Introduction
In
this article, we present results of a method that computes the upper
threshold for ambient concentrations of chemical elements and compounds
in San Francisco Estuary sediments. In light of the work described
below, the staff of the San Francisco Bay Regional Water Quality
Control Board (Regional Board) would consider chemical constituents
at concentrations equal to or below this upper threshold as ambient
values.
Background
Scientists
have studied sediments in San Francisco Bay and other estuaries
for many decades. Yet despite significant effort on a national level,
only general guidance has been available for assessing contaminated
sediments and dredged material quality. This project was intended
to support the Regional Board in its work in the area of assessment
and management of contaminated sediments. To better evaluate polluted
sites, Regional Board staff need data on ambient concentrations
of chemicals in sediments. For instance, sediments at a given site
in and along the Bay margin are often scrutinized for elevated concentrations
of elements, compounds, or classes of compounds. The thresholds
presented below can be used to determine whether sediments have
chemical concentrations greater than that of the current Bay ambient
condition.
Ambient
Values
Although
Bay sediments can be severely polluted, such as those found at a
state-listed "toxic hot spot", more often Bay sediments
are moderately contaminated. Since San Francisco Bay sediments are
not totally free of anthropogenic pollutants, ambient concentrations
for these compounds may be higher than that in pre-industrial sediments
(background concentration). It is therefore important to define
the typical range of concentrations that one would expect to find
in ambient sediments.
It
is often crucial to know how chemical concentrations in a given
sediment sample compare to those in the rest of the Bay. This is
especially true for habitat restoration projects where a newly restored
intertidal wetland would be subject to an influx of suspended sediments
from the daily tides. A restored wetland surface will have concentrations
at least as high as ambient levels because the new marsh substrate
will be comprised of sediment deposited by resuspension from ambient
sources.
Background
Values
The
concentrations of chemicals in sediments prior to the region's industrialization
are often relevant to sediment investigations. These pre-industrialization
concentrations are referred to as "background". However,
industrial activities carried out during and after the late nineteenth
century have had a profound deleterious effect on much of the San
Francisco Estuary. Industrial discharges continued uncontrolled
until the enactment of the Clean Water Act in the 1970s. Since then,
point source discharges of contaminants have steadily decreased.
Current factors controlling chemical contamination of surficial
sediments are point and non-point discharges, atmospheric deposition,
bioturbation, and resuspension of sediments by wave and current
action.
Recent
analysis of deep sediment cores by the United States Geological
Survey (USGS) has provided valuable information on pre-industrial
levels of several metals: copper, lead, mercury, silver, and zinc
(Hornberger et al., in press). Other metals in that report include
chromium, nickel, and vanadium. Hornberger found background metal
concentrations in deeper, pre-industrialized sediments to be lower
than those in the surficial (ambient) sediments.
Data
Sources and Considerations
Data
used to calculate ambient sediment concentrations were collected
as part of the Regional Monitoring Program for Trace Substances
(RMP) and the Bay Protection and Toxic Cleanup Program (BPTCP).
Sediments used in the ambient analysis were collected from sites
consistently shown to be representative of the cleanest portions
of the Bay (Table 4.4). The sampling
stations are located away from point and non-point pollution sources.
The data used in this analysis were gathered from the 1991 Pilot
RMP, ongoing RMP, and the BPTCP's 1995 Reference Site study. The
survey stations where sediments were collected for chemical analyses
are all within the San Francisco Estuary. The data used in the statistical
analyses consisted of 81 records for PAHs, PCBs, heavy metals and
metalloids, and selected chlorinated pesticides. Other analytes
were not analyzed due to the low number of detections. The station
names along with the sampling dates are listed in Table
4.4. River stations (noted as "BG") are located within
the Central Valley RWQCB's jurisdiction. In the Pilot RMP, some
of the Bay stations were located near potential sources of contamination
(e.g., marinas); these were removed from the database prior to analysis.
The "marsh" stations from the Pilot RMP were not included
in this database and the available database for "marsh"
sediments was not sufficiently large to warrant a separate analysis.
Sediment
Dynamics / Sample Type
In
shallow areas with fine-grained sediments, there is typically a
loose or "fluff" layer that hovers over the firm sediments.
The sediment samples discussed herein likely include a portion that
is periodically resuspended or recently deposited. Resuspension
of fine-grain material by wind-waves is a dominant force in shallower
regions, while current-driven bed-load transport of coarse material
is common in the deep channels (Schoellhamer and Burau, 1996). Data
for this project were obtained from sediment samples taken from
the upper five centimeters of the benthic substrate.
Removal
of Outliers
Outliers
were removed to prevent skewed results based upon only one or a
few values. Outliers were determined by visual observation of scatter
plots and searching for obvious breaks in the data clusters (Smith
and Riege, 1998). This process removed data from 19 stations.
Statistical
Approach for the Determination of Ambient Threshold Values
The
following summary presents a brief overview of the statistical data
analysis employed in the determination of ambient concentrations.
For a complete description of the statistical methods employed,
the reader is referred to Smith and Riege (1998).
Chemical
analyses of sediments from relatively clean locations yield a wide
range of concentrations for each element or compound. The aim of
the statistical analyses was to define a concentration at the upper
end of this data range to serve as an upper threshold for distinguishing
between concentrations representing ambient versus contaminated
conditions. One way to accomplish this is to define the threshold
as a percentile of the distribution. For example, one could define
the 85th percentile as the threshold. Here 85% of the data values
would fall below the threshold value, and 15% would fall above the
threshold. Since a relatively small set of ambient data was available,
the percentiles of the underlying distribution of ambient sediment
concentrations had to be estimated. The uncertainty in the estimate
is a function of the sample size. To incorporate the uncertainty
in the estimate of the threshold percentile, tolerance intervals
were computed instead of the percentiles. A tolerance interval is
the confidence interval bound of a percentile. The confidence interval
bound of the mean is widely used and understood. The tolerance interval
is similar except that it represents the confidence interval for
a percentile instead of the mean of a distribution. Figures
4.21 and 4.22 show the tolerance
intervals of selected percentiles for two different distributions.
The
size of the confidence interval around the threshold percentile
is related to the value chosen for the parameter a, where the size
of the confidence interval in percent is 100(1-a). Thus, if one
chose the 85th percentile as the threshold and a = 0.05, the upper
tolerance interval bound is the upper bound of the 95% confidence
interval of the 85th percentile of the ambient concentrations for
the chemical in question. If the statistical assumptions associated
with the method are correct, the tolerance interval bound would
be expected to cover the 85th percentile 95% of the time.
For
this project, the 85th percentile was selected as the threshold
concentration with an a= 0.05 used as the tolerance interval bounds.
For all chemicals, the tolerance interval bounds are reported for
40% fines and 100% fines. For PCBs, pesticides and metals, this
is a somewhat conservative procedure in that the chemical concentrations
tend to increase monotonically with finer sediment texture. Tolerance
interval bounds at 40% and 100% will be the highest bounds for coarse
and finer sediments, respectively. The results are presented in
this manner to avoid presenting multiple, continuously changing
bounds as a function of particle size.
Parametric
tolerance intervals are used when the data fit a normal distribution
or when the data can be transformed to a normal distribution. Otherwise,
non-parametric tolerance intervals are used. In this project, both
approaches were used, depending on the observed distribution of
the data sample. The statistical models are described in Smith and
Riege (1998).
Several
physical and chemical factors, such as total organic carbon, particle
surface area, and particle size distribution, are known to correlate
with chemical concentrations in sediments. These factors must be
considered when defining ambient concentrations. Grain size, as
measured by the percent fine, was selected as the main co-factor
for the data analysis, as it is easily measured, there is a known
interrelationship of other factors with grain size, and there is
a lack of data for many other parameters.
After
analysis of the distributions with respect to particle size, three
statistical models were employed. For the PAHs, two thresholds were
computed, one for coarse sediments (040% fines) and one for
finer sediments (>40% fines). The concentrations of pesticides,
metals, and PCBs tended to increase monotonically with decreasing
particle size. Tolerance interval bounds around regression lines
were used to account for the particle size-concentration relationship.
Results
The
results of the analysis of tolerance intervals at the 85th percentile
and a = 0.05 are presented in Table
4.5. The statistical analyses and the results for other tolerance
intervals can be found in Smith and Riege (1998).
Discussion
and Recommendations
We
recommend that the ambient level threshold for routine use be based
upon the bound for the 85th percentile as derived for sediments
at the 100% fines level (Table 4.5).
Most projects subject to regulatory scrutiny involve fine-grained
sediments (e.g., dredging projects and military base closure). Therefore,
the thresholds for fine material should prove more often useful
to various agencies. Coarse material analytical results should be
compared to the ambient value for coarse sediments. Given the uncertainties
of the data, it is appropriate that the threshold values for metals,
chlorinated hydrocarbons, and pesticides be based upon the upper
bound for coarse or fine-grained sediment. Coarse-grained ambient
sediments are essentially devoid of chlorinated compounds.
In
various site-specific sediment investigations, we have seen concentrations
of chlorinated organic compounds (e.g., PCB, DDT) above the ambient
upper thresholds. Some of these investigations were conducted in
locations that are well offshore and removed from suspected sources.
The fact that these sites have chlorinated organic compounds above
ambient levels is, in some cases, a reflection of the relatively
low detection limits used in the monitoring programs. In other words,
the detection of the chlorinated organic compounds is at, or near,
the limit of detection, and the occurrence of such compounds in
offshore sediments is heterogeneous. Therefore, the comparison of
project sediments to ambient thresholds for compounds, such as PCBs
and HCHs, will be essentially a comparison to non-detection. This
is a contrast to the detection of heavy metals and PAHs, which are
more widely distributed and have ambient thresholds well above routine
detection limits.
For
comparative purposes, we have also included effects range levels
(ERL, ERM) in Table 4.5 (Long et
al., 1995). The ambient threshold is the point at which one can
say with confidence that a given concentration is either within
the ambient (reference) population or elevated above it. The threshold
is not a sort of average around which there is a region of uncertainty
(error bar); rather, it is the edge of the reference envelope. Also,
the ambient thresholds do not speak to the potential toxicity of
these chemicals at low levels. The biological risk associated with
these chemicals at ambient levels is a question outside the scope
of this project. However, sediments that are swept into dredged
channels or onto newly formed marsh surfaces would be expected to
contain chemical concentrations similar to the ambient concentrations
presented in this report.
Several
metals were found at levels exceeding guidelines and thresholds
(e.g., ERLs). In some cases, for example nickel, this is partially
due to the mineralogy of the parent geologic material found in the
Estuary's watershed. In other cases, such as mercury, its occurrence
is mostly the result of anthropogenic activities. Metal concentrations
may be even higher in certain locations due to parent geologic materials.
For example, mercury concentrations are elevated in North Bay tributaries
as compared to the mid-Bay stations used in this study. We found
the threshold in fine sediment for mercury to be 0.43 ppm. However,
sediment samples taken in the Napa River and Novato Creek watersheds
show that mercury can be in the 2 ppm to 4 ppm range (Regional Board
Case Files: Corps of Engineers, Dredging applications, Napa River
Flood Control Project). These data are not covered in our analysis.
The
ambient concentration plots and thresholds should be used for evaluations,
and when possible, in concert with other measurements and endpoints.
If toxicity testing or bioassay data is available, those data should
also be considered during the decision-making process. In cases
where there is little reason to suspect polluted sediments, these
ambient thresholds may prove most useful as a "first-level
screen" in the decision-making process. In this way, ambient
concentrations can serve to define what is "elevated"
relative to the ambient sediments distributed throughout the Bay.
For projects involving sediment concentrations well above the ambient
thresholds, more sophisticated measurements of toxicity and estimates
of bioavailability should be considered.
It
is believed that the ambient values presented in this report will
be representative of conditions in San Francisco Bay for a number
of years. The RMP data collected over a four-year period shows that
the concentrations of contaminants in sediments at these mid-Bay
sites do not change substantially from one year to the next. For
this reason we recommend that the database be updated and the thresholds
recalculated on a triennial basis.
References
Hornberger,
M., S.N. Luoma, A. Van Green, C. Fuller, and A. Roberto. (in press).
Historical trends in metals in the sediments of San Francisco Bay,
California. Marine Chemistry.
Long,
E.R., D.D. MacDonald, S.L. Smith, and F.D. Calder. 1995. Incidence
of adverse biological effects within ranges of chemical concentrations
in marine and estuarine sediments. Environmental. Management. 19(1):81-97.
SFBRWQCB.
1998. Staff Report: Ambient concentrations of toxic chemicals in
San Francisco Bay sediments, May 1998. San Francisco Bay Regional
Water Quality Control Board, Oakland, CA.
Schoellhamer,
D.H. and J.R.Burau. 1996. Residual transport in Suisun Cutoff, San
Francisco Bay, California, during summer 1995: Part 2, Sediment
Transport. Abstract for the 1996 Fall American Geophysical Union
meeting, San Francisco, California, December 15-19, 1996.
Smith
R.W. and L. Riege. 1998. San Francisco Bay Sediment Criteria Project:
Ambient Analysis Report. Prepared for the San Francisco Bay Regional
Water Quality Control Board, Oakland, CA.
Document
Availability
Documents
that this article are based upon (listed below) are available from
the San Francisco Bay Regional Water Quality Control Board office
and can be found on the Regional Board Web Page at: http://www.rwqcb2.com/
San
Francisco Bay Regional Water Quality Control Board, Staff Report:
Ambient Concentrations of Toxic Chemicals in San Francisco Bay Sediments,
May 1998.
Smith,
R.W. and L. Riege. 1998. San Francisco Bay Sediment Criteria Project:
Ambient Analysis Report.
|
RMP:
RMP
Home | Program Information
| Publications | RMP
Data
