Clam , Macoma balthica
P A M E L A B . M C L E O D ,²
M A R T I N EJ . V A N D E N H E U V E L - G R E V E ,³
R I C H E L L EM . A L L E N - K I N G ,§
S A M U E LN . L U O M A ,|A N D
R I C H A R DG . L U T H Y *, ²
Department of Civil and Environmental Engineering,
Stanford University, Stanford, California 94305-4020,
National Institute for Coastal and Marine Management,
Fieldstation Middelburg, The Netherlands, Department of
Geology, State University of New York at Buffalo,
Buffalo, New York 14260, and U.S. Geological Survey,
Menlo Park, California 94025
We investigated the bioavailability via diet of spiked
52) fromdifferent carbonaceous (non-carbonate, carbon
containing) particle types to clams (Macoma balthica)
collected fromSan Francisco Bay. Our results reveal
significant differences in absorption efficiency between
compounds and among carbonaceous particle types.
Absorption efficiency for PCB-52 was always greater than
that for BaP bound to a given particle type. Among
particles, absorption efficiency was highest fromwood
and diatoms and lowest fromactivated carbon. Large
differences in absorption efficiency could not be simply
total organic carbon content. BaP and PCB-52 bound to
and were up to 60 times less available to clams than the
same contaminants associated with other types of
carbonaceous matter. These results suggestthatvariations
matter, whether naturally occurring or added as an
of hydrophobic organic contaminants. This has important
implications for environmental risk assessment, sediment
management, and development of novel remediation
aromatic hydrocarbons (PAHs) have been prioritized as
found that among Superfund sites, PCBs and metals were
the most common contaminants driving sediment environ-
mental risk, followed by PAHs and pesticides (1). Despite
strategies for PCB- and PAH-contaminated sediments. This
disagreement stems from the complex interactions of PCBs
and PAHs with sediment and the poorly understood effects
of these interactions on contaminant bioavailability to
benthic organisms at the base of the food web. Limited
understanding of the ways in which sediment biogeochem-
istry governs PCB and PAH bioavailability and fate hinders
the development of protective sediment quality guidelines
and innovative remedial measures for contaminated sites.
which they can (re)partition into overlying water and enter
bioavailabilitybyconsideringageneric organic matter(OM)
pool that sorbs the contaminants to an equal degree, based
on equilibrium partitioning (2, 3). Output from this organic
matter partitioning (OMP) model isused to judgetherisk to
water quality and animal exposure. Many researchers have
documented limitations of the OMP model in accurately
predicting the solid-water distribution of nonionic hydro-
phobic organic contaminants (e.g., refs 4-7). In practice,
the OMP model overpredicts aqueous PAH and PCB con-
centrations in equilibrium with a variety of field sediment
worked well for PCBs but overestimated concentrations of
phenanthrene and pyrene by as much as 250 and 30 times,
respectively. To the extent that pore water governs bioavail-
ability, the OMP model cannot dependably predict risk to
the food web from a given PCB or PAH sediment concentra-
Sediment is heterogeneous at grain and subgrain scales
(8). The limitations to the OMP model partly arise from that
heterogeneity. Specifically, different particle types within
carbonaceous matter can exhibit vastly different sorptive
affinities for PAHs and PCBs. For example, existing data for
phenanthrene reveal orders of magnitude differences in
observed partition coefficients for a suite of particle types
the OMP model should be replaced by a combination of
sorption isotherms in sediment (10). Although researchers
use different nomenclature, most of the devised schemes
consider linear absorption and nonlinear adsorption as the
dominant processes (ref 10 and references therein). Some
researchers propose that the new modeling paradigm
consider amorphous organic carbon and soot carbon as
separate sorbing matrixes in sediment (11-13). This idea is
supported by research that documents the ªsupersorbentº
capacity ofthesoot fraction and thecommon occurrenceof
Dey and Gschwend (13) showed that a two-compartment
model comprising soot carbon and amorphous organic
matter could predict sediment-porewater distribution coef-
In contrast, Jonker and Koelmans (14) recently found that
normalization to organic and soot carbon fractions did not
accurately describe PAH or PCB sorption to a suite of soot
and soot-like materials. This suggests that even a two-
compartment equilibrium model may oversimplify con-
taminant sorption in some sediment.
It is imperative to understand how sorption to different
particle types might influence the overall bioavailability of
PAHs and PCBs in sediment; but only limited work exists.
Ghosh et al. (9) showed that PCBs and PAHs in three
*Correspondingauthorphone: (650)723-3921;fax: (650)725-8662;
³National Institute for Coastal and Marine Management.
§State University of New York at Buffalo.
|U.S. Geological Survey.
Environ. Sci. Technol. 2004, 38, 4549-4556
10.1021/es049893b CCC: $27.50
Published on Web 07/24/2004
2004 American Chemical SocietyVOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY94549
contaminated harbor sediments were predominantly as-
sociated with certain types of carbonaceous matter in
sediment, including coal, tar, and cenophores. They found
that PAHs were more bioavailable for aerobic bioslurry
degradation when sorbed to semi-soft tar pitch than when
associated with coal-derived particles. Talley et al. (15)
employed bioslurry studies, earthworm uptake, and micros-
bioavailability of PAHs in sediment is directly related to
differences in the form of carbonaceous matter to which
PAHs are bound.
sorption on dietary exposure of PAHs and PCBs, although
particle ingestion can be an important exposure route for
contaminant sorption to certain types of carbonaceous
particles will result in lower absorption efficiencies in the
gut, butthishasnotbeen directly investigated.Itispossible,
for example, that the gut fluid in organisms is a strong
ofthecarbonaceousparticlesthemselvescan bedigested by
benthic organisms, leading to increased contaminant avail-
The objective of this research is to increase our under-
standing of PCB and PAH sediment biogeochemistry by
determining the effect of contaminant association with
different forms of particulate carbonaceous matter on
bioavailability via diet. The particles tested include forms of
carbonaceous matter with widely varying composition and
structure(such aswood, coke, char, and anthracite) and are
commonly found in sediments. We conducted feeding
experiments that measured the absorption efficiency for
spiked benzo[a]pyrene (BaP) and 2,2′,5,5′-tetrachlorobiphe-
nyl (PCB-52) to clams (Macoma balthica) from six different
particle types and diatoms, a usual form of food for these
clams. Results from our experiments suggest that the
bioavailability of PCBs and PAHs is highly dependent upon
thetypeofcarbonaceousmatterto which thecontaminants
Experim ental Section
Test Organism. The clam Macoma balthica was chosen
becauseofitsuniform presencein theintertidal mudflatsof
San Francisco Bay, the amount of existing data describing
the feeding behavior of M. balthica, and the relative ease of
handling and maintaining M. balthica under laboratory
conditions. It is usually a deposit feeder, but is also capable
time and other aspects of feeding behavior influential in
absorption are also well-known (19, 20). Two days prior to
each experiment, 60-80M. balthica werecollected by hand
Nature Reserve in south San Francisco Bay (station 6 in ref
20). Total PAH and PCB concentrations at the site were
(unpublished data), both representing background concen-
trations for San Francisco Bay. Pereira et al. found PAH
concentrations in relatively noncontaminated parts of the
Bay ranged from 0.04to 6.3ppm (21), and corestaken in the
<1-34 ppb (22).
Clams were rinsed in the field, transported to the
laboratory in Bay water, and maintained without feeding in
were chosen to mimic field conditions and minimize stress
to the organisms. The clam shells were brushed 1 d before
were similar in size (21-24 mm shell length) to minimize
at Santa Cruz), where the water is collected via pipeline,
sand-filtered, and pumped into storage tanks. Prior to use,
the seawater was additionally filtered (0.45 µm), diluted to
17 ppt with Milli-Q water, and stored at 13 °C.
Particle Preparation. The test particles were coke,
L. Throop Co. (Pasadena, CA). Coke and anthracite were
ground in a ball mill to produce smaller particles. Wood
particles were prepared by sanding white pine (Palo Alto
Hardware, Palo Alto, CA). Char was produced at the Johns
a nitrogen atmosphere at 650 °C/h to 525 °C. It was then
crushed using a mortar and pestle. Peat was purchased as
PahokeePeat Soil from theInternational Humic Substances
Society (St. Paul, MN). Fine-mesh activated carbon (TOG 50
× 200) was purchased from Calgon Carbon Corporation
(Pittsburgh, PA) and crushed using a mortar and pestle.
Each particletypewaswet sieved toobtain particleswith
the test organisms with a relatively uniform particle size
within the range clams typically ingest (19). The particles
were dried overnight (100-150 °C), transferred into 20-mL
vials, and stored at room temperature until use. Specific
surface area (SSA) was determined by Micromeritics, Inc.
(Norcross, GA) by adsorbing N2at liquid nitrogen temper-
atures using the BET method. SSA for char was measured at
the Johns Hopkins University. Total organic carbon (TOC)
was measured at Huffman Laboratories (Denver, CO) by
combustion ofacidified samples.Particlecharacteristicsare
summarized in Table 1.
Diatom Preparation. The diatom Phaeodactylum tricor-
nutum was obtained from SUNY Stonybrook (Stonybrook,
cultureswerekeptin Guillard'snutrientmedium (23) in 0.45
µm-filtered 35ppt seawater and grown with 10:14light:dark
for 10 d under artificial light. The cultures were suspended
Labeling of Particles and Diatoms with [3H]-BaP and
Piscataway, NJ) and 2,2′,5,5′-[14C]-PCB (sp. act. ) 14.4 mCi/
mmol, SigmaChemical, St.Louis, MO) weretransferredinto
amber vials with hexane upon receipt and stored at 4 °C
until use. The specific activity of BaP was reduced to 76.0
MO) to stock solutions.
Particles were simultaneously spiked with [3H]-BaP and
vials and evaporated to dryness by gentle hand swirling,
leaving the compounds adhered to the vial walls. Particles
50mgforall otherparticles) in 10mL deionized water.Spike
vials were sealed with Teflon septa, covered in aluminum
foil toprevent photodegradation, and mixed on arotatorfor
21d. Ten-day-old diatom cultures werespiked similarly but
were added as suspensions in 35 ppt seawater. Diatom
cultures were further grown at 10:14 light:dark for 7 d after
spiking and were suspended by hand-shaking twice daily.
Some photodegradation of BaP could have occurred during
this step. Unlabeled diatoms for use as cold feed were
maintained under the same conditions. Table 1 reports the
initial contaminant concentration on the particles and
diatoms as well as the initial activity present at the start of
the pulse-chase experiments.
45509ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
method used in previous studies of metal uptake by M.
balthica (24, 25) to determine clam absorption efficiencies
of [3H]-BaP and [14C]-PCB-52 from ingested particulate
materials and water. Glass beakers were used to minimize
sorption of organic compounds to the feeding apparatus.
anthracite, coke), experimentII(diatoms, wood, char, peat),
and experiment III (diatoms, activated carbon). All experi-
mentswereperformed in a controlled-temperatureroom at
Clams were randomly divided into groups of 13 per
treatment. Each clam was attached with Krazy Glue (Co-
of the rod when held vertically, and siphons were oriented
upward. Rods were pre-roughened with sandpaper to
promote strong attachment. After drying in air for 15 min,
each group of 13 rod-mounted clams were suspended from
Rodsweresecured with O-ringstoallow vertical adjustment
in the water column. Clams suspended from the mesh were
allowed to acclimate for 1 h in a 1-L beaker with 500 mL of
17 ppt seawater immediately prior to the hot feed.
filter to separate particulate from dissolved compounds in
the spiking water. The particles were introduced into the
500 mL of 17 ppt seawater. A total of 25 mL of unlabeled
clam feeding. The treatment with spiked diatoms did not
receive unlabeled diatoms. Particles and diatoms were kept
in suspension using a magnetic stirrer.
to feed for 8 h. Decho and Luoma (19) determined the
minimum gut passage time, defined by the first appearance
of radiolabel in feces, for M. balthica to be 9.6 ( 1.8 h, with
a range of 3-16 h. Thus, some clams may have produced
feces during our exposure period. However, we chose this
feeding time to ensure all clams fed, which was shown in
preliminary tests to decrease variability in clam absorption
efficiency as compared to hot feeds of shorter duration.
After the hot feed, clams were removed and rinsed. Ten
of the clams from each treatment were placed in unlabeled
unlabeled diatoms as feed. This allowed for separate feces
in calculated absorption efficiency as compared to tests in
which feces from several clams were pooled and analyzed
inert beads, M. balthica egested >95% of unabsorbed
radioactivity in 72-96 h (19). In our experiments, contami-
nants in the clam tissue after 88 h of depuration were
considered absorbed. Feces were collected on glass fiber
filters at 14, 26, 40, 64, and 88 h and analyzed for [3H]-BaP
and [14C]-PCB-52. An individual clam's feces from different
sampling times were sometimes combined to ensure mea-
surable radioactivity. In preliminary tests, negligible con-
taminant concentrations were detected in water samples
added to remove metabolic waste and encourage digestion
the shells, and shell length and wet weights were recorded.
Glass fiber filters placed under the clams during dissection
to capture internal fluids were also analyzed. Compounds
were found to adsorb to clamshells in preliminary tests, but
Estimating Partition Coefficients and Mass Balances.
In experiment III, additional samples were removed from
the feeding beaker before and after feeding to characterize
particle-water partitioning in our systems. Some samples
were filtered on a glass fiber filter, and the filter and filtrate
were analyzed. The sum of the radioactivity from the filter
and filtrate was operationally defined as the system total.
Other samples were centrifuged (Beckman L8-6000 Ultra-
centrifuge), and thesupernatant wasanalyzed to determine
particle-free aqueous concentration. These data were used
to estimate potential aqueous uptake and particle-water
on particles in the feeding beaker at the start of the hot feed
tothesum oftheactivityin particlesremainingin thebeaker
at the end of the hot feed, clam tissues, clam feces, and
methanol and hexane rinses of the feeding beaker. This
approach assumes negligible dissolved contaminant con-
carbon system. In the activated carbon test, 106.2% of the
BaP and 105.5% of the PCB-52 were recovered.
Sample Digestion and Extraction. Feces samples were
PA), 370 µL of concentrated HCl, and 12 mL of Scinti-Safe
were vortexed once daily for 5 d. Whole soft tissue of each
clam was chopped and digested in 2.8 mL of ScintiGest
(totaling 1 min) to aid digestion. A 0.6-mL subsample of
digested tissue was transferred to a 20-mL scintillation vial
TABLE1. FractionO rganic Carbon(fO C), Specific Surface Area (SSA), andInitial BaP andPCB-52 Concentrations for Particles in
Experim ents I-III
× 10-6, dpma
1032 564 ( 128
615 ( 45
864 ( 16
628 ( 39
701 ( 69
561 ( 69
561 ( 40
1.7 ( 0.4
3.0 ( 0.6
4.9 ( 0.9
801 ( 216
727 ( 42
1013 ( 18
659 ( 51
492 ( 55
785 ( 46
870 ( 39
3.5 ( 0.6
9.7 ( 0.9
15 ( 0.5
17.4 ( 4.0
16.2 ( 1.2
26.7 ( 0.5
12.4 ( 0.7
10.6 ( 1.0
14.8 ( 1.8
17.4 ( 1.3
1.02 ( 0.21
1.85 ( 0.39
3.22 ( 0.53
4.12 ( 1.11
3.18 ( 0.19
5.21 ( 0.09
2.16 ( 0.17
1.24 ( 0.14
3.44 ( 0.20
4.47 ( 0.20
0.35 ( 0.06
1.00 ( 0.09
1.48 ( 0.05
adpm ) disintegrations per minute; initial activity on particles in feeding beaker.bPhaeodactylum fOCas reported in ref 30.
VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 94551
with 80 µL of HCl and 14 mL of scintillation cocktail.
Subsamples were analyzed instead of whole tissue to
minimize tissue matrix effects and attain high counting
a tissue homogeneity test in which we fed clams following
the pulse-chase protocol and counted subsamples from
pooled, digested tissue. Coefficients of variation among
sampleswerelessthan orequal to0.01forbothcompounds,
verifying the homogeneity of tissue subsamples.
The digestion and extraction technique described above
was chosen after comparing the ability of several methods
to recover [3H]-BaP and [14C]-PCB-52 from spiked coke
for 21 d. The extraction methods tested were different
combinations of ScintiGest tissue solubilizer, scintillation
cocktail, and HCl. Particles were combusted at 750 °C in a
carbon combustion oven, and gases were captured in three
1 N NaOH traps in series. Samples from each trap were
analyzed and added. Combustion was assumed to recover
100% of the compounds from the particles. Measured
combustion recoveries for [3H]-BaP and [14C]-PCB-52 stan-
dardsin hexanewere96-101%.Theextraction methodused
in our experiments recovered 95% ( 2% of [3H]-BaP and
101% ( 3% [14C]-PCB-52 as compared to combustion.
Radiolabel Analysis. Radioactivity was determined by
liquid scintillation counting (model 6500, Beckman Instru-
Tissue and feces samples produced a yellow-amber color in
Thus, a quench curve for colored samples was established
for [3H]-BaP and [14C]-PCB-52 in colored samples were
100.02% ( 0.96 and 100.57% ( 0.96, respectively. All tissue
Noncolored samples were corrected for quench using the
manufacturer's quench curve. Background was subtracted
from all samples. Since the3H and14C counting windows
overlap, we also performed a3H:14C ratio test in colored
(tissue) and noncolored matrixes. Varying the ratio had no
significant effects on isotope recovery.
Absorption Efficiency. Absorption efficiency (AE) was
calculated for each clam using measured tissue and cumula-
tive feces activities and a mass balance approach:
where atissueis the activity in M. balthica soft tissue after 88
h of depuration and Σafecesis the sum of all activity in feces
losses of BaP and PCB-52 to depuration water, which was
confirmed in preliminary studies. No attempt was made to
correct for absorption of contaminants by the clams from
the water during the feeding period.
Researchers have used different terms to describe con-
28). In this paper, we use the term ªabsorption efficiencyº
to describe the proportion of organic contaminant physi-
ologically incorporated into soft tissues after ingestion and
depuration (19, 24, 25). In some treatments, the absorption
efficiency values included uptake from the aqueous phase.
As discussed later, the weak sorption of BaP and PCB-52 to
diatoms and wood resulted in a fraction of the spiked
compound partitioning into water during the feeding ex-
a bias might have affected interpretations.
Absorption Efficiency. Absorption efficiency resultsforBaP
and PCB-52 from experiments I-III are presented in Figure
the sorbents were equilibrated with spiked chemicals for
higher than that for BaP for a given particle type, from all
in all three experiments to assess the replicability and
seasonality of the data. The absorption efficiency for both
compounds was always highest from diatoms. Mean ab-
sorption efficiencies for BaP from diatoms in experiments
I-III were 35.2%, 50.7%, and 37.4%. Mean absorption
efficiencies for PCB-52 were 89.9%, 86.3%, and 83.3%. The
and allows us to directly compare results from experiments
I-III. Some photodegration of BaP during diatom spiking
significantly among different particle types, as shown here for
experiments I-III. Light columns represent BaP; dark columns
represent PCB-52. Error bars represent 95% confidence intervals.
Absorption efficiency of BaP and PCB-52 varies
% AE )
45529ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
could have contributed to the greater variability in the BaP
absorption efficiency from diatoms versus PCB-52.
Figure 2 presents the entire data set from experiments
I-III. For diatoms and wood, which were tested in more
than one experiment, the mean absorption efficiency was
determined as the average of all available data points.
Significant differences in absorption efficiency are seen
between particle types, with 30- and 60-fold differences
between activated carbon and diatoms for BaP and PCB-52,
respectively. Statistical analyses using ANOVA suggest the
associatedcontaminantbioavailability: (i) activatedcarbon;
(ii) coke and peat; (iii) anthracite and char; (iv) wood and
diatoms. Combining the data reiterates that contaminant
to which the contaminant is bound.
The differences in absorption efficiency we observed
among particles may stem from different contaminant-
material of biogenic origin following the scheme presented
by Allen-King et al. (10). Activated carbon, coke, anthracite,
and char represent thermally altered carbonaceous matter.
Sorption to wood is expected to be weaker than to particles
representingthermally alteredcarbonaceousmatter(10), so
the high absorption efficiency from wood is not surprising.
petrographic analysis) that impart highersorptivecharacter
than would be expected from a peat composed exclusively
of humic materials. This may explain the low absorption
efficiencies from peat measured in our experiment. Ad-
ditionally, it is not surprising that activated carbon showed
the lowest observed absorption efficiency, given its highly
sorptive tendency, a quality that has been exploited in the
water treatment sector for decades. Further investigations
ofthesorptivedifferencesamong coke, anthracite, and char
may be required for a more complete mechanistic under-
standing of the relationship between sorption to particles
and absorption efficiencies from these particles, although
this does not appear to be the only mechanism involved, as
Dietary absorption efficiencies of BaP and PCB-52 are
P. tricornutum to green mussels (Perna viridis) as 37.2 (
4.3%. The range of BaP absorption from P. tricornutum by
M. balthica reported herein are consistent with their (30)
from sediment to an amphipod (Diporeia spp.) and oli-
gochaete (Lumbriculus variegates) were 45% and 23-26%,
Higher absorption efficiency for PCB-52 than BaP is
observed in somebut not all studieselsewhere. In testswith
Diporeia, for example, Harkey et al. (31) found that BaP was
bioaccumulated less than 2,2′,4,4′,5,5′-hexachlorobiphenyl,
even though the two have comparable octanol-water
efficiency of BaP to the oligochaete L. variegates, ranging
from 23 to 75% for bacteria, algae, and sediment, when
absorption efficiency values could be calculated. For PCB-
52, they measured absorption efficiencies ranging from 21
to87%.In specific comparisons,BaP uptakesurpassedPCB-
52 uptake by the oligochaete. In chironomid larvae, by
contrast, thetrend wasreversed. Researchersstudying HOC
sorption to natural materialshaveobserved thetendency of
of similar hydrophobicity, including PCBs (5). It is logical
that stronger sorption might be related to lower bioavail-
ability, but apparently this is not true for all species.
and microscale food choices could confound such simple
Other researchers suggest that food quality influences
organic contaminant bioavailability. For example, Gunnars-
son et al. (33) report uptake of 3,3′,4,4′-tetrachlorobiphenyl
to brittle stars varied 5-fold depending on the nutritional
with labile TOC-like green macroalgae and least when
associated with lignin of terrestrial origin. It is possible that
nutritional value may influence the consistently high ab-
in our experiments. In contrast, therewerelargedifferences
their undoubtedly low nutritional value.
mentioned, a portion of BaP and PCB-52 repartitioned into
the aqueous phase in the wood and diatom systems. In
experiment III, we measured aqueous contaminant con-
centrations at the beginning and end of the feeding period
to assess the degree of repartitioning in the two extreme
system,virtuallyall(>99%) oftheBaP andPCB-52remained
sorbed to the carbon particles throughout the feeding
experiment.Calculatedlogarithmic distribution coefficients
(Kd, L/kgsorbent) ranged from 6.3 to 6.7 for both compounds.
These values are 2-3 orders of magnitude lower than those
reported by Jonker and Koelmans (14). These authors used
solid-phase extraction with polyoxymethylene (POM-SPE)
of thehighest partition coefficients to activated carbon ever
reported in the literature, by up to 2.5 log units. As these
authors suggest, phase separations using centrifugation, as
carbon particles in the aqueous phase, leading to an
underestimation of partitioning coefficients. Our purpose
here is not to precisely determine Kd but to confirm that
partitioning in the activated carbon system led to very low
In the diatom system, a significant fraction of both
compounds desorbed and was associated with the aqueous
phase. At the beginning of the feeding period, 81% of BaP
By the end of the feeding period, the percentage of sorbed
contaminants declined to 70% for BaP and 45% for PCB-52.
Calculated log Kd values at the beginning and end of the
experiments I-III. Light columns represent BaP; dark columns
represent PCB-52. Error bars show 95% confidence intervals. For
theparticles tested,absorptionefficiencyis lowestfromactivated
carbon and greatest fromwood and diatoms.
Absorption efficiency results for all particles in
VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 94553
experiment wereapproximately 3.4forBaP and 3.0forPCB-
52, several orders of magnitude below the octanol-water
partitioning coefficients for these compounds. The clams
would havefiltered all thewater in theexposurevessel once
over the 8-h exposure period, at an average filtration rate of
1.0 L/(g of dry tissue/d) (24). Given the partitioning to the
aqueous phase and the ventilation rate, it is probable that
the clams received a portion of their contaminant exposure
through the aqueous route.
We estimated potential uptake of BaP and PCB-52 from
theactivated carbon and diatom systemsusing theaqueous
contaminant concentrations measured in experiment III,
ventilation parameter values from Luoma et al. (24), and a
rangeof absorption efficiency from theaqueous phase(AEw
)0-0.6).Wangand Chow (30) calculated AEWforBaP tothe
green mussel Pe. viridis as 6.6-8.8%, and Bjork and Gilek
(34) found that PCB AEWto Mytilusedilusvaried between 10
and because the clams clearly did not remove all of the
radiolabeledcontaminantfrom solution,amaximum AEwof
0.6 was chosen. This still conservatively constrains the
assumption thatall uptakewastheresultofabsorption from
food. Equation 2, derived from the physiologically based
bioaccumulation model in Landrum et al. (35), was used to
estimate contaminant concentration in soft tissues after
depuration (Cclam-w, µg/g) that would result from aqueous
where Cw(µg/L) is the aqueous contaminant concentration
in the feeding beakers and V is the ventilation (filtration)
rate. The time of exposure (t) was 8 h. Calculated Cclam-w
in the laboratory. Estimates were made using the average of
initial and final feeding water contaminant concentrations.
The relative importance of aqueous uptake was calculated
as the ratio of estimated contaminant concentration from
aqueous uptake to the measured concentration of BaP or
PCB-52 in clam tissue.
contaminant uptake is depicted in Figure 3 for the range of
AEW. In the activated carbon test, particle ingestion would
IfAEWwere1%, 99% ofuptakewould befrom food; ifit were
system the aqueous route could account for a portion of
overall BaP and PCB-52 uptake. If AEW exceeds about 0.5,
more than 50% of the measured contaminant mass within
complexities could rearrange the order of purely dietary
absorption efficiencies from some of the particle types but
would not affect conclusions about the largest differences
(e.g., diatoms and other particle types vs activated carbon).
Sorbent Organic Carbon and Specific Surface Area.
Figure 4A,B shows that neither total organic carbon nor
specific surface area is a sufficient predictor of clam absorp-
tion efficiency for BaP or PCB-52 in these experiments. The
2-fold difference in organic carbon content among our
particle types cannot explain the 60-fold difference in
surface area did not correlate well with clam absorption
efficiency and has the highest surface area; its SSA is about
10times greater than char and 100-1000times greater than
theother particles tested. Within thelower SSA range(coke,
anthracite, peat, wood, and char), however, thereisno clear
relationship between SSA and absorption efficiency.
The OMP model assumes that amorphous, humic-like
oversimplifies the bioavailability of organic contaminants
from sediment. According to this model, TOC and bioavail-
researchers have observed the lowest accumulation factors
at low-TOC sites (36). Others showed increased nutritional
of sediment-sorbed contaminants (33, 37). Researchers
examining PAH and PCB sorption in field sediments docu-
mented significantly higher partitioning to the soot(-like)
carbon fraction than to thenormal organic carbon pool (38,
7). Our results quantify the extremely low bioavailability of
PAH and PCB from soot-like particles. To our knowledge
FIGURE 3. Estimated importance of aqueous uptake in activated
efficiencies fromwater, AEW. Solid lines represent BaP; dashed
lines represent PCB-52.
FIGURE 4. Clamabsorption efficiency is independent of (A) total
organic carbonand(B) surfacearea.Opensquares representdata
for BaP, and triangles represent data for PCB-52.
45549ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
these are also the first results that directly quantify the
extremely broad range of PAH and PCB bioavailability from
the individual particle assemblages representative of the
carbonaceous fractions of sediment.
Sorptive properties of sediment carbonaceous matter
appear to play some role in determining PAH and PCB
bioavailability, but it is not one that is simple to extrapolate
from all particletypesorto all species. Furtherinvestigation
of how such properties interact with the biological charac-
teristics affecting organic contaminant uptake by different
than the OMP model. Even if one assumes that sorption
efficiency and Kdis unlikely. Wehypothesizethat a quantity
correlated with Kd.Considering theparticlesingested by the
of particles in its feces, at equilibrium:
Mt is the mass of clam tissue, and Mc,t is the mass of
contaminant in clam tissue. Inverting the equation for
absorption efficiency and substituting mass for activity:
Substituting eq 4 into eq 3 and assuming that the ratio of
tissue mass to ingested particle mass is constant among
For the 60-fold variation in absorption efficiency observed
in our data, we would estimate a nearly 600-fold variation
in Kpc. Thus, a 1:1correlation between absorption efficiency
and Kdis probably not a reasonable assumption.
Implications. While there is more to learn about factors
about the applicability of TOC-based models to marine and
freshwater food webs and their usefulness in making policy
decisions(e.g., refs7, 31, and 36). They support researchers'
previous claims that additional sediment compartments
should beincluded in equilibrium partitioning models(11-
13). Instead of relying on traditional OMP framework to
predict risk at contaminated sites, engineers and scientists
should continue to develop models that incorporate both
absorption to natural organic matter and adsorption onto
particulate carbonaceous materials such as coke and coal.
The results of our experiments underscore the importance
of considering how the inherent grain and subgrain hetero-
geneity ofsediment could significantly impact contaminant
Our results also provide a strong basis for suggesting
activated carbon addition as a novel remediation technique
for PAH- and PCB-contaminated sediments. As shown in
these tests, BaP and PCB-52 bound to activated carbon are
up to 60 times less available to clams than the same
contaminants associated with other types of particulate
carbonaceous matter frequently found in sediment. West et
al. (39) found that the addition of Ambersorb (a synthetic
concentrations and bioavailability of eight PAHs, lending
credence to sorbent addition as a viable option for in situ
sediment remediation. Lebo et al. (40) compared the reduc-
tions in bioavailability of several hydrophobic organic
compoundsfrom laboratory-spikedsedimentby coarseand
fine versions of two different carbons using low-density
of activated carbon addition for in situ remediation of field
This work was supported under Stanford University's Bio-X
Initiative. The authors thank Sarah Rubinfeld for her as-
sistance in the laboratory and field; Dr. William Ball and
Roberta Brown for supplying the char; and Dr. Upal Ghosh
for assistance with data analyses. At the time of the
by Stanford University.
(1) Evison, L. U.S. Environmental Protection Agency, Proceedings
from A Workshop on Environmental Stability of Chemicals in
(2) Sediment Quality Criteria for the Protection of Benthic Organ-
isms: Phenanthrene; U.S. Environmental Protection Agency,
Office of Water: Washington, DC, 1993; EPA-882-R-93-014.
(3) DiToro, D. M.; Zarba, C. S.; Hansen, D. J.; Berry, W. J.; Swartz,
R. C.; Cowan, C. E.; Pavlou, S. P.; Allen, H. E.; Thomas, N. A.;
Paquin, P. R. Environ. Toxicol. Chem. 1991, 10, 1541.
(4) McGroddy, S. E.; Farrington, J. W. Environ. Sci. Technol. 1995,
(5) McGroddy, S. E.; Farrington, J. W.; Gschwend, P. M. Environ.
Sci. Technol. 1996, 30, 172.
(6) Jonker, M. T. O.; Smedes, F. Environ. Sci. Technol. 2000, 34,
(7) Bucheli, T.D.;Gustafsson, O.Environ.Toxicol.Chem.2001, 20,
(8) Luthy, R. G.; Aiken, G. R.; Brusseau, M. L.; Cunningham, S. D.;
Gschwend, P. M.; Pignatello, J. J.; Reinhard, M.; Traina, S. J.;
Weber, W. J. Jr.; Westall, J. C. Environ. Sci. Technol. 1997, 31,
2003, 37, 2209.
(10) Allen-King, R. M.; Grathwohl, P.; Ball, W. P. Adv. Water Resour.
2002, 25, 985.
P. M. Environ. Sci. Technol. 1997, 203.
(12) Accardi-Dey, A.; Gschwend, P. M. Environ. Sci. Technol. 2002,
(13) Accardi-Dey, A.; Gschwend, P. M. Environ. Sci. Technol. 2003,
(14) Jonker, M. T. O.; Koelmans, A. A. Environ. Sci. Technol. 2002,
(15) Talley, J. W.; Ghosh, U.; Tucker, S. G.; Furey, J. S.; Luthy, R. G.
Environ. Sci. Technol. 2002, 36, 477.
(16) Ma, X.;Bruner, K.A.;Fisher, S.W.;Landrum, P.F.J.GreatLakes
Res. 1999, 25, 305.
A. E.; Brownawell, B. J. Mar. Ecol. Prog. Ser. 2001, 212, 145.
(19) Decho, A. W.; Luoma, S. N. Mar. Ecol. Prog. Ser. 1991, 78, 303.
(20) Luoma, S.N.;Cain, D.;Johansson, C.Hydrobiologica 1985,129,
(21) Pereira, W. E.; Hostettler, F. D.; Luoma, S. N.; van Geen, A.;
Fuller, C. C.; Anima, R. J. Mar. Chem. 1999, 64, 99.
(22) Venkatesan, M. I.; de Leon, R. P.; van Geen, A.; Luoma, S. N.
Mar. Chem. 1999, 64, 85.
(23) Guillard, R. R. L. Culture of phytoplankton for feeding marine
W. L., Chanley, M. H., Eds.; Plenum Press: New York, 1975; pp
R. S.; Reinfelder, J. R. Environ. Sci. Technol. 1992, 26, 485.
(25) Lee, B.-G.; Luoma, S. N. Limnol. Oceanogr. 1998, 43, 1455.
(26) Penry, D. L. Environ. Toxicol. Chem. 1998, 17, 1633.
(27) Wang, W.-X.; Fisher, N. S. Environ. Toxicol. Chem. 1999, 18,
AE) 1 +Mc,p
VOL. 38, NO. 17, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 94555
(28) Kukkonen, J.; Landrum, P. F. Aquat. Toxicol. 1995, 32, 75.
(29) Karapanagioti, H. K.; Childs, J.; Sabatini, D. A. Environ. Sci.
Technol. 2001, 35, 4684.
(30) Wang, W.-X.; Chow, A. T. S. Environ. Toxicol. Chem. 2002, 21,
Toxicol. Chem. 1994, 13, 1445.
(32) Bott, T. L.; Stanley, L. J. Environ. Sci. Technol. 2000, 34, 4936.
(33) Gunnarsson, J. S.; Granberg, M. E.; Nilsson, H. C.; Rosenberg,
R.; Hellman, B. Environ. Toxicol. Chem. 1999, 18, 1534.
(34) Bjork, M.; Gilek, M. Environ. Toxicol. Chem. 1999, 18, 765.
(35) Landrum, P. F.; Lee, H. I. I.; Lydy, M. J. Environ. Toxicol. Chem.
1992, 11, 1709.
(36) Meador, J. P.; Casillas, E.; Sloan, C. A.; Varanasi, U. Mar. Ecol.
Prog. Ser. 1995, 123, 107.
(37) Gunnarsson, J. S.; Hollertz, K.; Rosenberg, R. Environ. Toxicol.
Chem. 1999, 18, 1149.
(38) Bucheli, T. D.; Gustafsson, O. Environ. Sci. Technol. 2000, 34,
(39) West, C. W.; Kosian, P. A.; Mount, D. R.; Makynen, E. A.; Pasha,
M. S.; Sibley, P. K.; Ankley, G. T. Environ. Toxicol. Chem. 2001,
(40) Lebo, J. A.; Huckins, J. N.; Petty, J. D.; Cranor, W. L.; Ho, K. T.
Chemosphere 2003, 50, 1309.
Received for review January 21, 2004. Revised manuscript
received June 2, 2004. Accepted June 10, 2004.
45569ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004