weight
Zn = concentration of zinc in surficial sediment, mg/kg dry weight
Cd = concentration of cadmium in surficial sediment, mg/kg dry weight.

However, there is no apparent reason why the concentration of total PCBs should increase with increasing concentrations of zinc, but decrease with increasing concentrations of cadmium. In most cases, the resulting predictive equations were not universally applicable between AOCs, probably as the result of site-specific variations among the study areas.

Because Microtox® was performed as both an indicator assay and a comprehensive analysis assay (along with many other analyses), the prediction of Microtox® toxicity could be evaluated. Predictive equations for the Microtox® test were calculated from the indicator analyses of the master stations, and then used to calculate EC50 values for each of the many reconnaissance station samples. For evaluation purposes, the calculated EC50 values were divided into three categories: low toxicity (>70 percent elutriate), intermediate toxicity (30-70 percent elutriate), and high toxicity (<30 percent elutriate), and comparisons were made between the calculated and measured values from each reconnaissance station. The comparison between the predicted Microtox®. EC50 values produced by the correlation equations and the measured EC50 values is presented in Table 4-3. The high percentage of "false positive" results (especially for the Buffalo River and Indiana Harbor study areas) suggests that the data do not provide sufficient resolution for predicting toxic and nontoxic conditions. Furthermore, the variability in regression equations among areas suggests that any one equation would have limited applicability. For example, the predictive equations for the Microtox®. test endpoint in the three AOCs were:

 Buffalo River Microtox®. EC50 = 119 - (587 x organobromine) - (0.127 x Pb)
(r2 = 0.9700; P = 0.0001)

Saginaw River Microtox® EC50 = 106 - (10.5 x Cd)
(r2 = 0.9943; P = 0.0001)

Indiana Harbor Microtox® EC50 = 59.3 - (0.117 x Cu) - (0.008 x Pb)
(r2 = 0.9432; P = 0.0032)

where:

Organobromine = concentration of organobromine in surficial sediment, mg/kg dry weight

Pb = concentration of lead in surficial sediment, mg/kg dry weight
Cd = concentration of cadmium in surficial sediment, mg/kg dry weight
Cu = concentration of copper in surficial sediment, mg/kg dry weight.

Hence, the predictive equation developed for one AOC would not enable accurate predictions to be made for the other AOCs. It should be noted that Microtox® also had a high number of significant correlations with other toxicity tests (Chapter 6).

Regression equations for other indicator variables were also highly variable among AOCs suggesting that site-specific differences and the strong potential for coincidental correlations severely limit the use of these predictive tools in decision-making. The regression equations are indicative of correlative rather than causative relationships among the variables.

Conclusions

In conclusion, each of the indicator analyses conducted in the ARCS Program was simple to perform and, except for metals by ICP-atomic emission spectroscopy (ICP/AES), required only one technician to operate. The TOC analysis, which was performed by several ARCS laboratories independently (LLRS, Battelle, and the National Oceanic and Atmospheric Administration [NOAA] Great Lakes Environmental Research Laboratory), yielded comparable data despite some differences in analytical procedure. Discussion with other laboratories suggests that TOC protocols can differ substantially, however, especially in the ratio of sediment:acid, strength of the acid, and oxidation time and conditions. In some cases, these differences may call interlaboratory data comparisons into question, which will have implications for risk assessments and sediment quality criteria development.

The sediment cores collected in the ARCS Program yielded information that would not have been available from surficial grab samples. Subsurface sediment layers, up to 13 ft beneath the sediment-water surface, exhibited high toxicity and high concentrations of metals and oils. Sediment buried at these depths may not pose substantial environmental or human health risks of exposure, but future sediment surveys should investigate the 3-dimensional distribution of contaminated sediments to provide reasonable estimates of the potential vertical extent of any remedial activities that may be required to address surface sediment contamination.

The indicator analysis approach provides a useful means of focusing the time-consuming and costly comprehensive chemical analyses and toxicity tests on selected samples that are most likely to be of concern from a study area. The interpretation of indicator analysis results, however, appears to be site-specific and without a firm scientific basis for making meaningful predictions for the outcome of more comprehensive assays. The combined indicator analyses, however, do provide a preponderance of evidence that can be used to make a decision on the existence or lack of concern about a specific study area. Further, some of the indicator analyses (e.g., Microtox®, metals by ICP, TOC) provide valuable, direct information.

The indicator analyses used in the ARCS Program, however, are more labor-intensive and have an overall higher cost than the screening-level analyses discussed in Summary of Screening-Level Methods. These screening-level analyses are efficient and provide interpretable data that can be used to make appropriate and timely decisions on how to focus resources on areas most likely to be remediated, as well as to identify less certain areas where comprehensive analyses are needed in a subsequent phase of investigation to make a final sediment assessment. Screening-level analyses show much promise for evaluating parameter concentrations (e.g., PAHs, PCBs, metals, toxicity) on many samples more quickly and less expensively than comprehensive chemical and biological analyses (Chapters 5, 6, and 7). Also, screening-level analyses are recommended over most indicator parameters because it is preferable to directly measure something rather than to model it. However, most of these screening-level analyses provide less selective and accurate results and should not be used to make decisions regarding the need for sediment remediation without confirmation by comprehensive chemical and biological analyses.

 


Local Navigation


Jump to main content.