III. PETROLEUM HYDROCARBONS IN SEDIMENTS AND RIBBED-MUSSELS (Geukensia demissa)

Ashok D. Deshpande^{9, 11}, Bruce W. Dockum^{9, 12}, and Amy M. Tesolin-Gee^{9, 10, 13}

Postal Address: ⁹National Marine Fisheries Serv., 74 Magruder Rd., Highlands, NJ 07732
Current Address: ¹⁰Dow Chemical Co., 2030 Dow Center, Midland, MI 48667
E-Mail Addresses: ¹¹Ashok.Deshpande@noaa.gov; ¹²Bruce.Dockum@noaa.gov; ¹³AMTesolin-Gee@dow.com

The goal of our study was to assess the effectiveness of the replanting effort for removing petroleum hydrocarbon contaminants from the Arthur Kill marshes, and to assess the usefulness of ribbed-mussels as possible biomarkers of petroleum-related spills. Our field protocol included collection of sediment and ribbed-mussel samples from the six saltmarsh sites in the Arthur Kill and from one saltmarsh site on Sandy Hook (Appendix Table B1). Samples from Sandy Hook provided a relatively uncontaminated regional reference for samples from the Arthur Kill marshes.

METHODS AND MATERIALS

Sediment Collection and Sectioning

Four stations were selected at each marsh site in the Arthur Kill. Each station was located 0.2 m above the mid-tide, and the stations were spaced within 2-20 m of each other (see description of the sampling transect in Chapter II, "Trace Metal Contaminants in Sediments and Ribbed-Mussels"). No cores were collected at the Sandy Hook site. All sediment samples were archived at the Howard Laboratory either at -20°C or at -80°C.

During September 1996, one sediment core was taken by hand at each station using a chrome-plated copper tube (3.8-cm o.d. × 22.9-cm length). A total of 24 sediment cores (i.e., 6 sites × 4 stations per site × 1 core per station) were collected. Sediment cores from Old Place Creek, Con Ed Tower, and Mill Creek were sectioned using a core-sectioning device built at the Howard Laboratory (Figure 11). The cores contained large amounts of diverse plant and other materials that prevented precise sectioning. Each core was sectioned into five individual sections, with each section being approximately 1-cm thick and weighing approximately 10 g. With four stations at each site and five sections per core per station, a total number of 60 core sections were prepared for the three marsh sites.

Diverse plant and other materials in the sediments prevented precise sampling of the Arthur Kill surface sediments. During May 1997, an approximately 1-cm section of surface sediment was collected at four stations at each of the six sites in the Arthur Kill using a stainless steel spoon. One surface sediment sample from Sandy Hook was collected during August 1997.

Mussel Collection, Processing, and Selection for Analysis

Mussels were collected randomly at each site in the Arthur Kill during September 1996 and May 1997. The number of mussels available for collection varied by site and sampling period. Thus, 17-34 mussels were collected at each site in September 1996, while only 6-15 mussels were collected at each site in May 1997. Sandy Hook mussels were collected during February 1997.

Live mussels were brought to the Howard Laboratory, and placed overnight in a 4°C, temperature-controlled room. Material for the determination of method detection limits (MDLs) in the mussels was obtained by overnight depuration of 12 additional Sandy Hook mussels in aerated seawater at 4°C. All mussels were dissected within 24 hr using implements cleaned with methylene chloride. After removal of extraneous materials from mussel shells (mud, barnacles, etc.), the physical characteristics were recorded for each specimen (Appendix Table B2). Mussel tissues were then excised, placed in precleaned glass containers, and archived at the Howard Laboratory at -80°C.

The length of an individual mussel was assumed to be related to its age, and possibly, the contaminant body burden. Since the length-frequency distribution of mussels varied by site and sampling period (Figure 12 and Figure 13), a length range of 55-74 mm (inclusive) -- visually identified to be common to all sampling periods -- was selected as the bin range for analyses.

Five mussels at each site in the Arthur Kill were targeted for analyses. The bin range was divided into five groups (i.e., 55-58 mm, 59-62 mm, 63-66 mm, 67-70 mm, and 71-74 mm), ensuring that mussels of different lengths were included in the analyses. Each mussel was assigned a random number. Mussels with the highest random numbers were sequentially selected from each group. When there were no mussels in one or more groups, mussels were selected in a two-step procedure. In step 1, the mussel with the highest random number within each group containing mussels was selected. In step 2, the mussel with the highest random number from all remaining mussels within the bin range was selected. Step 2 was repeated until the requirement of five mussels per site was met.

Seven out of the 12 undepurated mussels from Sandy Hook were targeted for analyses. The bin range was divided into seven groups (i.e., 55-56 mm, 57-58 mm, 59-60 mm, 61-62 mm, 63-64 mm, 65-66 mm, and 67-68 mm), and random numbers were assigned to each mussel. Mussels with the highest random number in each group were selected for analysis.

Extraction of Hydrocarbons in Sediments

Twenty core sections each from Old Place Creek, Con Ed Tower, and Mill Creek; four surface scoop samples each from Old Place Creek and Con Ed Tower; and a single surface scoop sample from Sandy Hook were processed in four extraction batches (Appendix Table B3). All sediment samples were dried with sodium sulfate before extraction. In the first batch, we extracted the sediments by shaking them with methylene chloride in an Erlenmeyer flask. The shaking/extraction procedure was manual and laborious, and did not save as much on extraction time or glassware as we initially thought. Therefore, Batches 2-4 samples were extracted with methylene chloride using automated Soxhlet extraction systems.

Batch 1 sediment samples were extracted by shaking sodium-sulfate-dried sediments with methylene chloride. Approximately 10 g of each Batch 1 sediment sample were placed into a mortar, then mixed by pestle with 60-80 g of anhydrous sodium sulfate until the mixture was dry. Approximately, 5 cc of activated copper were added to the sample for the bulk removal of elemental sulfur, and the mixture was transferred to an Erlenmeyer flask. Surrogate internal standard (20 µg of o-terphenyl) and any other spiking solutions, as appropriate, were added to the sample. Approximately 60 ml of methylene chloride were added to the Erlenmeyer flask, and the sample was shaken overnight. Methylene chloride was decanted, and the procedure was repeated two more times. The combined extract was concentrated to approximately 40 ml.

Batches 2-4 sediment samples were dried with sodium sulfate, and extracted with a Soxhlet extraction apparatus. Approximately 10 g of each sample from Batches 2-4 were placed into a mortar, then mixed by pestle with 60-80 g of anhydrous sodium sulfate until the mixture was dry. The sample was then transferred to a cellulose thimble. Surrogate internal standard (20 µg of o-terphenyl) and any other spiking solutions, as appropriate, were added to the sample. The thimble was then transferred to a labeled Soxhlet extraction apparatus. Hydrocarbons were extracted with methylene chloride for 18-24 hr. Activated copper gauze was placed in the extraction apparatus for the bulk removal of elemental sulfur. The methylene chloride extract was concentrated to approximately 40 ml.

Initially, we experimented with the volume of concentrate from each site to ensure that the flame ionization detector (FID) did not become overloaded during gas chromatographic (GC) analyses. We assumed that all potential interfering compounds in the 40-ml extract were at a very low level, and that they would not cause any chromatographic problems. Therefore, we injected 1 µL of this extract directly into a gas chromatographic column without any additional cleanup. The samples in which hydrocarbons were not detected were concentrated in a stepwise manner until the hydrocarbons were detected. After completing the screening of representative samples from each site, the methylene chloride extracts of all samples were subjected to silica-alumina glass column chromatographic removal of polar biogenic interferences. Column-cleaned extracts were then concentrated to appropriate final volumes for the GC analyses. GC internal standard (5--androstane) was added to each final extract before GC analyses.

Extraction of Hydrocarbons in Mussels

A total of 60 mussels from six Arthur Kill sites and seven mussels from one Sandy Hook site were processed in three extraction batches (Appendix Table B4).

Each mussel sample (3.2-15.4 g) was placed into a mortar, then mixed by pestle with 80 g of anhydrous sodium sulfate until the mixture was dry. The sample was then transferred to a cellulose thimble, and surrogate internal standard (20 µg of o-terphenyl) and any other spiking solutions, as appropriate, were added to the sample. The thimble was then transferred to a labeled Soxhlet extraction apparatus. Hydrocarbons in mussels were extracted with methylene chloride over 18-24 hr.

In Batch 1, we experimented with the volume of concentrate to ensure that the FID did not become overloaded during GC analyses. Initially, our total extract volume was 50 ml, and we used 10 ml of this extract for the lipid determination. We assumed that all potential interfering compounds in the remaining 40 ml of extract were at a very low level, and that they would not cause any chromatographic problems. Therefore, we injected 1 µL of this extract directly into the GC column without any additional cleanup. Contrary to our expectations based on the oil spill history in the Arthur Kill marshes, we barely observed any peaks in this dilute mussel extract. The remaining mussel extract was eluted through a silica-alumina glass column for the removal of polar lipids and other polar biogenic interferences, and the sample was concentrated to a volume of 5 ml. Surprisingly, a 1-µL injection of this extract did not overload the FID either, and the peak sizes were still very minute. This initial work suggested that our standard protocol of concentrating the sample to a final volume of 1 ml was also suitable for Arthur Kill and Sandy Hook mussels. All mussel extracts were then subjected to silica-alumina glass column chromatographic cleanup. The column-cleaned extract was concentrated to about 700-750 µL, the GC internal standard was added (200 µL of 5--androstane, 20 µg total), and the final sample volume was brought to 1 ml using methylene chloride.

Gas Chromatographic Analyses of Hydrocarbons

Instrument Operating Parameters

Sediment and mussel samples were analyzed for a total of 33 normal-chain hydrocarbons and two branched-chain hydrocarbons (Appendix Table B5) using a Hewlett Packard (HP) 5880A GC-FID. One microliter of the final sample extract was injected into a fused silica capillary column in splitless mode using an HP 7673A autosampler. Extracts of all mussel samples and extracts of sediment samples from Batches 1, 2 (excluding the nine samples from Con Ed Tower), and 3 were injected into an HP-5 (0.32-mm i.d. × 30-m length × 0.25-µm film thickness) capillary column. Extracts of Batch 2 sediment samples from the Con Ed Tower site and extracts of sediments from Batch 4 were injected into a J&W DB-5 (0.45-mm i.d. × 30-m length × 0.42-µm film thickness) capillary column. The injector port temperature was set at 300°C, while the detector temperature was set at 280°C. An initial purge time of 1 min was used to maximize the amount of higher-boiling hydrocarbons that reached the top of the GC column. The column oven temperature was held at 50°C for 1 min after sample injection, and then programmed to reach 310°C at a rate of 3°C/min. The oven temperature was held at the final temperature of 310°C for 30 min, resulting in a total run time of 120 min. We used 5--androstane as a time reference standard and as a GC internal standard for monitoring sample-to-sample variation in peak retention time and sample-to-sample variation in GC-FID response. We used o-terphenyl as a time reference standard and as a surrogate internal standard for assessing analyte recoveries. The chromatographic peaks were recorded with an HP 5880A Series GC terminal. The data in the electronic format were collected with a Perkin-Elmer Nelson 970 interface and Perkin-Elmer Nelson Turbochrom 4.0 chromatographic software. Generation of calibration curves, identification of peaks, and integrations were done with the Turbochrom software. The text files generated by Turbochrom software were imported into a Microsoft Excel spreadsheet for the determination of final analyte concentrations. The analyte concentrations are expressed as µg/g (ppm) on a wet-weight basis.

Chromatographic Performance Evaluation

New Jersey Department of Environmental Protection (NJDEP 1995) guidelines were employed in the evaluation of chromatographic performance. The evaluation criteria included chromatographic separation of analytes, resolution of critical pairs of peaks, mass discrimination, and curve correlation coefficients.

Separation of Individual Analytes and Internal Standards

The GC temperature program successfully separated all 35 hydrocarbon peaks (i.e., n-C₈ to n-C₄₀, including pristane and phytane) and two internal standard peaks in a mid-point calibration mixture (i.e., a mixture in which all hydrocarbon and internal standard concentrations are 10 ng/µL; Figure 14), and all identifiable hydrocarbon peaks and two internal standard peaks in a diesel fuel oil #2 standard (Figure 15).

Resolution of Critical Pairs of Peaks

Resolution (R) of critical pairs of peaks of hydrocarbons was calculated as:

R = 2(RT2-RT1)/(W1+W 2)

Eq. 1

where RT = retention time and W = baseline peak width of the respective hydrocarbon (NJDEP 1995). The R values for the n-C₁₇/pristane pair and the n-C₁₈/phytane pair were >0.8 for all mussel chromatographic batches. Resolutions for the two pairs of hydrocarbons for the sediment chromatographic batches were verified only visually and were found to be satisfactory.

Mass Discrimination

The NJDEP guideline for mass discrimination in the injector port (n-C₃₂ peak area/n-C₂₀ peak area >0.8) was met in all mussel and sediment chromatographic batches.

Calibration Curve Correlation Coefficients

Correlation coefficient squares (r²) for each analyte in the five-point (i.e., 2, 5, 10, 20, and 50 ng/µL; internal standards at 10 ng/µL) calibration curves were consistently >0.99 for all mussel and sediment batches.

Typical Gas Chromatographic Batch

A series of sequential steps were performed at the beginning of the study and prior to the analyses of a fresh batch of samples. First, we replaced the GC injection port septum, injection port liner, gold-plated seal, and the ring. Then we verified background cleanliness of overall instrument components using an instrument blank solution that contained only the GC internal standards. The GC column resolution check was followed by the verification of a minimum mass discrimination criterion using the ratio of the n-C₃₂ peak area to the n-C₂₀ peak area. Finally, a five-point calibration curve was generated for the identification and quantification of all detectable analytes.

Before beginning the analyses, column performance and detector stability were verified using a mid-point calibration solution that contained the analytes and GC internal standards at a concentration of 10 ng/µL. We then injected a set of 4-5 unknown samples. The instrument performance was verified after completing the analyses of these 4-5 samples. These steps, including the analyses of a set of unknown samples and the verification of instrument performance, were repeated until the remaining samples were analyzed. This sequence of steps provided a calibration chromatogram for every 4-5 samples or 10-12 hr of instrument operation. An HP 5880A controller program limited the maximum number of injections in a sequence to 26.

Quantification of Hydrocarbons in Sediments

The GC-FID chromatograms of sediment extracts were often complex, and exhibited areas of unresolved envelops (Appendix Figures E1-E14). Chromatographic complexities also created a potential for coelution of internal standards with unknown interfering compound(s). Sediment extracts were quantified using an external standard calculation method due to uncertainties in the identifications of internal standards.

A second-order curve equation was used to fit the hydrocarbon calibration data:

Y = C0+C1X+C2X2

Eq. 2

where, Y is the response of the analyte in the calibration sample, X is the amount of analyte in the calibration sample, and C₀, C₁, and C₂ are various curve coefficients for each analyte. For a given value of Y in an unknown sample, the Turbochrom software uses the first quadratic solution to Equation 2 to calculate the amount X near the origin of the curve:

X = {-C1+ [C12-4C2(C0 -Y)] ½}/2C2

Eq. 3

Calculations Using Individual Hydrocarbons

For each hydrocarbon analyte, a second-order curve given by Equation 2 was used to fit the calibration data. Values of r² >0.99 were obtained for each analyte fit.

In an unknown sample, the analyte amount X is calculated from Equation 3 given the measured analyte peak area Y. Amount X in this equation represents the analyte amount in 1 µL of the injected sample. The actual concentration of analyte in the sample is then calculated by multiplying X by a factor that incorporates the final extract volume and sample weight.

Calculations Using the Sum of Individual Hydrocarbons

In this method of calculation, X is defined as the sum of individual hydrocarbon amounts, and Y is defined as the sum of peak areas for individual hydrocarbons ranging from n-C₈ to n-C₄₀, including pristane and phytane. The X and Y values for each calibration solution were then used in Equation 2 to determine the new curve coefficients. Values of r² >0.99 were obtained for this fit.

The total alkyl hydrocarbon concentrations in an unknown sample were calculated using Equation 3 with a new set of curve coefficients and the value of Y defined as the sum of peak areas for individual hydrocarbons.

Calculations Using the Sum of All Peaks Eluting between n-C₈ and n-C₄₀

In this method of calculation, all peaks eluting between n-C₈ and n-C₄₀ were assumed to be a mixture of various normal-chain hydrocarbons, branched-chain hydrocarbons, and cyclic hydrocarbons. In Equation 2, X is defined as the sum of individual hydrocarbon amounts, and Y is defined as the sum of peak areas for the individual hydrocarbons n-C₈ to n-C₄₀, including pristane and phytane. The X and Y values for each calibration solution were used in Equation 2 to determine the curve coefficients.

For an unknown sample, the value for Y was calculated as Y = A-(B+C), where A is the sum of the areas for "all" peaks eluting between n-C₈ and n-C₄₀, B is the peak area for o-terphenyl, and C is the peak area for 5--androstane. We assumed that other hydrocarbons did not coelute with o-terphenyl and 5--androstane. The peak summation window began 15 sec before the retention time of the n-C₈ peak and ended 15 sec after the retention time of the n-C₄₀ peak. The value for X in Equation 2 is the sum of amounts of individual hydrocarbons, including n-C₈ to n-C₄₀, pristane, and phytane, in the individual hydrocarbon calibration mixture. The values for curve coefficients for the sum of individual hydrocarbons were used in Equation 3, to calculate TPH concentrations after applying an appropriate multiplication factor characteristic of sample weight and final volume of sample extract.

Quantification of Soil SRM 765

Soil SRM (i.e., SRM 765) obtained from Environmental Resource Associates was extracted with each sediment extraction batch. A one-point linear calibration was prepared using a standard solution of diesel fuel oil #2 from Restek Corporation. The linear equation Y = mX+C was used in this quantification. In this equation, Y was calculated as Y = A-(B+C), where A is the sum of areas for "all" peaks eluting between n-C₈ and n-C₄₀, B is the peak area for o-terphenyl, and C is the peak area for 5--androstane. We assumed that other hydrocarbons did not coelute with o-terphenyl and 5--androstane. The peak summation window began 15 sec before the retention time of the n-C₈ peak and ended 15 sec after the retention time of the n-C₄₀ peak. Also in this equation, X is the amount of Restek diesel fuel oil in a 1-µL injection, m is the slope of the line, and C is the intercept (where C = 0 because the line is forced through the origin). The concentration of diesel fuel oil in SRM 765 was calculated by multiplying X by a factor that incorporated the final extract volume and sample weight.

Quality Assurance for Sediment Analyses

Quality assurance criteria listed in Appendix Table C1 were used for evaluating the quality of sediment data. Results of quality assurance of sediment analyses are summarized for method detection limit, laboratory method blanks, surrogate internal standard recovery, matrix spike recovery, soil standard reference material analyses, and replicate sediment analyses.

Method Detection Limit

The target MDL value for TPH was 10 µg/g, and was based on the New Jersey Department of Environmental Protection's quality assurance document (NJDEP 1995). We spiked each replicate MDL sediment matrix with 20 µg of each hydrocarbon in approximately 10-g MDL replicate sediment samples. With 35 hydrocarbons used for spiking, and 20 µg spiked per hydrocarbon, the total spiked hydrocarbon amount was 700 µg, or 70 µg/g of sediment. The MDL for sediments was calculated as MDL = t; where is the standard deviation of seven replicate measurements and t is Student's t value of 3.143 with six degrees of freedom (EPA 1984). The MDL values for sediments varied from analyte to analyte, and ranged from 0.53 to 8.25 µg/g, wet weight, with a majority of MDL values between 1 and 2 µg/g (Appendix Table C2). Since n-C₈ was not detected in MDL samples, its MDL was not determined.

The EPA (1984) protocol for MDL determination recommends that the spiked amount be approximately 2-5 times greater than the target MDL. Our spiked amounts were 1.4-3.5 times greater than the EPA-recommended amounts in order to accommodate the poor sensitivity of the GC/FID. Most relative standard deviation (RSD) values for MDL determinations in the present study were around 10%, indicating a good precision in hydrocarbon determinations. [Relative standard deviation is the standard deviation divided by the mean, and is expressed as a percentage.]

Laboratory Method Blanks

For laboratory method blank samples, 134 out of 136 values were less than three times the MDL (data not provided). The blank criterion was not applied to n-C₈ because this compound was not detected in the MDL study.

Surrogate Internal Standard Recovery

Forty-four of 94 samples exceeded the surrogate internal standard recovery criterion (Appendix Table C3-C6), with the exceedances mostly occurring in the Con Ed Tower samples containing complex chromatograms. The high values probably resulted from the coelution of surrogate standards with interfering peaks which increased peak area of the surrogate standard.

Matrix Spike Recovery

Matrix spike recoveries for four sediment extraction batches are listed in Appendix Table C7. We added 100 µg of each hydrocarbon to the Batch 1-3 matrix spike samples, and 50 µg of each hydrocarbon to the Batch 4 matrix spike sample.

In Extraction Batch 1, 25 of 33 analytes met the matrix spike recovery criterion. Matrix spike recovery values were below the lower criterion value of 50% for n-C₉ and for n-C₃₄ to n-C₄₀, with recoveries ranging from 12.5 to 42.2%. Hydrocarbons n-C₈ and n-C₂₉ were not detected in any of the Batch 1 sediment samples.

In Extraction Batch 2, 20 of 35 analytes met the matrix spike recovery criterion. Matrix spike recovery values were below the lower criterion value of 50% for n-C₈ to n-C₁₆ and for n-C₃₅ to n-C₄₀, with recoveries ranging from 23.4 to 47.8%.

In Extraction Batch 3, 29 of 35 analytes met the matrix spike recovery criterion. The matrix spike recovery value was below the lower criterion value of 50% for n-C₈ (28.6%), while the recovery values were higher than the upper criterion value of 120% for n-C₃₁ (122%) and for n-C₁₉ to n-C₂₂, with recoveries ranging from 121 to 123%.

In Extraction Batch 4, 4 of 35 analytes met the matrix spike recovery criterion. Matrix spike recovery values were below the lower criterion value of 50% for n-C₈ to n-C₁₉,with recoveries ranging from 2.58 to 46.7%, and for n-C₂₄ to n-C₄₀, with recoveries ranging from 17.8 to 47.7%.

Soil Standard Reference Material Analyses

Except for Batch 1, the SRM analyses in all batches gave recovery values that were lower than the lower criterion recovery value of 70% (Appendix Table C8). The SRM analyses in Batch 2 and two of the three replicate SRM analyses in Batch 4 gave slightly lower recoveries (58.7-63.1%) than the lower criterion value of 70% recovery. The SRM analyses for Batch 3 and for the third SRM replicate in Batch 4 (a suspected outlier which was not included in any calculations) gave lower recoveries than the lower criterion recovery value of 70%. The average SRM recovery for five replicates was 58.4% with an RSD of 21.7%. Since the diesel fuel standard used in the preparation of soil SRM was not available for instrument calibration, a diesel fuel standard from Restek Corporation was used in instrument calibration. The difference in the types of two diesel fuels, and possibly their hydrocarbon contents, may have resulted in the lower recovery values.

Replicate Sediment Analyses

Seven replicates of spiked sediments were used in the MDL determination study (Appendix Table C2). Except for a slightly higher RSD value for n-C₉ (27.06%), all other hydrocarbons met the replicate analysis criterion of 25% RSD. Octane hydrocarbons (i.e., n-C₈) were not detected in any of the seven replicates. Three replicates of Soil SRM 765 were additionally extracted in Batch 4 (Appendix Table C9). One soil SRM replicate gave poor recovery (2.5%) of diesel oil, and was discarded from further discussion. The percentage difference for the two remaining soil SRM replicates was 3.9, based on the total diesel oil concentration. On an individual hydrocarbon basis, 13 of 35 hydrocarbons met the replicate criterion, 21 hydrocarbons were undetected, and although n-C₈ was detected, no MDL value was measured for this hydrocarbon.

Quantification of Hydrocarbons in Mussels

Mussel GC-FID chromatograms were considerably less complex than sediment chromatograms, and the internal standards were easily identifiable (Appendix Figure E15 and Appendix Figure E16). Hydrocarbons in mussel extracts were therefore quantified using the more accurate method of internal standard calculation.

A second-order calibration curve was used to calculate the concentrations of hydrocarbons in mussels. In Equation 2, Y is the ratio of the response of the analyte to the response of the internal standard in the calibration sample, and X is the ratio of the amount of the analyte to the amount of the internal standard in the calibration sample. For an unknown sample, the first quadratic solution to Equation 2 provides a value for X for a given Y (Equation 3). The final determination of analyte concentration required additional calculations.

Calculations Using Individual Hydrocarbons

In Equation 2, Y is the ratio of the individual hydrocarbon peak area to the GC internal standard (5--androstane) peak area, and X is the ratio of the individual hydrocarbon amount to the GC internal standard (5--androstane) amount. Values of r² >0.99 were obtained for each analyte fit.

For a measured-area-ratio Y in an unknown sample, the amount-ratio X is determined from Equation 2. The amount of analyte in an unknown sample is calculated by multiplying X by the amount of 5--androstane added to the unknown sample. Amount X in this equation represents the analyte amount in 1 µL of the injected sample. The actual amount of analyte in the unknown sample is calculated by multiplying X by a factor that incorporates the final extract volume, aliquot of sample extract taken for lipid determination, and sample weight.

Calculations Using the Sum of Individual Hydrocarbons

In Equation 2, Y is the ratio of the sum of peak areas for the individual hydrocarbons (n-C₈ to n-C₄₀, including pristane and phytane) to the peak area of 5--androstane, and X is the ratio of the sum of amounts of individual hydrocarbons (n-C₈ to n-C₄₀, including pristane and phytane) to the amount of 5--androstane. The curve coefficients were obtained by first calculating the values of X and Y for each calibration solution, and then by using these values in a fit. Values of r² >0.99 were obtained for this fit.

Total alkyl hydrocarbon concentrations were then calculated using Equation 3. The amount of 5--androstane added to the sample and other sample factors were used in the calculations.

Calculations Using the Sum of All Peaks Eluting between n-C₈ and n-C₄₀

The TPH concentrations for mussel samples were determined using a procedure similar to that used for sediment samples. The curve coefficients used in Equation 2 were the same as those calculated for the sum of individual hydrocarbons in mussels using the internal standard method.

For an unknown sample, the value for Y was calculated as Y = [A-(B+C)]/C, where A is the sum of the areas for "all" peaks eluting between n-C₈ and n-C₄₀, B is the peak area for o-terphenyl, and C is the peak area for 5--androstane. It was assumed that other hydrocarbons did not coelute with o-terphenyl and 5--androstane. The peak summation window began 15 sec before the retention time of the n-C₈ peak and ended 15 sec after the retention time of the n-C₄₀ peak. The calculated value for X from Equation 3 is the ratio of the sum of amounts of "all" hydrocarbons eluting between n-C₈ and n-C₄₀ to the amount of 5--androstane. When the value of X is calculated from Equation 3, the concentration is determined by multiplying X by the amount of 5--androstane in the sample and other sample-related factors.

Quality Assurance for Mussel Analyses

Quality assurance criteria listed in Appendix Table C1 were used for evaluating the quality of mussel data. Results of quality assurance of mussel analyses are summarized for MDL, laboratory method blanks, surrogate internal standard recovery, matrix spike recovery, diesel fuel spike recovery, mussel SRM analyses, and replicate mussel analyses. Possible anomalies are also covered.

Method Detection Limit

The MDL for mussels was calculated as MDL = t; where is the standard deviation of seven replicate measurements and t is Student's t value of 3.143 with six degrees of freedom (EPA 1984). The MDL values for individual hydrocarbon analytes ranged from 0.06 to 2.47 µg/g, wet weight, with numerous values around 0.1 µg/g (Appendix Table C10).

The MDL guideline for hydrocarbons in mussels is not specified in the NJDEP protocol. Since mussel extracts were expected to be relatively cleaner than sediment extracts, we assumed a target MDL of 1 µg/g for mussels, which was 10 times less than a target MDL of 10 µg/g for sediments.

We spiked mussels with higher-than-recommended amounts of hydrocarbons to accommodate the higher detection limits of GC-FID. A majority of RSD values around 10% indicated good precision, but the higher spiked amounts gave higher values for standard deviations that resulted in approximately five-times-greater MDL values.

Laboratory Method Blanks

For laboratory method blank samples, 95 out of 105 values were less than three times the MDL (data not provided).

Surrogate Internal Standard Recovery

Eighty-two of 86 internal surrogate values met the criterion for surrogate internal standard recovery (Appendix Tables C11-13).

Matrix Spike Recovery

Matrix spike recoveries for the three mussel extraction batches are listed in Appendix Table C14. For Extraction Batch 1, 26 of 35 analytes met the matrix spike recovery criterion. As expected from the relatively low boiling points of n-C₈, n-C₉, and n-C₁₀, poor recoveries were obtained for these three relatively volatile hydrocarbons. If these three hydrocarbons were not included in the data, 81% of values would meet the matrix spike recovery criterion.

For Extraction Batch 2, seven replicate mussel samples were spiked with individual hydrocarbons (total spiked amount per analyte = 4 µg) for the MDL determination. Similar to Batch 1 sample results, we decided not to include the data for n-C₈, n-C₉, and n-C₁₀. If these hydrocarbons are not included in the data, 180 of 224 values (80%) met the matrix spike recovery criterion.

For Extraction Batch 3, 26 of 35 analytes (74%) met the matrix spike recovery criterion. If recoveries for n-C₈, n-C₉, and n-C₁₀ are not included, 26 of 32 analytes (81%) met the matrix spike recovery criterion.

Diesel Fuel Spike Recovery

In Batch 2, diesel fuel oil #2 was spiked into Sandy Hook mussel homogenate. Chromatograms of background mussel extract, spiked mussel extract, and Restek diesel fuel oil #2 used in spiking the mussels are depicted in Figure 16. Matrix spike recovery was calculated for individual hydrocarbons as well as diesel fuel (Appendix Table C14). Recoveries of individual hydrocarbons were calculated by comparing the areas of hydrocarbons in mussel homogenate with the areas of hydrocarbons in the diesel fuel standard. Recoveries of hydrocarbons from n-C₁₁ to n-C₂₀, including pristane and phytane, ranged from 52% to 100%. Recoveries of other hydrocarbons did not meet the data quality objectives criterion due to interfering peaks.

To calculate the recovery of diesel fuel, the sum of areas of representative hydrocarbon peaks in the matrix spike sample was compared with that in the diesel fuel standard. We selected n-C₁₂, n-C₁₃, n-C₁₄, n-C₁₅, n-C₁₇, and pristane as representative hydrocarbons based on their GC-FID responses in the spiked sample and diesel fuel calibration standard, and minimal interference in the vicinity of these respective hydrocarbons. The matrix spike recovery of diesel fuel oil #2 was then calculated to be 76.8%.

Mussel Standard Reference Material Analyses

Concentrations of various hydrocarbons listed for NIST Mussel SRM 1974a are noncertified values, and range in the low ng/g (ppb) levels. NIST scientists determined the concentrations of these hydrocarbons using gas chromatography / mass spectrometry (GC/MS), and these concentrations are <10 times the MDLs of this study. In addition, hydrocarbon analyses in the present study were performed using GC/FID, which is 1-2 orders of magnitude less sensitive than the GC/MS. We concluded that SRM 1974a was not an appropriate SRM for the evaluation of the quality of our mussel data. Hydrocarbons detected above MDL were considered false positives based on relatively low values reported by NIST (Appendix Table C15). Quality assurance criteria other than those based on SRM were therefore used in the validation of mussel data.

Replicate Mussel Analyses

Thirty of 34 hydrocarbons met the replicate analysis criterion of 25% RSD in the mussel MDL determination study (Appendix Table C10). In addition, one large mussel from Mill Creek weighing 35.2 g was homogenized, and the homogenate was extracted in triplicate. None of the individual hydrocarbon values in the mussel homogenates were >10 times the MDL (Appendix Table C16). Therefore, replicate analysis criteria were not applicable to these mussel homogenates.

Possible Anomalies

One mussel sample from the Con Ed Tower site and one mussel sample from the Mill Creek site contained relatively higher concentrations of n-C₃₁. After examining the concentrations of other hydrocarbons in these mussels, as well as concentrations of n-C₃₁ in other mussels from these sites, it appeared that the higher concentrations of n-C₃₁ are possible anomalies. One additional mussel from Con Ed Tower appeared to have relatively elevated, but possibly anomalous, concentrations of n-C₂₁ and n-C₂₃. One mussel sample from Sandy Hook also appeared to have relatively elevated, but possibly anomalous, concentrations of n-C₂₉ and n-C₃₀.

We hypothesized that some of these elevated concentrations may have arisen from contributions of hydrocarbons from natural sources, including terrestrial plants, phytoplankton, and algae (Blumer et al. 1971, 1973; Prahl et al. 1980; Douglas et al. 1981; Sauer and Uhler 1994).

Indicators of Hydrocarbon Source and Weathering

Hydrocarbon patterns and ratios of certain hydrocarbons were used to examine the hydrocarbon source, weathering/biodegradative losses of spilled hydrocarbons, and contribution of biogenic hydrocarbons to the petrogenic hydrocarbons.

Farnsane (2,6,10-trimethyldodecane), 2,6,10-trimethyltridecane, nor-pristane, pristane, and phytane represent a class of branched-chain hydrocarbons that degrade slowly compared to normal-chain hydrocarbons (Wang and Fingas 1997; Atlas 1981; Atlas et al. 1981). Since farnsane, 2,6,10-trimethyltridecane, and nor-pristane were not included in the instrument calibration mixture, the discussion of hydrocarbon weathering was limited to pristane and phytane, the dominant hydrocarbons in partially weathered petroleum products (Broman et al. 1987). Typical ratios used as indicators of hydrocarbon source and weathering are given for the three petroleum products listed in Appendix Table C17.

Ratio of Pristane to n-C₁₇ and of Phytane to n-C₁₈

Ratios of pristane to n-C₁₇ and of phytane to n-C₁₈ indicate the extent of degradation of normal-chain hydrocarbons, with higher ratios suggesting greater losses of normal-chain hydrocarbons (Cripps 1989; reciprocals of these ratios used by Wang and Fingas 1995). Since natural sources of pristane (e.g., copepods) may alter the pristane to n-C₁₇ ratio in the sediments, this ratio should be interpreted with caution (NRC 1985; Douglas et al. 1996).

Ratio of Pristane to Phytane

Because of the resistance of pristane and phytane to biodegradation, the pristane-to-phytane ratio is used as a marker in measuring the early degradation rate of oil (Sauer and Uhler 1994). The ratio of pristane to phytane can also be used to examine if the hydrocarbon mixtures from different locations or from different sediment core sections originated from a common source. Since natural sources of pristane (e.g., copepods) may alter the pristane-to-phytane ratio in sediment, this ratio should be interpreted with caution. Also, pristane and phytane are lost at different rates in the later stages of biodegradation that may confound the identification of source oil (Douglas and Uhler 1993).

Carbon Preference Index

Carbon preference index (CPI) is a ratio of the sum of odd-numbered hydrocarbons to the sum of even-numbered hydrocarbons (Farrington and Meyers 1975; NRC 1985). Hydrocarbon mixtures originating from plant materials show a predominance of odd-numbered carbon chains with CPI values >5-7 (Farrington and Tripp 1977). A CPI value of 1.0 indicates a petrogenic origin of the hydrocarbons. Values of CPI >1.0 indicate the contribution of odd-numbered hydrocarbons of biogenic origin (Choiseul et al. 1998).

Weathering Index

The weathering index (WI) is a ratio of the sum of n-C₈, n-C₁₀, n-C₁₂, and n-C₁₄ to the sum of n-C₂₂, n-C₂₄, n-C₂₆, and n-C₂₈ (Wang and Fingas 1994; Wang et al. 1994). A lower value for WI indicates weathering losses of the lower-boiling hydrocarbons. We did not include n-C₈ in the calculation of WI because: 1) it was not recovered in the spiked replicates used in the MDL determination (Appendix Table C2 and Appendix Table C10), 2) coelution of n-C₈ with unknown interferences resulted in its inadequate quantification, 3) artifact concentrations of n-C₈ were not internally consistent with concentrations of other homologs for the Old Place Creek sediments, and 4) the ratios for Old Place Creek sediments were solely driven by n-C₈ concentrations.

Total Organic Carbon

Guida and Draxler in the following chapter, "Sediment Biogeochemistry," describe the determination of total organic carbon (TOC) in the surface sediments.

Reporting of Hydrocarbon Concentrations

In addition to calculating the concentrations of individual petroleum hydrocarbon components, the sums of the concentrations of such compounds were calculated for the following groups in each sample: 1) total of individual petroleum hydrocarbons (TIPH); 2) branched-chain hydrocarbons (i.e., pristane + phytane); 3) odd-numbered, normal-chain hydrocarbons; 4) even-numbered, normal-chain hydrocarbons (starting with n-C₁₀); 5) representative lower-boiling-point, normal-chain hydrocarbons (i.e., n-C₁₀ + n-C₁₂ + n-C₁₄); and 6) representative higher-boiling-point, normal-chain hydrocarbons (i.e., n-C₂₂ + n-C₂₄ + n-C₂₆ + n-C₂₈).

If an analyte was not detected in a particular sample, then that analyte was not included in the aforementioned summations nor in any of the subsequent hydrocarbon ratios. If an analyte concentration determined for a particular sample was less than the MDL, then that analyte is reported as "not detected" (nd). Only analyte values above the MDL are reported. For "not detected" analytes, a concentration value equal to one-half of the MDL value was used in the summations and subsequent statistical calculations. The core and station averages for sediments and the station averages for mussels for a given analyte are reported as "nd" if a given analyte was absent in all samples used for averaging. If the concentration of a given analyte was greater than the MDL in at least one sample, then one-half of the MDL value was used for the "not detected" samples in that particular group for the calculation of the average concentration value. If the average concentration value was less than the MDL, then it is reported as "<MDL." Only averages greater than the MDL are reported.

The MDL for TPH was determined as MDL = t; where is the standard deviation of seven replicate TPH measurements, and t is Student's t value of 3.143 with six degrees of freedom (EPA 1984). Those TPH concentrations below the TPH MDL are reported as "<MDL"; only TPH concentrations above the MDL are reported.

Group MDLs, such as those for TIPH, branched-chain hydrocarbons, odd-numbered normal-chain hydrocarbons, even-numbered normal-chain hydrocarbons, representative lower-boiling-point normal-chain hydrocarbons, and representative higher-boiling-point normal-chain hydrocarbons, were calculated by summation of individual MDLs in a given group.

Hydrocarbon analyses were performed on sections of sediment cores from the first collection period and on surface sediments from the second collection period. A slight uncertainty in the precise measurement of 1-cm-thick core sections and 1-cm-deep surface scoops was inevitable because of the complex nature of the sediment matrix in the Arthur Kill marshes. Since this uncertainty varies from station to station and from site to site in an unknown way, the correction factors to compensate for this uncertainty could not be determined. This uncertainty is presumed to be minimal and insignificant in the interpretation of the data.

Statistics

One-half of the MDL value was used for "not detected" values for the purpose of statistical analyses. Since replicate sediment samples were not collected for any given station, intrastation differences could not be examined. The nonparametric, Kruskal-Wallis, one-way, analysis-of-variance (ANOVA)-on-ranks test was used to detect differences among sites with respect to hydrocarbon concentrations. If differences were detected in the Kruskal-Wallis test, then pairwise, multiple-comparison tests (i.e., Dunn's and Student-Newman-Keuls) were performed, post hoc, to isolate the group(s) that differed from others. Correlation analyses were performed to examine if there existed any relationships between: 1) TPH and TOC in sediments, 2) TPH and lipid contents of mussels, 3) TPH in mussels and TPH in sediments, 4) TPH in mussels and TOC in sediments, 5) TPH in mussels and length of mussels, and 6) length of mussels and lipid content of mussels.

RESULTS

Hydrocarbons in Sediment Core Sections

The concentrations for individual and total hydrocarbons detected in the core sections from the cores collected in September 1996 were compared. These comparisons were done between sections of the same core and sections of cores from other stations from the site.

Old Place Creek -- Oiled and Replanted Site

When the GC-FID chromatograms were integrated over the entire envelope of peaks ranging from n-C₈ to n-C₄₀, the TPH concentrations ranged from "not detected" to 3280 µg/g (Figure 17). The TPH concentrations were highest for the bottom two core sections from Station D (Appendix Table D1; 3280 µg/g for the 3-4 cm deep section, and 2910 µg/g for the 4-5 cm deep section). The next highest concentration of about 1000 µg/g was for 2-3 cm deep section from Station D and the bottom-most section from Station B.

Stations A and C

With a few exceptions, most individual hydrocarbons were below the MDL values in sediment core sections from the first collection period. The hydrocarbon n-C₃₁ was detected in all core sections from Station C (Appendix Table D1).

Station B

Hydrocarbons n-C₁₄ to n-C_21, including pristane and phytane, were most consistently detected in the bottom four core sections from Station B (Appendix Table D1). Hydrocarbons n-C₃₁ and n-C₃₂ were detected in three out of five core sections at this station. Other hydrocarbons were detected only occasionally. The highest TPH concentration was generally found in the bottom core section; the TPH concentration in the top section was negligibly small. The TIPH and TPH values in the 2-3 cm deep core section and the 3-4 cm deep core section were similar, suggesting that these two sections are actually subsamples of one contiguous sediment section.

The CPI value of 1.22 for the 3-4 cm deep core section indicated a slight contribution of biogenic hydrocarbons to the petroleum hydrocarbons. Ratios of pristane to phytane, pristane to n-C_17, and phytane to n-C₁₈ were similar in the 2-3 cm deep core section and the 3-4 cm deep core section, indicating identical and approximately equally degraded hydrocarbon mixture in these core sections.

Station D

Hydrocarbons n-C₁₁ to n-C₂₁, including pristane and phytane, were most consistently detected in the bottom three core sections from Station D. There was no particular pattern related to the other hydrocarbons. The TPH concentrations in the 3-4 cm deep core section and the 4-5 cm deep core section were approximately three times greater than those in the 2-3 cm deep core section. The TPH concentrations in the top two core sections were negligible.

The CPI index for the 2-3 cm deep core section was significantly >1.0, indicating a contribution from the odd-numbered biogenic hydrocarbons. The CPI indices for the 3-4 cm deep core section and the 4-5 cm deep core section were approximately 1.0, indicating hydrocarbons of petrogenic origin. Ratios of phytane to n-C₁₈ in the bottom three sections were similar, indicating the presence of a similarly degraded hydrocarbon mixture. Ratios of pristane to phytane, and of pristane to n-C₁₇, were inconclusive, probably due to the biogenic contribution of pristane.

Con Ed Tower -- Oiled and Unplanted Site

Individual as well as TPH concentrations in core sections from the Con Ed Tower marsh were generally higher than those from all other sections analyzed in this study (Figure 18; Appendix Table D2). The lowest hydrocarbon concentrations were found in the top three sections from Station C, and n-C₃₆ was consistently absent in all core sections.

Stations A and B

With the exception of n-C₃₆, the target hydrocarbons were generally detected in sections of sediment cores from Stations A and B. The TPH concentrations increased with depth for Station A. A similar trend was observed for Station B except for the TPH concentration in the bottom-most section which was between the concentrations in the top two core sections (Figure 18; Appendix Table D2). The TPH concentrations in core sections from Station A and Station B were higher than the TPH concentrations in the corresponding sections from Station C and Station D.

A CPI value of about 1.0 in the top four core sections from Station A and the 0-1 cm, 1-2 cm, and 3-4 cm core sections from Station B indicated hydrocarbons related to a petroleum product. A CPI value of 1.0 is considered to have a petrogenic origin. The CPI value increases with contributions from the biogenic sources. A clear-cut differentiation between petrogenic and biogenic origins can be subjective, although a CPI value >3 is considered to be dominated by the biogenic sources (Farrington and Tripp 1977; Sauer and Uhler 1994; Ramirez 1997). Higher values of CPI in the bottom-most core sections indicated the contributions of hydrocarbons of biogenic origin. Based upon pristane:phytane ratios, Stations A and B appear to have experienced input of different petroleum products in different core sections. Both stations showed similar patterns of pristane:phytane ratios in core sections of similar depths.

Generally, lower WI values for the bottom sections compared to the top section indicated weathering of lighter hydrocarbons in the bottom core sections for Stations A and B.

Station C

With a few exceptions, individual hydrocarbons were not detected in the top two core sections from Station C (Appendix Table D2). With two exceptions, the hydrocarbons n-C₉ to n-C₁₄ were not detected in any core sections. The hydrocarbon n-C₁₇ was the only one detected in all core sections. The TPH concentrations increased with depth of the sediment core.

A CPI value of about 1.0 for the bottom three core sections indicated hydrocarbons of petroleum origin.

The low WI value of 0.19 for the bottom-most section indicated high degradation of lighter hydrocarbons.

Station D

Besides the total absence of hydrocarbons n-C₃₁, n-C₃₃, and n-C₃₆, the distribution of other hydrocarbons did not exhibit any particular pattern in the sediment core sections from Station D (Appendix Table D2). An increasing gradient of TPH concentrations was observed with depth.

A CPI value of near 1.0 for the middle three core sections indicated the presence of petroleum-related hydrocarbons in these sections. A higher value of CPI for the 4-5 cm deep core section indicated a contribution of hydrocarbons of biogenic origin.

Low WI values for the middle three sections indicated degradative losses of lighter hydrocarbons. The bottom-most core section seemed less weathered than the middle three sections.

Mill Creek -- Reference Site

With the exception of hydrocarbons n-C₂₉, n-C₃₁, n-C₃₃, and n-C₃₆, a large majority of hydrocarbons were not detected in the sediment core sections from Stations A-C (Figure 17; Appendix Table D3). The bottom three sections from Station D generally contained low concentrations of hydrocarbons n-C₁₃ to n-C₂₀, including pristane and phytane. The hydrocarbon n-C₁₇ was present in all core sections from Station D.

The TPH concentrations for Stations A-C were all below the MDL of 181 µg/g. The TPH concentrations in the bottom three sections from Station D were higher than the concentration in the 1-2 cm deep core section (Figure 17). The TPH concentration in the top core section from Station D was below the MDL.

Because many hydrocarbons were not detected, CPI and WI values were noncalculable, unreliable, or inconclusive.

Hydrocarbons in Surface Sediments

Surface sediments from Old Place Creek and Con Ed Tower marshes from the May 1997 collection were compared with the top core sections from sediment cores collected in September 1996. Mill Creek surface sediments were not analyzed for the 1997 collection, thus there is no comparison for that site. Concentrations of individual as well as total hydrocarbons in surface sediments from Old Place Creek and Con Ed Tower were generally similar to those in the top sections of sediment cores from the 1996 collection (Figure 18). The average TPH concentration from the top core sections from Con Ed Tower was higher than the average concentrations from all other top core sections and surface sediments (Figure 19).

Old Place Creek -- Oiled and Replanted Site

Except for a few isolated values, the concentrations of individual hydrocarbons in all surface sediment samples from Old Place Creek were below the MDL values (Figure 17; Appendix Table D1). The TPH concentration was above the MDL of 181 µg/g only for the surface sediment from Station B.

Con Ed Tower -- Oiled and Unplanted Site

Except for a few hydrocarbon values for Station A, and some isolated values for Stations C and D, the concentrations of individual hydrocarbons in surface sediments from Con Ed Tower were below the MDL values (Figure 18; Appendix Table D2). Except for Station A, the TIPH concentrations in surface sediments from all stations were below the MDL of 59 µg/g.

Sandy Hook -- Reference Site

One surface sediment sample from Sandy Hook was analyzed during sediment MDL determination. With the exception of hydrocarbons n-C₂₉, n-C₃₁, and n-C₃₂, the concentration of all other individual hydrocarbons in surface sediment samples from Sandy Hook were below the MDL values (Appendix Table D4). The TPH and TIPH concentrations were below the MDL of 181 and 59 µg/g, respectively.

Hydrocarbons in Ribbed-Mussels

Concentrations of individual hydrocarbons (TIPH) in almost all ribbed-mussels analyzed in this study were low, and the sum of n-C₈ to n-C₄₀ hydrocarbons was in the low µg/g range.

The average TPH concentrations in mussels from both collection periods are compared for each site in the Arthur Kill and Sandy Hook marshes (Figure 20). The TPH concentrations overall ranged from 20.6 to 541 µg/g. These TPH values included target hydrocarbon analytes and other unidentified compounds assumed to be a variety of branched-chain hydrocarbons and cyclic hydrocarbons. Although the significance of the method used to determine TPH concentrations in mussels is unclear, the method permitted the correlation analyses of mussel and sediment data.

Old Place Creek -- Oiled and Replanted Site

With the exception of one sample, hydrocarbons n-C₁₀ to n-C₁₆, n-C₁₈, n-C₂₀, n-C₂₂ to n-C₂₅, n-C₂₉, and n-C₃₀ were absent in mussels from both collection periods in the Old Place Creek marsh (Appendix Table D5). For the first collection period, the hydrocarbon patterns in mussels were dominated by the heavier hydrocarbons, suggesting exposure to highly weathered petroleum mixtures. Comparatively few heavier hydrocarbons were detected in mussels from the second collection period.

For the second collection period, the TPH concentration was above the MDL of 53.6 µg/g for only one mussel.

The CPI values >1.0 suggested the contribution of odd-numbered biogenic hydrocarbons in mussels from the first collection period.

Con Ed Tower -- Oiled and Unplanted Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₆, n-C₁₈, n-C₂₀, n-C₂₂, and n-C₂₄ to n-C₃₀ were absent in mussels from both collection periods from the Con Ed Tower marsh (Appendix Table D6). For the first collection period, the hydrocarbon patterns in mussels were dominated by heavier hydrocarbons, suggesting exposure to highly weathered petroleum mixtures. Comparatively few heavier hydrocarbons were detected in mussels from the second collection period.

The CPI values >1.0 in mussels from both collection periods suggested contributions of odd-numbered biogenic hydrocarbons.

Saw Mill Creek North -- Oiled and Replanted Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₅, n-C₁₈, n-C₂₀, and n-C₂₄ to n-C₃₀ were absent in mussels from both collections from Saw Mill Creek North (Appendix Table D7). For the first collection period, the hydrocarbon patterns in mussels were dominated by heavier hydrocarbons, suggesting exposure to highly weathered petroleum mixtures.

The CPI values >1.0 in mussels from both collection periods indicated contributions of odd-numbered biogenic hydrocarbons.

Saw Mill Creek South -- Oiled and Unplanted Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₆, n-C₁₈ to n-C₂₀, n-C₂₂ to n-C₂₆, n-C₃₀, n-C₃₄, n-C₃₇, and pristane were absent in mussels from both collection periods from Saw Mill Creek South (Appendix Table D8). For the first collection period, mussel averages for individual hydrocarbons were mostly below the MDL values.

The TPH concentrations were comparatively higher in mussels from the second collection period (Figure 20).

Higher CPI values for two mussels from the second collection period indicated contributions of odd-numbered biogenic hydrocarbons.

Tufts Point -- Reference Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₆, n-C₁₈, n-C₂₀, n-C₂₂ to n-C₂₅, n-C₃₀, pristane, and phytane, were absent in mussels from both collections from Tufts Point (Appendix Table D9). For the first collection period, hydrocarbon patterns in mussels were dominated by heavier hydrocarbons, suggesting exposure to highly weathered petroleum mixtures. Relatively few heavier hydrocarbons were detected in mussels from the second collection period.

The CPI values did not exhibit any particular trend for mussels in the first collection period. The CPI values >1.0 for four mussels from the first collection period indicated contributions of odd-numbered biogenic hydrocarbons.

Mill Creek -- Reference Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₅, n-C₁₈, n-C₂₀, n-C₂₃ to n-C₃₀, and pristane were absent in mussels from both collection periods from Mill Creek (Appendix Table D10). For the first collection period, hydrocarbon patterns in mussels were dominated by heavier hydrocarbons, suggesting exposure to highly weathered petroleum mixtures. Relatively few heavier hydrocarbons were detected in mussels from the second collection period.

The CPI values >1.0 in mussels from both collections indicated contributions of odd-numbered biogenic hydrocarbons.

Sandy Hook -- Reference Site

With a few exceptions, hydrocarbons n-C₁₀ to n-C₁₆, n-C₁₈, n-C₂₀, n-C₂₃ to n-C₃₆, pristane, and phytane were absent in mussels from Sandy Hook (Appendix Table D11).

A CPI value >1.0 indicated contributions of odd-numbered biogenic hydrocarbons in one mussel sample. Component hydrocarbons for the determination of CPI values were below the MDL values in other mussels from Sandy Hook.

Lipids in Mussels

Lipid contents of ribbed-mussels varied from mussel to mussel and ranged from 0.4 to 2.97% (Appendix Table B2). While the average lipid contents in mussels from Old Place Creek, Saw Mill Creek North, Tufts Point, and Mill Creek from the September 1996 collection were greater than those from the May 1997 collection, a reverse trend was observed for mussels from Con Ed Tower and Saw Mill Creek South.

DISCUSSION

Hydrocarbons in Sediments

Surface Sediments

The TIPH and TPH concentrations in surface sediments from three Arthur Kill marsh sites from both collection periods and in those from Sandy Hook varied by station, site, and collection period (Table 9), and the concentrations exhibited non-normal distributions (P = <0.001). Median TIPH and TPH values for Con Ed Tower surface sediments from the first collection period were higher than the corresponding values for all other sites and collection periods (Figure 21 for TPH). Since only one data point was collected for Sandy Hook, further statistical analyses were limited only to the Arthur Kill marshes.

In a nonparametric, Kruskal-Wallis, one-way, ANOVA-on-ranks test, the differences in median values among Arthur Kill sites were greater than would be expected by chance. Therefore, the median TPH values were considered significantly different (H = 10.396 with 4 degrees of freedom, P = 0.034). The Kruskal-Wallis, one-way, ANOVA-on-ranks test examines the hypothesis of no difference between several treatment groups, but does not determine which groups may be different, or the size of any differences.

Dunn's all-pairwise, multiple-comparison test was performed, post hoc, to isolate the sites that differed from the others. The TPH concentration in surface sediments from Con Ed Tower from the first collection period was found to be significantly different from surface sediments from Mill Creek (difference of ranks = 11.000, P = 5, Q = 2.968, and P <0.05). However, this difference was insignificant when the data were analyzed, post hoc, by the Student-Newman-Keuls all-pairwise, multiple-comparison test. The lower power of the Kruskal-Wallis, one-way, ANOVA-on-ranks test (P = 0.034) apparently resulted in contradictory results from two separate, all-pairwise, multiple-comparison tests. The numerical difference between these two sites was therefore considered to be a borderline significant difference. Statistical differences in TPH concentrations in surface sediments were not detected among other sites or other collections.

The TPH and TOC in surface sediments from Old Place Creek, Con Ed Tower, and Mill Creek had a correlation coefficient value (r) of 0.756 (P = 0.05) and a negative intercept on the Y-axis (Figure 22). Thus, total hydrocarbons in sediments increased with TOC, but in a proportion less than the corresponding increment in the TOC value.

Sediment Cores

Average TPH and TIPH concentrations for individual sediment cores from Old Place Creek, Con Ed Tower, and Mill Creek varied by station and site (Table 10). In contrast to the surface sediments, the TPH concentrations in sediment cores exhibited a normal distribution pattern (P = 0.057). The differences in median TPH concentrations among the three sites were greater than would be expected by chance in a parametric ANOVA test (P = 0.003), as well as in nonparametric Kruskal-Wallis, one-way, ANOVA-on-ranks test (H= 8.140 with 2 degrees of freedom, P = 0.005).

The average TPH concentration for sediment cores from Con Ed Tower was significantly higher than that for Mill Creek in a post hoc Dunn's all-pairwise, multiple-comparison procedure (P <0.05). The Con Ed Tower average concentration was significantly higher than both Mill Creek and Old Place Creek average concentrations in post hoc Tukey test (P <0.05) and Student-Newman-Keuls test (P <0.05).

A statistically significant difference in average TPH was not detected between sediment cores from Mill Creek and Old Place Creek (P <0.05).

Hydrocarbons in Mussels

The TIPH and TPH concentrations in mussels from six Arthur Kill marshes from both collection periods and in those from Sandy Hook varied by site and collection period (Table 11), and exhibited non-normal distributions (P = <0.001). In nonparametric Kruskal-Wallis, one-way, ANOVA-on-ranks test, the differences in median TIPH and TPH values for mussels from different marshes from different collection periods were greater than the differences that would be expected by chance. Median values of TIPH and TPH were thus found to be significantly different (TIPH: P = 0.025, H = 24.677 with 13 degrees of freedom; TPH: P = 0.041, H = 21.709 with 12 degrees of freedom). Since we analyzed five mussels at each site in the Arthur Kill, and seven mussels in the Sandy Hook marsh, the group sizes available for statistical comparison became unequal. In Dunn's all-pairwise, multiple-comparison test, the only available post hoc test for isolating groups of unequal size, no mussel groups were significantly different from one another.

The TPH concentrations in mussels from Tufts Point and Saw Mill Creek North covaried with lipid content (Tufts Point: r = 0.875, P = 0.05, Figure 23B; Saw Mill Creek North: r = 0.872, P = 0.05, Figure 24A). Similar correlation was not detected for mussels from other marsh sites. Surprisingly, there was no correlation between TPH concentrations in mussels and TPH concentrations in sediments for both collection periods (r = 0.101, P = 0.05). Also, there was no correlation between TPH concentrations in mussels and TOC concentrations in sediments (r = 0.084, P = 0.05 for September 1996 collection; r = 0.084, P = 0.05 for May 1997 collection). Except for a negative correlation between TPH concentrations in mussels and mussel length for Tufts Point (r = 0.746, P = 0.05, Figure 23A), there was no relationship between mussel length and TPH concentrations in mussels. Lipid content and mussel length correlated positively for Mill Creek (r = 0.707, P = 0.05, Figure 24C); however, they correlated negatively for Saw Mill Creek South (r = 0.787, P = 0.05, Figure 24B) and Tufts Point (r = 0.629, P = 0.05, Figure 23C).

CONCLUSIONS

The TPH concentrations in surface sediments from Mill Creek (i.e., a reference site) were numerically the lowest, those from Old Place Creek (i.e., an oiled and replanted site) were intermediate, and those from Con Ed Tower (i.e., an oiled but unplanted site) were the highest. Residual oil can easily be seen, felt, and smelled in the sediments at the latter site. The lower background levels at the Mill Creek and Old Place Creek sites may be due to oxidation and weathering of the oil, perhaps caused by the physical disturbance of planting (at Old Place Creek) and by the mineralization of oil by microbes around the roots of S. alterniflora. For the 1996 collection, surface sediments from Con Ed Tower and Mill Creek were statistically different in one post hoc test; however, the power of this test was considerably low (P = 0.034). Surface sediments from other sites were not statistically different from one another.

Hydrocarbon patterns and concentrations in sediment core sections varied by core section for a given station within a given site, suggesting heterogeneity of sediment composition, sediment deposition, and possibly, oil spillage chronology. Deeper core sections of Con Ed Tower sediments generally contained higher levels of hydrocarbons compared to the surface and subsurface core sections. The core average for TPH concentrations in sediment cores from Con Ed Tower was significantly higher than that in Mill Creek, and possibly to a smaller degree, than that in Old Place Creek.

The TPH concentrations in mussels from all Arthur Kill sites and the Sandy Hook marsh were at low levels, these concentrations were not significantly different, and there was no temporal trend for the two collection periods. When detectable concentrations were present, the mussel hydrocarbon patterns were dominated by heavier hydrocarbons, suggesting the exposure of these mussels to the highly weathered petroleum mixtures.

Lack of a distinct hydrocarbon pattern in any sediment or mussel sample may have resulted from a combination of factors, including extensive weathering of diesel fuel oil spilled in January 1990, and other reported and unreported oil spills in the Arthur Kill.

The TPH concentrations in sediments correlated with TOC concentrations in sediments with a correlation coefficient of 0.763. The TPH concentrations in mussels correlated with lipid content for Tufts Point and Saw Mill Creek North only. An absence of correlation between either TPH or TOC concentrations in sediments and TPH concentrations in mussels suggests a limited utility of this technique for the monitoring of old petroleum spills.

Except for Tufts Point mussels, the TPH concentrations did not correlate with mussel length, which contradicted our assumption that the hydrocarbon concentration is directly proportional to the mussel length and its age. Given that oil spills occur relatively frequently in the Arthur Kill, the coincidental timing of the sampling with the timing, location, and extent of an oil spill appears to be a major determining factor in finding hydrocarbon contaminants in mussels. The factor of chronic exposure of mussels to low levels of hydrocarbons in relatively pristine habitats that plausibly leads to gradual biomagnification of contaminants and a positive length (age) - contaminant relationship appears to be less significant for the Arthur Kill mussels.

The CPI of about 1.0 for the top sections of sediment cores from Con Ed Tower suggests petroleum origin, possibly from fresh input(s). The higher CPI values in the bottom sediments indicated biogenic hydrocarbon contributions.

Except for the first collection period for Tufts Point, the CPI for all ribbed-mussels was >1.0, indicating contributions of biogenic hydrocarbons.

Ratios of pristane to phytane, pristane to n-C₁₇, and phytane to n-C₁₈ indicated degradation of normal-chain hydrocarbons, and were useful in discerning petroleum origins in some sediment core sections.

Generally lower values of WI in the bottom sediment core sections indicated weathering losses of lower-boiling-point petroleum hydrocarbons.

REFERENCES CITED

Atlas, R.M. 1981. Microbial degradation of petroleum hydrocarbons: an environmental perspective. Microbiol. Rev. 45:180-209.

Atlas, R.M.; Boehm, P.D.; Calder, J.A. 1981. Chemical and biological weathering of oil from the Amoco Cadiz oil spillage within the littoral zone. Estuarine Coastal Mar. Sci. 12:589-608.

Blumer, M.; Ehrhardt, M.; Jones, J.H. 1973. The environmental fate of stranded crude oil. Deep-Sea Res. 20:239-259.

Blumer, M.; Guillard, R.R.L.; Chase, T. 1971. Hydrocarbons of marine phytoplankton. Mar. Biol. (Berlin) 8:183-189.

Broman, D.; Colmsjö, A.; Ganning, B.; Näf, C.; Zebühr, Y.; Östman, C. 1987. `Fingerprinting' petroleum hydrocarbons in bottom sediments, plankton, and sediment trap collected seston. Mar. Pollut. Bull. 18:380-388.

Choiseul, V.; Wilson, J.G.; Nixon, E. 1998. The distribution of hydrocarbons on the east and south-west Irish coasts and in the Liffey Estuary. Biol. Environ. 98B(2):75-86.

Cripps, G.C. 1989. Problems in the identification of anthropogenic hydrocarbons against natural background levels in the Antarctic. Antarct. Sci. 1:307-312.

Douglas, G.; Hall, P.B.; Bowler, B.; Williams, P.F.V. 1981. Analysis of hydrocarbons in sediments as indicators of pollution. Proc. R. Soc. Edinb. Sect. B (Biol.) 80B:113-34.

Douglas, G.S.; Bence, A.E.; Prince, R.C.; McMillan, S.I.; Butler, E.L. 1996. Environmental stability of selected petroleum hydrocarbon source and weathering ratios. Environ. Sci. Technol. 30:2332-2339.

Douglas, G.S.; Uhler, A.D. 1993. Optimizing EPA methods for petroleum-contaminated site assessments. Environ. Test. Anal. May/June 1993:46-53.

EPA [U.S. Environmental Protection Agency]. 1984. Rules and regulations. Appendix B to Part 136--definition and procedure for the determination of the method detection limit. Revision 1.11. Fed. Regist. 49(209):198-199.

Farrington, J.W.; Meyers, P.A. 1975. Hydrocarbons in the marine environment. In: Eglington, G., ed. Environmental chemistry. London, England: The Chemical Society; p. 109-136.

Farrington, J.W.; Tripp, B.W. 1977. Hydrocarbons in western North Atlantic surface sediments. Geochim. Cosmochim Acta 41:1627-1641.

NJDEP [New Jersey Department of Environmental Protection]. 1995. Quantification and identification of semi-volatile petroleum products in water, soil, sediment and sludge. Revision 2. N.J. Dept. Environ. Protect. Off. Qual. Assur. Anal. Meth. Doc. OQA-QAM-018-10/91.

NRC [National Research Council]. 1985. Oil in the sea: inputs, fates and effects. Washington, DC: National Academy Press; 601 p.

Prahl, F.G.; Bennet, J.T.; Carpenter, R. 1980. The early diagenesis of aliphatic hydrocarbons and organic matter in sedimentary particulates from Dabob Bay, Washington. Geochim. Cosmochim. Acta 44:1967-1976.

Ramirez, P., Jr. 1997. Environmental contaminants in the aquatic bird food chain of an oil refinery wastewater pond. U.S. Fish Wildl. Serv. Contam. Rep. R6/712C/97; 28 p. Available from: U.S. Fish and Wildlife Service, Cheyenne, WY.

Sauer, T.C.; Uhler, A.D. 1994. Pollutant source identification and allocation: advances in hydrocarbon fingerprinting. Remediation 5:25-46.

Wang, Z., Fingas, M. 1995. Differentiation of the source of spilled oil and monitoring the oil weathering process using gas chromatography-mass spectrometry. J. Chromatogr. A 712:321-343.

Wang, Z.; Fingas, M. 1997. Developments in the analysis of petroleum hydrocarbons in oils, petroleum products, and oil-spill-related environmental samples by gas chromatography. J. Chromatogr. A 774:51-78.

Wang, Z.; Fingas, M.; Sergy, G. 1994. Study of 22-year old Arrow oil samples using biomarker compounds by GC/MS. Environ. Sci. Technol. 28:1733-1748.

Wang, Z.D.; Fingas, M. 1994. Study of the effects of weathering on the chemical compositions of ASMB oil. In: Proceedings of the 17^th Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, June 8-10, 1994, Vancouver, BC, Canada. Ottawa, ON: Environment Canada; p. 133-172.

To Chapter 4