Mercury and Organic Matter Concentrations in Lake and Stream Sediments in relation to One Another and to Atmospheric Mercury Deposition and Climate Variations across Canada

This article focuses on analyzing the Geological Survey of Canada (GSC) data for total mercury concentrations (THg) in lake and stream sediments. The objective was to quantify how sediment THg varies by (i) sediment organic matter, determined by loss on ignition (LOI) at 500C, (ii) atmospheric Hg deposition (atm.Hgdep) as derived from the Global/Regional Atmospheric Heavy Metals Model GRAHM2005, and (iii) mean annual precipitation and mean monthly July and January temperatures (TJuly, TJan). Through regression analyses and averaging by National Topographic System tiles (NTS, 1:250,000 scale), it was found that 40, 70, and 80% of the sediment THg, LOI, and atm.Hgdep variations were, respectively, related to precipitation, TJuly , and TJan. In detail, lake sediment THg was related to atm.Hgdep and precipitation, while stream sediment THg was related to sediment LOI and TJuly . Plotting sediment THg versus sediment LOI revealed a curvilinear pattern, with highest Hg concentrations at intermediate LOI values. Analysing the resulting 10th and 90th log10THg percentiles within each 10% LOI class from 0 to 100% revealed that (i) atm.Hgdep contributed to the organic component of sediment THg and (ii) this was more pronounced for lakes than for streams.


Introduction
Mercury (Hg) concentrations in lake and stream sediments vary by many factors pertaining to geology, atmospheric Hg deposition, climate, vegetation, topography, and soil and sediment composition [2][3][4][5].Due to the continuing need to curb anthropogenic Hg emissions, it has become important to determine how these concentrations vary from location to location as influenced by these factors [6].In general, geogenic contributions to sediment THg dominate downslope from surface exposed Hg-containing mineral deposits (e.g., [7][8][9]).In contrast, atmospheric contributions accrue due to gradual Hg sequestration by vegetation on land and in water and through subsequent litter production and transport of organic matter by way of surface run-off and sedimentation [10][11][12][13].In detail, the extent of atmospheric Hg sequestration depends on (i) the rate of atmospheric Hg deposition [14], (ii) the extent of biological activities and vegetation cover (type, extent, and growth) on land and in water [15][16][17][18], (iii) the accumulation of sediment organic matter content [19][20][21], and (iv) regional to local climate variations as expressed by temperature, precipitation, and length of growing season [22,23].
Once incorporated into organic matter, reevaporation of Hg is limited by the extent of sunlit surface exposure and biological activity [24,25], with overall turnover times increasing from decades to thousands of years with increasing soil and sediment depth.This extended retention is due to strong Hg-S bonds in the form of organic sulfide groups and S-containing minerals, including HgS such as cinnabar [26][27][28].Due to decomposition and humification processes, this affinity also leads to a gradual increase of organically bound Hg concentrations in soils and sediments, as occurs with other elements such as S and N [29].
As changes in climate and human activities affect vegetation distributions and growth on land and in water, one would expect that these changes will not only affect the rates of atmospheric Hg emission and redeposition but would also affect the proportioning of organic versus mineral Hg in soils and sediments.Hence, there are attempts to discern this proportioning through watershed experimentation and isotopic Hg analyses (e.g., [30,31]).This article presents how the mineral versus organic proportioning of sediment Hg can be estimated through cross-referencing geospatial databases that inform about sediment composition, atmospheric Hg deposition (atm.Hg dep ), mean annual precipitation, and mean January and July temperatures ( Jan ,  July ), with Canada as a case study.
For this study, Canada-wide data layers were available pertaining to >250,000 sediment sampling points.The specific objectives were (i) to determine how the compiled values for lake and stream sediment THg and organic matter within these data layers relate to each other and also to atm.Hg dep , precipitation,  Jan and  July by way of regression analysis; (ii) to propose a simple model that can be used to determine the extent to which the organic matter contributions to sediment THg are affected by atm.Hg in particular and precipitation in general.
The working hypothesis was that atmospherically deposited Hg, as sequestered by vegetation, becomes part of soil organic matter through litter fall and biological processing including decay.The extent of this would generally decrease from the temperate forest regions along southeastern Canada and the Pacific coast to the snow-covered barrens and ice fields in the north and in alpine areas.This article supersedes earlier work on sediment THg distribution as documented by Rasmussen et al. [19] and later by Nasr et al. [3].That work focused on Canada-wide variations in geogenic Hg sources and how sediment THg varies topographically from uplands to lowlands, with sediment THg being higher in upland than lowland sediments, and generally decreasing with stream order and successive wetland and lake retention of upland-generated sediment contributions.The latter was revealed through case studies centered on the Selwyn Basin in the Yukon Territory, where surface exposure of black shales generates stream sediments with high THg concentrations while downstream sediment THg concentrations decrease with increasing upslope flow accumulation areas [3].Further case studies similar to this can be found in Nasr [32] for Nova Scotia, Quebec, and Bathurst Island (Nunavut).The advances to be described in this article refer to (i) using a nearly doubled database for sediment THg and sediment organic matter as determined by loss on ignition at 500 ∘ C (LOI) by including the sediment data for Quebec and Nova Scotia; (ii) relating sediment THg and LOI to the atmospheric deposition and climate variations across Canada; (iii) determining how the mineral to organic contributions to sediment vary as LOI increases from 0 to 100% within the context of increasing atmospheric Hg deposition loads.
Linear and nonlinear regression analyses were used to evaluate (i) the general trends regarding sediment THg, sediment LOI in reference to the Canada-wide variations in atmTHg dep versus precipitation,  Jan , and  July ; (ii) how the mineral and organic portions of sediment THg can, at least in part, be quantified in terms of the following model: where " = 1" and " = 2" refer to lakes and streams, "" refers to provinces, territories, and geological survey zones in Quebec, and   ,   , and   are the corresponding regression coefficients.With this model, it is assumed that (i) the mineral component of sediment THg is set to decrease from   to 0% as organic matter increases from 0 to 100%; (ii) the organic component of sediment THg is set to increase asymptotically as LOI increases from 0 towards   [1 − exp(−  LOI (%))] as an upper limit when LOI = 100%, that is, similar to THg versus LOI summarized by Munthe et al. [33]; essentially, organic matter contributions to sediment THg weaken somewhat with increasing organic matter because the weight ratio between THg and organic matter can be expected to decrease on account of (i) growth dilution, whereby the rate of Hg availability is constant but the rate of organic matter accumulation varies from being low to high, and (ii) increased loss of oxidized Hg from water-saturated organic soils and sediments due to enhanced rates of bio-chemical Hg(0) evasion [34][35][36]; parallel to this is the decreasing THg to dissolved organic matter concentration ratio as dissolved organic matter concentrations in stream and lake waters increase from low to high [37]; (iii) as a result, the sediment THg versus LOI profile should reach a maximum value at intermediate LOI values as determined by the relative strength of the mineral versus the organic sediment THg contributions; (iv) the   coefficient by which the organic component of Hg increases to an asymptotic value with increasing LOI is, as a simplification, set not to vary in principle from location to location, but may differ from streams to lakes.
The resulting best-fitted linear and nonlinear regression models are subsequently used to model and map lake and stream sediment THg and LOI, with Canada-wide rasters for atmTHg dep , precipitation,  Jan , and  July as predictor variables.

Data Compilation.
Bulk lake and stream sediment concentration data for THg and LOI were obtained from the open geochemical survey files of Natural Resources Canada, Government of Quebec, and Province of Nova Scotia [38].The procedures used for generating these data followed a standard protocol for sampling, sample preparation, and laboratory analysis [39].Lake sampling involved retrieving 30 cm deep sediment cores, with muddy tops removed.Stream samples were collected from active channels.Samples were collected and analyzed from region to region from 1960 onward to 2008.The files contained the following information: (i) the geographic location: longitude, latitude, National Topographic System (NTS) map tile number of each lake and stream sampling location; (ii) water and sediment characteristics of the sampled streams and lakes, notably lake area and depth, stream channel width, and depth, stream order and flow rate, water, and sediment colour; (iii) terrain type and landform; (iv) elemental composition, approximately 36 elements including heavy metals, notably Hg, Cu, Zn, Pb, Cd, As, and Au and other elements such as Fe, Mn, Se, and S (however, Se and S data were sparse); (v) LOI at 500 ∘ C [26].
The open files were compiled as part of an ArcMap project, with 235,943 sampling points containing data for both sediment THg and LOI.For the cross-referencing purposes, the THg and LOI entries were supplemented with point-extracted values from the following Canada-wide data layers [32]: (i) Bedrock geology polygon shapefile, including rock type, age, formation, and faults [40,41].
The resulting raster-extracted shapefile for the Hg survey points was exported as a text file for further processing, to enable data quality control, statistical analyses, and plotting, using Excel, Statview, and ModelMaker software.Quality assurance involved inspecting the compiled data and correcting for (i) faulty data alignments within and across individual files; (ii) numerical and typographical inconsistencies within each column; (iii) identifying, eliminating, and/or correcting data entries with typographical error, misspellings, order of magnitude outliers, and faulty locations: the GSC referenced longitude and latitude locations had to fall along already mapped or DEM-derived water courses and open surface water features (lakes).

Statistical Procedures.
The statistical analyses were done by province and territory (all of Canada) and by Quebec survey zone (Figure 12) to produce (i) basic summary tables, by provinces/territories and Quebec survey zones and by NTS tiles; (ii) scatterplots of log 10 THg versus LOI by lakes and streams; (iii) log 10 THg and LOI frequency distributions; (iv) the best-fitted linear regression results for GRAHM-2005-generated atm.Hg dep , sediment THg and sediment LOI using the PRISM rasters for precipitation,  Jan and  July as predictor variables ((3) to (6)); (v) 10th, 25th, 50th, 75th, and 90th percentile summaries of log 10 THg for each 10% LOI class from 0 to 100%, all split by provinces/territories and by geological survey zones in Quebec; (vi) the best-fitted nonlinear regression results for the   ,   , and   coefficients of (1) based on the plots of the 10th and 90th percentiles of log 10 THg per each 10% LOI class; (vii) the best-fitted trends of   versus atm.Hg dep and precipitation.
The regression analysis results were reported by compiling the least-squares fitted intercepts (if applicable), the regression coefficients, their standard errors of estimates, and corresponding  and  values.Variables with significant influence were selected as part of the stepwise procedure when (i) -values were >4.0, (ii) -values < 0.001, (iii) THg and LOI values remained positive, (iv) geographic location variables improved the best-fitted results significantly, and (v) choosing the most significant variable among partially correlated variables.Progress towards the best-fitted results was monitored by examining the actual versus best-fitted scatterplots at each step for the purpose of increasing the linearity between actual and best-fitted values while obtaining even (heteroscedastic) scatterplot distributions.Doing this included variable transformations, for example, THg to log 10 THg, precipitation to precipitation 0.5 , LOI to log 10 LOI, and redoing the analyses.

Mapping Procedures.
The best-fitted linear regression models were used to map lake and stream sediment THg and LOI across Canada, using the Canada-wide rasters for atm.Hg dep , precipitation,  Jan , and  July as predictive layers.This also included mapping the gain in sediment THg with increasing LOI using the following formulation: with "" referring to NTS-averaged values for atm.Hg dep or precipitation.

Results and Discussion
3.1.Sediment THg and LOI Data Presentation.The overview in Figure 2 identifies the extent and variations of sediment THg across Canada, by lake and stream sampling locations, which most frequently but not exclusively focused on western and eastern Canada, respectively.High to very high THg concentrations are generally associated with bedrock formations with mineral Hg exposures, such that they occur within the Selwyn Basin of the Yukon Territory [47] and in combination with sulfide mineral exposures due tectonic uplift (e.g., British Columbia [48][49][50]; Nova Scotia [51]; Quebec [52,53]), rift (Labrador Trough, [54]), mafic volcanic extrusions (Quebec [55]; Labrador [56]), and meteor impacts (Sudbury, Ontario [57]).Some of the THg anomalies are associated with sedimentary formations, especially where these are overlaying intruding or extruding formation or  2 and 3, respectively.where they are enriched with organic matter as in black shales or coal seams.Other anomalies refer to the accumulation of sediment Hg due to past and current mining activities and gold extraction activities in particular (see, e.g., [58]).The dependence of sediment log 10 THg versus sediment LOI followed curvilinear patterns (Figures 3 and 4), as to be quantified by way of (1) and as discussed by Rasmussen et al. [19,20].These plots are, however, generally less curved for lakes than for streams.In addition, some of stream plots have high THg values at low LOI and notably so for British Columbia, Yukon Territory, and some of the Quebec survey zones with high surface exposures of Hg-containing minerals.For the Grenville zones (zones 23 to 26, Figure 12) there is a clear separation of high to low stream sediment THg due to regional differences between high and low upslope Hg exposures.
Tables 1 and 2 summarize sediment THg and LOI by provinces/territories and by Quebec survey zones.These entries indicate that, on average, lowest THg values occur in Nunavut and Labrador streams.Also, on average, lake sediment THg is highest for southwestern Nova Scotia, while stream sediment THg is highest for (i) the Temiscamingue survey zone (Grenville zone 23) in Quebec, mainly due to exposure of mafic volcanic bedrock exposures and (ii) across the black shale formations of the Selwyn Basin within the Yukon Territory (Figure 2).

3.2.
Relationships between THg, LOI, atm.Hg  , Precipitation, T  , and T  .The linear regression analyses pertaining to the NTS-tile averaged values for atmospheric Hg deposition as well as sediment Hg and sediment LOI all revealed a strong dependence on the Canada-wide climate variations, as quantified by the following best-fitted regression equations, with fairly evenly distributed actual versus best-fitted and fairly evenly distributed scatterplots in Figure 6: In these equations, the numbers in brackets refer to the best-fitted least-squares regression coefficients and their ± standard estimates of error. 2 refers to the coefficient of determination.In detail, (3) quantifies how the GRAHM2005 results for atm.Hg dep are related to the Canada-wide rasters for precipitation,  July , and  Jan .In principle, net retention of atmospheric Hg is therefore facilitated by vegetationbased Hg uptake and sequestration, which increases from cold to warm and from dry to wet regions.In contrast, increasing January temperatures lower Hg retention likely through increased Hg volatility year-round [4].Note that the (3) implied dependency of atm.Hg dep on mean annual precipitation is not linear, likely due to increasing dilution with increasing precipitation.Along the high precipitation region along the Pacific coast, this dilution becomes even stronger, as captured by the introduction of the locator variable for the NTS tiles along the Pacific Rim (coded 1 when applicable and 0 otherwise).In addition, (3) introduces an adjustment for the increased presence and net retention of atmospheric Hg along arctic costs (coded 1 when applicable and 0 otherwise) due to oceanic Hg upwelling [2].Generally, stream LOI is lower than lake LOI, with the latter covering the 0 to 100% range but peaking at about 35%.Furthermore, stream LOI is most frequent below LOI ≈ 20%.In contrast, lake and stream sediment THg values are most frequent at about THg ≈ 100 ng g −1 , with stream THg somewhat more variable than lake THg (Figure 5).Equation ( 4) indicates that log 10 LOI for lake and stream sediments both increase with increasing precipitation and  July and decreasing  Jan .This is quite similar to the atmTH dep pattern, likely due to the parallel dependence on terrestrial and aquatic organic matter production (which generally increases with precipitation and  July ) and organic matter retention (which generally increases with decreasing  Jan [59,60]).Also note that the stream/lake variable (coded 1 for streams and 0 for lakes) sets LOI (streams) = 0.33 LOI (lakes); that is, the organic matter concentrations in stream sediments are, on average, two-thirds lower than in lake sediments.
Equations ( 5) and ( 6) specify that the dependency of sediment THg on atm.Hg dep differs from streams to lakes, with stream THg variations most strongly related to variations in LOI, while lake THg varies more directly with variations in precipitation and atmTHg dep .Direct relationships between increasing atmospheric Hg deposition and increasing sediment THg have been noted repeatedly for Figure 5: Frequency diagram for the compiled THg [log 10 (ng g −1 )] and LOI (%) data across Canada.lakes by, for example, (i) Muir et al. [23] across northern Quebec, (ii) Munthe et al. [33] across Scandinavia, and (iii) Hissler and Probst [21] downwind from industrial Hg emissions in France.Equations ( 5) and ( 6) also predict that decreases in atm.Hg dep due to reductions in Hg emissions should eventually lead to reductions in sediment THg, as documented by Kamman and Engstrom [61].Such reductions would be more readily observed for lake than for stream sediments.
Using (4), ( 5) and ( 6) produced the maps for sediment Hg and LOI in Figure 7. Overlaying the mean GSC data values for log 10 THg and log 10 LOI values per NTS-tile (dots) on these maps underscores the general conformance between the mapped and NTS-tile averaged log 10 THg values.This is further demonstrated in Figure 8 by way of cumulative frequency plots regarding the absolute differences between the mapped and NTS-tile averaged log 10 THg and log 10 LOI values.The extent of lake versus stream map-to-dot conformances is approximately the same, with 80% of the model projections falling within the log 10 Hg and log 10 LOI residual range of −0.2 to 0.2.
In principle, (3) implies that sustained precipitation increases by 0.1 m at 1 m per year would, on average, add 2.68 g m −2 to atm.Hg dep .Similarly, the temperature entries  5).
in (3) imply that an increase of 1 ∘ C in July temperature would, on average, increase atm.Hg dep by 0.80 g m −2 a −1 .A parallel increase of 1 ∘ C in January air temperatures, however, would compensate for some of that increase by 0.25 g m −2 a −1 , thereby resulting in a mean increase in atm.Hg dep of 0.55 ± 0.09 SD g m −2 .According to Barrow et al. [62], summer and winter temperatures will increase by about 5 to 6 ∘ C at, for example, Resolute Bay (southwest corner of Cornwallis Island; southeast of Bathurst Island, NU) by 2050.Based on (3), this translates into a net gain in atm.Hg dep rate of about 3.0 g m −2 a −1 .In turn, (5) implies that lake sediment THg would then increase as well by almost a factor of 10.Further south at Norman Wells (along the Mackenzie river, west of Great Bear Lake, NWT), the effect on atm.Hg dep would be somewhat smaller, with summer and winter temperatures expected to increase by about 2 to 3 ∘ C while the growing season would increase by about a month [62].
As indicated by the  2 values, the climate and regional predictor variables of (3) to ( 6) account for 80% of the atm.Hg dep variations, followed by 70% of the log 10 LOI

Analyzing and Interpreting the 10th and 90th
Percentiles of the THg versus LOI Scatter Plots.Determining and plotting the 10th and 90th percentiles of the sediment log 10 THg variations within each of the 10% LOI classes from 0 to 100% produced dots in Figure 9 for each of the provinces/territories and Quebec survey zones.These dots follow fairly consistent pattern, being highest at intermediate LOI values as implied by (1).
Curve-fitting the dots in Figure 9 with (1) produced the best-fitted lines and   ,   , and   coefficients in Table 4. Recognizing the similarity of these lines across the provinces/territories and Quebec survey zones led to a further best-fitted simplification by setting   (90th percentile) = 0.668 +   (10th percentile).This implies that, on average, THg (90th percentile) = 4.87 THg (10th percentile) when LOI = 0%.

Sensitivity of the 𝑎 𝑖𝑗 and 𝑏 𝑖𝑗 to Atmospheric Deposition.
Regressing the   values against the mean atm.Hg dep and precipitation values per NTS tile (Table 3) produces no significant trends.In contrast, the   coefficients in Table 4 Table 3: Mean annual precipitation, mean atmospheric deposition, and best-fitted   ,   , and   values (1) for the 10th and 90th sediment THg percentiles (lakes, streams), by provinces/territories and by Quebec survey zones (Figure 12).2)-( 4) projected map values in Figure 7.
are linearly related to atm.Hg dep and precipitation (Figure 10, Table 4), with  2 values near 0.50 for lakes, and less so for streams with  2 values at 12 to 23%.Hence, the organic contributions to sediment THg vary from being low to high as mean annual amounts for atm.Hg dep and precipitation vary from low to high.With the mineral and organic matter contributions to sediment THg quantified by way of the best-fitted   ,   , and   coefficients in Table 4 and (1), one can now estimate the organically induced gain of sediment THg from low to high atmospheric deposition environments via (2) as per Figure 10 and Table 1.For lake and stream sediments at LOI = 100%, the estimated gain amounts to a factor of 4.4 to 6.0 as mean annual precipitation increases from near zero to 1.4 m a −1 .Similarly, the estimated gain for lake sediments increases even further to a factor of about 7.8 to 10.5 when based on increasing atm.Hg dep from near zero to 26 g m −2 a −1 but remains low for streams at about 2.9 to 4.5 (Table 1).

Mapping the Atmospherically Induced Sediment THg
Gains across Canada.Using (4), (5), and (6) to calculate the organically induced gains for sediment THg via (2) produced the maps in Figure 11.These maps suggest lake sediments are particularly prone to atm.Hg dep -related THg acquisition across southeastern Canada and along the Pacific Rim.From there, this susceptibility is mapped to decrease towards the alpine areas and the boreal to arctic zones in the north.
The overlaid dots on the maps in Figure 11(a) are also generated from (2) but represent the sediment THg gains using NTS-averaged atm.THg dep (GRAHM 2005 raster) and GSC-surveyed LOI (%) values as gain predictors.These dotto-map patterns suggest a general conformance between the mapped and dotted THg gains.However, major differences occur as well: (i) Across northern Alberta, Saskatchewan, and Manitoba, lakes with high sediment organic matter content appear to be particularly sensitive to increasing atm.Hg dep .The more sensitive lakes occur in flat terrain (Figure 11(b); for details regarding the flat-area shading, see [3]).
(ii) In areas with rugged terrain, frequent streambed scouring leads to high mineral and low organic matter inputs into lake and stream sediments.This, in turn, would lower the atm.Hg dep induced gain of the organic matter contributions to sediment THg.(iv) Downstream from upland areas where sediment Hg would accumulate on account of (a) significant Hgcontaining mineral exposures and/or (b) air-or waterborne Hg emissions due to industrial activities.

Concluding Remarks
The linear regression results ( (3) to ( 6)) revealed a strong interdependence between the NTS-tile averaged values for sediment THg, sediment LOI, atm.Hg dep , and climate across  Figure 9: Plots of actual (dots) versus best-fitted 10th and 90th percentiles using (1) (lines) for lake and stream sediment log 10 THg (ng g −1 , log 10 scale) versus 0 < LOI < 100%, by provinces/territories (a) and Quebec survey zones ((b) Figure 12).Canada.This being so, the results suggest that lake and stream sediments become more enriched with Hg (i) as summers and winters become warmer, (ii) growing seasons become longer, and (iii) mean annual precipitation rates increase.The best-fitted nonlinear regression results for the percentile distributions of log 10 THg versus sediment LOI by survey zones are generally consistent with the following observations and suggestions: (i) Atmospheric deposition to lakes is direct whereas indirect to streams.While some of the lake-deposited Hg volatilizes, some of it is sequestered by dissolved and particulate organic matter (DOM and POM, resp.) and by aquatic organisms.In addition, some of the DOM flocculates thereby contributing further to the organically sequestered Hg portion within the sediments [64].  5.
(ii) Lake catchments are generally larger than stream catchments.Therefore, lake sediment THg is more reflective of area-wide atmospheric Hg deposition than stream sediment THg.
(iii) Stream sediments are generally closer to local upslope Hg sources than lake sediments and are therefore less diluted.
(iv) Stream sediments are subject to frequent relocation and scouring events, with sediments varying from being coarse to fine [65][66][67].In contrast, lake sedimentation is steady, cumulative, fine textured, and organically enriched [68].In addition, lake sediments remain largely undisturbed [69].(v) Stream sediments may lose some of the sediment THg by way of dissolved and particulate matter transport, with the finer particles generally carrying more Hg over larger distances than larger particles [65,67,70,71].(vi) Once entering lakes, finer particles [72] and some of the dissolved but organically bound Hg settles, especially after coagulation [73].
Note that all of the above results pertain to bulked lakes and stream sediment cores with no reference to sediment  depth other than lake samples being 30 cm deep.In detail and in contrast to LOI, THg changes strongly with increasing sediment depth, which is generally interpreted to result from (i) changes in the historical pattern of atmospheric Hg emission and deposition from preindustrial to modern time and (ii) changes in sedimentation processes [20,33,74].As such, the best-fitted models ((4), (5), and ( 6)) pertain to longterm trends only.Hence, these changes could be indicative of how current to future changes in climate affect atmospheric Hg deposition, as summarized by the International Joint Commission, Canada and United States [75].If so, all of this will affect not only THg and LOI, but also other correlated variables such as the bioavailability of methyl Hg and its trophic uptake by fish [20,74].
For detailed point-by-point examinations, one would need to increase the resolution of the analysis to address local Hg releasing or retaining situations as these vary upslope from each sediment sampling point.Such details would address the local variations in atmospheric Hg deposition and subsequent Hg transfer to streams and lakes.For example, (i) the dispersal of Hg contained in emission plumes depends on downwind air-flow pattern as affected by terrain and vegetation cover [76,77], (ii) the transfer of POM-bound Hg would increase with increasing slope and soil erosion, and (iii) the upslope transfer of DOM-bound Hg via run-off and stream flow would increase while POM-and mineral-bound Hg would decrease with increasing extent % of upslope wet area and wetland cover [3].While there could be differences due to region-specific biases in sediment sampling, preparation, and analysis, most of that has been addressed by employing the GSC standardized sediment surveying protocol, as described by Friske and Hornbrook [39].In fact, the above analyses do not reveal regional biases.Instead, there is a transregional conformance pattern between the GSC data for sediment THg and LOI and the best-fitted results, even though the GSC data were compiled from three distinct databases (Quebec, Nova Scotia, and rest of Canada) spanning several decades.

4 TFigure 1 :
Figure 1: Canada-wide projections for atmospheric Hg deposition, mean annual precipitation, and mean annual July and January temperatures, overlaid by the 1 : 250,000 National Topographic System tile pattern.

Figure 2 :
Figure 2: Point-by-point THg concentrations (ng g −1 , log 10 scale) in lake and stream sediments across Canada, as found in the open files of the Geological Survey of Canada, Quebec, and Nova Scotia.For data summary by provinces/territories and by survey zones in Quebec, see Tables2 and 3, respectively.

(
iii) Downstream from open nonforested areas much of the atmospherically deposited Hg would revolatilize.This would be the case in southeastern Quebec landscape where open field conditions dominate.

Figure 11 :
Figure 11: (a) Projected atmospherically derived gains (2) of sediment THg mapped from low (green) to high (red), based on (4), (5), and (6) estimates for LOI (%) and annual net atmospheric Hg deposition as predictor variables.Overlaid dots: the corresponding log 10 THg gain estimates per NTS tile, using (2), (5), and (6) in combination with the mean GSC-surveyed LOI values.(b) Close-ups for the Hg gains in lake sediments across Manitoba to Alberta (top) and in stream sediments across southwestern Quebec (bottom), overlaid on a hill-shaded digital elevation grid also showing the extent of low-lying and generally wet areas next to lakes and streams (shaded green).

Table 1 :
Mean sediment THg and LOI for streams and lake, by provinces and territories.
Calculated weighted mean and standard deviation (Std.Dev.) for locations with both THg and LOI values.

Table 2 :
Mean sediment THg and LOI for streams and lakes, by Quebec survey zones (Figure12).
(27)ulated weighted mean and standard deviation (Std.Dev.) for locations with both THg and LOI values.Opatica(27)represents the survey zone across the Opatica, Opinaca, and La Grande geological subprovinces.

Table 4 :
Trend analysis for the organic matter contributions to lake and stream sediment THg (10th and 90th percentiles) from low to high mean annual precipitation rate and atmospheric Hg deposition rates.