Marine Pollution Bulletin 58 (2009) 995–1006

Contents lists available at ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier .com/ locate /marpolbul

Recovery of interior brackish marshes seven years after the chalk point oil spill

Jacqueline Michel a,*, Zachary Nixon a, Jeffrey Dahlin a, David Betenbaugh a, Mark White a,Dennis Burton b, Steven Turley b

a Research Planning, Inc., 1121 Park Street, Columbia, SC 29201, United Statesb Wye Research and Education Center, University of Maryland, Queenstown, MD 21658, United States

a r t i c l e i n f o a b s t r a c t

Keywords:Oil spillWetlandsSpartina alternifloraSpartina cynosuroides

0025-326X/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.marpolbul.2009.02.015

* Corresponding author. Tel.: +1 803 256 7322; faxE-mail address: jmichel@researchplanning.com (J.

Seven years after the April 2000 spill of 140,000 gallons of a mixture of No. 6 and No. 2 fuel oils in thePatuxent River, Maryland, heavily oiled brackish marshes showed continuing effects. Stem density andstem height were significantly lower in oiled versus unoiled sites for Spartina alterniflora but not Spartinacynosuroides habitats. In contrast, belowground biomass was significantly lower in S. cynosuroides habi-tats but not S. alterniflora habitats. Total PAH concentrations were up to 453 mg/kg in surficial soils (0–10 cm) and 2921 mg/kg with depth (10–20 cm). The oil had lost 22–76% of its initial PAH content afterseven years, although the oil in marsh soils has undergone little to no additional weathering since Fall2000. Based on amphipod acute toxicity tests and sediment quality guidelines, 25% of the soils in themarsh are expected to be toxic (ESB-TUFCV values > 3.0; PMax > 0.65).

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

On 7 April 2000, an estimated 140,000 gallons of a mixture ofNo. 6 and No. 2 fuel oils were released into Swanson Creek, thePatuxent River, and downstream tributaries from a ruptured pipe-line to nearby Chalk Point Power Generating Station, Maryland. Anestimated 76 acres of brackish marsh dominated by Spartina alter-niflora, Spartina cynosuroides, and Typha spp. were affected. Clean-up methods varied widely. In the marsh adjacent to the leak site(Fig. 1), cleanup consisted of digging of trenches, low- to moder-ate-pressure and high-volume flushing, extensive manual removalusing sorbents, filling of the trenches, and replanting (Gundlachet al., 2003). In the heavily oiled interior marsh areas at the headof Swanson Creek, boardwalk pathways were used to provideworker access for recovery of the pooled oil using sorbents. Nutri-ents were applied manually and by helicopter several times in thesummer of 2000 in the interior areas of marshes east of the breaksite as part of a biostimulation program. Gundlach et al. (2003) re-ported that no cleanup was attempted in the interior marshes fur-ther to the west because of limited access.

In the Natural Resource Damage Assessment conducted by theresource trustees (Michel et al., 2002; NOAA et al., 2002), heavilyoiled interior vegetation was estimated to recover in 5–10 yearsand soils were estimated to recover in 10–20 years (shorter for S.alterniflora and longer for S. cynosuroides). Because of these pre-

ll rights reserved.

: +1 803 254 6445.Michel).

dicted long-term impacts, a study was conducted seven years afterthe initial spill to answer the following questions:

� What is the condition of the vegetation in the heavily oiledmarsh seven years post spill compared to unoiled marshes?

� What is the degree of weathering of oil in the marsh soils?� What are the sources of the PAHs in the marsh soils?� Is the oil in the marsh soils toxic? If so, at what PAH

concentrations?

2. Study methods

2.1. Study design

This study was conducted seven years after the Chalk Point oilspill, focusing on the most heavily oiled interior brackish marshhabitats of S. alterniflora and S. cynosuroides. The oil had pooledon the marsh surface, particularly in open, unvegetated areas cre-ated by muskrat grazing. We excluded the area immediately adja-cent to the pipeline break where aggressive cleanup wasconducted (Fig. 1). Sites were located in oiled marshes on eitherside of the release site where some cleanup efforts were conducted,using manual removal with sorbents by crews working fromboardwalks. Sites were also located at the very head of SwansonCreek where it was reported that little or no cleanup was at-tempted. Therefore, the results of this study are representative ofthese conditions.

The study site is characterized as a brackish-water marsh,with mixed stands of S. alterniflora, S. cynosuroides, Typha spp.,

Fig. 1. Station locations and spill location plotted on aerial photography taken in April 2000 shortly after the spill to allow visualization of the sites relative to the oiling at thetime of the spill. The four stations to the northeast were treated with nutrients in 2000. The four stations just to the northwest of the spill source were cleaned by workers onboardwalks using sorbents. The far western interior marshes were not cleaned.

996 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

and Scirpus spp., with minor occurrence of Polygonum spp. andAmaranthus spp. S. cynosuroides usually occurred as a continuousband of vegetation along creek banks and large zones through-out the marsh platform (comprising 49% of the marsh in Swan-son Creek). S. alterniflora generally occurred as smaller patchesinterspersed in the marsh (comprising 23% of the marsh inSwanson Creek), and it was harder to find mono-specific stands.Typha occurred mostly along the marsh fringe in the oiled hab-itats (comprising 27% of the marsh in Swanson Creek), so wasnot included in our study, which focused only on the interiorhabitats.

The study design consisted of an oiled versus unoiled compari-son of the following variables: vegetation health as indicated bystem density, stem height, and total (live and dead) belowgroundbiomass; oil fate and effects as indicated by polynuclear aromatichydrocarbon (PAH) concentration and weathering in soils; and tox-icity as indicated by sediment bioassay tests. A statistical analysisof the power of proposed study variables was conducted to deter-mine the minimum sample size to achieve 80% power given postu-lated effects sizes at the a = 0.2 significance level. This level ofsignificance was judged adequate given the uncertainty about ef-fects sizes and the effort available. The final distribution of sitesconsisted of:

S. alterniflora: 12 unoiled sites (site name AR for S. alternifloraand reference)10 oiled sites (site name AH for S. alterniflora and heavy oiling)S. cynosuroides: 12 unoiled sites (site name CR for S. cynosuroidesand reference)14 oiled sites (site name CH for S. cynosuroides and heavyoiling)

2.2. Site selection

Color infrared digital aerial photographs were classified intothree emergent marsh vegetation classes dominated by S. alternifl-ora, S. cynosuroides, and Typha spp. Oiled marsh areas were digi-tized from imagery acquired immediately after the spill. Ninerandom site locations were generated in oiled and unoiled interiorareas for both marsh types. These nine new sites, plus three sites ineach marsh type from previous investigations, were pooled to-gether for each group. The area east of the break site, where inten-sive cleanup methods were used, was excluded from this study.

Some of the pre-selected unoiled sites were located more thanseveral hundred meters from channels; in contrast, the oiled sitestended to be closer to marsh channels. Thus, some of the unoiledsites were located in the pre-selected areas but not as deep intothe marsh. Distance to the nearest tidal channel was 17.9 meters(m) for the unoiled sites and 12.1 m for the oiled sites. A nonpara-metric test (Wilcoxon Oneway Analysis) was used to determinethat the sites were not significantly different with respect to dis-tance from the nearest tidal channel, with a Z value of 0.0650.

If the vegetation was dominated by either S. alterniflora or S.cynosuroides, the site was used for that marsh type. If the sitehad a mixed species assemblage, the site was moved to the nearestpoint with a homogenous plant community dominated by the spe-cies of interest. For two of the S. alterniflora pre-selected sites, all ofthe adjacent vegetation was dominated by S. cynosuroides so theywere not moved, thus the final distribution of 14 oiled S. cynosuro-ides and 10 oiled S. alterniflora.

The oiled sampling sites are shown in Fig. 1, overlain onto theApril 2000 (shortly post spill) vertical aerial photograph. Unoiledsites were in adjacent creeks both north and south of the oiled

J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006 997

creek and in very similar settings. Salinities in these creeks rangefrom 7.04‰ in May to 11.6‰ in October. This part of the PatuxentRiver has large temporal and spatial variations in salinity; 24-hvariations in salinity at Benedict (near Swanson Creek) are 2–3.7‰ (Anderson et al., 1968).

2.3. Field methods

Stem height was measured for the five tallest individuals withina 1-m2 quadrat, and stem density was the number of stems of liveplants of each species within the quadrat. All coring and samplingequipment for sediment and toxicity samples was cleaned withmethylene chloride prior to each use. Toxicity samples were acomposite of the 0–10 centimeter (cm) intervals from two cores(occasionally three cores when needed to generate the required2 liters [L] of sediment) from opposite corners. Vegetative materialand animals were manually removed and the soil homogenized inthe field. A chemistry sample representing the 0–10 cm intervalwas taken from the toxicity container. Therefore, the 0–10 cm sam-ple for chemical analysis was a split of the well-homogenized sam-ple for toxicity testing. The 10–20 cm sample for chemical analysiswas a single sample from the center of the second core.

Belowground biomass cores (16-cm diameter) were extractedfrom the remaining corners of the quadrat. Each core was dividedinto two intervals, 0–10 cm and 10–20 cm and placed in Zip-Locbags. All samples were placed on ice upon completion of samplingat the site. Biomass samples were processed at the end of each dayby manual agitation and sieving with a 1.20-millimeter (mm)stainless steel mesh.

3. Laboratory analysis methods

Biomass samples were transferred into paper bags and placed ina 60 �C oven for a minimum of 3 h, cooled to room temperatureand weighed. This process was repeated until the weight was with-in 1 g of the previous weight.

PAHs were analyzed by gas chromatography/mass spectrome-try in the selected ion monitoring mode by B&B Laboratories (Col-lege Station, Texas) using standard operating procedure B&B1006.The analytes included 51 PAHs, 5 individual alkyl isomers, and 3hopanes/triterpane. The quality control (QC) criterion were as fol-lows: for blanks, no more than two target analytes exceeded 3X themethod detection limits; spike recoveries were between 40% and120%; the relative percent difference (RPD) for valid spiked dupli-cates was ±30%; RPDs for valid duplicates was ±30%; sediment SRMwas ±20% the laboratory derived mean; and reference oil SRM was±15% the laboratory derived mean. Surrogate solutions equivalentto 5–10X the method detection level were prepared for varioushydrocarbon analyses. The appropriate surrogate solution wasadded to every sample including QC samples. The data were cor-rected based on surrogate recovery up to 100%. The QC criteriafor surrogate recoveries were between 40% and 120%, exceptd12-perylene which must be greater than 10% and less than120%. Because many of the samples contained high amounts ofoil, extracts required dilution prior to instrumental analysis. Surro-gates were re-added to the diluted sample prior to analysis.

4. Toxicity study methods

Ten-day survival whole soil toxicity tests with the amphipodAmpelisca abdita were conducted on ten oiled soil samples thatrepresented the range of PAH concentrations (plus two referencesamples and an in-house control sediment). Tests were conductedvia Test Method 100.4:A. abdita, Eohaustorius estuarius, Leptocheirusplumulosus, or Rhepoxynius abronius 10-day Survival Test for Sedi-ments given in Section 11 of USEPA (1994).

Soil samples were press-sieved through a 1-mm mesh stainlesssteel screen to remove any indigenous organisms and debris notremoved during field collection. The test acceptability criterionfor the toxicity tests was a minimum mean control survival of90%. The soils from each marsh type were run as separate experi-mental units.

The amphipods used for the initiation of the 10-day tests were 3–5 mm. Organisms were purchased from Aquatic Research Organisms(Hampton, New Hampshire). Upon receipt of the organisms, amphi-pods were placed in in-house control sediment with 28‰ overlyingsea water (Instant Ocean�) and fed Isochrysis galbana at a density of�1 � 106 cells/milliliter (mL) twice daily for three days prior to test-ing. The organisms exhibited swimming behavior upon placement inthe holding containers and were opalescent pink in color.

Static tests were conducted in 1-L beakers containing 175 mLsediment and 800 mL water. Twenty organisms were randomlyplaced in each of five replicate test chambers. The water in the testchambers was well aerated Instant Ocean� at a salinity of 28‰.Tests were conducted in a water bath at 20 ± 2 �C under a constantphotoperiod (24-h L:0-h D) at 500–1000 lux. The animals were notfed during testing.

Temperature, pH, salinity, and dissolved oxygen were measuredat the beginning and end of the test. At the end of the 10-day expo-sures, the test materials were sieved through a 0.5 mm sieve to col-lect surviving organisms. Since A. abdita are tube-builders, thesieves were ‘‘slapped” forcefully against the surface of the waterto ensure that all of the amphipods and tubes were dislodged. Allliving animals were counted. Immobile organisms isolated beforesieving or from sieved material were considered dead. Percent sur-vival data were arc sine squared root transformed prior to statisticalanalyses. Survival of the unoiled soils was 98% and 99%, well withinthe test acceptability criterion of 90% or greater control survival.

PAHs were measured in the marsh soils to indicate the potentialtoxicity to benthic organisms, using two different approaches:

1. Equilibrium Partitioning Sediment Benchmark Toxic Unit FinalChronic Value (ESB-TUFCV) for PAH mixtures for the protectionof benthic organisms (USEPA, 2003) based on 34 PAHs in sedi-ments. The 34 PAHs summed in the calculation of this valueinclude most of the PAHs measured in the samples with theexception of benzothiophene, dibenzothiophene, and naph-thobenzothiophene (and their alkylated homologues), and C1-and C2-fluoranthenes/pyrenes, which are minor contributorsto the total PAHs in the samples. Two values are reported forthe range of reported total organic carbon (TOC) concentrationsfrom previous studies at the sites because TOC was not mea-sured in the individual samples collected in 2007.

2. Field et al. (2002) and USEPA (2005) developed a series of logis-tic regression models for 22 individual PAHs that quantify rela-tionships between the concentrations of sediment-associatedcontaminants and toxicity to the marine amphipods, A. abditaand R. abronius. In contrast to the sediment quality guidelines,logistic regression models provide specific sediment concentra-tions (Tp values) that would result in specific probabilities ofobserving toxicity (e.g., 20%, 50%, and 80%). PMax is the maxi-mum predicted PAH toxicity probability for a sample.

5. Results

5.1. Marsh soil chemistry

5.1.1. Total PAH concentrationsTotal PAH concentrations in marsh soils at each site are shown

in Table 1, along with field observations on oiling and a descriptionof the amount of oil inside the Zip-Loc bags used to store thebiomass samples. Oil could be seen along roots, rhizomes, and/or

Table 1Total PAHs in marsh soils by site.

Station Total PAH (mg/kg) Field observations

ACH-1 (0–10 cm) 45.8 Silver and rainbow sheen on water table; light stain on both biomass bagsACH-1 (10–20 cm) 1.90AH-1 (0–10 cm) 178.4 Black oil droplets on water table; biomass bag 0–10 cm moderate stain, 10–20 cm heavy

stainAH-1 (10–20 cm) 1011AH-2 (0–10 cm) 43.9 Open area with very stunted S. alterniflora; large patch of Scirpus in upper right; both

biomass bags no visible oilAH-2 (10–20 cm) 1.8AH-4 (0–10 cm) 2.07 No visible oil; both biomass bags no visible oilAH-4 (10–20 cm) 0.62AH-5 (0–10 cm) 156.4 Light rainbow sheen on water table; biomass bags no visible oilAH-5 (10–20 cm) 5.33AH-6 (0–10 cm) 8.51 Black oil 7–20 cm in core; black oil droplets on water table; biomass bag 0–10 cm clean,

10–20 cm heavy stainAH-6 (10–20 cm) 2921AH-7 (0–10 cm) 1.62 Silver sheen on water table; biomass bags no visible oilAH-7 (10–20 cm) 1.29AH-8 (0–10 cm) 151.1 Both biomass bags light stainAH-8 (10–20 cm) 188.8AH-11 (0–10 cm) 5.87 No visible oil; both biomass bags no visible oilAH-11 (10–20 cm) 2.1AH-12 (0–10 cm) 41.0 Black oil 10–20 cm in cavities; biomass bag 0–10 cm light stain, 10–20 cm heavy stainAH-12 (10–20 cm) 2084AR-5 (0–10 cm) 0.82 No visible oil; biomass bags no visible oilCA-1 (0–10 cm) 5.52 No visible oil; biomass bags no visible oilCA-1 (10–20 cm) 3.82CAH-10 (0–10 cm) 264.2 Both biomass bags light stainCAH-10 (10–20 cm) 0.68CAH-9 (0–10 cm) 9.9 Both biomass bags light stainCAH-9 (10–20 cm) 266.5CH-2 (0–10 cm) 4.72 Biomass bag 0–10 cm light stain, 10–20 cm heavy stainCH-2 (10–20 cm) 255.6CH-3 (0–10 cm) 5.07 Moved site 3 m to north because Scirpus dominated former quadrat sites; both biomass

bags no visible oilCH-3 (10–20 cm) 1.98CH-4 (0–10 cm) 0.68 No visible oil; both biomass bags no visible oilCH-4 (10–20 cm) 1.07CH-5 (0–10 cm) 10.85 Black oil in cavities 2–30 cm; biomass bag 0–10 cm moderate stain; 10–20 cm heavy stainCH-5 (10–20 cm) 796.7CH-6 (0–10 cm) 2.18 Black oil in cavities 0–30 cm in biomass core; no visible oil in chemistry cores; biomass

bag 0–10 cm v. light stain; 10–20 cm moderate stainCH-6 (10–20 cm) 23.52CH-7 (0–10 cm) 3.36 No visible oil; biomass bags no visible oilCH-7 (10–20 cm) 0.82CH-8 (0–10 cm) 91.86 Black oil 0–20 cm in first core; black oil in both biomass cores; biomass bag 0–10 cm

moderate stain; 10–20 cm very heavy stainCH-8 (10–20 cm) 729.4CH-9 (10–20 cm) 0.58 No visible oil in first 2 cores; biomass core rainbow sheen 10–20 cm; biomass bags no

visible oilCH-9 (10–20 cm) 0.91CH-10 (0–10 cm) 453.4 Biomass bag 0–10 cm moderate stain; 10–20 cm heavy stainCH-10 (10–20 cm) 11.09CH-11 (0–10 cm) 139.8 Black oil droplets on water table; biomass bag 0–10 cm moderate stain; 10–20 cm heavy

stainCH-11 (10–20 cm) 1384CH-12 (0–10 cm) 3.43 No visible oil but oily smell in first 2 cores; biomass core 0–10 cm rainbow sheen; both

biomass bags no visible oilCH-12 (10–20 cm) 0.64CR-11 (0–10 cm) 0.78 No visible oil; both biomass bags no visible oil

1 Pyrogenic Index = acenaphthelene + acenaphthene + anthracene + fluoranth-ene + pyrene + benz(a)antrhacene + benzofluoranthenes + benzopyrenes + perlyl-ene + dibenzoantrhacene + benzoperylene/C0–C4naphthalenes + C0–C3fluorenes + C0–C4phenanthrenes + C0–C4dibenzothiophenes + C0–C4chrysenes.

998 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

cavities. Total PAH concentrations varied widely, with depth, with-in a site, and among sites. We often noted in the field that theamount of visible oil was quite different among the 4–5 cores col-lected at a site. Because we followed a standard protocol in thesampling order (e.g., toxicity and chemical samples first, then thebiomass samples), sometimes the sample for chemical analysishad no visible oil whereas the biomass sample was visibly oiled.Therefore, the degree of oiling at a site is not always indicated bythe total PAHs measured in the soil sample analyzed for chemistry.

For most of the oiled cores, total PAHs varied by 1–2 orders ofmagnitude between the top and bottom intervals, reflecting highheterogeneity in oil distribution in the marsh soils, as well as thefact that the top sample was homogenized from 2 to 3 cores andthe bottom sample was a single sample from 1 core. However,there was a tendency for higher PAHs with depth; all of the sixsamples with over 500 milligrams per kilogram dry weight (mg/kg) total PAHs were from the 10–20 cm interval, equally from bothspecies. The maximum total PAH value in soils was 2921 mg/kg forsite AH-6 at 10–20 cm.

Total PAH concentrations were not normally distributed; themedian value was 7.19 mg/kg and the mean was 236 mg/kg, witha coefficient of variation of 238%. To reduce the effect of largedifferences in vertical distribution, we averaged the two intervalsto create a virtual 0–20 cm core, which reduced the coefficient ofvariation from 238% to 162%.

5.1.2. Oil source analysisTwo ratio approaches were used to infer sources of PAHs in the

marsh soils: The Pyrogenic Index,1 which is the ratio of the PAHsdominating in by-products from the combustion of fossil fuels tothe alkylated homologues that dominate in crude oil and refinedproducts, of Wang et al. (1999); and the fluoranthene/fluoranth-

Fig. 2. Petrogenic Index versus fluoranthene/fluoranthene + pyrene ratio in oiledmarsh soils, for different PAH concentration ranges. Samples with >8 mg/kg totalPAHs are dominated by oil from the spill; samples with <1 mg/kg containbackground pyrogenic PAHs; and samples with 1–8 mg/kg contain a mixture ofpyrogenic and petrogenic PAHs.

J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006 999

ene + pyrene ratio, Fl/(Fl + Py), which increases with the temperatureof combustion due to the thermodynamic stability of fluorantheneover pyrene (Yunker et al., 2002). Nine samples had total PAHs<1 mg/kg: two from unoiled sites, six from the deeper interval inthe core, and one from the surface interval. As shown in Fig. 2, theFl/(Fl + Py) ratio for samples with <1 mg/kg total PAH ranged from0.52 to 0.57; a ratio >0.50 indicates dominance by combustion ofbiomass and coal (Yunker et al., 2002), consistent with the presenceof the coal-fired power plant at Chalk Point. For these samples, thePyrogenic Index was 0.38–0.68, indicating that PAHs were domi-nantly from combustion sources. This index was <0.50 for samplesthat contained 10–30% naphthalenes, with the alkylated homologuesbeing most abundant, likely reflecting the heavy use of motorizedwatercraft in the area.

Seventeen samples had total PAHs from 1 to 8 mg/kg; thesesamples contained mixtures of both petrogenic and pyrogenichydrocarbons. The Fl/(Fl + Py) ratio in these samples ranged from0.30 to 0.45 (Fig. 2); Yunker et al. (2002) determined that a ratioof 0.40–0.50 indicated PAHs derived from liquid fuel combustionand <0.40 indicated PAHs derived from grass, wood, or coal com-bustion. The Pyrogenic Index for these samples ranged from 0.11to 0.43, further indicating that these PAHs originated from a vari-ety of sources.

Samples with more than 8 mg/kg total PAH had a Fl/(Fl + Py) ra-tio of 0.1–0.29; the source oil had a ratio of 0.16 (Fig. 2). The Pyro-genic Index for these samples ranged from 0.05 to 0.18; the sourceoil had a Pyrogenic Index of 0.03. These 24 samples, half of thosecollected, very clearly contained petroleum hydrocarbons derivedfrom fossil fuels, with a good match to the source oil (discussedin detail in the study report, Michel et al., 2008).

5.1.3. Oil weathering trendsAnalysis of PAH weathering trends was conducted for the 24

samples that contained more than 8 mg/kg because the PAHs inthese samples were dominated by petroleum hydrocarbons fromthe spilled oil, rather than complex mixtures from multiplesources. PAH depletion ratios were calculated following Douglaset al. (1996) using C2-chrysene as the conserved internal markerwithin the oil to act as a standard. The individual PAH depletion,corrected for oil loss, was determined by:

% PAH depletion ¼ ½1� ððPAH1=PAH0Þ� ðC2-chrysene0=C2-chrysene1ÞÞ� � 100

where PAH1 is the concentration of individual PAHs in the marshsoil sample, and PAH0 is the concentration of individual PAHs inthe source oil. C2-chrysene0 is the concentration of this PAH inthe source oil and C2-chrysene1 is the concentration of this PAHin the marsh soil sample. For the source oil, we used the averagePAH concentration of three source oil samples collected directlyfrom the pipeline break right after the spill, analyzed by the samelaboratory (B&B Laboratories) in 2000. The results, shown inFig. 3, indicated that the oil had lost 22–76% of its initial PAH con-tent in the seven years since the spill, with the exception of onesample (with 11.09 mg/kg) that had a negative depletion ratio, indi-cating that it contained other sources of PAH. There was a cleartrend with increased PAH depletion with decreasing total PAHconcentration.

Another factor affecting oil weathering rates is depth of pene-tration into the marsh soils. Fig. 3 shows that the oil at 10–20 cmwas generally less weathered than the top 0–10 cm. All sampleswith less than 55% total PAH depletion (indicating that they wereless weathered) were from the deeper interval of 10–20 cm. Therate of degradation of different groups of PAH compounds followedthe pattern noted by others (e.g., Douglas et al., 1996), namely thedegradation rate decreases with ring size. Naphthalenes (2-ringedPAHs) were the most degraded with 19 of the 24 samples at 90–99% depletion. Four samples with 60–80% naphthalene depletioncontained >40 mg/kg PAHs. The naphthalenes are the most volatileand the most readily degraded of all the PAH and, with naphtha-lenes comprising 50% of the total PAHs in the source oil, their deg-radation drives the total PAH depletion ratio.

The phenanthrenes/anthracenes (3-ringed PAH) were 25–80%depleted, although there was a broad range in the depletion ratioamong samples. Degradation of these compounds was also slowerwith depth; eight of the eleven deeper cores had depletion ratiosfor phenanthrenes/anthracenes of less than 35%. The depletion ra-tio for fluoranthenes/pyrenes and for chrysenes (4-ringed PAH)were generally less than 10% after seven years post spill for bothgroups of compounds.

5.2. Vegetation condition

Vegetation condition was indicated by three variables: stemdensity, stem height, and total belowground biomass. For eachvegetation variable, a two-sample pooled variance t-test was car-ried out comparing plot group means between oiled and unoiledareas. The data were assumed to be normally distributed, indepen-dent, and of equal variance. Where evidence for unequal variancesexisted, t-tests using unpooled variance were carried out. Sum-mary statistics for all variables are shown in Table 2; also includedare the data on stem density and stem height measured in 2000and 2001 (Michel et al., 2002).

Stem density of S. alterniflora in unoiled sites averaged223 stems/m2 in 2007 and was similar to that measured in 2001.In oiled sites, stem density in 2007 was averaged 141.2 stems/m2

and was reduced by 37% compared to that of unoiled sites. Averagestem density was significantly different (t = 2.999, df = 20, p =0.007) between oiled and unoiled sites in 2007. This decreasein stem density in S. alterniflora habitats has persisted since2001.

Stem density of S. cynosuroides in the unoiled sites averaged94 stems/m2 in 2007 and was similar to that measured in 2001but higher than in 2000. In the oiled sites, stem density in 2007was slightly higher (5% increase over the unoiled sites), averaging99.1 stems/m2. Average S. cynosuroides stem height was not statis-tically different between oiled and unoiled sites in 2007. Stem den-sity in the oiled S. cynosuroides marshes increased since 2000; inthe unoiled sites it initially decreased then remained steady, sug-gesting recovery of this metric.

-10

0

10

20

30

40

50

60

70

80

90

100

AH-2

(0-1

0 cm

)

CAH

-9 (0

-10

cm)

CH

-6 (1

0-20

cm

)

AH-6

(0-1

0 cm

)

CH

-8 (0

-10

cm)

AH-8

(10-

20 c

m)

CH

-5 (0

-10

cm)

ACH

-1 (0

-10

cm)

CH

-10

(0-1

0 cm

)

CH

-11

(0-1

0 cm

)

CH

-5 (1

0-20

cm

)

AH-5

(0-1

0 cm

)

CAH

-10

(0-1

0 cm

)

AH-1

2 (0

-10

cm)

AH-1

(0-1

0 cm

)

CH

-2 (1

0-20

cm

)

AH-8

(0-1

0 cm

)

AH-1

(10-

20 c

m)

CAH

-9 (1

0-20

cm

)

CH

-11

(10-

20 c

m)

AH-1

2 (1

0-20

cm

)

AH-6

(10-

20 c

m)

CH

-8 (1

0-20

cm

)

Dep

letio

n fo

r Tot

al P

AH

s (%

)

Station

0-10 cm 10-20 cm

CH

-10

(10-

20 c

m)

Fig. 3. Total PAH depletion ratios in samples collected in 2007 that contained >8 mg/kg.

1000 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

Stem height of S. alterniflora in unoiled sites averaged 125 cm in2007 and was slightly higher than in 2000. In oiled areas, stemheight averaged 106.3 cm, a reduction of 15% compared to unoiledsites. This decrease compared to unoiled sites has persisted since2001. There is moderate evidence (t = 2.286, df = 20, p = 0.033) forsignificant differences in average between oiled and unoiled sites.Due to evidence for unequal variances (Brown–Forsythe f = 4.559,df = 1, p = 0.045), testing using the unpooled variances was per-formed. Testing unpooled variances only slightly weakened(t = 2.181, df = 13.909, p = 0.047) the original conclusion.

Stem height of S. cynosuroides in unoiled sites averaged 208 cmin 2007, which is similar to that measured in 2000, but higher thanin 2001. In oiled sites, stem height was slightly lower (6% decreasecompared to unoiled sites), averaging 196.5 cm. Average stemheight was not statistically different between oiled and unoiledsites in 2007. Stem height has increased since 2000, suggestingrecovery of aboveground vegetation in the heavily oiled S. cynos-uroides habitats.

Belowground biomass in S. alterniflora habitats was similar forboth oiled and unoiled areas at both 0–10 cm or 10–20 cm depthin 2007. For sites in S. cynosuroides habitats, there is weak to mod-erate evidence for significant differences in belowground biomassbetween oiled and unoiled areas at the 0–10 cm (t = 1.984,df = 20, p = 0.059) depth range and moderate evidence at the 10–20 cm (t = 2.334, df = 20, p = 0.028) depth range. Both intervalswere about 20% lower in the oiled sites compared to the unoiledsites.

5.3. Soil toxicity studies

Copper and nickel were measured in the soil samples used inthe toxicity tests because the Chalk Point power plant had beenimplicated in early studies as a source of localized copper pollutionfrom copper and nickel condenser tubing (Wright and Zamuda,1991). The lowest concentration for both metals had the highestsoil toxicity, which correlated with the highest total PAH concen-trations. Copper concentrations did not exceed the Effects Range-

Low of Long et al. (1998). Nickel concentrations did not exceedthe Effects Range-Medium in any of the marsh samples. Copperand nickel concentrations did not exceed T80 or T50 toxicity prob-ability concentrations in any marsh sample. Copper did not exceedthe T20 in any marsh sample. These multiple lines of evidence donot suggest a meaningful contribution of copper and nickel to tox-icity of marsh soils.

Mean amphipod survival in the two unoiled soils was 98% and99% for S. alterniflora and S. cynosuroides, respectively. In-housecontrol survival was 96%. In Table 3, the soil toxicity results arecompared with two sediment quality guidelines used to screencontaminants in sediments for potential toxicity.

TOC concentrations were not measured in the samples collectedin 2007; however, TOC measurements of samples of oiled marshsoils in 2000 ranged from 8.7% to 11.6% (B&B Laboratories, 2001,unpublished data) and in 2002 ranged from 5.6% to 26.8% (Men-delssohn and Slocum, 2004). The actual TOC concentration in aspecific sample significantly affects the calculated toxicity, thuscalculations were made using the minimum and maximum valuesfor TOC measured in previous samples from the site.

The ESB-TUFCV predictions of toxicity with TOC of 5.6% in themarsh soils closely matched the measured amphipod toxicity. Val-ues less than 1.0 are not expected to have toxic effects, and all thesamples with ESB-TUFCV values less than 1.0 had >90% survival. Thefour samples with <3% survival had ESB-TUFCV values of 2.35–7.70.Two samples had intermediate survival: CH-8 had 66% survival andESB-TUFCV of 1.45, and AH-5 had 85% survival and ESB-TUFCV of2.56. The ESB-TUFCV values calculated using the upper value of26.8% TOC at did not correlate well with the actual measured tox-icity in the soils. Only one sample was >1.0, even though toxicitywas observed in six of the samples. The TOC values for previoussamples include the roots and rhizomes; therefore, the lowerTOC value of 5.6% is likely to be representative of the fine-grainedorganic carbon associated with the soils that affects the dissolutionof PAHs in pore water.

Fig. 4 shows the calculated ESB-TUFCV values for the 26 ‘‘virtual0–20 cm cores” using the low and high TOC values; however, the

Table 2Summary statistics for stem density, stem height, and belowground biomass for heavily oiled and unoiled areas in S. alterniflora and S. cynosuroides dominated habitats.

Date No. of Sites Mean stem density (#/m2) Mean stem height (m) Mean belowground biomass 0–10 cm (g/m2)

Mean belowground biomass 10–20 cm (g/m2)

Oiled Unoiled Oiled Unoiled Oiled Unoiled Oiled Unoiled

S. alterniflora7/2000 3 261 150 1.30 1.17 – – – –7/2001 3 95 233 1.02 1.02 – – – –8–9/2007 10–12 141* 223 1.06* 1.25 3163 3343 3291 31958–9/2007 10–12 CV = 46% CV = 28% CV = 23% CV = 11% CV = 22% CV = 33% CV = 27% CV = 32%

S. cynosuroides7/2000 3 45 141 1.56 2.09 – – – –7/2001 3 77 94 1.35 1.62 – – – –8–9/2007 12–14 99 94 1.96 2.08 3784* 4823 3670* 46188–9/2007 12–14 CV = 39% CV = 36% CV = 19% CV = 15% CV = 37% CV = 26% CV = 27% CV = 23%

CV = coefficient of variation.* significant difference between oiled and unoiled sites in 2007.

Table 3Comparison of sediment quality guidelines for PAHs in marsh soils. Samples with ESB-TUFCV > 1 are bolded.

S. alterniflora AR-5 (Ref.) AH-11 AH-6 AH-2 AH-5 AH-1

Amphipod survival (%) 98 100 97 96 85 0Total PAHs (mg/kg) 0.82 5.87 8.51 43.9 156.4 178.4PMax (predicted probability of toxicity) 0.16 0.28 0.31 0.57 0.73 0.80ESB-TUFCV (5.6% TOC) 0.02 0.09 0.13 0.61 2.56 3.17ESB-TUFCV (26.8% TOC) 0.00 0.02 0.03 0.13 0.53 0.66

S. cynosuroides CR-11 (Ref.) CH-6 CH-3 CH-5 CH-8 CH-11 CAH-10 CH-10

Amphipod survival (%) 99 94 97 91 66 3 1 0Total PAHs (mg/kg) 0.78 2.18 5.07 10.85 91.86 139.7 264.2 453.4PMax (predicted probability of toxicity) 0.17 0.23 0.35 0.29 0.64 0.70 0.80 0.85ESB-TUFCV (5.6% TOC) 0.02 0.03 0.07 0.17 1.45 2.35 4.31 7.70ESB-TUFCV (26.8% TOC) 0.00 0.01 0.02 0.04 0.30 0.49 0.90 1.61

J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006 1001

values for the 5.6% TOC are considered to be the better predictor oftoxicity. ESB-TUFCV values for the oiled soils, assuming 5.6% TOC,ranged from 0.01 to 30.48, with the following distribution and pre-dicted effects based on the toxicity test results:

� 54% (13/24) were <1.0; these soils were not toxic in amphipodtoxicity tests and are not expected to exhibit toxic effects.

� 8% (2/24) were between 1.0 and 2.0; one toxicity test in thisrange had 66% amphipod survival, thus these soils could haveeffects on sensitive organisms.

Fig. 4. ESB-TUFCV values (the sum of 34 PAHs averaged for the 0–20 cm interval) for threported for the marsh soils.

� 38% (9/24) were between 2.0 and 3.0; the two toxicity tests inthis range had 3% and 85% amphipod survival, thus these soilsare expected to have significant effects on sensitive organismsand effects on many organisms.

� 25% (6/24) were >3.0; the three toxicity tests with values greaterthan 3.0 had 0–1% amphipod survival; these soils are expectedto be toxic to many organisms.

In Fig. 5, plots of the percent amphipod survival test results ver-sus the calculated ESB-TUFCV values (top) and the maximum pre-

e 26 marsh soil samples using both the low and high range of TOC concentrations

Fig. 5. Amphipod survival in the bioassay tests versus ESB-TUFCV values calculatedas the sum of 34 PAHs (top) and the maximum predicted probability (PMax) toxicityusing the logistic modeling regression as the sum of 22 PAHs (bottom), for thevirtual 20-cm cores by species, using 5.6% TOC.

Fig. 6. Maximum predicted probability (PMax) toxicity values for the 26 marsh soilsamples (assuming 5.6% TOC) for the individual 0–10 cm and 10–20 cm intervalsand the virtual 0–20 cm interval, listed in order of increasing PMax for the 0–20 cmvirtual interval.

1002 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

dicted probability (PMax) of toxicity using the logistic modelingregression (bottom) for the ‘‘virtual 20-cm” cores clearly demon-strate the utility of both approaches, particularly with such a lim-ited dataset. The logistic modeling regression approach accuratelypredicted the toxicity observed in the S. cynosuroides soils for PMax,though overestimated the toxicity of some of the moderately con-taminated S. alterniflora soils. The overestimation may be attribut-able to the uncertainty in the actual TOC concentrations in theSpartina-dominated marsh soils. Tp values were derived from alarge database in which the average TOC concentration was1.97% (USEPA, 2003). Fig. 6 shows the calculated PMax values (cal-culated at 5.6% TOC) for the marsh soils for which PAHs were mea-sured, with different bars for the 0–10 cm, 10–20 cm and the‘‘virtual 0–20 cm” intervals. Six of the 24 samples (25%) had a PMax

value greater than 0.65. According to Field (pers. comm., 2008),PMax values = 0.65 had approximately 50% incidence of toxicity tomarine amphipods in the national database used by Field et al.(2002) to derive the logistic regression model for sediment chem-ical mixtures. These same six samples also had ESB-TUFCV values>3.0. Thus, the soils at 25% of the sampling sites contained PAHsat concentrations that pose significant toxic risks to sensitive ben-thic organisms.

6. Discussion

Based on analysis of the PAH results, marsh soil samples can bedivided into three groups:

1. Those with <1 mg/kg total PAH that are dominated by pyro-genic hydrocarbons plus naphthalenes and represent‘‘background.”

2. Those with 1–8 mg/kg total PAH that are a mixture of multiplesources of pyrogenic and petrogenic hydrocarbons, includingthe source oil from the Swanson Creek spill.

3. Those with >8 mg/kg that are dominated by petrogenic hydro-carbons that match the source oil from the Swanson Creek spill.

Half of the samples (24 out of 48) collected in 2007 fall into thethird group, with about equal representation from both 0–10 cmand 10–20 cm intervals. Highest total PAH concentrations werein soils from 10–20 cm, with the six samples containing>700 mg/kg being from the lower interval. At nine oiled sites, totalPAH concentrations increased with depth, usually by 1–2 orders ofmagnitude, indicating lower overall weathering rates with depth.At six oiled sites, total PAH concentrations decreased with depth,with concentration in the lower interval below 8 mg/kg, indicatingthat little oil penetrated beyond 10 cm into soils at these sites. Fourof the sites were near the edge of the oiled areas, so perhaps theyreflect lower initial oil loading to the surface. Five of these siteswere in areas where there was no cleanup. No differences in PAHconcentrations, distribution with depth, or degree of weatheringwere detected in samples from marshes that were manuallycleaned or nutrients applied versus those that were reportedlynot cleaned (at the very head of Swanson Creek), probably becauseof the high degree of spatial variability in the oil distribution bothvertically and horizontally in the marsh soils.

Four sites in the oiled area had low total PAHs (less than about2 mg/kg in both depth intervals) and no visible oil or only rainbowsheens observed in the field: AH-4, AH-7, CH-4, and CH-9. All ofthese sites were located close to the edge of visibly oiled areas(Fig. 1) and likely were less oiled initially. Five sites in the oiledarea had total PAHs in both core intervals that did not exceed6 mg/kg and no visible oil or only rainbow sheen observed in thefield: AH-11, CA-1, CH-3, CH-7, and CH-12. Two of these sites(CA-1 and CH-3) were established in 2000 in obviously oiled areas,and chemical analysis of the soils showed high levels of oil andPAH contamination in both 2000 and 2001 (Michel et al., 2002).It is likely that the oil at these sites has undergone significant nat-

Fig. 7. Mean and standard deviation PAH depletion ratios for total PAH, naphtha-lenes, and phenanthrenes/anthracenes in the Swanson Creek oiled marshes in Fall2000 (n = 9), Summer 2001 (n = 14), and Summer 2007 (n = 24) that contain morethan 8 mg/kg.

J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006 1003

ural removal, that there is significant spatial variation in the distri-bution of the oil in the marsh soils, or a combination of both.

In a study of the effectiveness of nutrient addition in stimulat-ing microbial degradation of the oil in two marsh areas, Entrix(2002) sampled the top 5 cm at six stations twice during 2000 (Julyand September) and at 21 stations five times during 2001 (May,June, August, September, and October). Total PAH concentrationsin 2000 were 1800–2600 mg/kg, whereas they averaged 482 mg/kg in October 2001 (range 1.34–3068 mg/kg). Entrix (2002) notedthat most of the compositional weathering of the oil occurred bySeptember 2000, in the first six months after the spill. This weath-ering, dominated by loss by evaporation of the semi-volatile PAHsbut also some microbial degradation, likely occurred while the oilwas pooled on the marsh surface. Fig. 7 shows the average andstandard deviations for depletion ratios for total PAHs and individ-ual PAH groups for the samples collected in 2000 and 2001 and

Fig. 8. Comparison of total PAH depletion ratios

those from 2007 that contain greater than 8 mg/kg. By Spring2001, there were few changes in the overall degree of PAH weath-ering (compared to September 2000). By October 2001, there wasagain little change in overall PAH weathering or depletion ratios,but there was more variability within areas. There were no differ-ences in PAH weathering in the surficial soils between those sitestreated with nutrients compared to those without nutrient addi-tion, indicating that nutrients were not a limiting factor in micro-bial degradation in these marshes. Although there are manydifferences in the sampling locations and depths for these threeperiods, the results support the conclusion that the oil in marshsoils has undergone little to no additional weathering since Fall2000.

Fig. 8 shows a plot of total PAH% depletion versus total PAH con-centrations in two sets of marsh soil samples: 21 samples from 0 to5 cm collected in October 2001 as part of the nutrient biostimula-tion monitoring and 24 samples collected in August/September2007 that exceeded 8 mg/kg in this study from both 0–10 cm and10–20 cm intervals. It should be noted that 15 of the 21 samplescollected in 2001 were from the same general areas sampled in2007. Even though they are from different depths for the two sam-pling periods, the similarity in the degree of PAH weathering isremarkable. Both sample sets exhibit a clear relationship betweentotal PAH concentrations and the degree of PAH degradation. Thereappears to be more degradation of moderately oiled soils in 2001when the sampling interval was 0–5 cm compared to 2007 whenthe sampling interval was 0–10 cm, suggesting that degradationprocesses were faster at the surface. Entrix (2002) reported thatpercent depletions for n-alkanes from n-C12 to n-C34 were greaterthan 95% by October 2001, which indicates that microbial degrada-tion was occurring in these surface soils.

Based on the PAH data reported by Entrix (2002) and discussedhere, it appears that there has been very little weathering of PAHsin marsh soils since Fall 2000. What are the limiting factors to nat-ural weathering processes in marsh soils? There are likely two:slow physical removal processes and low oxygen availability. Theinterior marsh habitat is flooded by daily tides through many smallchannels. During spring tides, there can be 20–30 cm of water inthe marsh. The marsh surface has extensive micro-topographywith low areas between dense clumps of stems that hold poolsof water during low tide. Sediments in these low areas are soft

for samples collected in 2001 versus 2007.

1004 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

and water saturated. Obviously, during spring low tides, the marshsoils do drain as low as 30 cm, because the oil penetrated to thesedepths in some areas. The falling tide drains through dense vegeta-tion. Tidal flushing may have been an initial mechanism for re-moval of bulk oil stranded on the surface; however, it would notbe effective at mobilizing oil from below the marsh surface. Thereare few bioturbating benthic organisms in these marshes. Photo-oxidation does not occur below ground. The only other removalmechanism would be microbial degradation.

Mendelssohn and Slocum (2004) measured the soil oxidation–reduction intensity (Eh) in Swanson Creek soils as part of theirstudy of soil cellulose decomposition. At seven of eight sites, Ehat 2 cm averaged �41 milliVolts (mV) (range of 31 ± 62 to�130 ± 3 mV); at 15 cm it averaged �84 mV (range of 0 ± 24 to�137 ± 30 mV). Thus, the soils are highly anaerobic. Zhu et al.(2001), in the US Environmental Protection Agency guidance man-ual for bioremediation of oil in freshwater wetlands, state that‘‘. . .in freshwater wetland environments, petroleum degradationis likely to be limited by oxygen availability.” In some of the casestudies cited by Zhu et al. (2001), no enhanced oil degradation oc-curred in anaerobic soils even when nutrients were added. Men-delssohn and Lin (2002) showed that nutrient addition to S.alterniflora marsh cores only increased PAH degradation in thetop 5 cm in treatments where oxidants were applied in additionto bioremediation agents under drained conditions. There wereno differences when the cores were flooded. Reddy et al. (2002)found that the oil from the West Falmouth spill persisted in marshsoils at depths of 6–28 cm for 30 years, and only n-alkanes and themore water-soluble and volatile compounds had been lost.

With few mechanisms in the marsh interior for physical re-moval of the oil, and the slow rate of degradation expected underanaerobic conditions, it is likely that the oil will persist in theseareas of the marsh for decades. Overall degradation rates in thesurface soils may be faster where respiration by roots and tidalflushing can be a source of oxygen.

The impact of the oil on vegetation condition seven years postspill varied by species. For S. alterniflora, stem density and stemheight were significantly lower in oiled sites compared to unoiledsites, but no differences were seen for belowground biomass. For S.cynosuroides, the results were opposite, with belowground biomasssignificantly lower in oiled sites but not stem density and stemheight. The reasons for these differences may be related to the rel-ative distribution of above-versus belowground biomass and thetypes of biomass for each species.

The study by Schubauer and Hopkinson (1984) of both speciesin a brackish marsh in Georgia provides interesting insights intopossible factors. Because S. cynosuroides has more and larger rhi-zomes and the rhizome biomass has a peak at 10–20 cm, this spe-cies may be more exposed to oil that is at higher concentrationsand more persistent in the cavities along the rhizomes. Most ofthe black oil that we observed in the cores occurred along rhi-zomes, some of which were hollow and dead. Roots and rhizomesin the soil would grow until they encountered zones of oil thatwould slow growth and could eventually lead to death. S. alternifl-ora has about an equal proportion of roots to rhizomes and the rhi-zomes are smaller, so any reductions in the biomass of therhizomes may have had a lesser effect on the overall belowgroundbiomass. Alternatively, the lower belowground biomass of S. alter-niflora may be in less direct contact with the oil. In the only pub-lished study found on the effects of oil on S. cynosuroides, Ferrellet al. (1984) conducted greenhouse experiments with both S. alter-niflora and S. cynosuroides. They suggested that the impacts of oiladded to the substrate to both species were caused by decreasedroot and rhizome growth, but they did not note any differences be-tween S. alterniflora and S. cynosuroides in the short-term test re-sults. With live root turnover of about 1–2 years (Schubauer and

Hopkinson, 1984), reduced belowground biomass observed in S.cynosuroides habitats would reflect current stress from exposureto the residual oil.

Several studies have determined the oil concentrations thatcould affect salt marsh species, mostly in controlled experiments.Most of these studies report total oil loading either as the amountof oil mixed into the soils prior to transplanting, or applied to thesoil surface and thus reported as volume/area. Few studies re-ported the PAH concentrations. However, it is clear it can take rel-atively high levels of oil in marsh soils to affect plants. Alexanderand Webb (1987) found that it took greater than 10,500 mg/kg ofa light crude oil to affect stem density of S. alterniflora. Krebs andTanner (1981) found that 2000 mg/kg of a No. 6 fuel oil spill inthe Potomac River had no effect or a slight positive effect on S. alt-erniflora aboveground metrics, but 10,000 mg/kg resulted inmortality.

Lin et al. (2002) homogeneously mixed fresh No. 2 fuel oil intomarsh soils and transplanted S. alterniflora stems that were free ofsoil and rhizomes into different oil dosages. They found significantreductions in belowground biomass (about 50% decrease) afterthree months of exposure at oil dosages greater than 29,000 mg/kg dry soil, which equated to PAH concentrations greater than1600 mg/kg. However, they only saw impacts to aboveground bio-mass, stem density, and shoot height at above 57,000 mg/kg oil.They observed significant stimulation of belowground biomass atthe 7000 mg/kg oil dose, which equated to about 400 mg/kg PAH.Other studies have also reported stimulated growth (Hershnerand Moore, 1977; Lin and Mendelssohn, 1996). However, it is dif-ficult to compare exposure to a No. 2 fuel oil that was well-mixedin the soil in their greenhouse experiment with the highly variableoil distribution in the Swanson Creek marsh soils. The measuredPAH concentrations in the collected sample from a site are only ageneral indication of the degree of oiling in marsh soils; it is notrepresentative of the dose to the roots and rhizomes because theoil was highly concentrated in some cavities. Thus, only a fractionof the roots and rhizomes are actually exposed, and the exposurewould be high enough to cause mortality and prevent re-growth.

Mendelssohn and Slocum (2004) did not find any effect of oil ondecomposition rates of organic matter in the marsh soils in Swan-son Creek, as measured using the cellulose (cotton) strip techniquein a study conducted in August 2002, two years after the spill. Theirstudy consisted of seven oiled sites of varying degrees of oiling andone reference site. Total PAH concentrations were comparable tothose found in 2007, although they reported a significant decreasein oil with depth at their sites. Other naturally occurring environ-mental factors appeared to control decomposition rates, such assalinity, pH, and depth. Our study also suggests that decompositionrates have not been affected in that the belowground biomass islower in the oiled areas. To be more definitive, measurements ofboth live and dead belowground biomass would be needed.

Oiled S. alterniflora sites had reduced stem density and stemheight compared to unoiled sites whereas S. cynosuroides did not.Reduced stem density and height are commonly reported in the lit-erature for oiled S. alterniflora (e.g., Krebs and Tanner, 1981; Linand Mendelssohn, 1996). The denser, shorter S. alterniflora vegeta-tion may be more sensitive to oil exposure, compared to the talland rigid culms of S. cynosuroides.

The fourteen amphipod toxicity tests correlated well with pre-dicted toxicity using ESB-TUFCV values, with samples showingsome toxicity at values between 1.0 and 2.0 and 0% survival at val-ues >3.0. Extrapolating predicted toxicity from ESB-TUFCV valuesfor the average PAHs in the 0–20 cm interval to the 24 heavily oiledsites, 11 sites or 46% had a value >1.0 and thus are likely to causetoxic effects. However, 25% of the sites had a value >3.0 and thuscontained PAH at concentrations that could cause chronic toxicityeven seven years post spill. The toxicity tests also correlated well

Table 4Comparison of the predicted services present as of 2007 and the 2007 actual results for the marsh and soil services in the interior, heavily oiled habitats.

Resource category Predicted services present in 2007 2007 Results (ratio = value oiled/unoiled sites)

Vegetation– S. alterniflora interior heavy 100% Stem density = 63%

Stem height = 85%– S. cynosuroides interior heavy 83% Stem density = 105%

Stem height = 94%

Soils– S. alterniflora interior heavy 85% ESB-TUFCV values in soil samples

40% >1.040% >2.030% >3.0Belowground biomass = 98%

– S. cynosuroides interior heavy 57% ESB-TUFCV values in soil samples50% >1.036% >2.021% >3.0Belowground biomass = 78%

J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006 1005

with the maximum predicted toxicity, PMax, using the logisticregression model approach. These results are of particular valuefor long-term monitoring studies of oiled marshes; first, becausetoxicity is seldom monitored, and second, because surficial soils(0–10 cm) in some areas still exhibit toxicity after seven years.With slow rates of physical removal and microbial degradation ofthe oil that penetrated into the marsh soils, it is likely that the tox-icity will also persist.

It is of value to compare the 2007 study results with the pre-dicted service losses for the vegetation and soils that were devel-oped during the natural resource damage assessment. The 2007study sites were located only in heavily oiled, interior habitats.The recovery curve inputs from Michel et al. (2002) were convertedto the predicted services present for the two marsh types in 2007,shown in Table 4. The 2007 study results are summarized as the ra-tio of the mean value for oiled sites to that for the unoiled sites,without any consideration for statistical significance. It seems thatthe predicted service losses for S. alterniflora vegetation recoverywere underestimated and for S. cynosuroides were overestimated.The predicted soil service losses appear to be supported by the2007 results in that S. alterniflora habitats showed no reductionsin belowground biomass but 40% of the sites are predicted to stillhave some toxicity. The S. cynosuroides habitats, however, haveboth a reduced belowground biomass and slightly larger percent-age of sites with toxicity.

Acknowledgments

Funding was provided by the Estuary Restoration Act throughthe NOAA Office of Response & Restoration. Jim Hoff, Kate Clark,and Jay Field of NOAA provided technical support and comment.Mitch Keiler, Krissy Rusello, and Ian Hartwell assisted in the fieldwork. The fourteen technical reviewers who provided commentson the draft report are thanked for their contributions.

References

Alexander, S.K., Webb Jr., J.W., 1987. Relationship of Spartina alterniflora growth tosediment oil content following an oil spill. In: Proceedings of the 1987 Oil SpillConference. American Petroleum Institute, Washington, DC, pp. 445–449.

Anderson, R.R., Brown, R.G., Rappleye, R.D., 1968. Water quality and plantdistribution along the upper Patuxent River, Maryland. Chesapeake Science 9(3), 145–156.

Douglas, G.S., Bence, A.E., Prince, R.C., McMillen, S.J., Butler, E.L., 1996.Environmental stability of selected petroleum hydrocarbon source andweathering ratios. Environmental Science and Technology 30 (7), 2332–2339.

Entrix, 2002. Swanson Creek incident post-biostimulation data report for 2001sampling season. Prepared for the Potomac Electric Power Company andSupport Terminals Operating Partnership, L.P. Prepared by Entrix, Inc.,Wilmington, DE.

Ferrell, R.E., Seneca, E.D., Linthurst, R.A., 1984. The effects of crude oil on the growthof Spartina alterniflora Loisel. and Spartina cynosuroides (L.). Journal ofExperimental Marine Biology and Ecology 83 (1), 27–39.

Field, L.J., MacDonald, D.D., Norton, S.B., Ingersoll, C.G., Severn, C.G., Smorong, D.,Lindskoog, R., 2002. Predicting amphipod toxicity from sediment chemistryusing logistic regression models. Environmental Toxicology and Chemistry 21(9), 1993–2005.

Gundlach, E.R., Baca, B., Barry, D.C., 2003. Emergency marsh restoration as part ofthe response to the Swanson Creek (Maryland) oil spill. In: Proceedings of the2003 International Oil Spill Conference. American Petroleum Institute,Washington, DC, pp. 1–9.

Hershner, C., Moore, K., 1977. Effects of the Chesapeake Bay oil spill on salt marshesof the lower bay. In: Proceedings of the 1977 Oil Spill Conference. AmericanPetroleum Institute, Washington, DC, pp. 529–533.

Krebs, C.T., Tanner, C.E., 1981. Restoration of oiled marshes through sedimentstripping and Spartina propagation. In: Proceedings of the 1981 Oil SpillConference. American Petroleum Institute, Washington, DC, pp. 375–384.

Lin, Q., Mendelssohn, I.A., 1996. A comparative investigation of the effects of southLouisiana crude oil on the vegetation of fresh, brackish, and salt marshes.Marine Pollution Bulletin 32 (2), 202–209.

Lin, Q., Mendelssohn, I.A., Suidan, M.T., Lee, K., Venosa, A.D., 2002. The dose-response relationship between No. 2 fuel oil and the growth of the salt marshgrass, Spartina alterniflora. Marine Pollution Bulletin 44, 897–902.

Long, E.R., Field, L.J., MacDonald, D.D., 1998. Predicting toxicity in marine sedimentswith numerical sediment quality guidelines. Environmental Toxicology &Chemistry 17 (4), 714–727.

Mendelssohn, I.A., Lin, Q., 2002. Development of Bioremediation for Oil SpillCleanup in Coastal Wetlands. OCS Study MMS 2002-048, U.S. Department of theInterior, Minerals Management Service, Gulf of Mexico OCS Region, NewOrleans, LA, 84 pp.

Mendelssohn, I.A., Slocum, M.G., 2004. Relationship between soil cellulosedecomposition and oil contamination after an oil spill at Swanson Creek,Maryland. Marine Pollution Bulletin 48, 359–370.

Michel, J., Smith, K., Keiler, M., Rizzo, A., Ayella, R., Hoff, J., 2002. Injury to WetlandsResulting from the Chalk Point Oil Spill. Administrative Record Document I.D.Number 2035. <www.darp.noaa.gov/neregion/chalkpt>.

Michel, J., Nixon, Z., Dahlin, J., Betenbaugh, D., White, M., Burton, D., Turley, S., 2008.Monitoring of Recovery of Marshes Impacted by the Chalk Point Oil Spill. NOAAOffice of Response & Restoration, Silver Spring, MD. 68 pp. + appendices.

National Oceanic and Atmospheric Administration (NOAA), Maryland Departmentof Natural Resources, Maryland Department of the Environment, U.S. Fish andWildlife Service, 2002. Final Restoration Plan and Environmental Assessmentfor the April 7, 2000 Oil Spill at Chalk Point on the Patuxent River, Maryland.NOAA Damage Assessment Center, Silver Spring, MD, 136 pp.<www.darp.noaa.gov/neregion/chalkpt>.

Reddy, C.M., Eglinton, T.I., Hounshell, A., White, H.K., Xu, L., Gaines, R.B., Frysinger,G.S., 2002. The West Falmouth oil spill after thirty years: the persistence ofpetroleum hydrocarbons in marsh sediments. Environmental Science andTechnology 36, 4754–4760.

Schubauer, J.P., Hopkinson, C.S., 1984. Above- and belowground emergentmacrophyte production and turnover in a coastal marsh ecosystem, Georgia.Limnology and Oceanography 29, 1052–1065.

U.S. Environmental Protection Agency (USEPA), 1994. Methods for Assessing theToxicity of Sediment-associated Contaminants with Estuarine and MarineAmphipods. EPA/600/R-94/025. USEPA, Office of Research and Development,Washington, DC, 140 pp.

USEPA, 2003. Procedures for the Derivation of Equilibrium Partitioning SedimentBenchmarks (ESBs) for the Protection of Benthic Organisms: PAH Mixtures. EPA-600-R-02-013. Office of Research and Development. Washington, DC, 156 pp.

USEPA, 2005. Predicting Toxicity to Amphipods from Sediment Chemistry. EPA/600/R-04/030. National Center for Environmental Assessment, Washington, DC, 192pp.

1006 J. Michel et al. / Marine Pollution Bulletin 58 (2009) 995–1006

Wang, Z.D., Fingas, M., Shu, Y.Y., Sigouin, L., Landriault, M., Lambert, P., 1999.Quantitative characterization of PAHs in burn residue and soot samplesand differentiation of pyrogenic PAHs from petrogenic PAHs – The 1994Mobile Burn Study. Environmental Science and Technology 33 (18), 3100–3109.

Wright, D.A., Zamuda, C.D., 1991. Copper contamination in the Patuxent River,Maryland. Hydrobiologia 215, 31–41.

Yunker, M.B., MacDonald, R.W., Vingarzan, R., Mitchell, H., Goyette, D., Sylvestre, S.,2002. PAHs in the Fraser River basin: a critical appraisal of PAH ratios of PAHsource and composition. Organic Geochemistry 33, 489–515.

Zhu, A., Venosa, A.D., Suidan, M.T., Lee, K., 2001. Guidelines for the Bioremediationof Marine Shorelines and Freshwater Wetlands. U.S. Environmental ProtectionAgency, Office of Research and Development, National Risk ManagementResearch Laboratory, Cincinnati, OH. 156 pp.