The Environmental Significance of Sediment Surface Area as a Controlling Factor in the Preservation of Polychlorinated Dibenzo-P-Dioxins and Dibenzofurans (PCDD/PCDF) in Sediments Adjacent to Woodfibre Pulp Mill, Howe Sound, British Columbia

A sediment core was retrieved from an area adjacent to a Pulp and Paper Mill in Howe Sound, British Columbia, in order to examine the accumulation dioxins (PCDDs) and furans (PCDFs). Downcore distribution of TOC in the bulk samples is relatively uniform (0.5–1.7 wt. %). Bulk PCDD/F concentration shows selective enrichment and depletion at specific sediment horizons, and a low to moderate correlation with surface area (r2 = 0.23–0.54). TOC in size fractionated sediments ranges from 0.3–11 wt. % and shows a moderate correlation with surface area (r2 = 0.51). The relationship between PCDD/Fs and surface area is congener specific, ranging from no significant correlation (TCDD; r2 = 0.05), to a good correlation (i.e., OCDF; r2 = 0.74). Results indicate that both dioxin and furan concentrations are related to organic matter concentration, molecular chlorination and sediment surface area.

4 spectrometer operating at 10,000 resolving power or higher during the entire chromatographic run; (2) the retention time of the specific peaks was within three seconds to the predicted time obtained from analysis of authentic compounds in the calibration standards; (3) the peak maxima for both characteristic isotopic ions of a specific congener coincided within two seconds; (4) the observed isotope ratio of the two ions monitored per congener were within +15% of the theoretical isotope ratio; (5) the signal-to-noise ratio resulting from the peak response of two corresponding ions was greater or equal to three for proper quantification of the congener.  Sample preparations were carried out in the regional dioxin laboratory at the Institute of Ocean Sciences (IOS), Sidney, BC, Canada. in order to reduce any potential sample contamination prior to organochlorine analysis. Bulk sediment samples were firstly analyzed to determine the PCDD/PCDF concentrations. Bulk sediment samples were homogenized unfrozen and 2 g aliquots were removed for moisture determinations. Analytical samples, approximately 10 g wet weight (w.w.), were dried with 125 g Na 2 SO 4 in a mortar and transferred into the glass thimble of the Soxhlet where they were spiked with a mixture of 13 C 12 -labeled PCDD/Fs, a surrogate internal standard supplied by Cambridge Isotope Laboratories (Andover, MA, USA). The composition of the surrogate internal standard mixture and the concentrations are given in Table 1. The spiked samples were Soxhlet extracted for 16 h with 350 mL of toluene/acetone (80:20); washed with 40 mL of KOH, 80 mL of high performance liquid chromatography (HPLC) grade water and subsequently with 10 mL of H 2 S0 4 . The solvents were finally removed by rotary evaporation and the samples were reconstituted in 10 mL of DCM/hexane (1:1). This analytical procedure was repeated for size fractionated sediment samples. Table 1. Composition of internal standard surrogate mixtures used to spike all samples analysed.
The mass spectrometer was operated at 10,000 resolution under positive EI conditions (35 eV electron energy) and data were acquired in the Single Ion Resolving Mode (SIR). Two or more ions, M + and M 2+ in most cases, of known relative abundances, were monitored for each molecular ion cluster representing a group of isomers, and two for each of the 13 C 12 -labeled surrogate standards. Compounds were identified only when the GC/HRMS data satisfied all of the following criteria: (1) two isotopes of the specific congeners were detected by their exact masses with the mass spectrometer operating at 10,000 resolving power or higher during the entire chromatographic run; (2) the retention time of the specific peaks was within three seconds to the predicted time obtained from analysis of authentic compounds in the calibration standards; (3) the peak maxima for both characteristic isotopic ions of a specific congener coincided within two seconds; (4) the observed isotope ratio of the two ions monitored per congener were within +15% of the theoretical isotope ratio; (5) the signal-to-noise ratio resulting from the peak response of two corresponding ions was greater or equal to three for proper quantification of the congener.
The concentrations of identified compounds and their minimum detection limits (MDLs) were calculated by the internal standard method using mean relative response factors determined from calibration standard runs made before and after each batch of samples were run. The specific compounds analyzed are listed in Table 2. The criteria for identification and quantification, and the quality assurance and quality control measures undertaken for the sample workup and the GC/HRMS analysis of all the analytes of interest were based on procedures described in the Environment Canada [41,42] protocols. Fractionation was carried out using a combination of wet sieving and settling techniques based on Stokes law [43]. Selected samples were chosen, based on highest bulk PCDD/PCDF concentration, for size fractionation into five size fractions (<2 µm, 2-10 µm, 10-20 µm, 20-63 µm and >63 µm). These selected samples for geochemical analyses are from the core depths of 14-16 cm, 20-22 cm, 34-36 cm and 46-48 cm. Three other sediment intervals, of similar depth to those used for organochlorine analysis, were separated using wet sieving techniques [43] in order to examine down core mass distribution, and to choose suitable size fractions that contained a mass sufficient for all the analyses. Samples were sieved at 63 µm, and the remainder was analysed using a Micromeritics ® Sedigraph 5100 (Micromeritircs, GA, U.S.A) and results were combined to produce mass profiles for the four samples ( Figure 2).
The sieve and spatulas used for sample fractionation were all stainless steel, and were triple solvent rinsed (Acetone 3 times, Toluene 3 times, Hexane 3 times) and baked at 150 • C for 1 h etween samples. All glassware used during the settling procedure was washed, triple solvent rinsed, baked overnight at 350 • C, and triple solvent rinsed again and dried inside a fume hood. All equipment used was covered/sealed in hexane rinsed aluminum foil immediately after cleaning. Narrow bore, flexible Teflon tubing (internal diameter of 2 mm), in conjunction with a disposable Pasteur pipette connected via rubber tubing to a water aspirator and was used to siphon off different size fractions during settling. The coarsest fraction, >63 µm, was sieved off, and the remaining solids, suspended in distilled water, were transferred to 2 L glass beakers for settling. Once separated, sediment size fractions were stored in 250 mL solvent rinsed glass jars with Teflon lined lids, and frozen prior to analysis. To remove particles from suspension in the <2 µm fraction, the water was centrifuged at 3500 rpm using a Beckman 64R centrifuge (Beckman Coulter, Indianapolis, IN, U.S.A). Four, 1 liter polycarbonate centrifuge bottles were adapted by removing the neck of the bottle, so that they could be lined with hexane rinsed aluminum foil to prevent contamination (the polycarbonate centrifuge bottles could not be solvent cleaned in the same manner as glassware). Any leakage through the foil was discarded. The water was then stored in solvent rinsed 4 liter bottles and stored in the fridge at 4 • C.

6
Indianapolis, IN, U.S.A). Four, 1 liter polycarbonate centrifuge bottles were adapted by removing the neck of the bottle, so that they could be lined with hexane rinsed aluminum foil to prevent contamination (the polycarbonate centrifuge bottles could not be solvent cleaned in the same manner as glassware). Any leakage through the foil was discarded. The water was then stored in solvent rinsed 4 liter bottles and stored in the fridge at 4 °C.
PCDDs, PCDFs were analyzed by GC/HRMS. The instrument was a VG Autospec high resolution mass spectrometer (Micromass, Manchester, UK) equipped with a Hewlett-Packard model 5890 Series II gas chromatograph and a CTC A200S autosampler (CTC Analytics, Zurich, Switzerland). The GC was operated in the splitless injection mode with a 60 m DB-5 fused silica capillary column (0.25 mm I.D. with 0.1 µm film thickness) from J&W Scientific (Folsom, CA, USA). The temperature program for the PCDD/PCDF analysis was as follows the initial column temperature was held at 100 °C for 2 min after injection and increased at 20 °C/min to 200 °C, then at 1 °C/minute to 215 °C, held for 7 min followed by a ramp of 4 °C/minute to 300 °C where it was held for 3 min.

Carbon and Nitrogen Analyses
Total carbon (TC; wt. %) and total nitrogen (TN; wt. %) of bulk and fractionated sediment samples was determined using a Carlo Erba NA-1500 CHN analyzer [44]. The standard error was +/− 1%. Macroscopic organic debris was removed prior to analysis under a binocular microscope. Total organic content (TOC) was determined by difference after the subtraction of inorganic carbon in carbonate form. PCDDs, PCDFs were analyzed by GC/HRMS. The instrument was a VG Autospec high resolution mass spectrometer (Micromass, Manchester, UK) equipped with a Hewlett-Packard model 5890 Series II gas chromatograph and a CTC A200S autosampler (CTC Analytics, Zurich, Switzerland). The GC was operated in the splitless injection mode with a 60 m DB-5 fused silica capillary column (0.25 mm I.D. with 0.1 µm film thickness) from J&W Scientific (Folsom, CA, USA). The temperature program for the PCDD/PCDF analysis was as follows the initial column temperature was held at 100 • C for 2 min after injection and increased at 20 • C/min to 200 • C, then at 1 • C/minute to 215 • C, held for 7 min followed by a ramp of 4 • C/minute to 300 • C where it was held for 3 min.

Carbon and Nitrogen Analyses
Total carbon (TC; wt. %) and total nitrogen (TN; wt. %) of bulk and fractionated sediment samples was determined using a Carlo Erba NA-1500 CHN analyzer [44]. The standard error was +/− 1%. Macroscopic organic debris was removed prior to analysis under a binocular microscope. Total organic content (TOC) was determined by difference after the subtraction of inorganic carbon in carbonate form.
The amount of carbon in carbonate form in each sample was determined using a coulometric titration technique on a CO 2 coulometer (Coulometrics Incorporated, Model 5010, Chattanooga, TN, U.S.A.). The standard error was +/− 5%.

Surface Area
Samples were first oxidized using a combination of hydrogen peroxide (30%) and sodium pyrophosphate (0.1 M) [45] in order to remove organic matter from the samples. This process was carried out in a water bath (70-80 • C) for 48-96 hrs; aliquots of H 2 O 2 were added twice daily until oxidation was complete as marked by the cessation of CO 2 evolution. Samples were then rinsed twice in distilled water to remove any residual inorganic salts and dried. Freeze drying, oven drying and oven drying under vacuum were tested as methods of sample drying prior to analysis; good agreement was found between all three methods.
A Micromeritics ASAP 2010 surface area analyzer was used to measure surface area by N 2 adsorption, using both single and multi-point BET (Brunauer, Emmett, Teller) methods [10,46]. Prior to analysis, samples were degassed at 350 • C for a minimum of 6 hrs. The standard error of the surface area ranged from 2-4%.

210 Pb Dating
Dating of the sediment core was carried out using the 210 Pb method of [47] modified after [48]. One to two grams of sediment was weighed out into acid cleaned microwave digestion vessels and spiked with 208Po to allow for normalization of differences between samples during counting and plating [49]. The samples were then microwave digested using a combination of HNO 3 , HCl and HF, to remove organic material, aluminosilicates and silica, and to liberate 210 Po from the sediment matrix [49]. The digested residue was dried overnight under heat lamps, resuspended in 10% HCl and digested again. The pH was then brought up to 6-9 using NH 4 OH, forming an iron precipitate containing 210 Po, which was stored in 210 mL Nalgene bottles until plating (HDPE).
To plate the samples, the iron precipitate was redissolved in 10% HCI, centrifuged, and the supernatant decanted into acid cleaned 250 mL glass beakers. Sodium citrate (25%, 2 mL), NH 2 OH.HCl (20%, 5 mL) and Bi(NO 3 ) 3 (10%, 1 mL) were added to the supernatant and the pH was raised to 1.5-2.5 using NH 4 OH. The solution was stirred and heated to a temperature range of 90-100 • C, and a polished silver disk in a Teflon holder was inserted into the solution. Plating of 210 Po onto the disks occurred for 4-5 hrs, after which the silver disks were cleaned with concentrated HCl to improve counting efficiency, washed in double distilled water and allowed to air dry. The disks were counted for at least 24 h in an Ortec 576A Multi-channel Analyser (Ortec, Oak Rdige, TN, U.S.A) and Alpha-counter.
The initial 210 Pb dating of the homogenized bulk samples taken from the study core failed to yield a linear chronology. The dating procedure was modified slightly, changing the ratio of sample to 208 Po spike, and repeated. Unfortunately, significant fluctuation in age was found with increasing core depth, which ultimately prevented meaningful temporal correlation with other data.
This dating technique has been successfully applied to other samples recovered from different areas in Howe Sound [33], and reasons for the apparent failure in this case are unclear, but are further discussed in later sections.

Organic Carbon and Nitrogen Analyses
Total organic carbon (C org ) concentrations range from 0.5-1.7 wt. % and TN ranges between 0.04-0.1 wt. %. (Table 3). The highest concentrations of organic carbon occur at the base of the cored sediments (50 cm below the sediment-water interface) and C org concentrations decline to the middle of the cored sediment at 30 cm depth ( Figure 3A). The total nitrogen (TN) shows a similar trend of decreasing concentration from 50 cm to 30 cm depth ( Figure 3B). The C org and TN concentrations both increase from 30 cm to 18 cm depth and no clear down-core trends exist between the depths of 0-18 cm. Table 3. Organic Carbon (C org ), Total Nitrogen (TN), Carbon/Nitrogen Ratio (Cor/N) and Surface Area for bulk sediment samples from the core taken adjacent to Woodfibre Pulp Mill, Howe Sound, British Columbia, subsampled every centimeter for the first ten centimeters and every two centimetres thereafter.

Sediment Interval (cm)
C org (wt. %) TN (wt. %) C org /N Surface Area (m 2 /g)  (Table 3). When plotted against depth ( Figure 4), a similar increasing trend from 50 cm to 30 cm to that seen in Figure 3 is evident. There is no clear trend seen in C org /N values from 0 to 30 cm depth. The spread of values for C org /N is relatively narrow, and a slight overall increase in ratio can be seen with increasing depth (Figure 4).

Organochlorine Compounds
Total PCDD concentrations remains low and constant throughout the cored sediment ( Figure  6a) with the exception of two concentration spikes at 14-16 cm (200 pg/g) and at 46-48 cm (700 pg/g) horizons (Table 4). TCDD (Tetra-), PeCDD (Penta-), HxCDD (Hexa-), HpCDD (Hepta-) and OCDD (Octa-) homologues all share the same concentration profile as total PCDD (Figure 7). HxCDD is present at significantly higher concentrations than the other dioxin homologues ( Figure  7, Inset). If HxCDD is ignored, a positive relationship exists between increasing homologue concentration and the number of chlorine atoms present in the homologue molecular structure ( Figure 7). Total PCDF ( Figure 6) has an identical depth profile to TCDF (Figure 8 Inset). Overall concentrations of furan homologues are on the order of four times lower than their dioxin counterparts (Table 4). TCDF is the most concentrated furan homologue, present in concentrations approximately six times that of the next most concentrated homologue, HpCDF (Table 4). Unlike the dioxin homologues, furan homologues do not show the same relationship between relative concentration of a given homologue and its molecular chlorination (

Total Organic Carbon and Surface Area
Surface area values for bulk sediment samples range from 2.5-12 m 2 /g ( Table 3). Although peaks in surface area values do correlate with some increased TOC concentrations at certain depth intervals, a plot of TOC versus surface area ( Figure 9) shows a moderate positive correlation (r 2 = 0.4) with surface area increasing with organic carbon concentration. Data scatter above and below the regression line, suggests the inorganic component (i.e., mineral grains) does influence the surface area characteristics of the sediments.

PCDD/Fs and Surface Area
When plotted against surface area, total PCDD, total PCDF and the five homologues analysed for each sample show a low to no correlation with r 2 varying from 0.23 to 0.54.

Organochlorine Compounds
Total PCDD concentrations remains low and constant throughout the cored sediment (Figure 6a) with the exception of two concentration spikes at 14-16 cm (200 pg/g) and at 46-48 cm (700 pg/g) horizons (Table 4). TCDD (Tetra-), PeCDD (Penta-), HxCDD (Hexa-), HpCDD (Hepta-) and OCDD (Octa-) homologues all share the same concentration profile as total PCDD (Figure 7). HxCDD is present at significantly higher concentrations than the other dioxin homologues (Figure 7, Inset). If HxCDD is ignored, a positive relationship exists between increasing homologue concentration and the number of chlorine atoms present in the homologue molecular structure (Figure 7). Total PCDF ( Figure 6) has an identical depth profile to TCDF (Figure 8 Inset). Overall concentrations of furan homologues are on the order of four times lower than their dioxin counterparts (Table 4). TCDF is the most concentrated furan homologue, present in concentrations approximately six times that of the next most concentrated homologue, HpCDF (Table 4). Unlike the dioxin homologues, furan homologues do not show the same relationship between relative concentration of a given homologue and its molecular chlorination ( Figure 8). The downcore concentration profiles of individual PCDD/F (Figures 7 and 8) homologues as well as those for total PCDD/F ( Figure 6) concentration are similar to that of TOC ( Figure 3A), with corresponding peaks in concentration at 15, 35 and 45 cm.

Total Organic Carbon and Surface Area
Surface area values for bulk sediment samples range from 2.5-12 m 2 /g ( Table 3). Although peaks in surface area values do correlate with some increased TOC concentrations at certain depth intervals, a plot of TOC versus surface area ( Figure 9) shows a moderate positive correlation (r 2 = 0.4) with surface area increasing with organic carbon concentration. Data scatter above and below the regression line, suggests the inorganic component (i.e., mineral grains) does influence the surface area characteristics of the sediments.

PCDD/Fs and Surface Area
When plotted against surface area, total PCDD, total PCDF and the five homologues analysed for each sample show a low to no correlation with r 2 varying from 0.23 to 0.54.

Carbon and Nitrogen
TOC values for the four samples selected for size fractionation, ranged from 0.3-11 wt. % (Table 5). TN concentrations range from 0.03-0.2 wt.% (Table 5). For the four samples analyzed, the >63 µm fraction contains the highest concentration of both TOC (5-11 wt. %) and TN (0.15-0.2 wt. %; Figure 10a). The <2 µm fraction contains the second highest concentration of TOC and TN, ranging from 1-3 wt. % and 0.8-0.2 wt. % respectively (Table 5; Figure 10a,b). C org /N values for size fractionated samples range from 13-53, with a mean of 23, indicating a dominant terrestrial signature for the organic matter present in the samples. The C org /N ratio increases with increasing grain size (Table 5). A significant increase in C org /N values occurs between the 20-63 µm fraction and the >63 µm fraction. Figure 11 shows TOC plotted against TN with the >63 µm fraction grouped separately from the remaining size fractions. The separation of the >63 µm data from the other size fractions, which all fall on a single regression line (r 2 = 0.8; Figure 11), is the result of significantly higher organic carbon contents in that fraction due to particulate organic matter (OM).
The TOC concentrations for individual size fractions plotted against surface area show data clustering occurs amongst each size fraction (Figure 12a). Two end members exist, the <2 µm fraction which has high surface area values compared to TOC content (Table 5), and the >63 µm fraction which has high TOC content and low surface area values ( Table 5). The remaining size fractions, 2-10 µm, 10-20 µm and 20-63 µm, cluster between the two end members. Figure 12a suggests that no relationship between surface area and TOC content exists (r 2 < 0.1), and that organic matter is not adsorbed to sediment particle surfaces. However, if data from the >63 µm fraction is considered as outliers (because organic matter in this fraction is probably detrital, occurring as discrete particles, and therefore independent of surface area), then the correlation between TOC content and surface area is significantly improved. Figure 12b shows the same data with the >63 µm fraction omitted, producing and r 2 value of 0.51 which suggests a relationship between surface area and TOC concentration via adsorption of organic matter.

Surface Area of Size Fractionated Samples
Surface area values for size fractionated samples range from 1.5-22 m 2 /g, with highest values found in the <2 µm fraction (16-22 m 2 /g) and lowest values in the 20-63 µm fraction (1.5-2.4 m 2 /g; Table 5).
The TOC concentrations for individual size fractions plotted against surface area show data clustering occurs amongst each size fraction (Figure 12a). Two end members exist, the <2 µm fraction which has high surface area values compared to TOC content (Table 5), and the >63 µm fraction which has high TOC content and low surface area values ( Table 5). The remaining size fractions, 2-10 µm, 10-20 µm and 20-63 µm, cluster between the two end members. Figure 12a suggests that no relationship between surface area and TOC content exists (r 2 < 0.1), and that organic matter is not adsorbed to sediment particle surfaces. However, if data from the >63 µm fraction is considered as outliers (because organic matter in this fraction is probably detrital, occurring as discrete particles, and therefore independent of surface area), then the correlation between TOC content and surface area is significantly improved. Figure 12b shows the same data with the >63 µm fraction omitted, producing and r 2 value of 0.51 which suggests a relationship between surface area and TOC concentration via adsorption of organic matter.

Organochlorine Compounds
The majority of trends observed in individual dioxin and furan congeners with respect to grain size and concentration are represented by their parent homologues (Table 2). Thus, concentrations are discussed here with reference to homologues rather than to each individual congener. The concentrations of each homologue and total homologue (PCDD, PCDF) for both dioxins and furans are shown for a selection of sample depths (DF 14-16 cm; DF 20-22 cm; DF 34-36 cm and DF 46-48 cm) and their size fractionated samples (Table 6 and Figures 13-16).  19 Figure 13. Plot showing the changes in concentration of dioxins (A), furans (B) and total homologues (C) between different size fractions at the 14-16cm depth interval. The overall trend is of decreasing concentration with progressively finer grain size.    The selected sediment intervals presented in Table 6

Organochlorine Compounds and Surface Area
Relationships between surface area and homologue concentration in size fractionated samples are highly variable with majority of size fractionated samples and homologues showing no correlation to surface area (Table 7). Only sample DF 46-48, shows a (positive) correlation between OCDF, HpCDF, OCDD and HpCDD with surface area (Figure 17), however the remaining homologues or total dioxin/furan concentrations do not show any correlation with surface area at this depth. DF 46-48 does have the highest concentration of homologues in the sample suite and this be a factor in showing a relationship with surface area. The lack of correlation within the entire sample suite illustrates that surface area of mineral matter does not appear to be a significant factor in the accumulation of dioxins or furans in this sediment. Table 7. Regression co-efficient (r 2 ) values for dioxin/furan/total homologue concentration (pg/g) versus surface area (m 2 /g). Negative values indicate a negative correlation between the two parameters. The number of size fractions indicates how many data points were used to generate the regression coefficient; r 2 values using four size fractions exclude the >63 µm fraction from the calculation, r 2 values using five size fractions include all the fractions.

Organochlorine Compounds and Surface Area
Relationships between surface area and homologue concentration in size fractionated samples are highly variable with majority of size fractionated samples and homologues showing no correlation to surface area (Table 7). Only sample DF 46-48, shows a (positive) correlation between OCDF, HpCDF, OCDD and HpCDD with surface area (Figure 17), however the remaining homologues or total dioxin/furan concentrations do not show any correlation with surface area at this depth. DF 46-48 does have the highest concentration of homologues in the sample suite and this be a factor in showing a relationship with surface area. The lack of correlation within the entire sample suite illustrates that surface area of mineral matter does not appear to be a significant factor in the accumulation of dioxins or furans in this sediment.

PCDD/Fs and Organic Matter
A good correlation exists between PCDD/F and total organic carbon (Corg) concentration in both bulk and size fractionated sediments. Figure 18 shows TCDD through OCDD plotted against Corg concentration for size fractionated samples. TCDD has the weakest correlation, r 2 = 0.28, whilst HpCDD has the strongest correlation, r 2 = 0.63. Figure 19 depicts the same relationship, without the

PCDD/Fs and Organic Matter
A good correlation exists between PCDD/F and total organic carbon (C org ) concentration in both bulk and size fractionated sediments. Figure 18 shows TCDD through OCDD plotted against C org concentration for size fractionated samples. TCDD has the weakest correlation, r 2 = 0.28, whilst HpCDD has the strongest correlation, r 2 = 0.63. Figure 19 depicts the same relationship, without the >63 µm fraction, which was omitted based on previous data that suggested that unlike finer fractions, it is not related to mineral surface area. Regression co-efficient (r 2 ) values are higher for HxCDD through OCDD (r 2 = 0.8; Figure 19), whilst the strength of the relationship decreased for TCDD and PeCDD.

Organic Matter, Total Nitrogen and Surface Area
The concentration of TOC and TN (0.5−1.7 wt. % TOC; 0.04−0.1 wt. % TN; Table 3) in the bulk sediment samples is typical for a paralic environment such as Howe Sound, and is similar to TOC and TN concentrations found to the north in the Squamish Delta sediments [40]. However, the range of TOC concentrations is smaller than values obtained in other cores of comparable length taken further south along Howe Sound [33]. Unlike the cores taken by [33], the core sample taken in this study does not show a progressive decrease of TOC to a background concentration with increasing sediment depth ( Figure 3A), as is often the case [11]. Reasons for this are unclear. Upon visual inspection, the core did not seem to be significantly affected by bioturbation, and this was confirmed during subsampling. The location of the core, chosen based upon its proximity to the Woodfibre Pulp Mill, could also be a contributing factor to the observed TOC profile. The area receives not only sediments from the north, supplied from the distal reaches of the Squamish Delta, but also from the south. A return gyre affects this area, supplying sediments from other areas in the Sound [50]. The proximity of the core to the Woodfibre Pulp Mill might result in a TOC signature that is directly linked to temporal changes in the discharge and composition of mill effluent. As a result, it is likely that a combination of these factors may be the cause of the irregular downcore changes in TOC concentration. This may also explain the ambiguity that arose from 210 Pb dating, where multiple sediment sources could have disrupted the steady accumulation and decay of 210 Pb from a single sediment source [47].
The TOC and TN trends observed in the size fractionated sediments are identical (Figure 10a, b), with the >63 µm fraction containing 55-70% of the TOC and 40-50% of the TN in each sample. The <2 µm fraction contains 15-20% of the TOC and 25-30% of the TN in each sample, with the remainder being distributed between the three intermediate size fractions. The enrichment of TOC in the >63 µm fraction is attributed to the presence of particulate rather than adsorbed OM on the surfaces of mineral grains, and is therefore independent of surface area [40]. However, the OM in the <2 µm fraction, which probably represents approximately one fifth of the OM present in the sample, is most likely adsorbed to mineral surfaces [40] and represents the recalcitrant portion of the sample that is no longer susceptible to remineralization [11].
The C org /N ratio trend for the bulk sediment samples shares a similar downcore profile as TOC ( Figure 3A). C org /N values change from 13 to 23 over the length of the core (Figure 4), which suggests a dominant terrestrial source for the organic matter [51]. However, minor scatter around the regression line in Figure 5 suggests the presence of either an additional type of OM, or a secondary source. C org versus TN relationship for the size fractionated sediments is also variable (Figure 11), particularly in the >63 µm fraction, which appears to have consistently higher TOC than TN concentrations, as shown by the >63 µm regression. This is similar to results from sediment samples collected from the Squamish Delta [40], where the >106 µm fraction falls on a more "carbon rich" regression line. This is attributed to the dominance of particulate or "detrital" OM over adsorbed OM in the coarsest fraction [40]. However, in this study, the >63 µm fraction probably contains both detrital OM and some adsorbed OM due to the larger variation in grain size within this fraction.
Linear regression of surface area and TOC for bulk sediment samples ( Figure 9) explains 40% of the variance. Many studies have found significantly better correlation between these parameters, especially with samples collected from continental shelves [11,52]. Larger degrees of variation between surface area and TOC content is not uncommon in paralic settings [40], and is attributed to variable proportions of detrital OM, and a much larger range of grain sizes in the bulk samples. The relationship between surface area and TOC in size fractionated samples highlights the effects of detrital OM and grain size on the correlation (Figure 12a,b). There is little relationship between surface area and TOC when all size fractions are considered, primarily due to the four samples from the >63 µm, which have higher than expected TOC contents for their surface areas (Figure 12a). These samples also have high surface area values (Table 5) when compared to intermediate size fractions, which may be the result of macroscopic mica flakes (probably muscovite) that are more common in this fraction. The remaining size fractions show a stronger relationship between surface area and TOC (r 2 = 0.51; Figure 12b). However, three of the four <2 µm fraction samples have lower than expected TOC contents than the regression predicts, a phenomenon not uncommon in paralic settings where TOC loadings are often lower than sediments of similar grain size from continental shelf regions [11,53]. Excluding these outliers from Figure 12b further improves the regression, but is probably not representative of the actual relationship between TOC and surface area in the samples.

Controls on Organochlorine Concentrations
As this study is focussed on a single cored interval within the Howe Sound, the discussion on spatial distribution is limited to stratigraphic distribution and will not discuss lateral distribution of the pollutants within the sediments.
Downcore concentration trends in polychlorinated organic pollutants show selective enrichment at specific horizons rather than gradual enrichment or depletion with increasing depth, as is usual with organic-rich sediments [11,54,55]. Thus, concentrations of total PCDD and PCDF are relatively constant except for the 10-25 cm interval and below 42 cm, where large increases in concentration occur (Figure 6a,b). Unfortunately, problems with 210 Pb dating prevented derivation of a chronology for the core. Previously published data [33] from other parts of Howe Sound shows peak concentrations of 8 pg/g, 65 pg/g, and 30 pg/g (TCDF) and 660 pg/g, 480 pg/g and 340 pg/g circa 1970, 1982 and 1989 respectively, but without more accurate information on sedimentation and accumulation rates, and potential surface mixing within the sediments, it is impossible to correlate these dates with peak concentrations in the core. Bioturbation was observed in a similar core retrieved from the eastern arm of Howe Sound [33], suggesting that the sediments in the study core may have been mixed, but visual inspection upon retrieval and subsampling did not reveal evidence of bioturbation. It is also possible that any mixing that may have occurred resulted from the coring procedure and subsequent subsampling, although every effort was made to minimize disturbance during retrieval and processing. However, the complex sedimentary environment that was the likely cause of the disrupted 210 Pb results may also be the reason for the unusual response observed in the organochlorine compounds. This is likely, especially if the organochlorine compounds are related to surface area in the same manner as TOC. Without accurate temporal data it is not possible to relate current concentration levels of dioxins and furans with the initial concentrations in either effluent discharged from the Woodfibre Pulp Mill or atmospheric inputs from other sources. Thus, any change in concentration from the time of initial input cannot be taken into account when assessing preservation of dioxins and furans and their relationship with surface area.
Downcore trends in individual homologues (Figures 7 and 15) are very similar to those of PCDD and PCDF. In both the dioxin and furan families, hexachlorinated homologues are present in much higher concentrations than other homologues. When HxCDD is considered separately, a correlation between chlorination and increasing concentration is observed (Figure 7), suggesting that preservation potential may also be a function of the number of chlorine atoms present in the molecule. This phenomenon is not observed in furans (Figure 15), where the abundances of TCDF and OCDF are reversed, OCDF being the least concentrated of the five homologues, and TCDF being the most concentrated. The lipid content of the TOC may also be a factor in the concentration of homologues as increasing chlorination of PCDD and PCDF increases their lipophilicity [5]. Higher polychlorinated organic compounds may be preferentially concentrated in the sediments due to higher lipid content of the organic matter in the sediment. More research is needed on the composition of the organic matter to determine if this is the case.
There is evidence in the size fractionated sediment data to suggest that organochlorine compounds are selectively adsorbed to inorganic particle surfaces, and that their concentration in the <2 µm fraction increases with depth. A commonality that exists in all four of the size fractionated samples is the high abundance of organochlorine compounds in the >63 µm fraction (Figures 13-16). This is attributed to the presence of particulate OM that is not adsorbed to particle surfaces. DF 14-16 shows depletion of both total and individual homologues in both dioxins and furans with progressively finer grain size, which strongly suggests that surface area adsorption is not the mechanism by which OM and organochlorine compounds are associated at this sediment depth ( Figure 13). However, other homologues from sample DF 14-16 show an opposite trend, with increasing concentrations in finer size fractions. OCDD, HxCDF, HpCDF and OCDF all show minor enrichment in the <2 µm fraction versus the 2-10 µm fraction (Figure 13a,b). Samples DF 20-22 and DF 34-36 show progressive enrichment from the 10-20 µm through to the <2 µm fraction in total PCDD and PCDF, and in a number of individual homologues as well (Figures 14 and 15). This suggests an increasing association of organochlorine compounds with finer sediment fractions and thus increasing importance of surface area as the mechanism by which these compounds are preserved. This trend is more pronounced in sample DF 46-48, which shows increasing concentrations of total PCDD, HxCDD, HpCDD, OCDD, and all of the furan homologues except for TCDF, from the 20-68 µm fraction to the <2 µm fraction ( Figure 16). The increase between the 2-10 µm and <2 µm fractions is significantly greater than increases between coarser fractions, which highlights the significant increase in surface area in the <2 µm fraction compared to coarser fractions. DF 46-48 has the largest concentration of organic carbon, compare the other samples and this organic matter would contain large internal surface area which would adsorb organochlorine compounds.
It is evident from Table 4 ( Figures 6-8) that neither a downcore enrichment nor depletion trend exists for homologue or total homologue concentration in dioxins of furans. However, when concentrations are compared to surface areas, the highest concentrations, which are found in the 14-16 cm, 20-22 cm, 34-36 cm and 46-48 cm intervals, do correspond to the highest observed surface area values in three of the four intervals, which suggests some link between surface area and concentration. Correspondingly, depth intervals 4-5 cm, 24-26 cm and 40-42 cm which contain the lowest measured concentrations for the majority of the compounds analysed also have the lowest surface area values. Regression analyses of the concentration of organochlorine compounds and surface area are poor and mineral surface area is not the only controlling factor in concentration of these pollutants.
The relationship between organochlorine compound concentrations and surface area are poor in all samples. The relationship does improve for sample DF 46-48 which may be due to the higher dioxin and furan concentrations. The majority of the homologues from the 46-48 cm interval have a moderate positive correlation with surface area (Table 7; Figure 17). There are strong positive correlations that exist between the TOC (Corg) content and dioxin and furan concentrations ( Figure 18) with some homologues preferentially absorbing in this fraction (Tetra-Pentachlorinated PCDD/F). The r 2 values also increase for hexa-through octachlorinated dioxin/furan when the >63 µm was removed (0.4/0.6 to 0.8 dioxins, 0.4/0.7 to 0.8, furans; Figure 19). Correspondingly, r 2 values decreased in tetra-and pentachlorinated dioxin/furan when the >63 µm was removed (0.6/0.3 to <0.2, Dioxins, 0.7 to 0.6, Furans; Figure 19). This suggests an intrinsic relationship between tetra-and pentachlorinated dioxin/furan and the >63 µm fraction, whilst other homologues appear to be directly related to finer fractions. The improvement of the relationship between OM and homologue concentration in sediments <63 µm (Figure 19), as well as the strength of the relationship (r 2 ≈ 0.8, PCDD/F) suggests a strong link between OM and homologue concentrations. Similarities in downcore concentration profiles of dioxins and furans to that of TOC ( Figures 3A, 7 and 8) also support this conclusion. When placed in the context of surface area, which is a significant factor in controlling homologue preservation in sediments (Table 7; Figure 17), it appears likely that the relationship is somewhat symbiotic. The organochlorine compounds adsorb to the OM, which can be adsorbed to inorganic particle surfaces or as discrete particles within the sediment.
Although sediment toxicity was not the focus of this study, it is interesting to note that the toxicity of bulk sediment samples ranges from 0.4 ng/kg TEQ to 20.6 ng/kg TEQ, with only four of the bulk samples analysed exceeding the ISQG value of 0.85 ng/kg TEQ. These four samples were those selected for size fractionation: 14-16 cm, 20-22 cm, 34-36 cm and 46-48 cm. None of the TEQ values for these four samples exceeded the PEL value of 21.5 ng/kg TEQ. TEQ values for size fractionated samples fell above the ISQG value, but were usually less than the PEL value. However, several size fractions did exceed the PEL value, most notably in the 46-48 cm depth interval, where the >63 µm fraction and <2 µm fraction had TEQ values of 67 and 40 respectively, suggesting significant contamination.

Conclusions
Analysis of bulk sediment samples taken from a core in Howe Sound, adjacent to the Woodfibre Pulp and Paper Mill, showed organic matter concentrations typical for a paralic environment. The average C org /N ratio of 15 suggests that the dominant source of sedimentary organic matter is terrestrial. Further analyses revealed the presence of tetra-through octachlorinated dioxins and furans, which were likely sourced directly from Woodfibre Pulp Mill via effluent discharge into Howe Sound.
The primary source of organic and inorganic sediments for the Woodfibre area is the Squamish River, which discharges into the north end of Howe Sound via the Squamish Delta. However, complex currents in this area often rework sediments, and can also transport sediments from secondary sources further south in Howe Sound. Other sources of organic and inorganic sediments include the Woodfibre Mill effluent, the barges that deliver unprocessed woodchips to the mill and material directly from the steep terrestrial slopes surrounding Howe Sound. This complex sedimentary setting is the main causal factor in the unusual down core trends seen in TOC and TN concentration, C org /N ratio, chlorinated organic compound concentration, and the failure of 210 Pb dating in producing a reliable chronology for the core.
Surface area values for bulk sediments were typical of a paralic environment, show moderate correlation with organic carbon and organochlorine pollutants, suggesting in turn that sediment surface area is a factor in the retention of both naturally occurring organic material and man-made organic pollutants in the sedimentary record.
Results from the analysis of selected size fractionated samples showed significant enrichment of both TOC and industrial pollutants in the coarsest and finest sediment fractions. Organic matter and associated pollutants present in coarse grained sediments is likely particulate in nature and not adsorbed to particle surfaces. A strong relationship between OM and dioxin/furan concentrations also occurs which supports the theory that pollutants absorb onto organic matter surfaces and remain as discrete particles of both the pollutant and OM adsorption on to mineral surfaces. Some homologues (tetra-and pentachlorinated) are directly related to the coarsest sediment fraction (> 63 µm), whilst other homologues (hexa-through octachlorinated) are concentrated in the finer sediment fractions (<63 µm).
The strength of the relationship between surface area and dioxin and furan concentration is governed by organic matter concentration and relative chlorination. Increasing organic matter concentration increases the concentration of pollutants; while dioxin and furan compounds with higher numbers of chlorine atoms in their molecular structure e.g., hex-octachlorinated PCCD/F, appear to be better preserved through adsorption to particle surfaces. The lack of variation observed in the geochemical composition and mineralogy of sediments in the upper reaches of Howe Sound [40] indicates that the composition of the inorganic sediment fraction is not a factor in the adsorption of organic matter or pollutants on to particle surfaces. This observation was also found by other workers [56,57].
The fate of industrial pollutants in inland waterways and paralic environments has been the focus of intense scrutiny over the past four decades. The relationship between sediment surface area and pollutant concentration demonstrated in this study is intrinsically related to the relationship between OM and surface area which is also observed by other workers [11,13]. It is unlikely that this relationship is confined only to dioxins and furans, but is probably ubiquitous throughout many other industrial chemical classes such as PCBs, polychlorinated insecticides et cetera, as well as other types of pollutants such as heavy metals from the mining industry. It is not clear, however, as to the extent of surface area preservation within other chemical classes, or whether there is any potential for desorption and release of pollutants back into the environment following changes in sedimentological or redox conditions within the sediments.
Author Contributions: G.C. and R.A. co-wrote this research article. A.B. and M.B. conceptualized key themes of this study. Formal analyses were performed by R.A.
Funding: This research received no formal funding.