Comparison of mercury and zinc profiles in peat and lake sediment archives with historical changes in emissions from the Flin Flon metal smelter, Manitoba, Canada

Science of the Total Environment 409 (2011) 548–563

Contents lists available at ScienceDirect

Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v

Comparison of mercury and zinc profiles in peat and lake sediment archives with historical changes in emissions from the Flin Flon metal smelter, Manitoba, Canada P.M. Outridge a,⁎, N. Rausch b,c, J.B. Percival a, W. Shotyk b, R. McNeely a a b c

Geological Survey of Canada, Natural Resources Canada, 601 Booth St., Ottawa K1A 0E8, Canada Institute of Earth Sciences, University of Heidelberg, Im Neuenheimer Feld 236, D-69120 Heidelberg, Germany European Commission Joint Research Center, Institute for Transuranium Elements, P.O. Box 2340, D-76125 Karlsruhe, Germany

a r t i c l e

i n f o

Article history: Received 18 May 2010 Received in revised form 21 October 2010 Accepted 22 October 2010 Available online 20 November 2010 Keywords: Mercury Zinc Smelter Lake sediments Peat Flin Flon

a b s t r a c t The copper–zinc smelter at Flin Flon, Manitoba, was historically the largest single Hg point-source in Canada, as well as a major source of Zn. Although emissions were reported by industry to have declined significantly since the late 1980s, these reductions have never been independently verified. Here, the histories of Hg and Zn deposition over the past century or more were determined at five lake sediment and three peat study sites in the surrounding region. At sites spanning the range from heavy to minor pollution, lake sediment Hg and Zn concentration and flux profiles increased significantly in the early 1930s after the smelter opened. Two of the three peat archives were wholly or partially compromised by either physical disturbances or biogeochemical transitions which reduced their effectiveness as atmospheric metal deposition recorders. But the remaining peat records, including a detailed recent 20 yr record at a moderately polluted site, appeared to show that substantive reductions in metal levels had occurred after the late 1980s, coincident with the reported emission reductions. However, the lake sediment results, taken at face value, contradicted the peat results in that no major declines in metal concentrations or fluxes occurred over recent decades. Mercury and Zn fluxes have in fact increased substantially since 1988 in most lakes. We suggest that this discrepancy may be explained by catchment soil saturation by historically deposited metals which are now mobilizing and leaching into lakes, as has been reported from other smelter polluted systems in Canada, whereas the upper sections of the peat cores reflected recent declines in atmospheric deposition. However, further research including instrumented wet and dry deposition measurements and catchment/lake mass balance studies is recommended to test this hypothesis, and to provide definitive data on current atmospheric metal deposition rates in the area. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved.

1. Introduction Mercury (Hg) in the environment continues to be a major focus of international scientific and policy concern because of its intrinsic toxicity and its propensity to bioaccumulate and biomagnify in the human food chain. In Canada, a site of national concern from a Hg emission perspective is the copper–zinc smelter at Flin Flon, Manitoba. It is believed that the smelter, which opened in 1931, was historically the largest single source of Hg air pollution in Canada, and remained the largest emitter in 2000 despite substantial reductions during the 1990s (CCME, 2000). A study of soil humus metal concentrations around Flin Flon found an anthropogenic contribution of 95–100% of total humus Hg within 10 km of the smelter, and 36% at 50–85 km distance; background Hg levels in soil humus were attained approximately 85 to 105 km down-wind (Henderson et al., 1998; McMartin et al., 1999; Sim

⁎ Corresponding author. Tel.: +1 613 996 3958. E-mail address: [email protected] (P.M. Outridge).

et al., 1999). The amount of Hg emitted prior to the 1990s is somewhat uncertain. McMartin et al. (1999) reported that 583 t of particulate Hg was released between the smelter opening in 1931 and 1995, based on industry-supplied estimates. But more recently, extrapolation of the assumed constant rate of Hg release of 19.9 t/yr prior to 1993 (Nilsen, 2003), suggests that N1000 t of particulate Hg may have been emitted since the start of operations. The smelter is also a major source of Zn and other metals. Although not of the same concern from a policy perspective in Canada, the amount of Zn emitted (total of 120,900 t from 1931 to 1995) makes it the largest single pollutant metal at the smelter, representing 26.4% of historical dust emissions (Nilsen, 2003). Based on data submitted by Hudson Bay Mining and Smelting Co. (HBMS) to Environment Canada, metal emissions to the atmosphere from the Flin Flon smelter declined sharply in the late 1980s and early 1990s because of technology improvements. The reductions in Hg amounted to about 90% or 18 t/yr, from 19.9 t/yr prior to 1993 to an average of 1.4 ± 0.2 t/yr in 1995–2002 (particulate Hg only; Nilsen, 2003, Table 4.2; CCME, 2000). As these emission figures were based on sampling of electrostatic precipitator dust only, gaseous elemental Hg

0048-9697/$ – see front matter. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2010.10.041

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

(GEM) releases to the atmosphere were not monitored (Nilsen, 2003). The reduced emissions still constituted half of the 2.8 t/yr emitted in 2000 by the entire Canadian base metal smelting sector, and ~ 12% of the 12 t/yr emitted into the air by all human activities in Canada (CCME, 2000). The reported decrease at Flin Flon significantly influenced the national Hg emission inventory. Environment Canada reported that “between 1990 and 1995, Canadian anthropogenic Hg emissions dropped from approximately 35 to 11 tonnes [per year] primarily as a result of process improvements in the base metal mining and smelting industry.” (Environment Canada, 2004). It can be deduced, therefore, that the ~18 t/yr decrease reported from the Flin Flon smelter accounted for most of the 24 t/yr national reduction between 1990 and 1995. Zinc emissions from Flin Flon also declined substantially in the late 1980s, amounting to an ~80% reduction from 1575 t/yr in 1989 to between 60 and 470 t/yr in 1995–2002 (Nilsen, 2003). Surprisingly, however, these reported reductions have never been independently verified, and the historical and recent rates of atmospheric metal flux in this area have never been quantified. Metals in precipitation were measured over one year in the 1970s, but did not include Hg (Franzin et al., 1979). Interpretation of an early lake sediment core study (Harrison et al., 1989; Harrison and Klaverkamp, 1990) was hampered by the absence of sediment chronologies and sedimentation rates with which to calculate metal fluxes. A reliable reconstruction of the history of Hg and Zn deposition around the smelter is needed, first, to reveal the impact of the smelter on deposition rates in the surrounding environment, and second, as an independent verification of the effectiveness of emission control measures that were implemented beginning in the late 1980s. This study addressed these aims by using lake sediment and peat cores from the surrounding area to reconstruct changes in regional atmospheric Hg and Zn deposition from the mid- to late 19th century

549

to the early 2000s, and by comparing the changes over recent decades against the emissions reductions reported by industry. We determined the down-core trends of Hg and Zn concentrations and fluxes in sediment cores from five sites in four lakes, and three peat cores from peat bogs within two of the lake catchments. This combined approach using spatially clustered sediment and peat archives is uncommon (although see Lamborg et al., 2002), but it was used here because each type of archive presents advantages and disadvantages with respect to reconstructing metal deposition histories (Shotyk, 1996; Biester et al., 2007). 2. Materials and methods 2.1. Area description The area of this study covers about a 0.5° latitude by 1.5° longitude area around Flin Flon, Manitoba (~54.5°N, 101.8°W; Fig. 1) which was established in 1930 as a service town (currently ~7500 people) for the local mines and smelter. The annual average temperature is −0.2 °C, ranging from a mean of 17.8 °C in July to −21.4 °C in January (Canadian Climate Normals, 1971–2000; Flin Flon “A” station). Annual precipitation is about 470 mm, with 30% falling as snow. Winds are predominantly from the northwest from September to February, but tend to be most often southerly from March to August. The surrounding landscape is boreal forest and lakes. Occurring as it does on the Precambrian Shield margin, the area is geologically complex, being underlain by sandy-clay tills derived from Precambrian greenstone-granites and gneisses in the northern part as well as Mesozoic sedimentary and Paleozoic carbonate rocks in the south (McMartin et al., 1996; Henderson et al., 1998). The area was glaciated by ice flowing from the north and northeast, and after ice retreat was entirely inundated by proglacial Lake Agassiz.

Fig. 1. Location map of the study sites near Flin Flon, Manitoba. (NAD83, UTM Zone 14 projection).

550

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

1600 20 Hg

1400

Zn

16

1200

14

1000

12

800

10 600 8 400 6

Zn Emissions (t/yr)

Hg Emissions (t/yr)

18

to 1993 are estimates based on the 2 spot samples in 1977 multiplied by the annual total dust emissions, and should be regarded as less reliable than the post-1993 data (J. Nilsen, pers. comm. to J.B. Percival, Nov. 16, 2006). The significant reductions in particulate Hg releases since 1993 were achieved because of the technical changes in the Zn plant, described earlier, which now has no atmospheric emissions (CCME, 2005). Since 1995, Hg and Zn emission trends have diverged, with Hg showing a consistently low emission rate of 1.4 ± 0.2 t/yr, whereas Zn fluctuated substantially because of technical difficulties or improvements at the Cu smelter's dust bag-house (Nilsen, 2003). 2.3. Study design

200

4

0

2 0 1988 1990 1992 1994 1996 1998 2000 2002

Year Fig. 2. Mercury and Zn emission data for the Flin Flon smelter, 1988–2002. (Source: Nilsen, 2003. Hg data prior to 1993 are regarded as less reliable than those after 1993).

2.2. Smelter Hg and Zn emission history The smelter opened in 1931 to service numerous surrounding mines; 15 mines have operated since then, of which 5 (Flin Flon, Trout Lake, Callinan, 777, and Konuto) were either still operational or under development in 2003 (Nilsen, 2003). The smelting complex presently consists of an ore concentrator, a zinc plant and a copper smelter. Since 1993, Zn metal has been extracted and refined by a sulphuric acid leachate and electrolytic extraction method; prior to that a roastleach-electrowinning process was employed which involved gas and dust emissions. Copper metal is smelted, the process being powered by coal until 2000 when propane burners were installed. Waste gases and dusts are discharged after emission control treatments from the 250 m high stack which was constructed in the early 1970s. Smelter dust emissions were initially curtailed in the late 1980s using textile or Goretex® dust bags, which were replaced in 2000 by electrostatic precipitators (ESP). These were further upgraded in 2002 to increase dust trapping efficiency. Total atmospheric dust emissions from 1931 to 1995 were estimated at 458,100 t but have declined substantially since 1990 (Nilsen, 2003). Since 1993, the Cu smelter accounts for most of the atmospheric metal emissions from the complex (J. Nilsen, pers. comm. to J.B. Percival, Nov. 16 2006). Measurements of particulate Hg and other metal releases from the smelter started sporadically in the 1970s. Mercury was measured in 2 samples of bag-house dust in 1977, and again in 1993; thereafter, regular annual Hg determinations on baghouse and ESP dust samples were carried out. A slightly longer history of Zn emission monitoring is available, because Zn determinations were made on the dust as early as 1988. Fig. 2 shows the putative changes in Hg and Zn emissions from the Flin Flon smelter complex from 1988 to 2002. The dramatic reduction in Zn releases after 1989 from ~1600 t/yr to ~ 300 t/yr was attributed to the implementation of dust trapping bags on stack exhaust (Nilsen, 2003). Mercury data prior

The sampling strategy was to sample lakes and peat bogs spanning the range of smelter pollution from heavily polluted to minimally polluted according to lake sediment Hg and Zn data (Harrison et al., 1989) and surface peat metals data (Zoltai, 1988). Two peat bogs which exhibited ombrotrophic plant species on their surfaces were selected as study sites (Table 1), one 30 km northeast of Flin Flon near Kotyk Lake, and the second about 85 km northwest of Flin Flon near Lake “Sask4”. These were selected because they represent areas of moderate and low levels of smelter contamination, respectively. No actively growing peatlands were found within 30 km of the smelter, and so peat sampling in the most heavily contaminated area was not possible. Single cores were taken from two different peat bogs within the Sask4 catchment (designated SK4_1 and SK4_5), and one core from the Kotyk Lake catchment. Four lakes were chosen for sampling (Table 2); in addition to Sask4 and Kotyk, representing areas of low and moderate contamination, we included Meridian Lake which was among the most heavily contaminated of the lakes studied by Harrison and Klaverkamp (1990), and Persian Lake (also known as Cleaver) which represented a moderately heavy level of contamination. The bathymetry of each lake was determined with a boat-mounted depthsounder and GPS unit, and the deepest points noted for sampling. Because of two distinct basins in Meridian Lake, two sampling sites were located in this lake (designated Meridian North and Meridian South). None of the study lakes were obviously disturbed by recent land-use changes, extensive housing, or shoreline erosion, and thus occur in relatively undisturbed, forested catchments. 2.4. Sample collection and preparation — peat The surfaces of both the Kotyk Lake and Lake Sask4 peatland were dominated by Sphagnum bog species, which are indicative of ombrotrophic (i.e. solely atmospherically fed) peat conditions. A modified titanium Wardenaar corer, measuring 15 × 15 × 100 cm, was used for sampling, employing coring techniques previously described (Givelet et al., 2004). In August 2003, peat monoliths were taken from peat hummocks lying several hundred metres from the nearest road. SK4-5 was a better developed bog compared to SK4-1, but was located within a few hundred metres of a small, unpaved airstrip which is used on average once per day; this may be an additional source of dust deposition at SK4-5. The core monoliths were frozen at − 18 °C within a day of collection and air-transported to the University of Heidelberg, Germany.

Table 1 Coordinates and basic descriptions of peat core sampling sites. Coring site

Latitude; longitude

Peat bog type

Distance/bearing from smelter

Core length (cm)

Kotyk Lake

54° 52.82′N; 101° 27.25′W 55° 21.40′N; 102° 39.18′W 55° 16.93′N; 102° 45.27′W

Sphagnum, mixed ombro-minerotrophic

30 km, NE

90

50

Sphagnum, mixed ombro-minerotrophic

88 km, NW

82

61

Sphagnum, Ombrotrophic

85 km, NW

75

N100

Sask Lake 4, Site 1 Sask Lake 4, Site 5

Water table (cm depth)

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

551

Table 2 Coordinates, water depths and focusing factors of lake sediment coring sites (distances/bearings and surface areas from Harrison et al., 1989). Lake

Distance/bearing from smelter

Surface area (ha)

Coring locations (latitude/longitude)

water depth (m)

Excess 210Pb flux (Bq/m2/yr)

Focusing factor (210Pb-based)

Meridian North

9 km, W

123a

5.1

8 km, W

a

Persian (Cleaver)

23 km, SE

58

Kotyk

30 km, NE

127

Sask4

83 km, NW

25

95 (core 97 (core 129 (core 160 (core 163 (core 124 (core 156 (core 143 (core 136 (core 133 (core

0.6 (both cores)

Meridian South

54° 45.18′ N 101° 59.94′ W 54° 44.71′ N 101° 58.83′ W 54° 42.75′ N 101° 33.08′ W 54° 53.87′ N 101° 28.84′ W 55° 21.13′ N 102° 38.85′ W

a

5.2 6.8 5.2 5.5

1) 2) 1) 2) 1) 2) 1) 2) 1) 2)

0.8 (core 1) 0.9 (core 2) 1.0 (core 1) 0.7 (core 2) 0.9 (core 1) 0.8 (core 2) 0.8 (both cores)

Meridian North and South are different basins within Meridian Lake, which has a total area of 123 ha.

In the laboratory, samples were prepared following the detailed protocol described elsewhere (Givelet et al., 2004). The frozen monoliths were cut into 1 cm slices using a stainless steel band saw. The outer 1 cm edges of each slice were trimmed away using a ceramic knife and Plexiglas template and discarded to avoid possible “smearing” of the peat profile during coring. Following a consistent sub-sampling strategy, core slices were sampled for different types of analyses, and then freeze-dried and homogenized in a stainless steel mill. 2.5. Sample collection and preparation — sediment The study lakes were located next to gravel roads because of the need for boat access, and the sediment coring sites were situated at the lakes' deepest points which were varying distances (~100 to 800 m) from the road. Sampling occurred in July 2004. At the coring sites (which were registered by GPS, Table 2), divers removed duplicate cores from within 3 m of each other. The divers pressed 12 × 60 cm acid-washed Perspex® core tubes into the sediment to a depth of about 40 cm, capped the tops of the tubes, and then dug them out by hand. The tube bottoms were also capped before the tubes were removed. Two divers were used to remove the cores simultaneously so that disturbed sediment from one did not affect the other core. The capped core tubes were kept upright and taken by boat to shore, where they were extruded on site into sliced intervals of 0.5 cm (from 0 – 10 cm depth), 1 cm (10–30 cm) or 2 cm (N30 cm). The samples were transferred quantitatively to plastic Whirlpaks®, kept cool on ice, and frozen at the end of each day. Frozen samples were transported to Department of Fisheries and Oceans laboratories in Winnipeg, Manitoba, where they were freeze-dried. 2.6. Chemical analyses — peat All chemical determinations on peat were carried out at University of Heidelberg laboratories. 2.6.1. Age dating The upper sections of peat cores were analyzed at the University of Heidelberg for 210Pb and 137Cs by low background gamma-spectrometry. About 1 g of dried, milled peat from each slice was packed into plastic containers and counted for 22 h on an HPGe detector (GCW4028, Canberra). 210Pb-based ages and peat accumulation rates were calculated for each core slice down to the depth of excess 210 Pb extinction, using the Constant Rate of Supply model (Oldfield and Appleby, 1984), and checked against the nominal peak in 1963 of nuclear bomb-related 137Cs. 2.6.2. Mercury analysis Mercury concentrations in peat were determined by cold-vapour atomic absorption spectrometry (AAS) on 200 mg of homogenized peat matter using a Leco AMA 254 Hg Analyser (see Roos-Barraclough

et al., 2002). Method detection limit based on repeated blank determinations was 0.01 ng Hg, equivalent to 0.05 ng/g DW in a 200 mg sample. The working range of the instrument was 0.05 to 600 ng Hg; samples exceeding this mass were reanalysed at a reduced sample weight. Analytical accuracy and precision were determined by regularly measuring the following standard reference materials (SRMs), with certified Hg values between 30 and 141 ng/g DW): NIST 1570a Spinach Leaves (N = 17), NIST 1515 Apple Leaves (N = 33), and NIST 1633b Coal Fly Ash (N = 15). Measurements were accurate to within 5% of the certified values, and analytical precision calculated as relative standard deviations (R.S.D.) was within ±5% of the mean measured values. 2.6.3. Other elements The concentrations of Ca, Fe, Zn and Ti were measured directly on 1 g of dried, milled peat powder by X-ray fluorescence, with the TITAN XRF spectrometer (Cheburkin and Shotyk, 2005). Detection limits were 10 μg/g DW for Ca, 5 μg/g for Fe, and 1.5 μg/g for Zn and Ti, which were many times lower than sample concentrations. The analyses were checked for accuracy and precision using NIST 1633b Coal Fly Ash; accuracy was within 10% of certified values with precisions (RSDs) of ±2, ±5, ±5 and ±12% of mean measurements for Ca, Fe, Zn and Ti, respectively. 2.7. Chemical analyses — sediments 2.7.1. Age dating Sediment samples were radiometrically dated at Department of Fisheries and Oceans Canada, Winnipeg (DFO), by measuring 210Pb and 137Cs activity as a function of depth (Oldfield and Appleby, 1984). 210 Pb-based ages and sedimentation rates were calculated for each core slice down to the depth of excess 210Pb extinction, using the Constant Rate of Supply model (Oldfield and Appleby, 1984); overall accuracy of dating was checked against the nominal peak in 1963 of 137 Cs. 2.7.2. Mercury analysis Total Hg in sediment subsamples (~ 0.3 g DW) was analyzed by DFO laboratories using cold vapour AAS analysis of acid-digested samples (see Lockhart et al., 1998). Interspersed analyses of five SRM sediments (LKSD-1 (Canadian Centre for Materials and Energy Technology, N = 11), LKSD-3 (N = 17), LKSD-4 (N = 16), MESS-2 (National Research Council of Canada, N = 16) and PACS-2 (National Research Council of Canada, N = 5)) which spanned the range of sample Hg concentrations gave overall mean accuracies across all analytical batches within 2–11% of certified values, with RSDs of 6– 18% of mean measured values. 2.7.3. Other elements At Geological Survey of Canada laboratories, Ottawa, dried sediment (1 g) was completely digested with nitric, perchloric and

552

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

Metal fluxes in archives were calculated as the product of their concentrations and slice-specific sedimentation or net peat accumulation rates determined from 210Pb profiles. To help distinguish smelter-related metal deposition in peat and sediment archives from other pollution sources, and to evaluate the effectiveness of emission reductions at the smelter, concentration and flux profiles were divided into four segments: pre-1900, 1900–1930 (to determine average fluxes immediately prior to smelter opening), 1931–1988 (from smelter opening to the implementation of control technologies), and post-1988 (after emission controls were in place). Peat flux data theoretically represent atmospheric fluxes directly. However, to be directly comparable to the peat fluxes, mean sediment fluxes were corrected for element focusing by dividing the calculated fluxes by each core's focus factor. The focus factor of sediments was determined by dividing the measured 210Pb flux at the top of each core by a soil 210 Pb value of 175 Bq/m2/yr reported for a similar latitude (Experimental Lakes Area, northwest Ontario) by Lockhart et al. (1998). Significant differences between the mean metal concentrations, Tinormalized concentration values, and fluxes in different time periods were determined by a non-parametric Kruskall–Wallis analysis of variance on ranks followed by a Dunn's multiple comparisons test of means with unequal sample sizes (SigmaStat v. 2.03, SPSS Inc). Use of a non-parametric test in this way minimizes possible erroneous outcomes from non-normal data distributions or unequal variances. Significant differences in the text are at P b 0.05 unless stated otherwise.

hydrofluoric acids over heat, allowed to go to dryness and subsequently made up to 25 mL with 5% ultrapure HNO3. Analyses were by ICP-optical emission spectroscopy for Zn and TiO2; Ti was calculated from its oxide concentration. Procedural blanks and SRM sediments (LKSD-3 (N = 12) and LKSD-4 (N = 12), Canadian Centre for Materials and Energy Technology, Ottawa) were also carried through the procedures. Analytical detection limits were 5 μg/g DW for Zn and 12 μg/g DW for Ti, which were at least several-times lower than the lowest sample concentration. Analytical accuracy for Ti was within 1% of the certified LKSD-3 value, but 21% lower than certified for LKSD-4. Precision was ±3% RSD of the mean measured values. For Zn, accuracy was within 9% for both SRMs, with precisions of better than 2% RSD.

2.8. Data handling and statistics Similar to the approach used by Farmer et al. (2009) and Landers et al. (2008) for peat and lake sediments, respectively, metal concentrations in our archives were normalized against Ti. There were two reasons: first, to correct trends in absolute concentrations for any variations of inputs of natural Hg and Zn in aeolian dust and eroded soil, and thus reduce variability in the metal profiles; and second, to allow cross-archive comparison against the reported smelter emissions. Titanium is a conservative, lithogenic element in the sense that its minerals are resistant to chemical weathering, and its distribution in sediment and peat profiles is taken to reflect the abundance of natural dust and soil inputs from erosional and aeolian sources (Sageman and Lyons, 2003; Farmer et al., 2009). One important qualification in using Ti in this way is that the smelter should not be a source of Ti to the surrounding area. No data on Ti releases are available, however, Ti in humus was not elevated in the vicinity of the smelter (McMartin et al., 1996), which argues against the smelter being a significant Ti source.

0 10

90

3.1. Peat profiles The Kotyk Lake peat core showed a distinct layer of blackened, oxidized plant matter at a depth of 25–30 cm (see ESI Fig. S1) which probably represents a fire passing through the bog in the past. This

Bulk Density (g/cm3)

Moisture (%) 85

3. Results

95

0.0

0.2

0.4

A

0.0

0.5

1.0

1.5

B

Kotyk

Fe (ug/g DW)

Ca (% DW)

0.6

2.0

0

C

25x10 3

100x10 3

D

Depth (cm)

20 30 40 50 60 70 80 85 0 10

Depth (cm)

20

90

E Sask 4

95

0.0

0.2

0.4

0.6

0.0

F

0.4

0.8

1.2

1.6

2.0

G

0

2500 5000 7500 10000

H

SK4_1 SK4_5

30 40 50 60 70 Fig. 3. Background physical and chemical data on the peat cores. (Top panels are for the Koytk site, bottom panels for the Sask4 site).

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

event likely had an impact on the 210Pb and metal profiles at this point in the core and possibly higher and lower in the profile through volatilization of Hg, concentration of non-volatile elements in ash, and making more nutrient elements such as Ca, Fe and Zn available for subsequent plant growth. To try to minimize the effects of this event on the study, metals data only from above 20 cm and below 35 cm depth will be discussed. The surface appearance of the peat cores suggested that all three were ombrotrophic, but Kotyk and SK4_1 contained minerotrophic peat starting at different depths. Core SK4_5, however, appeared ombrotrophic throughout. Physical parameters (moisture content and bulk density) and Ca concentrations supported these observations (Fig. 3). Minerotrophic bogs exhibit lower moisture contents and 210

0

Depth (cm)

0

higher bulk densities than ombrotrophic bogs because of inorganic material inputs from groundwater (Shotyk, 1996). Calcium is a mobile lithogenic element which can be used to indicate the trophic status of peat bogs; Ca contents above ~ 1% dry weight (DW) suggest extensive groundwater contributions of elements from nearby rock–water contact (Weiss and Shotyk, 1997; Roos-Barraclough et al., 2006). The transition from minerotrophy to ombrotrophy in Kotyk occurred at 10–11 cm depth. Only above 9 cm was Ca in the Kotyk core b0.4%, typical of an air-fed bog (Shotyk et al., 2001), whereas below 10 cm and again below 35 cm Ca was N1% (Fig. 3C) which is consistent with minerotrophy. Core SK4_1 exhibited a sharp transition to lower moisture and higher bulk density below 30–35 cm depth (Fig. 3E and F). Calcium contents increased between 10 and 30 cm and was N0.6%

Calendar year

Pb (Bq/kg)

100

200

300

1900 1920 1940 1960 1980 2000 0

A Kotyk

5

5

10

10

15

15

20

20

1975

25

30

40

35

210

Pb

137

Cs

40

45

45 0

50

0

100

100

150

200

200

300

1900 1920 1940 1960 1980 2000

0

508

5

Depth (cm)

0

D Sask4

5

SK4_1

10

10

1985

15

15

20

20

25

25

30

30

35

35

40

C Sask4

40

45

SK4_1

45

50 0

50

0

Depth (cm)

B Kotyk

25

30 35

0 5

100 100

150 200

200

50 250

300

1900 1920 1940 1960 1980 2000

E Sask4

0

F Sask4

SK4_5

5

SK4_5

10

10

15

15

20

20

25

25

30

30

35

35

40

40 0

50

100 137

150

200

250

Cs (Bq/kg)

Fig. 4. 210Pb (unsupported) and

137

553

Cs profiles and age–depth plots for the three peat cores.

554

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

for most of the core below 10 cm (Fig. 3G), which suggests this section may have been receiving groundwater inputs and was intermediate between ombro- and minerotrophic peat. Core SK4_5 generally had b0.4% DW Ca throughout, consistent with ombrotrophy (Fig. 3G). The presence of distinct redox boundaries in peat cores, as indicated by peaks in redox-sensitive elements such as Fe, may influence the distributions of some metals. In the Kotyk peat profile, a sharp peak of 13.3% Fe at 16 cm depth may indicate a distinct redox transition (Fig. 3D), or excessive plant Fe uptake after the fire. In the two Sask4 peat cores the absence of any dominant Fe peaks like that in Kotyk suggests an absence of redox boundaries (Fig. 3H). 210 Pb dating of the cores, and resulting age–depth curves (Fig. 4), suggested that the upper 35–45 cm of the cores represented about the last 100 years, with no cores datable before 1890 with this method. There were substantial differences in net peat accumulation rates over time, as indicated by deviations from the expected inverse exponential decay curves for 210Pb especially in the Kotyk core. In this core, the presumed loss of organic matter and spike in 210Pb associated with the fire (25–30 cm depth) means that the observed 210Pb profile may not be reliable. Therefore the calculated chronology and peat accumulation rates for Kotyk may be compromised and should be treated cautiously. Peaks in 137Cs in Kotyk occurred at 1975 (23 cm) instead of at the expected date of 1963. In SK4_1, the 137Cs peak occurred at 1985 whereas in SK4_5, 137Cs continued to increase to the top of the peat profile. These patterns probably reflect the mobility of 137 Cs in these bogs. Errors in 210Pb dating due to smearing of the 210Pb profile in the layer of living plant matter (Biester et al., 2007) may also cause chronological errors in peat cores, but this possibility cannot be evaluated here. During 1930–1989, when smelter emissions were reportedly highest, Hg and Zn concentrations in the Kotyk peat core (Fig. 5A, B) were

Hg (ug/g DW) 0.0 0

0.1

0.2

over two times higher than in the Sask4 cores (Fig. 5E, F) which were in a less polluted area. However, Ti-normalization of Hg and Zn resulted in overall similar values at Kotyk and Sask4 (Figs. 5C, D vs. 5G, H; Table 3). This is because Ti concentrations in Kotyk peat significantly increased after 1900 (i.e. before the smelter opened) and remained at this level after 1930, while there were no significant Ti changes in core SK4_5 (Table 3). The timing of the Ti increase at Kotyk and the lack of additional increases after the smelter opened suggest that the elevated Ti was not due to emissions (see also McMartin et al., 1996). As human settlement and land clearing began after 1900 in the area around Flin Flon, it is possible that the Ti increase reflects elevated soil dust deposition. Therefore, in Kotyk peat the raw Hg and Zn concentrations may better reflect the timing of atmospheric inputs of smelter metals than Ti-normalized data. Of the two Lake Sask4 peat cores, increases of Hg and Zn levels in the SK4_5 core showed the best agreement with the documented opening of the smelter in 1931 (Fig. 5E, F). The previously described geochemical boundary at about 30 cm depth in SK4_1 coincided with the smelter opening, and with a substantial decrease in Hg concentrations from pre-1930 levels which confounded the mean Hg values in the pre-1900 and 1900–30 periods. Elevated Hg concentrations in SK4_5 below 55 cm depth were eliminated from statistical comparison, which allowed the pre-1900 values to be calculated for this core. Thus, as core SK4_5, the entirely ombrotrophic core, appears to provide a less problematic record of Hg in the Lake Sask4 catchment than SK4_1, mean metal concentrations and statistics were only calculated for this core (see Table 3). Both metals in the Kotyk peat core were generally at their highest levels during 1931–89, showing significant increases above the pre1900 period (Table 3). However, the Zn increase began at about 40 cm depth, prior to 1900 according to the 210Pb-based chronology

Hg/Ti

Zn (ug/g DW) 0

A

50 100 150 200 250

B

0.5

1.0

Zn/Ti 1.5

0.0

0.4

0.8

1.2

1.6

D

C

10

1989

Kotyk Depth (cm)

20 30

1930 1900

40 50 60 70 80 0.00 0 10

E

0.04

Sask 4

0.08

0.12

0

20

40

60

0.5

80 100

G

F

1.0

1.5

0.0

H

0.4

0.8

1.2

1.6 1989 4_1 1989 4_5

Depth (cm)

20 1930 4_5 1930 4_1 1900 4_5

30 40

1900 4_1

50 60 70

SK4_1 SK4_5

Fig. 5. Mercury and Zn concentrations, and Hg/Ti and Zn/Ti profiles in the three peat cores. (Dates shown on right hand side are 210Pb-based, see Fig. 4).

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

555

Table 3 Summary of Hg and Zn concentrations, and Ti-normalized metals data, during different time periods in sediment and peat cores around Flin Flon, and Hg/Ti increases since smelter opening in 1931 (relative to the 1900–1930 period). (Concentration units for both archive types are %DW for Ti, and μg/g DW for Hg and Zn. Data shown as mean ± S.D.; for sediments, the values include data from duplicate cores. Zn/Ti values have been divided by 102. Different superscript letters indicate significant differences among means at P b 0.05). Lake sediments Ti

Zn

Zn/Ti

Hg

Hg/Ti

Hg/Ti increase (%)

Meridian North Pre-1900 1900–1930 1931–1988 Post-1988

0.240A ± 0.003 0.240A ± 0.006 0.224B ± 0.007 0.214C ± 0.004

172A ± 18 321A ± 85 1790B ± 660 1940B ± 120

6.8A ± 1.9 13.4A ± 3.8 80.2B ± 30.3 90.8B ± 4.8

0.19A ± 0.03 0.25A ± 0.03 1.20B ± 0.50 1.87C ± 0.39

0.79A ± 0.14 0.98A ± 0.28 5.38B ± 2.33 8.75C ± 1.76

– – 450 795

Meridian South Pre-1900 1900–1930 1931–1988 Post-1988

0.214A ± 0.005 0.213A ± 0.065 0.197B ± 0.005 0.192B ± 0.004

309A ± 104 696A ± 133 2360B ± 500 2210B ± 220

14.5A ± 5.1 33.1A ± 6.5 120.1B ± 25.9 115.5B ± 9.3

0.19A ± 0.04 0.33A ± 0.05 1.91B ± 0.60 2.45B ± 0.43

0.91A ± 0.20 1.56A ± 0.23 9.76B ± 3.16 12.80B ± 2.25

– – 524 718

Persian Pre-1900 1900–1930 1931–1988 Post-1988

0.131A ± 0.005 0.128A ± 0.005 0.138B ± 0.007 0.141B ± 0.006

103A ± 11 146A ± 40 684B ± 289 864B ± 56

7.9A ± 0.9 11.4A ± 2.8 48.9B ± 19.2 61.1B ± 3.5

0.06A ± 0.03 0.10A ± 0.04 0.37B ± 0.24 0.67B ± 0.26

0.46A ± 0.26 0.74A ± 0.31 2.59B ± 1.64 4.77B ± 2.04

– – 250 544

Kotyk Pre-1900 1900–1930 1931–1988 Post-1988

0.055A ± 0.016 0.046B ± 0.003 0.056A ± 0.004 0.054A ± 0.001

106A ± 22 182A ± 59 446B ± 103 518B ± 16

22.4A ± 5.3 38.6AB ± 9.9 78.7B ± 15.8 95.9C ± 3.0

0.11A ± 0.03 0.17A ± 0.05 0.36B ± 0.12 0.46B ± 0.09

1.75A ± 0.55 2.89A ± 0.39 5.01B ± 1.42 6.61B ± 1.77

– – 73 114

Sask4 Pre-1900 1900–1930 1931–1988 Post-1988

0.295A ± 0.005 0.284B ± 0.003 0.290B ± 0.004 0.285B ± 0.003

128A ± 6 131A ± 4 147B ± 6 150B ± 5

4.4A ± 0.2 4.6AB ± 0.2 5.1BC ± 0.2 5.3C ± 0.2

0.14A ± 0.02 0.19B ± 0.01 0.23C ± 0.02 0.23C ± 0.02

0.46A ± 0.09 0.66AB ± 0.04 0.79B ± 0.06 0.80B ± 0.09

– – 20 22

Ti

Zn

Zn/Ti

Hg

Hg/Ti

0.011A ± 0.004 0.017AB ± 0.009 0.019B ± 0.006 0.018AB ± 0.004

14.5A ± 14.5 118.6BC ± 32.1 149.4C ± 29.0 72.3B ± 44.0

12.0A ± 1.0 69.5B ± 16.5 84.1B ± 31.1 40.5B ± 19.8

0.06A ± 0.04 0.13AB ± 0.02 0.17B ± 0.04 0.10A ± 0.02

5.32A ± 2.56 7.56AB ± 0.92 9.15B ± 2.38 5.61A ± 1.23

19 − 27

0.011A ± 0.005 0.009A ± 0.002 0.011A ± 0.004 0.008A ± 0.002

14.0A ± 8.4 37.9AB ± 2.6 72.8B ± 14.0 58.3B ± 13.1

19.2A ± 20.2 46.6AB ± 12.4 70.8B ± 25.9 72.6B ± 18.8

0.06A ± 0.02 0.04A ± 0.00 0.07AB ± 0.02 0.09B ± 0.02

0.59A ± 0.19 0.42A ± 0.11 0.64A ± 0.25 1.11B ± 0.29

– – 51 163

Peat bogs

Kotyk Pre-1900 1900–1930 1931–1988 Post-1988 Sask4 (core SL4_5) Pre-1900 1900–1930 1931–1988 Post-1988

(Fig. 5B), whereas Hg increases began at around 50 cm depth, or early in the 19th century (estimated). In core SK4_5, only Zn was significantly elevated above pre-1900 levels during 1930–89 (Fig. 5F) and its increase also began prior to 1900. Possible explanations for these patterns are either that the chronologies in the lower cores are in error to varying degrees, or post-depositional mobility of metals occurred, or diffuse sources of metal pollution from outside the Flin Flon region were present prior to 1930. Significant decreases in peat Zn concentrations occurred after 1989 (i.e., above 10 cm) in the Kotyk core (see Table 3 and Fig. 5). The coincidence between documented improvements at the smelter in the late 1980s and the timing of decline in Zn concentrations suggests that the chronology of the upper Kotyk core, representing the last two decades, may be relatively accurate. The minimum peat Zn value (43.9 μg/g) occurred at 1995 (6 cm depth), and was followed by another increase up to 66.3 μg/g by 2002, a pattern which agrees to within ± 1 year of the fluctuating emissions reported from the smelter during this period (see Fig. 2). The decrease to 1995 was ~75% down from the peak value in 1989, which is also similar to the reported ~80% reduction of Zn emissions (see Fig. 2). These changes in Zn concentrations in the upper 10 cm of the core occurred immediately above a shift from miner-

Hg/Ti increase (%) – –

otrophy to ombrotrophy, based on the Ca concentration profile (see Fig. 3C). However, as Zn was not elevated in the minerotrophic section below 40 cm depth, it appears unlikely that the change in trophic status interfered with the Zn profile in the upper core. This section of the core is also well above the fire-affected zone at 20–35 cm. However, the magnitude of changes in atmospheric inputs seen in peat archives should be interpreted with care from raw metal concentration data alone, because of possible changes in metal concentrations brought about by coincidental variations in plant growth rate, organic matter decomposition, and compaction. The recent Hg profile in Kotyk, on the other hand, was not as easily compared with emission records, because of the assumed constant rate of emissions of 19.9 t/yr prior to 1993 (Nilsen, 2003). Thus the observed changes of Hg in peat during the late 1980s and early 1990s may reflect emissions at the smelter, but this cannot be checked. Mercury concentrations during the post-1989 period were significantly lower than in 1930–89 (Table 3), and the timing of the decline in peat (beginning in 1990) occurred earlier than the reported emission reductions beginning in 1993 (see Fig. 2). In Sask4 peat core SK4_5, Ti concentrations did not change upcore, and so Ti-normalized data may be used in conjunction with raw

556

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

concentrations. As in Kotyk peat, Zn concentrations in SK4_5 decreased after 1989. However, the decline began from peak values of 88–94 μg/g in 1976–79, shortly after the tall stack at Flin Flon was installed. Zinc content reached a minimum of 38.0 μg/g in 1998 (18–20 cm depth) before again increasing to 70.7 μg/g in 2003, a fluctuating pattern which is similar to Zn in Kotyk and to the technical history of the smelter. Ti-normalized trends (Fig. 5H) generally followed those of raw Zn concentrations. Mercury concentrations and Hg/Ti in SK4_5 showed more subtle increases after 1930 than those of Zn and Zn/Ti, with no significant increase in 1930–1989 compared to pre-1900 or 1900–30 mean values (Table 3). Peak Hg and Hg/Ti were attained in 1997–98, and decreased thereafter to the top of the core in 2003. Because of the putative chronology errors lower down the Kotyk core, which would have affected calculated net peat accumulation rates, and the previously described problem with the Hg profile of core SK4_1, atmospheric Hg and Zn fluxes (accumulation rates) will be presented here only for SK4_5 (Fig. 6; see ESI Fig. S2 for the Kotyk and SK4_1 profiles). Zinc flux increases in SK4_5 started in the 1930s and attained a maximum (91 mg/m2/yr) in 1991. Subsequently, the flux declined rapidly to a minimum of 33 mg/m2/yr in 1997 before rising again to 62 mg/m2/yr in 2002. Zinc flux was significantly higher during 1930–89 and post-1989 compared to 1900–30 in SK4_5 (Table 4), but the magnitude of the overall increase in flux was influenced by increasing net peat accumulation rate over time as well as by changes in Zn concentrations (see Fig. 6A). The Zn and Hg fluxes were significantly correlated with net mass accumulation rate (P b 0.001). Although metal fluxes are calculated based on metal concentrations and peat accumulation rate, the significant association between peat accumulation and metal flux (see also ESI Fig. S2) suggests that biological factors related to organic matter growth and/ or decomposition may have affected the Zn and Hg flux reconstructions from this peat core. 3.2. Lake sediment profiles Radiometric profiles (210Pb and 137Cs) and age–depth plots derived using the CRS Model are presented in Fig. 7 for each lake sediment sampled in 2004. In most cases, the radioisotopes did not exhibit the monotonic declines in 210Pb expected of undisturbed sediments with a constant sedimentation rate. Except for Persian Lake, in most cores both radioisotopes displayed flattened profiles in their upper sections, with Sask4 and Kotyk lakes being most severely affected. Two possible explanations are: physical and/or biological mixing of the upper cores, or increases of sedimentation rates in recent years which diluted the 210Pb concentrations. The former

Net peat accumulation (kg/m2/yr) 0.0 2000

A

0.4

0.8

process would likely result in inaccurate chronologies and calculated sedimentation rates. An aid in checking the accuracy of the 2004 chronologies is provided by a companion paper (Percival and Outridge, in preparation), which compared metal profiles in the 2004 sediment cores against those sampled from the same lakes in 1985 by Harrison et al. (1989). Although the 1985 cores were not 210Pb-dated, by re-plotting their Zn profiles against our 2004 data so that the top of their cores corresponds to a 210Pb date of 1985, we can see whether the two profiles are superimposed on each other or not. If the 2004 chronologies were seriously in error, or if significant diagenesis/mobility of the Zn profiles had occurred between 1985 and 2004, then the two sets of profiles would not overlap because the position of 1985 in the 2004 cores would be incorrect. In a few cases, especially for high Zn concentration samples in Meridian Lake, reanalysis of 1985 sediments showed errors in the original 1985 data. But for Meridian South, Kotyk, Persian and Sask4 lakes, the 1985 and 2004 core profiles were in good agreement with each other in terms of the timing of initial concentration increase in the 1930s and the shape of the subsequently increasing profiles (see ESI Fig. S3). This suggests that the 210 Pb-derived dates for the 2004 cores are accurate to within ±1 cm or 2–3 years in those lakes, given the rates of vertical sediment accumulation per year. This agreement also suggests that the effect of diagenesis or upward migration of metals was negligible. However in Meridian North, the shape of the Zn profile in 2004 was parallel to that from 1985 but offset by about 2–4 cm, with the 1985 data appearing lower in the 2004 profile than it should have (i.e. older by about 5–8 years depending on the slice depth). Intra-basin heterogeneity could be part of the reason for this difference, and for some of the small differences among other lakes, because the 1985 sampling locations within each lake were unspecified. Harrison et al. (1989) also sampled only one location in Meridian Lake whereas we cored two distinct basins (Meridian North and South), and it is unknown which one was sampled in 1985. But the most conservative explanation for the difference is that 210Pb dates of the 2004 Meridian North cores are too recent by 5–8 years. This helps to constrain the sediment dating errors to not more than 8 years in Meridian North, and less than 3 years in the other lakes. As with the peat data, the conservative lithogenic element Ti was employed as a normalizing element to assist in reducing the variability in sediment profiles introduced by naturally occurring Hg and Zn. Titanium concentrations showed minor and mostly nonsignificant fluctuations up-core in all of the study lakes, varying by a few percentage points across different time periods depending of the lake (Table 3). In contrast, the profiles of Hg and Zn concentrations, and of Hg/Ti and Zn/Ti, in most lake sediments showed consistent and

Zn flux (mg/m2/yr) 1.2

0

20

B

40

60

Hg flux (ug/m2/yr) 80

0

25

50

75

100

125

150

C

1990 1980 1970

Year

1960 1950 1940 1930 1920 1910 1900 1890 Fig. 6. Peat mass accumulation rates and Zn and Hg fluxes (accumulation rates) calculated for the peat cores. (Dates on y-axes are 210Pb-based, see Fig. 4).

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563 Table 4 Summary of Zn and Hg fluxes during different time periods in sediment and peat cores near Flin Flon, and flux increases since smelter opening in 1930 (relative to the 1900– 1930 period). (Kotyk and SK4_1 peat data are excluded because of problems associated with dating, and/or trophic changes down-core (see text). Flux values shown as mean ± S.D.; for lake sediments these values include duplicate cores. Units are mg/m2/yr for Zn and μg/m2/yr for Hg in both types of archives. Sedimentary fluxes are focus-corrected, see Table 2 for focus factors. Different superscript letters indicate significant differences among means at P b 0.05). Lake sediments Zn flux

Zn increase (%)

Hg flux

Hg increase (%)

Meridian North Pre-1900 1900–1930 1931–1988 Post-1988

125A± 38 219A ± 47 604B ± 216 858C ± 83

– – 176 292

145A±57 176A ± 48 405B ± 169 832C ± 201

– – 130 372

Meridian South Pre-1900 1900–1930 1931–1988 Post-1989

177A ± 73 116A ± 43 545B ± 173 742C ± 80

– – 372 542

112A ± 47 202A ± 32 452B ± 195 826C ± 170

– – 124 309

Persian Pre-1900 1900–1930 1931–1988 Post-1988

35.4A ± 22.1 28.6A ± 6.4 107B ± 38.3 156C ± 6.5

– – 273 445

15.5A ± 8.4 19.1A ± 10.0 56.1B ± 34.8 122C ± 49.9

– – 194 538

Kotyk Pre-1900 1900–1930 1931–1988 Post-1988

8.4A ± 2.9 10.9AB ± 2.0 40.6B ± 16.4 87.8C ± 12.8

– – 272 704

8.3A ± 3.1 10.5AB ± 1.4 33.3B ± 15.8 78.6C ± 18.3

– – 218 650

Sask4 Pre-1900 1900–1930 1931–1988 post-1988

29.9A ± 13.2 25.1A ± 6.2 33.4A ± 5.2 54.5B ± 7.3

38.1A ± 15.6 35.8A ± 9.9 52.5A ± 9.9 81.9B ± 11.0

– – 47 129

– – 33 117

Peat bogs Zn flux Sask4 (SK4_5) Pre-1900 – 1900–1930 7.6A ± 1.4 1931–1988 43.4B ± 16.8 Post-1988 55.4B ± 25.4

Zn increase (%) – – 478 639

Hg flux – 8.0A ± 1.2 38.6A ± 15.6 82.1B ± 28.9

Hg increase (%) – – 381 924

dramatic increases after 1930 compared to the pre-1900 and 1900–30 periods (Fig. 8 and Table 3). The mean metal concentrations during 1930–89 and the increase in mean Hg/Ti and Zn/Ti above 1900–30 declined with increasing distance from the smelter (Table 3), thus implicating smelter emissions as the primary factor behind the increases. Only in Lake Sask4, the most distant from Flin Flon, were there relatively small (but still significant) Hg and Zn concentration increases above the 1900–30 average. The increases of Zn and Hg in all lakes ceased or slowed in the mid1970s, and between 1976 and 1981 declining or stabilizing patterns of concentrations began in Meridian, Persian and Kotyk lakes. Surprisingly, in Persian and Meridian cores in the post-1988 period, consistently higher Hg concentrations and Hg/Ti were found, although the increase above 1930–88 was only statistically significant in Meridian North. The maximum Hg concentrations found throughout the Meridian North and South cores (3.0 and 3.5 μg/g DW, respectively) were recorded during the post-1988 period, with mean concentrations of 1.87 ± 0.39 and 2.45 ± 0.43 μg/g DW, respectively. These sites, and Persian Lake, also recorded marked spikes in Hg and Hg/Ti during the late 1990s or early 2000s (see Fig. 8A–I), in contrast to the suggested smelter emission history. Zinc concentrations in most lakes, on the other hand, were generally lower post-1988, except for Sask4 in which no obvious decline occurred.

557

None of the changes in metal concentrations and Hg/Ti and Zn/Ti were correlated with concentrations of total organic carbon or redoxsensitive elements such as Fe and U (see ESI Fig. S4), which indicates that the variations in sediment Hg and Zn profiles were probably not influenced by changes in organic matter inputs or by post-depositional mobility related to redox transitions as sometimes occurs (e.g., see Outridge et al., 2005). The sedimentary fluxes of Hg and Zn displayed significant increases following the opening of the smelter in 1931 up to 1988, and again after 1988 in most cores (Fig. 9 and Table 4). The exception to this pattern was Lake Sask4 which exhibited significant increases after 1988 but not during 1930–88. Increasing sedimentation rates mostly account for the significant post-1988 increase of Hg and Zn flux in Sask4. Hg/Ti increased during this period by only about 20% compared to the pre-1930 period (see Table 3) whereas Hg and Zn fluxes increased by ~ 120% (see Table 4). There was an unexplained discrepancy between sedimentation rates in the lower part of the Sask4 profiles (Fig. 9M) which led to corresponding differences between the duplicate cores in Hg and Zn fluxes. Similarly, increasing sedimentation rates in Kotyk Lake after 1980 account for much but not all of the significant increases of Hg and Zn fluxes post-1988, which were 650–700% above the 1900–30 fluxes in this lake (see Table 4). Sedimentation rates increased by about 300% from pre-1930 levels to the 1990s–2000s (Fig. 9J), but Hg/Ti in Kotyk post-1988 was only elevated by 114% (see Table 3). Sedimentation rate increases in Meridian and Persian lakes were considerably more muted than in Sask4 and Kotyk, and thus explained less of the increase in Hg and Zn fluxes. Persian Lake did not exhibit marked increases after 1930 in its sedimentation rate, and displayed Hg flux increases of 190% in 1931– 88 and 540% post-1988 relative to the 1900–30 period (Table 4). 4. Discussion 4.1. Comparison of the natural archives with emission records The key question being addressed by this study is whether there is evidence in the lakes sediments and peat bogs around Flin Flon of the reported large reductions in Hg and Zn emissions from the smelter after the late 1980s. Reconstructions of atmospheric metal deposition history from these natural archives offer the only possible independent check of those reports. The sediment study sites spanned the range of smelter pollution from heavy (Meridian and Persian lakes) to moderate (Kotyk) to near-background (Sask4). Coincident peat study sites in the Kotyk and Sask4 catchments allowed a combined approach to reconstructing metal deposition history, which has an advantage when evaluating the smelter's emission records because it may overcome limitations associated with either single archive. The sources of metal entering each archive overlap but are different. Sediments reflect varying degrees of focusing of particle-bound elements (both natural and anthropogenic) from their catchments as well as direct atmospheric deposition, whereas ombrotrophic bogs receive inputs only from the atmosphere. As expected (see Lamborg et al., 2002; Biester et al., 2007), the lake sediments in this study contained higher concentrations of Hg than peat, as well as Zn and Ti (see Table 3). There is also the potential in each archive type for different postdepositional mechanisms to drive changes in trace metal or 210Pb and 137 Cs profiles so that the chronology and amounts of atmospheric deposition are obscured. In sediments these mechanisms may include chemical diagenesis, change in redox state, altered water column productivity, and erosional processes in catchments and within-lake sediment slumping, whereas in peat bogs they include changes in plant species, micro-topography, or redox state, particle through-fall, organic matter humification and decomposition, trophic transitions, and surface disturbance by fire or other physical processes (Shotyk, 1996; Lamborg et al., 2002; Bindler et al., 2004; Outridge et al., 2005;

558

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

Activity (Bq/g) 0.0 0

0.2

0.4

0.6

Activity (Bq/g) 0.8

1.0

Depth (cm)

A

0.0 0

0.2

20

Cs-137

C

20

Cs-137

Pb-210

Depth (cm)

Core 1

Meridian North

Meridian North - Core 2 40 0

E

F

10

10

10

20

20

20

30

30

30

Meridian South - Core 1

40

Meridian South - Core 2

0

0

G

40

I

10

10

10

20

20

20

30

30

30

40 0

J

Persian

Persian - Core 2

Persian - Core 1 0

40 0

K

L

10

10

10

20

20

20

30

30

30

Kotyk - Core 1

40 0

Kotyk - Core 2

40

O

N

5

5

5

10

10

10

15

15

15

20

20

20

Sask 4

Sask 4 - Core 2

Sask 4 - Core 1 25

Kotyk

40 0

0

M

Meridian South

0

H

40

Core 2

30

40 0

D

20

Pb-210 30

40

Depth (cm)

1.0 1850 1875 1900 1925 1950 1975 2000 0 10

Meridian North - Core 1

Depth (cm)

Date 0.8

B

30

Depth (cm)

0.6

10

10

40 0

0.4

25

25 Fig. 7. 210Pb (unsupported) and

137

Cs profiles and age–depth plots for the five sediment cores.

Biester et al., 2007). Because of problematic chronologies, historical fire damage, and/or changes of trophic status in some peat records, the most reliable peat archives around Flin Flon are believed to be Sask4 core SK4_5 and the upper 10 cm of the Kotyk core. Metal fluxes calculated for the ombrotrophic SK4_5 profile were strongly related to net peat accumulation rates, which means that these fluxes may be unreliable overall as an independent check of the smelter emissions. However, confidence in the accuracy of the most recent flux data (post-1988) at Sask4 is increased by the finding that the Sask4 sediments and peat core SK4_5 gave mean Hg and Zn fluxes for the post-1989 period within a few percent of each other (see Table 4).

Prior to this, however, there is increasing deviation between sediment and peat for both Hg and Zn fluxes, with peat showing progressively lower mean values relative to sediments, as has been previously found (Lamborg et al., 2002; Biester et al., 2007). The chronologies for most sediment archives appeared to be accurate to within 3 years, except for Meridian North where the error may be up to 8 years. Therefore, the evaluation of the emission records will be based on the SK4_5 and upper Kotyk peat profiles, and the five sediment archives. Except for Hg in SK4_5 peat, which was most remote from the smelter, these archives agreed that historically there had been a substantial impact from smelter emissions on Hg and

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

Hg (ug/g) 0

Depth (cm)

0

1

Zn (ug/g)

2

3

0

1000

2000

Zn/Ti

Hg/Ti 3000

0

B

A

559

3

6

9

12

15

18 0

C

2500 5000 7500 10000

D

1989

10

1930

20

Core 1 Core 2

30

Meridian N 40 0

Depth (cm)

0

1

2

1000

E

2000

3000

0

3

6

9

12

15

18

0

G

F

2500 5000 7500 10000

H

1989 1930

20 30

Meridian S 0.0

0

Depth (cm)

0

10

40

0.5

1.0

0

500

1000

0

3

6

9

12

15

J

I

18

0

2500 5000 7500 10000

L

K

1989 Core 2 1989 Core 1

10

1930

20 30

Persian

40

0.0 0

Depth (cm)

3

0.2

0.4

0.6

0

250

500

0

N

M

3

6

9

12 15 18

0

O

2500 5000 7500 10000

P

10

1989

20

1930 Core 2

30

1930 Core 1

Kotyk

40 0.0

0.1

0.2

0.3

120

140

160

0

3

6

9

12

15

18

0

2500 5000 7500 10000

Depth (cm)

0

1989

10 1930 20 30 40

Q

Sask4

R

S

T

Fig. 8. Mercury and Zn concentrations, and Hg/Ti and Zn/Ti profiles in the sediment cores. (Dates shown on right hand side are 210Pb-based, see Fig. 7)

Zn levels in the surrounding area, consistent with previous studies in various media (Zoltai, 1988; Harrison et al., 1989; Harrison and Klaverkamp, 1990; McMartin et al., 1999). The SK4_5 peat core and the lake sediment cores generally accurately recorded the timing of initial increases of Zn and Hg concentrations and fluxes during the 1930s, which continued to increase up to the 1970s or beyond depending on the archive. The sediments exhibited the expected spatial pattern of declining maximum Hg and Zn concentrations, Hg/ Ti, Zn/Ti and metal fluxes with increasing distance from Flin Flon. The reductions or stabilizations of metal concentrations that occurred in most sediment cores between 1976 and 1981 followed construction of the 250 m high stack at the smelter in the early 1970s, which would be expected to distribute metal contamination over a wider area and reduce local metal deposition rates. Thus, there are several inflection

points in the pre-1988 sediment and peat records which coincide with the documented changes in smelter operations. Both Hg and Zn in sediments, and Zn in peat core SK4_5, exhibited subtle increasing trends in the decades prior to smelter start-up, which may reflect deposition of anthropogenic Hg and Zn transported from distant sources. The independent validation of the sediment chronologies argues that these increases, which occurred in all the lakes, were not simply an artifact of inaccurate dating of the 1930 point in the cores. The increases started at least 10 cm lower than the 1930 point, and below the point at which 210Pb dating was possible but which would have been about the mid-19th Century or later. Elevated Hg concentrations and fluxes in peat bogs (Benoit et al., 1994; Roos-Barraclough et al., 2006) and lake sediments (Engstrom and Swain, 1997), possibly related to regional coal burning, were

560

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

Sed. rate (g/m2/yr) 0 2000

200

400

600

Zn flux (mg/m2/yr) 800

0

Meridian N

A

Hg flux (ug/m2/yr)

100 200 300 400 500 600

0

B

200

400

600

800

1000

200

400

600

800

1000

C Smelter

1980

Year

1960 1940 1920 1900 Core 1

1880

Core 2

1860 0 2000

200

400

D

600

800

0

200

400

600

800 0

F

E

Meridian S

Smelter

1980

Year

1960 1940 1920 1900 1880 1860 0 2000

50

100 150 200 250 300

G

0

25

75 100 125 150 175 0

50

H

Persian

50

100

150

200

I Smelter

1980

Year

1960 1940 1920 1900 1880 1860 0 2000

40

80

120

160

200

0

J

20

40

60

80

100

0

20

40

60

80

100

L

K

Smelter

1980

Year

1960 1940 1920 1900 1880

Kotyk

1860 0 2000

100

200

300

400 0

M

10

N

20

30

40

50

60 0

20

40

60

80

O Smelter

1980

Year

1960 1940 1920 1900 1880

Sask 4

1860 Fig. 9. Sedimentation rates and Hg and Zn fluxes calculated for the sediment cores. (Dates on y-axes are

210

Pb-based, see Fig. 7).

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

already detectable by the late 19th Century in the northeastern USA, and there is abundant evidence that anthropogenic Hg fluxes increased in temperate and sub-Arctic regions of North America after 1930 (e.g., Engstrom and Swain, 1997; Lockhart et al., 1998; Belzile et al., 2004; Roos-Barraclough et al., 2006; Muir et al., 2009). However, compared with the smelter inputs after 1930, these distant and diffuse sources were of negligible importance around Flin Flon. Muir et al. (2009) estimated an average focus-corrected anthropogenic Hg flux in the subArctic of Canada of 7.5 ± 5.9 μg/m2/yr for recent decades. If this value is accurate and applicable to Flin Flon, then long-range Hg is likely to have contributed b1% of the ~830 μg/m2/yr Hg flux to Meridian Lake sediments, b7% of the 120 μg/m2/yr at Persian Lake, and b10% of the ~80 μg/m2/yr at Kotyk and Sask4 lakes in the decades after1988 (see Table 4). The massive tonnage of Zn (N120,800 t; Nilsen, 2003) emitted at Flin Flon over six decades after 1930 also argues against distant sources of Zn playing a major role around Flin Flon. 4.2. Deviations after the 1980s In the period after 1988, the peat and sediment archives diverged in terms of metal concentration and flux trends. The two peat cores agreed with each other in terms of the timing of Zn deposition changes during this period. The upper 10 cm of the Kotyk peat core, and the upper SK4_5 core, both displayed declines of Zn concentrations and Zn/Ti consistent with the timing of reported emission declines after the late 1980s (cf. Figs. 2 and 5A, D). Subsequently, increases of Zn and Zn/Ti in peat corresponded to the increase of Zn emissions due to technical problems at the smelter during 1995– 2001. The timing of the Hg decline in the upper Kotyk peat core (beginning in 1990) occurred earlier than the reported emission reductions beginning in 1993 (see Fig. 2), but as there were no measurements (only estimates) of Hg emissions prior to 1993 it is uncertain whether this means the archive in erroneous on this point. The Hg record in Kotyk peat did not show an increase during 1995– 2001, unlike Zn, which also agrees with the stable Hg and increasing Zn particulate emissions after 1995. Despite the influence of increasing peat accumulation rate over decades on metal fluxes, the timing of the Zn flux patterns in SK4_5 matched the reported emission trends as well, including the decline after 1989 to a minimum in 1997 and the subsequent increase, agreeing to within ±2 years with the Kotyk record. Although the timing of emission changes appeared to be reflected by the peat cores, the accuracy of peat-based reconstructions of emission reductions based on either raw metal concentration data or fluxes should be interpreted with care. Possible changes in concentrations may be brought about by coincidental variations in plant growth rate, organic matter decomposition, and compaction (Shotyk et al., 2003). Concentration data ideally should be converted to fluxes for this purpose but, as discussed, the fluxes in SK4_5 appear to be influenced by net peat accumulation rate. The magnitude of reduction of Zn flux in SK4_5 in the early 1990s was about 65% from its peak in 1991, while Hg and Hg/Ti declined by about 50% from peak values. In Kotyk peat, Zn, Zn/Ti, Hg, and Hg/Ti values declined by ~75% for Zn and Zn/Ti, and by ~ 50% for Hg and Hg/Ti, to minimum values post1988. The relative magnitude of these reductions of metal fluxes and concentrations after 1988 is somewhat less than the emission reductions reported by Hudson Bay Mining and Smelting Co. which amounted to N90% for Hg and ~80% for Zn (Nilsen, 2003; see Fig. 2). Again, these estimates of the magnitude of reductions must be treated with caution. But another explanation for these smaller than expected reductions could be the fact that the peat sampling sites were relatively remote from the smelter, with larger reductions possibly occurring closer to the smelter because of the rapid fall-out of larger dust particles. In contrast to peat, the lake sediments in most cases did not display significant decreases in concentrations or fluxes following their

561

stabilization in the mid-1970s. This initial stabilization followed construction of the higher dust exhaust stack which likely would have reduced metal loadings in the vicinity of the smelter. The coincident inflection point in sediment metal concentration profiles suggests that the sediments were reliably tracking changes in atmospheric deposition up to at least the 1970s. However, subsequent to the additional N90% and ~ 80% decreases in Hg and Zn emissions after 1993 and 1988, respectively (Nilsen, 2003), sedimentary metal concentrations and fluxes did not correspondingly decline. Instead, substantial increases of Hg and Zn fluxes and in some cases increases of concentrations occurred. The two alternative explanations for this finding are: either the smelter's emission records are wrong, or the sediments were not faithfully tracking changes in atmospheric metal deposition after 1988. At this stage of our understanding, the uncertainties surrounding the peat core chronologies and fluxes in particular prevent a definitive explanation. The present results from the two peat study sites (Kotyk and Sask4), which were relatively remote from the smelter compared to some of the lake sediment sites, support the idea that smelter emissions did decline after the late 1980s, although perhaps not by the same relative amounts as reported. However, the peat dataset consists of single cores at these sites, and reliable flux data could not be calculated at either site. Although increased sedimentation rates played a part in the recent sediment flux increases, they did not account for the results in Meridian and Persian lakes where sedimentation increases were minimal, nor for the increases of Hg concentrations in most lakes. The monotonic decline in 210Pb with depth in Persian Lake suggests that there was no significant mixing of sediment layers in that lake, and that sedimentation rates were consistent over recent decades. This lake displayed similar relative temporal patterns of Hg and Zn fluxes or concentrations to the other lakes. Mercury concentrations and Hg/Ti in Persian Lake increased by about 4-fold after 1990, while the Hg flux doubled and Zn flux rose by about 10%. Upward migration of remobilized metals in sediment porewaters does not appear to explain these trends, given that a comparison of the metal profiles in this study against the original data of Harrison et al. (1989) indicates that the pre-1985 profiles of Hg, Zn and other metals have been stable over the intervening 19 years (Percival and Outridge, in preparation). The absence of significant Fe and U peaks in the sediment profiles also argues against a redox-mediated upward mobilization of subsurface metals (see ESI Fig. S4). There are virtually no other datasets available to independently verify the sediment or peat archive data. The only instrumented measurements of metal deposition in the Flin Flon area come from an early precipitation sampling study, which included Zn but not Hg (Franzin et al., 1979). These data suggest that the Zn fluxes reconstructed from lake sediment data in the present study were of similar magnitude to those in precipitation at the time. During 1976–77, the Zn wet flux at Douglas Lake (2 km east of Meridian Lake) was 760 mg/m2. By comparison, the focus-corrected Zn flux from 1975 to 1980 was 775 ± 82 mg/m2/yr in Meridian North, and 677 ± 147 mg/m2/yr in Meridian South. Although this agreement is encouraging, the fact that the wet deposition measurements did not include dry deposition is problematic. Dry deposition would be incorporated in the sediment flux values. Given the large amount of particulate Zn emitted over time, the total (wet + dry) atmospheric Zn flux must have been larger than that measured in precipitation, suggesting that the sediment data may have under-estimated total atmospheric Zn deposition rates. Further comparisons with this study are precluded by the fact that no other study lake or peat bog is located close to the precipitation sampling sites, which exhibited significant variations over relatively short distances. For example, at Patmore Lake, located ~ 5 km southwest of Douglas Lake, annual Zn wet flux was only 182 mg/m2 (Franzin et al., 1979). Unfortunately, no instrumented Hg flux measurements are available from this region.

562

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563

4.3. On-going impact of the smelter on sediment metals levels In studies around other base metal smelters on the Canadian Shield, metal concentrations and fluxes in lake sediments were generally found to reliably track the historical increases and subsequent declines of emissions of Hg, Zn and other metals (e.g., Belzile et al., 2004; Couillard et al., 2008; Tropea et al., 2010). Here, however, the stability or continuing increases of lake sediment metal concentrations and fluxes after 1988 indicate on-going loadings of Hg and Zn to the lakes at similar or even higher rates to those prevailing during the 1930–88 period, in contrast to the peat archives. One possible explanation for this finding, proposed by Nriagu et al. (1998) for severely contaminated lakes near Sudbury, Ontario, is soil “saturation” with historically deposited metals in lake catchments. Nriagu et al. (1998) suggested that these metals were slowly remobilized from soil and till reservoirs by groundwater flows, wind blown dust, and surface runoff, so that catchment soils recently have begun to act as major sources of metals to sediments. Similarly, modeling by Selin et al. (2010) indicated that the cycling of historical pollution can be an important constraint on how Hg levels in aquatic ecosystems respond to reductions in atmospheric emissions. In the present study, leaving aside Kotyk and Sask4 lakes in which recent increases in sedimentation rates strongly influence metal fluxes, catchment saturation may be an explanation for recent trends in the more heavily polluted Meridian and Persian Lakes. Soil metals data support this hypothesis. Smelter Hg and Zn inputs were estimated to comprise a maximum of 65% and 82%, respectively of total Hg and Zn in humus out to 50 km distance (McMartin et al., 1999). Within 25 km of the smelter, the contributions were up to 83% Hg and 91% Zn. There was also evidence that humus Hg and Zn were mobile and leaching into subsurface tills within 10–25 km of the smelter. In most areas, humus Zn was predominantly in a soluble organic fraction whereas N95% of the humus Hg was in insoluble fractions, possibly bound in sulfides, silicates and Fe-oxides (Henderson et al., 1998). A key question regarding this hypothesis is why the erection of the tall smelter stack in the mid-1970s, which presumably caused atmospheric metal loadings in the immediate area of the smelter to decline, resulted in stable or declining Hg and Zn inputs to local lake sediments, whereas the putative major emission cuts in the 1980s and 1990s did not. A possible explanation, which requires testing, is that there was a time lag of several decades between the start of heavy metal loadings and the beginning of metals leaching from catchments. One means of testing the earlier hypothesis, and of assessing smelter emissions at the present time, would be mass balance studies of metal inputs and outputs in surface waters, runoff, groundwater and deposition around key study lakes, incorporating wet and dry metal deposition measurements. Persian and Meridian lakes would be preferred study sites. Wet and dry deposition monitoring of Hg and other trace metals in the area is particularly important, because it would provide a direct (i.e. not reconstructed) dataset of atmospheric metal fluxes. The Hg flux data could be compared to data from southern Canadian sites available from the Mercury Deposition Network (www.nadp.sws.uiuc.edu) and from northern background sites at Churchill, Manitoba, and in northern Alberta (Sanei et al., 2010), to determine how polluted the Flin Flon atmosphere presently is above regional background levels. Supplementary materials related to this article can be found online at doi:10.1016/j.scitotenv.2010.10.041. Acknowledgements This study was funded by the Metals in the Environment Program of the GSC and the German Bundesministerium für Bildung und Forschung, and received material support from the Department of Fisheries and Oceans, Winnipeg. We thank the following: Tommy Nørnberg and Jutta Frank for peat coring; the divers, Jeremy Stewart

and Chris Hauser; Andriy Cherbukin and Paul Wilkinson for peat and sediment radiometric dating, respectively, also Andriy for XRF measurements; Gail Boila for sediment Hg analyses; Nicole Betrand for other sediment metals analyses. Hamed Sanei supplied the organic carbon data. The Manitoba Geological Survey (Tom Heine) kindly provided accommodation in Flin Flon, and Joel Nilsen provided HBMS' stack emission estimates and comments. We thank Jack Klaverkamp and Carolyn Ransom for their 1985 sediment samples, and Isabelle McMartin for her constructive critique of an earlier version of this paper. This paper is ESS Contribution no. 20100052. References Belzile N, Chen Y-W, Gunn JM, Dixit SS. Sediment trace metal profiles in lakes of Killarney Park, Canada: from regional to continental influences. Environ Pollut 2004;130:239–48. Benoit JM, Fitzgerald WF, Damman AWH. Historical atmospheric mercury deposition in the mid-continental U.S. as recorded in an ombrotrophic peat bog. In: Watras CJ, Huckabee JW, editors. Mercury pollution: integration and synthesis. Boca Raton: Lewis; 1994. p. 187–202. Biester H, Bindler R, Martinez-Cortizas A, Engstrom DR. Modeling the past atmospheric deposition of mercury using natural archives. Environ Sci Technol 2007;41: 4851–60. Bindler R, Klarqvist M, Klaminder J, Förster J. Does within bog spatial variability of mercury and lead constrain reconstructions of absolute atmospheric deposition rates from single peat records? The example of Store Mosse. Glob Biogeochem Cycle 2004;18:GB3020. doi:10.1029/2004GB002270. CCME. Canada-wide standards for mercury emissions. Canadian Council of Ministers of the Environment, endorsed by CCME Council of Ministers, June 2000, Quebec City. www.ccme.ca/assets/pdf/mercury_emis_std_e1.pdf accessed Mar 4, 2010. CCME. Canada-wide standards for mercury emissions (mercury emissions, mercurycontaining lamps and mercury for dental amalgam waste) — a report on progress. Canadian Council of Ministers of the Environment; 2005. June 2005, 41 pp. [www. ccme.ca/assets/pdf/joint_hg_progress_rpt_3.pdf; accessed Nov 2, 2009]. Cheburkin AK, Shotyk W. Energy-dispersive XRF spectrometer for Ti determination (TITAN). X-ray Spectrom 2005;34:69–72. Couillard Y, Cattaneo A, Gallon C, Courcelles M. Sources and chronology of fifteen elements in the sediments of lakes affected by metal deposition in a mining area. J Paleolimnol 2008;40:97-114. Engstrom DR, Swain EB. Recent declines in atmospheric mercury deposition in the Upper Midwest. Environ Sci Technol 1997;31:960–7. Environment Canada. Mercury and the environment – sources of mercury – Canadian releases. The Green Lane™ website. updated Feb 4 2004.[www.ec.gc.ca/mercury/ sm/en/sm-cr.cfm accessed Mar 22, 2010]. Farmer JG, Anderson P, Cloy JM, Graham MC, MacKenzie AB, Cook GT. Historical accumulation rates of mercury in four Scottish ombrotrophic peat bogs over the past 2000 years. Sci Total Environ 2009;407:5578–88. Franzin WG, McFarlane GA, Lutz A. Atmospheric fallout in the vicinity of a base metal smelter at Flin Flon, Manitoba, Canada. Environ Sci Technol 1979;13:1513–22. Givelet N, Le Roux G, Cheburkin AK, Chen B, Frank J, Goodsite ME. Suggested protocol for collecting, handling and preparing peat cores and peat samples for physical, chemical, mineralogical and isotopic analyses. J Environ Monit 2004;6:481–92. Harrison SE, Klaverkamp JF. Metal concentration in liver and muscle of northern pike (Esox lucius) and white sucker (Catostomus commersoni) and in sediments from lakes near the smelter at Flin Flon, Manitoba. Environ Toxicol Chem 1990;9: 941–56. Harrison SE, Dutton MD, Hunt RV, Klaverkamp JF, Lutz A, Macdonald WA, et al. Metal concentrations in fish and sediment from lakes near Flin Flon, Manitoba. Can Data Rep Fish Aquat Sci 1989;(747), 74 pp. Henderson PJ, McMartin I, Hall GE, Percival JB, Walker DA. The chemical and physical characteristics of heavy metals in humus and till in the vicinity of the base metal smelter at Flin Flon, Manitoba, Canada. Environ Geol 1998;34:39–58. Lamborg CH, Fitzgerald WF, Damman AWH, Benoit JM, Balcom PH, Engstrom DR. Modern and historic atmospheric mercury fluxes in both hemispheres: global and regional mercury cycling implications. Glob Biogeochem Cycle 2002;16:1104. doi: 10.1029/2001GB001847. Landers DH, Simonich SL, Jaffe DA, Geiser LH, Campbell DH, Schwindt AR, et al. The fate, transport and ecological impacts of airborne contaminants in western National Parks (USA). Report EPA/600/R-07/138. Corvallis: U.S. Environmental Protection Agency, Western Ecology Division; 2008. (on-line: www.epa.gov/uheerl/wacap). Lockhart WL, Wilkinson P, Billeck BN, Danell RA, Hunt RV, Brunskill GJ, et al. Fluxes of mercury to lake sediments in central and northern Canada inferred from dated sediment cores. Biogeochemistry 1998;40:163–73. McMartin I, Henderson PJ, Nielsen E, Campbell JE. Surficial geology, till and humus composition across the Shield Margin, north-central Manitoba and Saskatchewan. Geol Surv Can Open File; 1996. no.3277. McMartin I, Henderson PJ, Nielsen E. Impact of a base metal smelter on the geochemistry of soils of the Flin Flon region, Manitoba and Saskatchewan. Can J Earth Sci 1999;36: 141–60. Muir DCG, Wang X, Yang F, Nguyen N, Jackson TA, Evans MS, et al. Spatial trends and historical deposition of mercury in eastern and northern Canada inferred from lake sediment cores. Environ Sci Technol 2009;43:4802–9.

P.M. Outridge et al. / Science of the Total Environment 409 (2011) 548–563 Nilsen J. Status of the environment report. Hudson Bay Mining and Smelting Co. Ltd; 2003. Nriagu JO, Wong HKT, Lawson G, Daniel P. Saturation of ecosystems with toxic metals in Sudbury basin, Ontario, Canada. Sci Total Environ 1998;223:99-117. Oldfield F, Appleby PG. Empirical testing of Pb-210 dating models for lake sediments. In: Howarth EV, Lund JWG, editors. Lake sediments and environmental history. Minneapolis, USA: U. of Minnesota Press; 1984. Outridge PM, Stern GA, Hamilton PB, Percival JB, McNeely R, Lockhart WL. Trace element profiles in a varved Arctic lake sediment. Geochim Cosmochim Acta 2005;69:4881–94. Percival JB, Outridge PM. The stability of metal profiles in lake sediments near the Flin Flon smelter, Manitoba: evaluating changes over recent decades. Manuscript in preparation. Roos-Barraclough F, Givelet N, Martinez-Cortizas A, Goodsite ME, Biester H, Shotyk W. An analytical protocol for the determination of total mercury concentrations in solid peat samples. Sci Total Environ 2002;292:129–39. Roos-Barraclough F, Givelet N, Cherbukin AK, Shotyk W, Norton SA. Use of Br and Se in peat to reconstruct the natural and anthropogenic fluxes of atmospheric Hg: a 10 000-year record from Caribou Bog, Maine. Environ Sci Technol 2006;40:3188–94. Sageman BB, Lyons TW. Geochemistry of fine-grained sediments and sedimentary rocks. In: Holland HD, Turekian KK, editors. Treatise on geochemistry; sediments, diagenesis, and sedimentary rocks, Vol. 7; Mackenzie FT, Series Ed. Oxford: Elsevier-Pergamon; 2003. pp. 115–58. Sanei H, Outridge PM, Goodarzi F, Wang F, Armstrong D, Warren K, et al. Wet deposition mercury fluxes in the Canadian sub-Arctic and southern Alberta, measured using an automated precipitation collector adapted to cold regions. Atmos Environ 2010;44:1672–81.

563

Selin NE, Sunderland EM, Knightes CD, Mason RP. Sources of mercury exposure for U.S. seafood consumers: implications for policy. Environ Health Perspect 2010;118: 137–43. Shotyk W. Peat bog archives of atmospheric metal deposition: geochemical evaluation of peat profiles, natural variations in metal concentrations, and metal enrichment factors. Environ Rev 1996;4:149–83. Shotyk W, Weiss D, Kramers JD, Frei R, Cheburkin AK, Gloor M, et al. Geochemistry of the peat bog at Etang de la Gruère, Jura Mountains, Switzerland, and its record of atmospheric Pb and lithogenic trace elements (Sc, Ti, Y, Zr, Hf and REE) since 12,370 14C yr BP. Geochim Cosmochim Acta 2001;65:2337–60. Shotyk W, Goodsite ME, Roos-Baraclough F, Frei R, Heinmeier J, Asmund G, et al. Anthropogenic contributions to atmospheric Hg, Pb and As accumulation recorded by peat cores from southern Greenland and Denmark dated using the 14C “bomb pulse curve”. Geochim Cosmochim Acta 2003;67:3991–4011. Sim BL, Agterberg FP, Beaudry C. Determining the cutoff between background and relative base metal smelter contamination levels using multifractal methods. Comput Geosci 1999;25:1023–41. Tropea AE, Paterson AM, Keller W, Smol JP. Sudbury sediments revisited: evaluating limnological recovery in a multiple-stressor environment. Water Air Soil Pollut 2010;210:317–33. Weiss D, Shotyk W. Atmospheric lead deposition from 12,400 to ca. 2000 yrs BP in a peat bog profile, Jura Mountains, Switzerland. Water Air Soil Pollut 1997;100: 311–24. Zoltai SC. Distribution of base metals in peat near a smelter at Flin Flon, Manitoba. Water Air Soil Pollut 1988;37:217–28.