JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
Related Collections
Right arrow Soil Microbiology
Right arrow Wetland Soils
Right arrow Biogeochemical Processes
Right arrow Field-Scale Studies
Right arrow Heavy Metals
Published in J. Environ. Qual. 32:2085-2094 (2003).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS

Heavy Metals in the Environment

Are Methylmercury Concentrations in the Wetlands of Kejimkujik National Park, Nova Scotia, Canada, Dependent on Geology?

Steven D. Siciliano*,a, Al Sangsterc, Chris J. Daughneyd, Lisa Losetob, James J. Germidaa, Andrew N. Renczc, Nelson J. O'Driscollb and David R. S. Leanb

a Dep. of Soil Science, Univ. of Saskatchewan, 51 Campus Drive, Saskatoon, SK, Canada S7N 5A8
b Dep. of Biology, Univ. of Ottawa, 30 Marie Curie, Ottawa, ON, Canada K1N 6N5
c Geological Survey of Canada, 601 Booth Street, Ottawa, ON, Canada K1A 0E8
d Institute of Geological and Nuclear Sciences, P.O. Box 30368, Lower Hutt, New Zealand

* Corresponding author (Siciliano{at}sask.usask.ca).

Received for publication October 17, 2002.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUMMARY
 REFERENCES
 
In the relatively pristine ecosystem in Kejimkujik Park, Nova Scotia, methylmercury (MeHg) concentrations in loons, Gavia immer, are among the highest recorded anywhere in the world. This study investigated the influence of bedrock lithology on MeHg concentrations in wetlands. Twenty-five different wetland field sites were sampled over four different bedrock lithologies; Kejimkujik monzogranite, black sulfidic slate, gray slate, and greywacke. Soil samples were analyzed for ethylmercury (EtHg), MeHg, total Hg, acid-volatile sulfides (AVS), organic matter, and water content as well as the biological parameters, mercury methyltransferase (HgMT) activity, sulfate reduction rates, fatty acid methyl ester (FAME) composition, and acidity. Methylmercury concentrations in the wetlands were highly dependent (P < 0.08) on lithology with no significant difference between bogs, fens, and swamps. Methylmercury concentrations in wetland soils developed on Kejimkujik monzogranite averaged 900 ng kg-1 compared with only 300 ng kg-1 in wetland soils developed on black sulfidic slate. Fatty acid methyl ester composition was also lithologically dependent (P < 0.001) with biomarkers for Desulfobulbus spp. discriminating between sites containing high and low MeHg concentrations. Levels of MeHg in wetlands were predicted mainly (41% of the sum of squares) by HgMT activity that differed (P < 0.009) between wetlands, with activity in bogs almost three times that present in swamps. Wetland MeHg concentrations are highly dependent on the lithology on which they have developed for largely biological reasons.

Abbreviations: AVS, acid-volatile sulfides • EtHg, ethylmercury • FAME, fatty acid methyl esters • HgMT, mercury-methyltransferase • MeHg, methylmercury • Q3, third quartile • SAM, S-adenosylmethionine • SCM, surface complexation model • SRB, sulfate-reducing bacteria • U, one unit of HgMT activity was defined as the formation of 1 nmol of tetrahydrofolate formed in response to the addition of 1 ng of Hg to the reaction vessel


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUMMARY
 REFERENCES
 
IN CONTRAST to areas that have received direct inputs of MeHg and Hg to the ecosystem, Kejimkujik National Park in eastern Canada is a relatively pristine, headwater ecosystem; yet, the blood of common loons in the Kejimkujik area contains the highest Hg levels of any loons in North America (Burgess et al., 1998). The increased Hg concentrations present in adult loons were closely reflected in the Hg content of chick blood, which in turn was tightly correlated with Hg levels in fish (Burgess et al., 1998). There are no known point sources of Hg within the park and atmospheric deposition is similar to that found elsewhere in North America, (about 7 µg m-2 yr-1) (O'Driscoll et al., 2001). Interestingly, MeHg concentration in the biota and lake water correlates with bedrock lithology. Methylmercury concentrations in lakes on the Kejimkujik monzogranite lithology are two to 10 times greater than that observed in lakes overlying other lithologies and total Hg in fish was 1.4 to 2.1 times greater (O'Driscoll et al., 2001). Mercury concentrations in the lakes were also dependent on the geology with lakes on granitic geology having higher total Hg than lakes overlying greywacke (Vaidya and Howell, 2002). A Hg mass balance study of the Big Dam West watershed in Kejimkujik indicated that of the 160 g ({sigma} = 38.4) of Hg inputs directly to the lake, 92% (147 g, {sigma} = 36.7) was due to inflow from wetlands, 7% (11 g, {sigma} = 0.0) was due to wet precipitation and, 1% (2 g, {sigma} = 1.7) was due to ground water inflow (N.J. O'Driscoll, S.D. Scilliano, and D.R.S. Lean, unpublished data, 2001). The surface inflow results from over 184 g ({sigma} = 20.2) Hg yr-1 being deposited from the atmosphere to the terrestrial catchments comprising 27 x 106 m-1. Mercury and MeHg deposition is influenced by forest canopy in other ecosystems (St Louis et al., 2001) and its variation across the Kejimkujik area is unknown. However, atmospherically deposited MeHg is not a significant source to current levels of MeHg in wetlands (Branfireun et al., 1998) so we have assumed that the formation of MeHg from inorganic Hg in the terrestrial zone is a critical process influencing MeHg transfer from terrestrial zones to their adjacent water bodies. Since the underlying lithology of a terrestrial zone is an important determinant of water chemistry and water chemistry plays a central role in Hg transport (Daughney et al., 2002) and methylation (Branfireun and Roulet, 2002), it seemed reasonable to investigate the role of lithology in Hg fate in the terrestrial landscape.

Wetlands play a major role in the export of MeHg to the watershed (St Louis et al., 1994; Waldron et al., 2000; Watras et al., 1995). The type and hydrological flow (Branfireun and Roulet, 2002) present in a wetland influences how much MeHg is exported to the watershed. In wetlands, the methylation of inorganic Hg is thought to occur primarily through the activity of sulfate reducing bacteria (SRB) with Hg methylation rates linked to the composition and activity of SRB (Devereux et al., 1996; King et al., 2001; Macalady et al., 2000) as well as the availability of sulfate (Heyes et al., 2000). By definition, the biological methylation of Hg (II) can only occur to that fraction of Hg that is bioavailable. In an environmental setting, the fraction of Hg sorbed to biological colloids is determined by the chemistry of the surface water (Daughney et al., 2002), which in turn is controlled (in part) by the surrounding geology, watershed geometry, and hydrology (Kolka et al., 1999). Thus, while the hydrology of a wetland will undoubtedly play a major role in the methylation of Hg, the underlying lithology will also influence MeHg formation by modifying the biological and chemical factors, which interact with hydrological processes.

We hypothesized that MeHg concentrations in Kejimkujik's wetlands could be, in part, predicted by the lithology of that site. The general till cover over the bedrock terrain is very thin, commonly only 1 to 3 m and there is only limited (<1 km) down-ice transport of bedrock materials (Finck et al., 1994). This till contains between 0.5 and 16 µg kg-1 for surface rocks and around 60 µg kg-1 for soil (Sangster et al., 2001) and thus may act as a source of Hg to the ecosystem. The weathering of Hg from till will likely be dependent on pH, acid volatile sulfides and organic matter because of their influence on Hg transport and bioavailability (Barkay et al., 1997; Benoit et al., 2001; Benoit et al., 1999a; Skyllberg et al., 2000). Atmospheric deposition may play a large role in the Hg cycle in Kejimkujik Park as witnessed by the importance of atmospheric input in the Big Dam West mass balance (N.J. O'Driscoll, S.D. Scilliano, and D.R.S. Lean, unpublished data, 2001), but the primary role of atmospheric deposition in this system may be as a source of inorganic Hg because atmospheric input of MeHg in other ecosystems has been found to be an insignificant source compared with other MeHg sources (Branfireun et al., 1998).

We suggest that the type of lithology present in the terrestrial landscape will influence the type of plants and microbes in the larger ecological area as has been seen in agricultural research (Dunfield and Germida, 2001). The wetlands integrate this regional effect because ground water from the surrounding lithology, with its associated microbial and plant community, will collect in the wetland. These regional factors will influence MeHg production by modulating parameters known to be important for MeHg formation, biological sulfate reduction (Devereux et al., 1996; King et al., 2001; Macalady et al., 2000), inhibitory anions (Chen et al., 1997), or HgMT activity (Siciliano and Lean, 2002). The position in the landscape and hydrological conditions will alter the export of MeHg from individual wetlands (Branfireun and Roulet, 2002) but we hypothesized that differences between large populations of similar type, that is, bog, fen, and swamp, wetlands would be dependent on lithology and thus, detectable by our experimental design. We tested our hypothesis by first assessing if wetland MeHg concentrations were statistically dependent on lithology and then determining the lithological dependence of factors known to modify Hg bioavailability and MeHg formation.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 SUMMARY
 REFERENCES
 
Geological Framework
The geological substrate in the Kejimkujik Area consists of greywacke, black slate, gray slate, and monzogranite overlain by glacial deposits of various types (Fig. 1) . The chemical characteristics of these units are summarized in Table 1. The greywacke (feldspathic sandstone), and gray and black slates are part of the Meguma Group of marine sedimentary rocks that were deposited on an ocean floor between 550 and 475 million years ago (Schenk, 1995). The greywacke is a massive to thickly bedded, gray sandstone composed mainly of quartz (SiO2) and feldspar (K, Ca, Na aluminosilicate minerals) with minor accessory mica (sericite [K-, Na-rich aluminosilicate], and biotite [Km Fe, Mg aluminosilicate]). This lithology is relatively resistant to oxidation and weathering.



View larger version (68K):
[in this window]
[in a new window]
  		Fig. 1. Bedrock geology of study area with sampling sites indicated. The lower panel indicates the watershed boundaries for the sampling area.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Mineral and element composition of rock samples obtained on the different bedrock lithologies.

 
The black slates are mineralogically similar to the greywackes but with a finer grain and a higher content of micaceous minerals and less quartz. The main distinguishing feature is a content of 0.5 to 4% organically derived C and accompanying 1 to 5% sulfide minerals including pyrite [FeS2] and pyrrhotite [Fe1-xS]. The slate has a well-developed fracture cleavage that results in greater porosity and permeability for meteoric fluids than other rock types in this area. The high sulfide content of many of the black slates enhances their weathering and breakdown in the surficial environment through the production of dilute sulfuric acid from the oxidation of the sulfide minerals. In subsurface black slates seen in drill core, it is common to see oxidation on joints and cleavage surfaces ten's of meters below bedrock surface. The unit is a significant contributor to acid rock drainage throughout its area of exposure (Fox et al., 1997) and may be a provider of S and leachable trace elements to the biological community.

The gray slates are transitional in character, their composition is similar to the black slates but contain very little organic C. They also possess the well-developed fracture cleavage found in the black slates but lack the sulfide mineral content that enhances the chemical weathering of the black slate.

The granitic (strictly monzogranite) rocks are part of the South Mountain Batholith complex that intruded the metasedimentary rocks about 370 million years ago. They are medium to coarse grained and contain orthoclase (K-rich) and plagioclase (Ca, Na-rich) feldspar, an average of about 15% biotite with minor quartz. The plagioclase (Ca, Na) feldspar commonly occurs as megacrysts that may form 10 to 15% of the rock. The rock is massive and very coarsely jointed. Weathering is restricted to a rind of feldspar kaolinization and chloritization of biotite that may be a few centimeters to a few meters thick.

The granititic rocks form the upland areas with the greywackes and slates forming a topographically lower undulating plain. Unconsolidated glacial sediments form a thin but pervasive cover over all rock types. Rock outcrop only occurs in erosional settings such as lakeshores and riverbeds. In southern Nova Scotia, the glacial sediment deposited directly by glaciers has typically been transported by glaciers a few hundred meters from the source bedrock. Over granite and greywacke exposure, soils developed a till composed of rock flour and as such are sandy, porous, and well drained. Soils over the slates contain a higher clay content and tend to be less well drained. In addition, impervious clays were deposited in periglacial lakes and ponds over all lithologies and are now the agent that is controlling the location of the wetlands.

Sampling Design
Sites were selected so that wetlands from each lithology (Fig. 1) and wetlands classified on the basis of dominant vegetation (Group, 1988). In Kejimkujik Park, bogs were classified by a high degree of sphagmum moss, sparse stands of black spruce [Picea mariana (Mill.) Britton et al.] and no open water. Sedge grasses characterized fens with white pine (Pinus strobus L.) growing throughout and an open stream or water source. Swamps had no grass present, small amounts of moss but were heavily forested with white pine and black spruce and contained an open water source. Sample Sites 1 to 5 are on higher ground over Kejimkujik monzogranite with Sites 1 to 3 being small wetlands (1 bog, 1 fen, 1 swamp) of <100 m diameter and Sites 4 and 5 (1 bog, 1 swamp) forming a large wetland complex marginal to a small lake and stream. Sites 6 to 15 are swamps (n = 7) and fens (n = 3) over gray silty slate bedrock. Clays deposited during a higher stand of the lake underlie these wetlands (data not shown). Lithological boundaries in this area are poorly known. The area of Sites 6 to 10 contains gray and gray-green slates with some greywacke whereas the area of Sites 11 to 15 is more consistently gray slate. Sites 14 and 15 are underlain by slates containing a higher C content (i.e., black slate) but low sulfide content and therefore are chemically similar to the gray slates. Sites 16 to 20 are swamps (n = 2) and fens (n = 3) over high sulfide carbonaceous black slate and Sites 21 to 25 (2 bog, 2 fen, 1 swamp) are over greywacke (feldspathic sandstone). The bog at Site 23 was a large (>1 km diameter) bog but the bog at Site 24 was more typical of the bogs in the area with a diameter of <100 m. In total, four bogs, 12 swamps, and nine fens were sampled. Unfortunately these wetland types were not evenly distributed across lithologies but rather represent those wetland types accessible by foot or all terrain vehicles. Altogether wetlands in the Kejimkujik area represent on average 7% of watershed areas due to the flat topography of the park.

Surface drainage water (n = 2) and soil samples (n = 5) were collected from each wetland site. The soil replicates at each location formed a pentagon with the apexes 1 m apart. If present at bog sites, overlying sphagnum was carefully cut and removed. Samples (500 g) of soils were collected from the top 15 cm using clean-room grade PVC gloves and samples placed in sterile polypropylene bags. The pH (5:1 water/soil) of these samples (n = 125) ranged from 3.7 to 6.2 (median = 5.03, 1 quartile = 0.29), contained 25 to 95% (median = 86, 1 quartile = 2.3) water (w/w) and 7 to 96% (median = 68, 1 quartile = 19) organic matter (loss on heating to 350°C for 48 h). Soil samples were transported back to base camp, flushed twice with Ar, sealed in a second bag, frozen, and kept on ice until returned to the laboratory.

Water sampling for Hg anions and cations was performed following procedures developed at the Geological Survey of Canada (Hall et al., 2002). Briefly, a 50-mL syringe (Delta Scientific 6602-2016, Delta Scientific Inc., Ivyland, PA) was first vigorously rinsed twice with the water to be sampled. The syringe was then filled and fitted with a polypropylene Millipore Sterivex-HV 0.45-µm cartridge filter (Millipore SV-HU010RS, Millipore, Billerica, MA) and water was extruded through the filter into prenumbered 60 mL Nalgene high density polyethylene (HDPE) wide mouth bottles (Fischer 03-313-4B, Fischer Scientific, Pittsburgh, PA). Each of three sample bottles was vigorously rinsed with filtered water from the sample site. While in the field, samples were kept in a dark cooler containing ice. Daily, on returning to base from the field, 1 mL of BrCl (USEPA Method 1631, USEPA, 2001) was added as a preservative to the filtered sample for total Hg analysis. Non-filtered samples for total Hg were sampled in the same manner except that no filters were attached to the syringe. Field blanks consisting of deionized water filtered as above were added to the sample population in the field. Samples were refrigerated at the field base and shipped in refrigerated coolers to the laboratories of the Geological Survey of Canada where they remained refrigerated until analysis.

Chemical Analysis
Methylmercury and EtHg in soil was determined by capillary gas chromatograph-atomic fluorescence spectrometry following an acidic-KBr/dichloromethane extraction (Cai et al., 1996; 2000). Blanks and matrix spikes were used to assess daily analytical performance. Recoveries averaged 80% with a percentage of deviation of triplicates (n = 12) of 19% and blanks contained no detectable inorganic or organic Hg. Total Hg in freeze-dried soil and moss was analyzed using USEPA method 7437 (USEPA, 1998) . This method employs a Milestone AMA 254 analyzer (Milestone, Monroe, CT), which detects Hg by pyrolysis followed by Au amalgamation atomic absorption spectrophotometry. Detection limits of 0.1 ppb were achieved by this method using a 100-mg sample.

Acid volatile sulfide content, thought to immobilize divalent Hg, was determined on 10 g of wet soil, acidified with 15 mL of anoxic 2 M HCl and the released sulfides trapped in 100 mL of 0.1 M NaOH (Cline, 1969). Trapped sulfides were quantified by mixing 0.8 mL of Cline's Reagent (1.1 g FeCl3 and 0.92 g N,N dimethyl p-phenylene diamine in 125 mL of concentrated HCl diluted with 125 mL water), incubating for 20 min and measuring absorbance at 670 nm.

Total Hg in water was analyzed by using SnCl2 reduction in cold vapor-atomic fluorescence spectroscopy as outlined in USEPA Method 1631 (USEPA, 2001). The method detection limit (3{sigma} of all blanks) for total Hg was 0.71 ng L-1 (n = 8) and a percentage of recovery of 97 ± 5% (n = 8 recoveries).

Characterization of the Microbial Community
The microbial community was characterized by analyzing the FAMEs present in soil (Cavigelli et al., 1995). Briefly, 5 g of soil is saponified for 30 min at 100°C with 5 mL of 3.75 M NaOH in 50% methanol and then methylated at 80°C for 10 min with 10 mL of 3.25 M HCl in 46% methanol. The methylated fatty acids are extracted at 130 rpm with 1.5 mL of hexane and methyl-tert butyl ether (1:1) for 10 min and the extracts washed at 130 rpm (rotary) for 10 min with 3 mL of 0.3 M NaOH, and centrifuged at 1000 x g before GC/FID analysis. Fatty acids are described using the nomenclature "number of carbons/number of unsaturations" followed by double bond locations referenced from the omega or aliphatic end of the molecule with "c" or "t" referring to a cis or trans configuration of the double bond (Macalady et al., 2000). The FAME procedure cannot separate all the phospholipids present in soil with unresolved fatty acids reported as Sum of Feature. However, this method has been used successfully by numerous laboratories to study soil microbial communities (Ritchie et al., 2000).

Sulfate reduction was quantified as described previously (Ulrich et al., 1997) except that a single-step Cr reduction method was used to extract sulfides (Fossing and Jørgensen, 1989). Briefly, 3.7 x 10-4 Bq (1 µCi) of carrier-free Na235SO4 was added to 2 g of wetland soil, incubated for 24 h and 8 mL of 1 M Cr (II) in 0.5 M HCl as well as 4 mL of 12 M HCl injected. The released sulfide was trapped in 10% zinc acetate and the incorporated 35S activity was determined by liquid scintillation counting.

Mercury-methyltransferase activity (Siciliano and Lean, 2002) was assessed by extracting protein from soil (Ogunseitan, 1997) and determining HgMT (Drummond et al., 1995). Briefly, 80 µL of 1 M K2PO4 buffer, pH 7.2, was added to 338 µL sterile, distilled water, followed by 40 µL of dithiothreitol (500 mM), 4 µL of S-adenosylmethionine (SAM) (3.8 mM), 10 µL of HgCl2 (100 µg L-1), 200 µL of the enzyme mixture to be assayed, and 80 µL hydroxocobalamin (500 µM). After 5 min of incubation, 48 µL of MeTHF (250 µM) was added and the mixture incubated for 20 min, at 22°C in the dark. The reaction was quenched by the addition of 200 µL of 5 M HCl in 60%(v/v) formic acid and THF transformed to methenyltetrahydrofolate by heating at 80°C for 10 min. Methenyltetrahydrofolate was assessed at 350 nm using a 96 well microplate (PRO-BIND, Becton Dickinson, Franklin Lakes, NJ) with a standard curve between 0 and 196 µM tetrahydrofolate used to interpolate results. Mercury specific methyltransferase activity was assessed by comparing methyltransferase activity in presence of 10 µL of 100 µg L-1 HgCl2 to assays with only 10 µL of distilled water (Siciliano and Lean, 2002). One unit (U) of HgMT activity was defined as the formation of 1 nmol of tetrahydrofolate formed in response to the addition of 1 ng of Hg to the reaction vessel.

Methyltransferase activity can happen by a cobalamin dependent or independent process. Coblamin dependent methyltransferase requires SAM whereas the cobalamin-independent process does not. To determine the dependence of enzyme extracts on SAM, enzyme extracts from soil were dialyzed (Spectrapor 4 standard cellulose dialysis tubing, 1200–14000 molecular weight cut off; Spectrum Laboratories, Inc., Rancho Dominguez, CA), three times against 10 volumes of 100 mM K2PO4 buffer, pH 7.2 for 24 h at 4°C. The HgMT activity of these extracts was assessed in the presence or absence of added SAM. S-adenosylmethionine dependent activity was calculated by subtracting the amount of activity observed in the absence of SAM from the activity observed in the presence of SAM. because of technical difficulties, we were only able to process seven samples, four from Kejimkujik Monzogranite and three from gray slate.

Statistical Analysis
Data was tested for normality using the Anderson–Darling test for normality and homogeneity of variance assessed by Bartlett's test (Sokal and Rohlf, 1995). Log transformed MeHg, HgMT, AVS, and arc-sin transformed percentage of organic matter data were normally distributed and the variance amongst the four lithologies was homogenous allowing the use the of a general linear model (GLM) with lithologies and wetland type as fixed factors. This analysis allows one to assess, statistically, the lithological dependence of MeHg concentrations in wetland soils. In contrast, EtHg and total soil Hg were nonnormally distributed and no tested transformation achieved normality. However, total soil Hg had homogenous variances between lithologies and thus, a general linear model was used for analysis. In contrast, EtHg had heterogenous variances and we used the Moods Median Test (Sokal and Rohlf, 1995) to test for the dependence of geology and report the median of the observed values.

Regression analysis was performed first by determining the optimal combination of predictor variables using best-subsets regression and then performing the indicated linear regression. This method corrects for the increase in fit that occurs naturally as additional parameters are included in the regression equation (Sokal and Rohlf, 1995). Means for each site were used in the regression because the purpose of this regression equation was to explain site differences and not differences between subsamples at a site. Principle component analysis was performed on the FAME data using the covariance matrix and the resulting scores analyzed by a one-way analysis of variance using wetlands or lithology as a factor (Glimm et al., 1997). The weightings used for the biomarkers were determined from the predictor coefficients obtained for each principle axis. Determinant analysis was performed on this data by first grouping the sites into three groups; high, replicates with MeHg concentrations in the fourth quartile of the distribution; medium, MeHg concentrations in the second and third quartile; low, MeHg concentrations in the first quartile. A linear discriminant function was extracted in which the coefficients of the various biomarkers were determined.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 SUMMARY
 REFERENCES
 
Methylmercury concentrations in wetland soil were dependent on lithology (GLM, P < 0.078) with wetlands on Kejimkujik monzogranite having the highest, 920 ng kg-1, levels compared with the other lithologies (Fig. 2) . There was no difference in MeHg concentrations between bogs, swamps, and fens (GLM, P < 0.238). In general, MeHg concentrations were <0.25% of the total Hg. As expected, total soil Hg concentrations were also dependent on lithology (GLM, P < 0.001) but did not predict MeHg concentrations. For example, total Hg in soils over the gray slate was similar, 180 µg kg-1, to that seen in soils over Kejimkujik monzogranite, 190 µg kg-1, but the MeHg over Kejimkujik monzogranite was 70% or 300 ng kg-1 higher compared with the gray slate. The reverse trend was observed for total Hg in rock samples over the different lithologies with total Hg being double in rocks obtained over the black sulfidic slate lithology, 1.5 µg kg-1, and compared with the Kejimkujik monzogranite, 0.8 µg kg-1 (Table 1). These results highlight the importance of wetland soils in the Hg cycle with Hg being present in 100 times greater concentrations in bulk wetland soil samples compared with rock samples.



View larger version (37K):
[in this window]
[in a new window]
  		Fig. 2. Organic matter and methylmercury (MeHg) (open bars) as well as pH and total Hg (closed bars) in wetlands found on different lithologies. Closed circles indicate acid volatile sulfide or surface complexation model estimate of percentage of total Hg sorbed to bacteria in different lithologies. Note the 1000-fold difference in the scales for MeHg and Hg. Error bars indicate standard error of the estimate.

 
The method used in this manuscript is one of the few analytical methods capable of detecting EtHg, so these results are some of the first reports of EtHg in wetlands. Ethylmercury levels were dependent on lithology (Moods Median Test [Sokal and Rohlf, 1995], P < 0.001) with the highest levels observed in the black sulfidic slate (25 ng kg-1, median = 0, third quartile [Q3] > 32) and gray slate (27 ng kg-1, median = 0, Q3 > 39) compared with greywacke (13 ng kg-1, median = 0, Q3 > 0) and Kejimkujik monzogranite (2.1 ng kg-1, median = 0, Q3 > 0). There was no correlation between EtHg and MeHg (r = 0.204) for individual samples. A discriminant analysis identified soil Hg concentrations as the most important parameter separating sites (105 squared distance between groups, 100% success rate) that contained EtHg (227 µg Hg kg-1, SE = 19.6) from those sites were no EtHg (247 µg Hg kg-1, SE = 17.3) was detected.

To account for differences in bioavailability that are dependent on lithology, we compared organic matter, pH, and AVS across the lithologies (Fig. 2). Kejimkujik monzogranite lithology had the highest percentage of organic matter (91%) of all lithologies and the lowest average soil pH (4.5) but neither of these variables was correlated with MeHg concentrations (r = 0.02 and r = 0.193, respectively) and combined, these variables predicted practically none (r2 = 0.02) of the MeHg variation. Similarly, there was no (P < 0.538) lithological dependence in AVS and no difference between wetlands (P < 0.712) in AVS levels.

We have recently developed a surface complexation model (SCM) that predicts 62% of the variation in Hg bound to particulate organic C in Kejimkujik (Daughney et al., 2002). This model which uses dissolved organic matter, particulate organic matter, total Hg and Cl levels as input parameters can be considered as a surrogate for Hg bioavailability because it is bacteria that methylate Hg and sorption to the bacterial surface is the first step in bioavailability. The model explicitly models the sorption of neutral complexes to the cell surface. There were no significant differences in filtered Hg between lithologies with approximately 4.2 ng L-1 present in filtered wetland water. Based on the SCM, Hg bioavailability to microorganisms was dependent on lithology (P < 0.163) with the greatest bioavailability (74%) in Kejimkujik monzogranite lithology. It should be noted that recent works suggests that microorganisms may actively take up Hg from solution (Golding et al., 2002) contradicting earlier assumptions of Hg bioavailability (Barkay et al. 1997). In either case, sorption to the cell surface is necessary so the SCM is a good first estimate of bioavailability.

The HgMT activity was highest in bog wetlands (P < 0.009) with activity double that of fens or swamps (Fig. 3) . There was no detectable lithological dependence when all wetlands were combined (P < 0.557) and due to unbalanced samples, we were unable to collectively test interactions between wetlands and geology. When considered separately, levels of HgMT activity in the two Kejimkujik monzogranite bogs was much higher (60 U g-1 soil, SE = 4.9) than that seen in the greywacke (34 U g-1 soil, SE = 4.9). In contrast to HgMT, sulfate reduction rates were highest (P < 0.023, Moods Median Test) in fens compared with bogs and swamps (Fig. 3) with no lithological dependence (P < 0.743, Mood's Median Test [Sokal and Rohlf, 1995]).



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3. Influence of wetland type on mercury-methyltransferase (HgMT) activity (closed bars) and sulfate reduction rate (closed circles). Error bars are standard error of methyltransferase activity and one quartile for sulfate reduction rate. U, one unit of HgMT activity was defined as the formation of 1 nmol of tetrahydrofolate formed in response to the addition of 1 ng of Hg to the reaction vessel.

 
The composition of the microbial community was dependent (P < 0.001) on lithology with the communities present on Kejimkujik monzogranite clearly separating on the first two principle components from the other lithologies (Fig. 4) . Fatty acid methyl esters considered as biomarkers (17:1, 15:1, and 15:0) (Macalady et al., 2000) for Desulfobulbus, were primarily responsible for the separation of Kejimkujik monzogranite communities from the other three lithologies as depicted by the insert in Fig. 4. Of these biomarker FAMES, 17:1 w7c and 15:1 w8c had a combined importance of 24% for principle Component 1 and 27% for principal Component 2 and this gives rise to the vector drawn in the insert present in Fig. 4. The microbial community composition, especially Desulfobulbus biomarkers, was also an effective (76% accuracy) method of grouping sites into areas of high, medium, and low MeHg concentrations in wetland soil (data not shown). The linear discriminant function responsible for 40% of the variance had the Desulfobulbus biomarker 15:0 anteiso as the most important predictor followed closely by a Desulfovibrio marker iso 16:0.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 4. Principle component analysis of fatty acid methyl esters (FAME) present in wetland soil from Kejimkujik monzogranite (diamond), greywacke (circle), black sulfidic slate (square) or gray sulfidic slate (triangle). Open symbols are the average principle component scores for each lithology and closed symbols are the average principle component scores for each site. Inserted graph is the relative weighting of established Desulfobulbus FAME biomarkers for Principle Component 1 and 2.

 
Including all the delineated parameters allows us to predict 84% (P < 0.282) of the MeHg variation in wetlands in the Kejimkujik area:

(Table 2). Adjusting the predictive power of the equation for the number of parameters included in the equation reduces this estimate to 45% predictive power and identifies the best predictive parameters for MeHg levels in wetlands. The majority of this predictive power was obtained from HgMT activity that accounted for 41% of the sequential sum of squares of the multiple linear regression model (Table 3). Total soil Hg was the second most powerful predictor as determined by best subsets regression and accounted for 19% of the sequential sum of squares in the model. Previous work has indicated that the form of HgMT, cobalamin independent or dependent correlates with methylation of Hg (Siciliano and Lean, 2002). A higher percentage of the observed HgMT activity (55%, n = 4) on Kejimkujik monzogranite was cobalamin independent compared with gray slate (15%, n = 4).


View this table:
[in this window]
[in a new window]
  		Table 2. Best-subsets regression analysis of predictive parameters of methylmercury (MeHg) concentrations in wetlands.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Multiple linear regression model to predict wetland methylmercury levels.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
SUMMARY
 REFERENCES
 
There was a strong dependence of MeHg concentrations on bedrock lithology in wetland soils. This was tested by a two-way ANOVA with wetlands and lithology as factors. We found that there was little difference in total Hg present in soils but that other chemical factors combined to increase Hg sorption to particulate organic matter as modeled by the SCM. Furthermore, the enzyme postulated to be responsible for MeHg formation was also higher in the Kejimkujik monzogranite. Thus, there is greater bioavailability and methylation activity in Kejimkujik monzogranite, which explains the high MeHg levels found in the wetland soils. It may be that this sequence of events, greater Hg bioavailability and methylation activity, is brought about by factors other than lithology, such as wetland hydrological characteristics, but if so, these factors are correlated with the bedrock lithologies of southern Nova Scotia.

We observed that the dependence was related to an interaction between inorganic Hg and levels of biological methylation activity. Total soil inorganic Hg was the most powerful predictor of MeHg concentrations of all the chemical parameters measured. The concentration of total Hg will be one of the important parameters controlling the bioavailability of inorganic Hg for methylation. Divalent Hg is strongly bound to soil organic matter with a log KSOC (KSOC = HgSOC/[Hg2+]) of around 23 (Skyllberg et al., 2000) and this binding is dominated by reduced S present in the organic matter. These authors postulated that reduced sulfide would play little role in soils because it is always in excess and variations in levels would not alter how much Hg was sorbed to organic matter. Our work supports the interpretation of Skyllberg et al. (2000) that at the large regional scale, reduced sulfide in wetland soils had little role in Hg methylation.

In the case of Hg bioavailability, reduced S, pH, organic matter, and particulate matter are all parameters (Benoit et al., 1999b; Jay et al., 2000; Loux, 1998; Skyllberg et al., 2000) known to not only be important in regulating bioavailability but also to interact with each other. Thus, a comprehensive bioavailability model of Hg in wetland soil will likely require development of a soil Hg-SCM such as the one recently developed for aerobic waters (Daughney et al., 2002). This aerobic-water SCM was able to accurately estimate Hg sorption to POC carbon in Kejimkujik surface waters and in this study, application of the SCM illustrated a strong lithological dependence of Hg sorption to POC in surface waters. The original SCM was developed for conditions prevalent in Kejimkujik area surface waters, that is, low dissolved sulfide and competing mineral sorbants. These assumptions are violated in the soil system and in the case of sulfide; it is known that this will alter the speciation of Hg considerably (Jay et al., 2000). Mineral sorbants in the soil system cannot only alter the bioavailability of Hg via simple adsorption but also the reactivity of Hg toward methylation (Farrell et al., 1998). We did not perform clay mineralogical analysis on our samples but bedrock lithologies often differ in mineral composition and this may be one of the primary routes by which geology acts as a determinant in the Kejimkujik system.

Wetland MeHg concentrations were most successfully predicted by HgMT activity. As expected, the HgMT activity was highest in bog samples that also contained the most MeHg (800 ng kg-1, {sigma} = 350) compared with fens (393 ng kg-1, {sigma} = 108) and swamps (578 ng kg-1, {sigma} = 136). The levels of HgMT activity observed in Kejimkujik bogs were very similar to the peak activities observed at Mer Bleu bog in Ontario, Canada (Siciliano and Lean, 2002). While there were differences between lithologies in HgMT activity with Kejimkujik monzogranite having the highest HgMT (31 U g-1), followed by greywacke (26 U g-1), gray slate (26 U g-1), and black sulfidic slate (17 U g-1), these differences were not significant and not of the same order of magnitude as MeHg concentrations. There are two possible explanations for this discrepancy. As seen for a small subset of samples, the difference between cobalamin dependent and independent activity corresponded with an increase in MeHg levels across two lithologies. In addition, our sampling campaign was limited to a 14-d period in the early spring when there was still snow on the ground and water temperatures were below 10°C. As was seen in Mer Bleu (Siciliano and Lean, 2002), there is a strong temporal component to methylation activity and our sampling campaign may have missed the peak MeHg formation period in Kejimkujik.

There was a striking lithological dependency on microbial community composition with differences in Desulfobulbus biomarker FAMEs being primarily responsible for this separation. In sediments, Desulfobulbus abundance is linked with high MeHg concentrations (Devereux et al., 1996). Biomarkers should not be considered as definitive for that organisms since these biomarkers are based on the concept that there is more 15:0 anteiso present in Desulfobulbus compared to Desulfovibrio but there is still 15:0 anteiso present in Desulfovibrio (Macalady et al., 2000). Thus, our data indicates a predominance of Desulfobulbus biomarkers at sites with high MeHg but alternatively, this may arise if there are significantly more SRBs in general at these sites. This alternative interpretation is not supported by the low sulfate reduction rates observed and that sulfate reduction rates were not related to MeHg concentrations in the wetlands.

The sites on the four different lithologies present in Kejimkujik National Park are dominated by different forest types with 51% of the forest on Kejimkujik monzogranite sites being coniferous compared with only 34% on the Greywacke sites. Forest composition expressed as percentage of the forest that was coniferous had little power (r2 = 0, P < 0.410) to predict the MeHg levels in wetlands. As expected, the forest composition data was correlated (r > 0.56) with the soil parameters organic matter, soil pH, and AVS. Given the ability of vegetation to alter microbial community composition (Staddon et al., 1997), it is feasible that the lithological dependence we observed was related to plant community composition. Other investigators have found that plant community composition can affect Hg flux from the atmosphere (St Louis et al., 2001) and thus, Hg deposition may have differed between the lithologies. However, other authors suggest that whereas atmospherically deposited Hg(II) is important to wetland Hg processes, deposited MeHg has little influence on MeHg levels present in catchments (Branfireun et al., 1998). There was no lithological difference in dissolved Hg in our study suggesting that atmospheric deposition of Hg did not differ between lithologies but we have no direct evidence that atmospheric deposition is similar or different between lithologies. The interaction between atmospheric deposition, forest composition and wetland processes, which in turn may be partially controlled by hydrology and lithology, requires additional research.


   SUMMARY
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUMMARY
REFERENCES
 
This study demonstrated that MeHg concentrations in wetland soil differ between lithologies. Common environmental parameters such as pH, AVS, sulfate reduction activity, as well as inhibitory anions were not important in determining MeHg concentrations in soil. Rather, this dependence was weakly related to total Hg concentrations, presumably via bioavailability and strongly related to HgMT activity and microbial community composition. The causal chain linking lithology to the environmental parameters that control MeHg formation was not elucidated from our study. Our results suggest that geology influences Hg bioavailability and the biology responsible for the methylation reaction thereby giving rise to the observed lithological dependence of MeHg in soil as well as wildlife.


   ACKNOWLEDGMENTS
 
The authors thank the staff of Kejimkujik National Park for their help in accessing remote locations. This work was supported by Toxic Substance Research Initiative Project #124 grants to D.R.S. Lean, A. Rencz, and A. Sangster. N. O'Driscoll was supported by NSERC. This is Geological Survey of Canada Contribution #2002153.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 SUMMARY
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Siciliano, S. D.
Right arrow Articles by Lean, D. R. S.
Related Collections
Right arrow Soil Microbiology
Right arrow Wetland Soils
Right arrow Biogeochemical Processes
Right arrow Field-Scale Studies
Right arrow Heavy Metals

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome