Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition

Atmospheric Environment xxx (2016) 1e10

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition H. Cathcart a, *, J. Aherne a, D.S. Jeffries b, K.A. Scott c a

Environmental and Resource Studies, Trent University, 1600 West Bank Drive, Peterborough, Ontario, K9J 7B8, Canada Environment Canada, National Water Research Institute, Burlington, Ontario, Canada c Saskatchewan Ministry of Environment, Regina, Saskatchewan, Canada b

h i g h l i g h t s
Regression kriging was used to map critical loads of acidity for ~90,000 lakes in northern Saskatchewan.
A region greater than 13,500 km2 was predicted to be very sensitive to acidic deposition.
Predicted critical loads are low, with 5938 lakes <5 meq m2 yr1 and 23, 043 lakes <10 meq m2 yr1.
An estimated 12% of lakes are in exceedance of their critical loads under 2006 modelled sulphur deposition.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 29 February 2016 Received in revised form 15 August 2016 Accepted 16 August 2016 Available online xxx

Atmospheric emissions of sulphur dioxide (SO2) from large point sources are the primary concern for acidic deposition in western Canada, particularly in the Athabasca Oil Sands Region (AOSR) where prevailing winds may potentially carry SO2 over acid-sensitive lakes in northern Saskatchewan. A novel catchment-scale regression kriging approach was used to assess regional sensitivity and critical loads of acidity for the total lake population of northern Saskatchewan (89,947 lakes). Lake catchments were delineated using Thiessen polygons, and surface water chemistry was predicted for sensitivity indicators (calcium, pH, alkalinity, and acid neutralizing capacity). Critical loads were calculated with the steady state water chemistry model using regression-kriged base cations, sulphate, and dissolved organic carbon concentrations modelled from surface water observations (n > 800) and digital landscape-scale characteristics, e.g., climate, soil, vegetation, landcover, and geology maps. A large region (>13,726 km2) of two or more indicators of acid sensitivity (pH < 6 and acid neutralizing capacity, alkalinity, calcium < 50 meq L1) and low critical loads < 5 meq m2 yr1 were predicted on the Athabasca Basin. Exceedance of critical loads under 2006 modelled total sulphate deposition was predicted for 12% of the lakes (covering an area of 3742 km2), primarily located on the Athabasca Basin, within 100 km of the AOSR. There have been conflicting scientific reports of impacts from atmospheric emissions from the AOSR; the results of this study suggest that catchments in the Athabasca Basin within 100 km of the AOSR have received acidic deposition in excess of their critical loads and many of them may be at risk of ecosystem damage owing to their sensitivity. © 2016 Published by Elsevier Ltd.

Keywords: Acid rain Sulphur dioxide Critical loads Regression-kriging Steady state water chemistry (SSWC) model Athabasca Oil Sands

1. Introduction Anthropogenic emissions of acidifying compounds such as sulphur dioxide (SO2) have contributed to an acid rain problem in Canada for decades, causing biological damage to aquatic life and terrestrial ecosystems in eastern Canada (Schindler, 1988; Neary

* Corresponding author. E-mail address: [email protected] (H. Cathcart).

et al., 1990; Jeffries et al., 2003). As a result, SO2 emissions in the east have declined steadily since 1985 owing to policy changes; in contrast, by 2006 emissions in the western provinces (British Columbia, Alberta, Saskatchewan, and Manitoba) overtook eastern Canada, driven primarily by emissions from large point sources such as smelters and mines (Environment Canada, 2014). The largest point source of SO2 and nitrogen oxides (NO) in Canada, the Athabasca Oil Sands Region (AOSR), is located in northern Alberta within 50 km of the Saskatchewan border.

http://dx.doi.org/10.1016/j.atmosenv.2016.08.048 1352-2310/© 2016 Published by Elsevier Ltd.

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

2

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

Average AOSR SO2 emissions were approximately 116,000 kT annually between 2002 and 2012 (Environment Canada, 2014) in a region with otherwise little background atmospheric SO2. Prevailing winds are from the west, and the emissions footprint from the AOSR has been shown to extend into the north, and eastward into Saskatchewan (Hazewinkel et al., 2008; Kurek et al., 2013). Downwind of the AOSR in northern Saskatchewan lie more than 90,000 lakes, rivers, and other bodies of water, many of which sit on extremely acid-sensitive granitic bedrock overlain with coarse, sandy soils. Historic surveys have broadly mapped this region as acid-sensitive or highly acid-sensitive (e.g., Liaw and Atton, 1981; Shewchuk, 1982; Saffran and Trew, 1996) based on indicator surface water chemistry such as pH, calcium (Ca2þ), alkalinity (ALK), and acid neutralizing capacity (ANC). Large-scale regional surveys since 2006 have reinforced the suggestion that these lakes are predominately acid-sensitive; 60% of 259 lakes in northern Saskatchewan were considered sensitive with an ANC below 200 meq L1 (Scott et al., 2010). Furthermore, lakes receiving sulphur (S) deposition in excess of their critical loads of acidity (defined as the “quantitative estimate of exposure to acidic (S and nitrogen [N]) deposition below which significant harmful effects on specified sensitive elements of the environment do not occur according to present knowledge” (after Nilsson and Grennfelt, 1988)), have been found across northern Saskatchewan, but exceedances, i.e., lakes receiving acidic deposition in excess of their critical load, were greatest closer to the Alberta-Saskatchewan border (Jeffries et al., 2010). The growing concern of deleterious ecosystem impacts from AOSR emissions (e.g., Environment Canada, 2011) has sparked regional investigations into the acid status of lakes. While these regional surveys provide information on the acid status of lakes in northern Saskatchewan, they represent <1% of an estimated 90,000 lakes, and leave large geographic gaps. In the current study, a geospatial approach was proposed to assess the total lake population owing to the large number of lakes in the region, and because the large range in reported lake sensitivity suggested an influence from landscape features (Jeffries et al., 2010). Geospatial techniques for assessing acid sensitivity are not new and have been popular in regional soil assessments but specialized spatial regression techniques such as regression kriging have not yet been applied to surface waters or critical loads assessments. A common approach is to use multiple linear regression models to estimate water chemistry from observed data (e.g. Berg et al., 2005; Sullivan et al., 2007; Wolniewicz et al., 2011; Povak et al., 2014); however, regression kriging has the added benefit of incorporating the stochastic component of the regression as well as modelling spatial autocorrelation, performing better than regression or spatial interpolation alone (Hengl et al., 2007). With an abundance of geochemical and landscape attribute data for northern Saskatchewan available in databases and digital maps, regression kriging is an intriguing new tool for predicting surface water chemistry at a regional scale. The goal of this study was threefold; firstly, to determine if catchmentscale regression kriging was a viable approach to developing critical loads of acidity for ~90,000 lakes in northern Saskatchewan; and secondly, to estimate critical loads of acidity and exceedance for the total lake population in northern Saskatchewan under total S deposition during 2006. Lastly, the overarching objective of the paper was to determine if acid sensitive lakes in northern Saskatchewan are at risk from acidic deposition. 2. Material and methods 2.1. Study area The defining feature of northern Saskatchewan (for the

purposes of this study, an area of approximately 280,000 km2 above 54 latitude) is its multitude of small lakes; more than 200,000 water bodies dominate the landscape (Government of Canada, 2007). Most of northern Saskatchewan sits on the Precambrian Shield, an acid-sensitive bedrock structure. Within the Shield is the Athabasca Basin (see Fig. 1), which is comprised of Athabasca sandstone (Ramaekers, 1990). Note that portions of the region located off the Precambrian Shield were excluded from the study area due to underrepresentation in the surface water data set and to bring focus to the acid-sensitive Precambrian Shield (see Fig. 1). The boreal forest covers the majority of northern Saskatchewan and forest stands are comprised primarily of jack pine (Pinus banksiana), black spruce (Picea mariana), tamarack (Larix laricina) and trembling aspen (Populus tremuloides) with a ground cover of lichens (Wiken, 1986). Fens and bogs occupied by dense sphagnum moss and black spruce are found in poorly drained areas throughout northern Saskatchewan. While exposed bedrock is present in all regions on the Precambrian Shield, it is more prominent in the Taiga Shield (see Fig. 1), where it is accompanied by thin soils and frequent fens and bogs (Wiken, 1986). Soils are predominately sandy throughout and in particular on the Athabasca Basin (Fig. 1), which features exposed sand dunes; in contrast, the Boreal Plain is covered with loamy to clayey glacial till (Wiken, 1986). The AOSR is located approximately 50 km west of the Saskatchewan border at approximately 57 latitude with eastward prevailing winds. Northern Saskatchewan is a remote region with few observations of atmospheric deposition; the only monitoring station in the study region, located near Cree Lake, was discontinued during 1992. The most recent comprehensive acidic deposition estimates are from Environment Canada's A Unified Regional Air Quality Modelling System (AURAMS, Moran et al., 1998) and were modelled using 2006 emissions inventory data. Sulphur dioxide emissions did not change substantially between 2006 and 2013 in the AOSR or Saskatchewan (Fig. 2). However, the downturn in the oil and gas industry beginning in 2014 resulted in a large drop (38% from 2013) in AOSR SO2 emissions. While the future of oil sands production is uncertain, the 2006 data are representative of emissions during the greater part of the last decade, and generally illustrate the regional and provincial spatial trends from 2006 to 2013. Modelled total sulphate (SO2 4 ) deposition estimated by AURAMS ranged from 0.1 to 5 kg ha1 in northern Saskatchewan during 2006, except where elevated near large point sources in the AOSR and Flin Flon, Manitoba, where extreme values between 60 and 90 kg ha1 were predicted (see Supporting Information Fig. SI.1). The deposition footprint of the Hudson's Bay smelter in Flin Flon was evident within the 2006 dataset (Fig. SI.1) and at the time was the greatest point source of emissions in Saskatchewan, but by 2010 its production was greatly scaled back. 2.2. Catchment delineation A single-lake catchment scale was chosen to model water chemistry and determine critical loads of acidity for the total lake population of northern Saskatchewan. Small lakes (below 0.02 km2), large lakes (above 40 km2) and water bodies considered impermanent or explicitly defined as non-lakes in the National Hydro Network (NHN) database (Government of Canada, 2007) were considered unrepresentative of the general lake population and removed from the dataset. Traditional catchment delineation typically involves using a Digital Elevation Model (DEM) to compute flow direction and outline the hydrologically contributing area. The low relief of the terrain and coarse resolution of the DEM available for northern Saskatchewan, as well as the large number of lakes, made such

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

3

Fig. 1. Saskatchewan and Alberta with the study area in northern Saskatchewan outlined. Major geologic features (Precambrian Shield, Athabasca Basin, Western Canada Sedimentary Basin) are illustrated in crosshatching and ecozones (Boreal Plain, Boreal Shield, Taiga Shield) are shaded. An encircled 50 km radius north of Fort McMurray represents the Athabasca Oil Sands Region (AOSR).

shores of neighbouring lakes. Several catchments (<20) which were known to be large rivers were manually excluded on visual inspection; otherwise, catchments were presumed to be the hydrologically contributing areas of single lakes. 2.3. Regression models

Fig. 2. Sulphur dioxide (SO2) emissions during 2006e2014 for Saskatchewan and the Oil Sands industry in Alberta (Environment Canada, 2014).

methods impractical; an alternative method was implemented using Thiessen polygons to delineate single-lake catchments for the total lake population. While Thiessen polygons are conventionally used to delineate areas contributing to rain gauge stations, they have been used in automated catchment delineation processes with the advantage of lower computational requirements and data inputs (i.e., Johnston et al., 2009). Lake polygon lines were converted to vertices to generate Thiessen polygons, which were then dissolved together by their seed lake to produce catchments with borders representing the line halfway between its shores and the

A database containing 75 landscape characteristics (see Supporting Information, Table SI.1) that may influence lake chemistry was derived from digital map layers (Table 1). Mean (climate, EVI, NDVI, DEM, CAO, atmospheric deposition) or percent cover (soil, land cover, geology) summaries were calculated for each catchment. Lake observations were available from >800 lakes in surveys carried out from 2007 to 2008 (Scott et al., 2010) and 2007e2009 (Jeffries et al., 2010). In addition, pH data were available for 8688 lakes from eight surveys carried out by the National Geochemical Reconnaissance between 1981 and 1994 (Table 1). Water chemistry observations were averaged by lake if more than one year of sampling was present. Database geoprocessing was carried out using QGIS v2.0 (QGIS, 2014) and GRASS v.6.4 (GRASS, 2012). Multiple linear regression models were produced for dissolved organic carbon (DOC: 845 lakes), base cations (BC: 818 lakes) and SO2 (846 lakes) for inclusion in critical load calculations (as 4 described below). Models were also produced for four common chemical indicators of acid sensitivity: ANC (842 lakes), ALK (810 lakes), Ca2þ (846 lakes) and pH (9534 lakes). . Logistic transformation was used for all dependent variables to ensure unbiased back-transformation of predicted values after regression kriging (Hengl et al., 2007). Multicollinearity was assessed using an automated function that calculated the Variance Inflation Factor (VIF)

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

4

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

Table 1 Source and resolution of the digital map layers used to derive catchment characteristics; their derived predictor variables used in the regression analysis are outlined in detail in the Supporting Information (Table SI.1). Layer

Resolution

Source

Lake polygons Soil polygons Calcium oxide (CAO)

1:250,000 1:1,000,000 500 m

National Hydro Network (Government of Canada, 2007) Soil Landscapes of Canada (The Canadian System of Soil Classification, 1995) Interpolated using ordinary kriging from lithogeochemical surveys (Geological Atlas of Saskatchewan, 2012a; Geological Atlas of Saskatchewan, 2012b; Campbell, 2002; Campbell, 2005; Card et al., 2011; Card and Bosman, 2012a; Card and Bosman, 2012b; McMartin et al., 2007; McMartin et al. 2008). Annual climate variables (ANUCLIM) and bioclimatic variables (BIOCLIM) for 1971e2000 (McKenney et al., 2006) Global Land Cover Characterization Database (United States Geological Survey, 1998)

Climate 10 km Categorical land 1 km cover Vegetation indices 500 m Geology

1:250,000

Enhanced Vegetation Index (EVI) and Normalized Difference Vegetation Index (NDVI) from the Moderate Resolution Imaging Spectroradiometer (United States Geological Survey, 2000) Geological Atlas of Saskatchewan (2013); reclassed into five categories (agrillaceous, siliceous, felsic, mafic, and calcareous) based on Sullivan et al. (2007). Digital Elevation Model (Government of Canada, 2000)

Topography (DEM) 1:100,00 e1:250,000 Atmospheric 42 km  42 km AUnified Regional Air Quality Modelling System (AURAMS) 2006 (Moran et al., 1998) deposition Water chemistry >800 lakes ANC, ALK, BC, Ca2þ, pH, DOC and SO2 4 (Jeffries et al., 2010; Scott et al., 2010); pH only (Coker and Ellwood, 1981; Coker et al., 1981; Geological Survey of Canada, 1987, Hornbrook and Friske, 1988a; Hornbrook and Friske, 1988b; Hornbrook and Friske, 1988c; Friske et al., 1994a, Friske et al. 1994b).

for each variable and removed the variable with the highest value until all variables fell below a threshold of 5. A forward and backward stepwise comparison was performed using the remaining variables and the model with the best fit was chosen based on the significance of its variables (p < 0.01) and lowest Akaike's Information Criterion (AIC). Regression kriging is an interpolation method wherein the stochastic and deterministic components are summed to predict the value of a variable at a geographic location, yielding better results than traditional regression or interpolation (Hengl et al., 2007; Bishop and McBratney, 2001). The regression kriging portion of the analysis was performed using the GSTAT package in R (Pebesma, 2004). Spatial correlation was modelled using semivariograms automatically fitted by GSTAT and reviewed and adjusted manually in cases of anisotropy.

2.4. Critical loads of acidity for surface waters The predicted maps of water chemistry were used to calculate critical loads of acidity using the Steady-State Water Chemistry (SSWC) model (Henriksen and Posch, 2001). Saskatchewan is far inland and water chemistry was not corrected for sea salt following Henriksen et al. (2002). Nitrate was excluded due to low observed values (averaging 0.8 meq L1). The SSWC model determines critical loads of acidity as: CLac ¼ Q([BC0] e [ANC]limit) Hydrological runoff is represented by Q, which was interpolated from points generated using the MetHyd model (Bonten et al., 2016) and meteorological data (New et al., 2000). The preindustrial concentration of base cations, BC0, was derived using:

[SO2 4 ]0 ¼ a þ b[BCt] where a and b, which are coefficients for atmospheric and geologic sources, were based on ranges from Scandinavian lakes (a ¼ 19, b ¼ 0.07; Posch et al., 1997). The ANClimit is defined as the lowest ANC concentration that does not affect aquatic biota (generally fish), or conversely the ANC concentration above which no damage to the chosen indicator biota occurs (Nilsson and Grennfelt, 1988). As per Posch et al. (2012), the ANClimit is generally chosen based on regional-scale assessments of an indicator species. In absence of regional assessments of indicator species in Saskatchewan, a commonly used ANClimit of 20 meq m3 was chosen, which was originally established through empirical relationships between lake water chemistry and fish status in Norway (Lien et al., 1996) and adopted in other parts of the world (i.e., Duan et al., 2000; Boggero et al., 1998; Moiseenko, 1994). However, because organic acids influence ANC, a variable ANClimit taking into account each lake's concentration of DOC (mg L1; generated via regression kriging) was used in the present study following Posch et al. (2012) and Lydersen et al. (2004). The ‘organic-acid-adjusted’ ANC limit, ANCoaa.limit, was set to a target aquatic biota species for protection; a commonly used value of 8 meq L1 for trout (Lydersen et al., 2004) was adopted: ANClimit ¼ ANCoaa.limitþ 3.4  DOC The risk of ‘harmful effects’ was quantified by the exceedance of critical loads of acidity; exceedance of critical load was determined using total S deposition during 2006 from the AURAMS model (see Fig. SI.1): Ex ¼ CLac e Sdep

2e BC0 ¼ BCt  F(SO2e 4 t  SO4 0)

3. Results and discussion The present-day concentration of base cations and sulphate (BCt and SO2e 4 t), were generated for each catchment using regression kriging. The F-factor (the ratio of change in non-marine BC concentrations) was derived according to Henriksen and Posch (2001) using a base cation concentration of 400 meq L1 for the value at which F is equal to 1 (see Brakke et al., 1990). The pre-industrial sulphate concentration ([SO2 4 ]0) was calculated using:

3.1. Catchment delineation A total of 89,947 catchments were generated using Thiessen polygons with a mean area of 2.79 km2. While the boundaries were based on distance, the generated catchments were comparable in area to those delineated using a DEM (Gibson et al., 2010) that

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

encompassed a single lake; Thiessen polygon catchments averaged 2.4 km2 and DEM delineated catchments averaged 2.0 km2 for 10 lakes compared between the datasets. Assessing suitability of lakes for inclusion or exclusion in the analysis was in some instances problematic as a result of undefined water body types in the NHN database, and therefore some catchments may belong to rivers, wetlands, or complex inflows and outflows at the mouths of rivers, and other water bodies with difficult or incorrect delineation or definition within the database. However, this affected only a small portion of the total lake population and overall, lakes were well represented by their Thiessen polygon boundaries.

3.2. Regression models Regression model fit ranged between an R2 value of 0.29 (SO2, 4 Table 2) and 0.42 (ANC; see Supporting Information Table SI.2) with satisfactory residual normalcy. Predictor variables that appeared in multiple models included elevation, which was significant (p < 0.01) in all models except lake SO2 4 , where slope was the sole significant topographic variable (Table 2). The Enhanced Vegetation Index (EVI) and the Normalized Difference Vegetation Index (NDVI) were significant in all models except BC, and may be representative of a range of underlying landscape characteristics relating to geology, soil, and topography; however, they were not highly correlated with those variables. The presence of calcium oxide, derived from surveys of glacial till and geochemical boreholes, was significant in the ANC, BC, Ca2þ and DOC models. Bedrock geology (either as felsic, siliceous or mafic types) was also a significant predictor in all models except for SO2 4 and generally indicates geographic presence on the Precambrian Shield or off the Shield. The SO2 4 model (Table 2), in contrast, was influenced heavily by atmospheric variables, with five out of eight predictor variables attributed to climate. The predicted SO2 map showed a clear relationship 4

5

between lake SO2 4 concentrations and proximity to point sources (see Figs. SI.1 and SI.2), such as the Hudson's Bay smelter in Flin Flon, and also smaller active and decommissioned mines. This was driven by the inclusion of total S deposition as a significant predictor. The semivariogram fit for the SO2 4 map was somewhat weak, possibly as a result of conflicting atmospheric and geologic 2 contributions of SO2 4 . However, predicted SO4 closely followed 2þ observed values, as did Ca (Fig. 3). Other variables (particularly ALK and BC) lost predictive power at high concentrations (Fig. 3); those variables are influenced by rare observations of lakes with high ALK or BC. While data quality issues can affect the quality of regression itself, coarse or aging data layers or inconsistent sampling and laboratory protocols can be somewhat overcome by inclusion of the stochastic component of regression kriging. Inclusion of regions south of the Precambrian Shield on the geologically distinct Western Canada Sedimentary Basin may contribute to higher

Table 2 Regression models for base cations (BC), dissolved organic carbon (DOC), and sulphate (SO2 4 ) and their significant predictor variables. These models were the basis for the regression kriging and calculation of critical loads. See Supporting Information (Table SI.2) for additional models of surface water chemistry used for acid sensitivity assessment (alkalinity (ALK), acid neutralizing capacity (ANC), calcium (Ca2þ), and pH). Model

Predictor

Estimate

Std. Error

T-value

P-value

R2 adj.

BC

(Intercept) CAREA ELEV CAO FELSIC BROAD GWDTS PDRYQ GDDP2 SLOPE (Intercept) ELEV CAO FELSIC PCOLDQ EVI NVDI (Intercept) ISO TRANGE PDRYQ GDDP2 SLOPE EVI NVDI SO4

5.878 0.001 0.005 0.047 0.005 0.006 0.003 0.043 0.012 0.103 2.317 0.003 0.029 0.002 0.027 0.010 0.010 7.803 20.408 0.174 0.025 0.007 0.220 0.022 0.010 0.005

0.478 0.001 0.001 0.010 0.001 0.001 0.001 0.006 0.001 0.018 0.335 0.001 0.009 0.001 0.004 0.003 0.002 1.639 2.566 0.026 0.005 0.001 0.023 0.004 0.002 0.001

12.302 3.773 16.515 4.650 8.721 4.423 3.957 7.496 10.774 5.782 6.921 9.692 3.156 4.917 6.460 3.197 5.376 4.760 7.954 6.556 4.636 4.771 9.580 5.228 3.975 5.437

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.01 <0.001 <0.001 <0.01 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

0.38

DOC

SO24

0.36

0.29

Fig. 3. Observed versus predicted (regression-kriged) Acid Neutralizing Capacity (ANC), base cations (BC), alkalinity (ALK), dissolved organic carbon (DOC), calcium (Ca2þ), sulphate (SO2 4 ) and pH.

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

6

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

variance as a result of a stronger right skew in variable distribution than on the Shield (see also Liaw and Atton, 1981). While logistic transformation offers unbiased back transformation for the target variables, variance is unable to be similarly back-transformed and therefore only relative variance can be determined; regions to the south of the shield and west of the Manitoba border experienced the highest relative variance (see Supporting Information Fig. SI.3). Off-shield regions were included in the interest of providing spatial coverage for catchments in northwestern Saskatchewan in proximity to the AOSR, however few observed surface water data were available for off-shield catchments east of 109 longitude (see Supporting Information Fig. SI.4), which may contribute to uncertainty in that area. 3.3. Sensitivity indicator maps There is a common region on the Athabasca Basin and the Precambrian Shield where predicted values of ANC, ALK, Ca2þ and pH were low (Fig. 4). While a pH below 6 was not predicted for the Athabasca Basin, a large number of lakes (28,241 or 70,364 km2) had pH values between 6 and 6.5 on the Athabasca Basin central and southern Shield regions. A portion of the economy of northern Saskatchewan is based on commercial and sport fishing, and a low pH can signal decline in fish populations; for example, streams in the Adirondacks experienced loss of half the total fish species with

a pH below 6 (Kretser et al., 1989). However, a single indicator alone does not provide sufficient information for determination of acid sensitivity (i.e., pH does not provide information about buffering capacity) and so a number of indicators are generally presented together for assessment (e.g. Wolniewicz et al., 2011). Catchments that had two or more indicators of sensitivity were assessed under two thresholds (Fig. 4): sensitive criteria (pH < 6.5 and ANC, ALK, Ca2þ < 100 meq L1) and very sensitive criteria (pH < 6 and ANC, ALK, Ca2þ < 50 meq L1). In both cases, two major regions located on the Athabasca Basin show sensitivity (Fig. 4); approximately 13,726 km2 or 5657 lakes were characterized as very sensitive, compared to 33,758 sensitive lakes with an area of 79,972 km2. Broad historical assessments using geologic regions as the sensitivity indicator have also identified the Athabasca Basin as very sensitive with similar criteria (Shewchuk, 1982; ESSA et al., 1987). Few studies have evaluated the sensitivity of the total lake population; rather, previous studies were generally designed to be statistically representative of the total lake population. In Canada, an assessment of sensitivity of Nova Scotian lakes (Wolniewicz et al., 2011) used linear regression (without spatial interpolation) on 204 lakes to estimate the chemistry of the entire lake population (6104 lakes), with stronger predictive regression models ranging from 38 to 73% (R2) suggesting that 70% of lakes exceeded one or more sensitivity thresholds set to protect biota (pH < 6 and, or ANC < 20 meq L1). Under these criteria, 5% of lakes in the current

Fig. 4. Mapped predicted values of indicators of acid sensitivity (ANC, ALK, Ca2þ, pH) in lakes produced via regression kriging and represented at catchment scale. Catchments identified as possessing two or more indicators of sensitivity are identified: ANC, ALK, Ca2þ each under 50 meq L1, pH below 6.0 (light red) or ANC, ALK, Ca2þ each under 100 meq L1, pH below 6.5 (dark red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

study are below at least one threshold. Compared with the regional Saskatchewan assessment of headwater lakes, 39% (versus 60% in Scott et al., 2010) of northern Saskatchewan lakes in the current study are considered sensitive with an ANC of 50e200 meq L1. An effect seen in all the maps is the influence of the Carswell Structure, an impact crater in the Precambrian Shield located near the Alberta border and south of Lake Athabasca (Fig. 1). It is approximately 40 km in diameter and, in contrast to the surrounding granitic Shield, is comprised of primarily calcareous bedrock (Ramaekers, 1990). The calcareous nature of the lakes in the Carswell Structure as well as its effect on spatial interpolation causes the region of sensitivity and low critical loads to be pushed farther east than seen in assessments based on soils (Aherne and Watmough, 2006; Aherne, 2008), and away from the AOSR. The boundaries of the structure are well defined and it is likely that catchments adjacent to the crater on the Alberta-Saskatchewan border east of the AOSR are of higher sensitivity and lower critical loads than portrayed due to edge effect. 3.4. Critical loads and exceedances Very low critical loads of acidity, i.e., <5 meq m2 yr1 (Fig. 5), were predicted in 5938 catchments (18,294 km2) corresponding with the region of high sensitivity on the Athabasca Basin (Fig. 4). A larger region (23,043 catchments or 25% of the total lake population) was predicted to have critical loads < 10 meq m2 yr1. Critical loads ranged between 0.12 and 344 meq m2 yr1, with more extreme values than previous Saskatchewan estimates (9e163 meq m2 yr1; Environment Canada, 2004), although this study is geographically restrained to northern Saskatchewan and excludes high values found in the southern portion of the province. Predicted critical loads are among the lowest ranges calculated for lakes across Canada. Only Nova Scotia (<0e292 meq m2 yr1) is similarly low; in contrast, the upper ranges of other provinces were predicted to extend up to 11,000 meq m2 yr1 (British Columbia; Environment Canada, 2004). The mean critical load of acidity for the entire lake population (20.6 meq m2 yr1) was slightly lower than estimates from previous regional lake water surveys in Saskatchewan (24.2 meq m2 yr1 in Scott et al., 2010; which is consistent with Whitfield et al., 2016, and 25.6 meq m2 yr1 in Jeffries et al., 2010) and much lower than estimates in eastern

Fig. 5. Mapped estimates of lake critical loads based on the Steady State Water Chemistry (SSWC) model and catchments receiving 2006 total sulphur deposition in exceedance of their critical load, represented at catchment scale (see Fig. SI.1 for AURAMS modelled sulphur deposition).

7

provinces (i.e., 92.8 meq m2 yr1 in Nova Scotia). In contrast, several states in the northeastern US, have similarly low mean values (Rhode Island ¼ 10.8 meq m2 yr1, Vermont ¼ 38.2 meq m2 yr1; Dupont et al., 2005). In Western Canada, mean critical loads of acidity for lakes in nearby Manitoba were also comparable (18e40 meq m2 yr1; Jeffries et al., 2010), while lakes in British Columbia had higher mean critical loads at 70 meq m2 yr1 (Strang et al., 2010). Critical loads of acidity for lakes in Alberta ranged from 176 to 356 m2 yr1 (Gibson et al., 2010). Exceedances of critical loads occurred in approximately 12% of the total lake population under 2006 S deposition (see Fig. 5 and Fig. SI.1), but particularly in the sensitive region on the Athabasca Basin within 100 km of the AOSR. Under 2002 S and N deposition, a similar estimate of 11.6% of lakes in eastern Canada and the northeastern US were predicted to exceed their critical loads of acidity, compared with 32% for acid-sensitive lakes in Nova Scotia (Dupont et al., 2005) and 18% of lakes surveyed in British Columbia (Strang et al., 2010). It is important to note that exceedance in eastern provinces and British Columbia was driven by higher S deposition than in the central provinces; e.g., regional comparisons in Aherne and Jeffries (2015) showed S-only exceedance in British Columbia was 6.5% under average S deposition of 12.2 meq m2 yr1; in contrast, Saskatchewan had more (10.7%) catchments in exceedance under less S deposition (9.2 meq m2 yr1). Lake surveys in Manitoba and Saskatchewan identified exceedances in all sampling groups, with the largest exceedances seen downwind of the AOSR and near the Hudson's Bay smelter in Flin Flon (Jeffries et al., 2010). The influence of the Flin Flon smelter is seen in regions of exceedence in the southern portions of the study area (Fig. 5), which was active in the modelled deposition year (2006) but has since reduced production. It is important to note that in the current study N deposition has been excluded from the critical load (and exceedance) calculations, which are based solely on S contributions, as the SSWC considers only acidity from S. Emissions of NO in the AOSR are growing and expected to exceed SO2 emissions in the future, and current NO emissions estimates are likely under-represented as a result of exclusions in reported emissions, e.g., the mobile mining fleet (Environment Canada, 2014). Modelled AURAMS N deposition in 2006 averaged 140 eq m2 yr1 in the 50 km2 zone surrounding the AOSR, or 20% of the total acidic (S and N) deposition. Despite increases in emissions of NO relative to SO2, and historic NO levels, it seems likely that catchment uptake will mitigate much of the N deposition (see Laxton et al., 2010; Vitt et al., 2003). Their exclusion from these calculations means that acidic contributions may be greater than suggested in this study; however, impacts on surface waters in Canada have primarily been driven by S deposition (Jeffries et al., 1995). While diatom assessments in the current study region suggest there has been little evidence to indicate changes due to historic acidification (Laird et al., 2013), the impacts may have been masked by wider scale climatic changes in the diatom record (Kurek et al., 2013). Ecosystem interactions outside the scope of the SSWC model may also have a mitigating effect on S deposition. Prevalence of wetlands, catchment retention, and the ambiguity of lake status in this region may mean that wetlands reduce a considerable portion of deposited SO2 (Schindler et al., 1986; Bayley et al., 1986). 4 Muskeg is prevalent across northern Saskatchewan and in some areas, such as the northeast, accounts for 25e50% of landcover (Wiken, 1986). However, wetland sinks may also turn to a source of acidic ions during periods of dryness (Bayley et al., 1986), as such it represents a temporary buffer only.

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

8

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

3.5. Lakes at risk Eastern Canada has witnessed damage to aquatic and terrestrial ecosystems as the result of elevated S and N deposition during the past four decades, as evidenced through long-term ecosystem monitoring (Jeffries et al., 2003). In contrast, little background or temporal data exists to assess changes in Saskatchewan's lake status with regards to acidification. Although exceedance is predicted across 12% of the lakes in northern Saskatchewan under 2006 total S deposition, it is important to note that the steady state estimates do not provide a timeline for acidification, which can take months to years or decades to manifest as damage to ecosystems. Consequences of acidification in Saskatchewan include impacts on fishing activities. More than $496 million dollars was invested in the province in 2010 by anglers (Government of Saskatchewan, 2010), and the direct and indirect impacts of sport fishing in the province were estimated at $29 million in GDP (Derek Murray Consulting, 2006). Fishing represents a major portion of the economy in the north and is sometimes the primary or sole source of income for northern and remote communities (Derek Murray Consulting, 2006), who also may rely on fishing for subsistence purposes (Berkes, 1990). Major sport fishes in northern Saskatchewan include northern pike (Esox lucius), walleye (Sander vitreus), and lake trout (Salvelinus namaycush), and population distributions of these species may be limited at a pH under 6.4 (Wales and Beggs, 1986). Predicted exceedances under 2006 deposition, while influenced by AOSR emissions, are also impacted by long-range transport from all S emissions sources downwind in Alberta, as well as by other point sources, such as the Hudson's Bay smelter (which is under reduced production since 2010) in Flin Flon whose deposition footprints are evident in the predicted lake SO2 4 map (see Fig. SI.2) and in the higher critical loads in the southern areas of the study area (Fig. 5). However, emissions of SO2 from the AOSR have declined since the oil and gas downturn in 2014 (see Fig. 2) and it is likely that fewer lakes are exceeded under present deposition levels. 4. Conclusion The regression kriging approach undertaken in this study provided maps for assessing regional acid sensitivity and critical loads of acidity at a detailed, catchment-based scale. Regression kriging may be of use for developing critical loads in similarly remote regions where surveying is costly or difficult. The single-lake catchment delineation using Thiessen polygons proved to be especially appropriate in this region with tens of thousands of closely-spaced lakes under limited terrain differentiation. The study highlighted a considerable number of lakes in northern Saskatchewan that are highly acid-sensitive, having critical loads below 5 meq m2 yr1, and predicted to exceed their critical loads under 2006 modelled total S deposition estimates. These catchments are likely at risk from elevated SO2 emissions originating from the AOSR and 12% were predicted to receive acidic deposition beyond their critical loads. However, the SSWC does not provide a timeline for acidification, rather it identifies lakes receiving acidic deposition in excess of their neutralizing capacity, i.e., the population under risk of impacts. Dynamic modelling may provide further insight into the acid status and timeline of impacts to lakes in northern Saskatchewan; however, current data sources are limited to adequately constrain dynamic model simulations. Acknowledgements This research was undertaken, in part, thanks to funding from

the Canada Research Chairs Program, and an NSERC Discovery grant. In addition, this project was undertaken with financial support of the Government of Canada through the federal Department of the Environment. We also gratefully acknowledge field support from Jim Johnson, Kevin Adkinson, and Chris Jones. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2016.08.048. References Aherne, J., 2008. Critical Load and Exceedence Estimates for Upland Forest Soils in Manitoba and Saskatchewan: Comparing Exceedance under Multiple Deposition Fields. Final Report. Canadian Council of Ministers of the Environment. Retrieved from: http://www.ccme.ca/files/Resources/air/acid_rain/pn_1408_ clab.pdf. Aherne, J., Watmough, S., 2006. Calculating Critical Loads of Acid Deposition for Forest Soils in Manitoba and Saskatchewan. Final Report: Data Sources, Critical Load, Exceedance and Limitations. Canadian Council of Ministers of the Environment. Retrieved from: http://www.ccme.ca/files/Resources/air/acid_rain/ pn_1408_clab.pdf. Aherne, J., Jeffries, D., 2015. Critical load assessments and dynamic model applications for lakes in North America. In: In de Vries, W., Hettelingh, J.P., Posch, M. (Eds.), Critical Loads and Dynamic Risk Assessments: Nitrogen, Acidity and Metals in Terrestrial and Aquatic Ecosystems, vol. 25. Springer Netherlands, Dordrecht. Bayley, S.E., Behr, R.S., Kelly, C.A., 1986. Retention and release of S from a freshwater wetland. Water, Air Soil Pollut. 31, 101e114. Bishop, T., McBratney, A., 2001. A comparison of prediction methods for the creation of field-extent soil property maps. Geoderma 103 (1e2), 149e160. Berkes, F., 1990. Native Subsistence Fisheries: a synthesis of harvest studies in Canada. Arctic 43 (1), 35e42. Berg, N.H., Gallegos, A., Dell, T., Frazier, J., Procter, T., Sickman, J., Grant, S., Blett, T., Arbaugh, M., 2005. A screening procedure for identifying acid sensitive lakes from catchment characteristics. Environ. Monit. Assess. 105 (1), 285e307. Boggero, A., Barbieri, A., De Jong, J., Marchetto, A., Mosello, R., 1998. Chemistry and critical loads of Alpine lakes in Canton Ticino (southern central Alps). Aquat. Sci. 60, 300e315. Bonten, L.T.C., Reinds, G.T., Posch, M., 2016. A model to calculate effects of atmospheric deposition on soil acidification, eutrophication and carbon sequestration. Environ. Model. Softw. 79, 75e84. Brakke, D., Henriksen, A., Norton, S., 1990. A variable F-factor to explain changes in base cation concentrations as a function of strong acid deposition. Int. Ver. fuer Theoretische Angew. Limnol. Verhandlungen 24 (1). Campbell, J., 2002. Phelps Lake 2002: till Geochemistry (<0.063 Mm Size Fraction) Keseechewun Lake Area (64M-NE). Data file. Retrieved from: http://economy. imaginesask.ca/DF28. Campbell, J., 2005. Peter Lake Domain Project 2003e2004: Parts of NTS Map Sheets 64E-9, -10, -15 and -16. Data file. Retrieved from: http://economy.imaginesask. ca/DF27. Card, C., Bosman, S., 2012a. Data File Report 30: Geochemical Analyses of Athabasca Group Outcrops in Saskatchewan (NTS 64L, 74F to 74K, and 74N to 74P) Supplementary to Data File Report 29. Data file. Retrieved from: http:// economy.imaginesask.ca/DF30. Card, C., Bosman, S., 2012b. Data File Report 31: Geochemical Analyses of Athabasca Group Drillholes in Saskatchewan (NTS 64L, 74F to 74K, and 74N to 74P) e Supplementary to Data File Reports 24, 29, and 30. Data file. Retrieved from: http://economy.imaginesask.ca/DF31. Card, C., Bosman, S., Slimmon, W., Zmetana, D., Delaney, G., 2011. Data File Report 29: Geochemical Analyses of Athabasca Group Outcrops in Saskatchewan (NTS 64L, 74F to 74K, and 74N to 74P). Data file. Retrieved from: http://economy. imaginesask.ca/DF29. Coker, W.B., Ellwood, D.J., 1981. National Geochemical Reconnaissance, Northern Manitoba, Southeastern Northwest Territories and Northeastern Saskatchewan (54L, M; 64I, J, K, N, O, P and Parts of 64L, M; 65A, B, C. Geological Survey of Canada, Open File 742, (23). Retrieved from: http://geogratis.gc.ca/api/en/ nrcan-rncan/ess-sst/20a6e964-3525-5edb-a253-1dfea59e0006.html. Coker, W.B., Ellwood, D.J., Hornbrook, E.H.W., 1981. National Geochemical Reconnaissance, East-central Saskatchewan (63M, 64D and Parts of 63K, L, N and 73I, O, P and 74A). Geological Survey of Canada, Open File 744. Retrieved from: http://geogratis.gc.ca/api/en/nrcan-rncan/ess-sst/5957c395-6581-54d7-b4d0a67475ea88a5.html. Derek Murray Consulting, 2006. Final Report: Economic Evaluation of Saskatchewan's Commercial Non-outfitted Sport Fishing. Saskatchewan Environment. Retrieved from: http://www.publications.gov.sk.ca/details.cfm?p¼24510. Duan, L., Hao, J., Xie, S., Du, K., 2000. Critical loads of acidity for surface waters in China. Sci. Total Environ. 246 (1), 1e10. Dupont, J., Clair, T.A., Gagnon, C., Jeffries, D.S., Kahl, J.S., Nelson, S.J., Peckenham, J.S., 2005. Estimation of critical loads of acidity for lakes in northeastern United

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10 States and Eastern Canada. Environ. Monit. Assess. 109, 275e291. Environment Canada, 2004. Canadian acid Deposition Science Assessment 2004. Retrieved from:. Environment Canada, Ottawa http://publications.gc.ca/site/ eng/9.650140/publication.html. Environment Canada, 2011. Integrated Monitoring Plan for the Oil Sands: Expanded Geographic Extent for Water Quality and Quantity, Aquatic Biodiversity and Effects, and Acid Sensitive Lake Component. Retrieved from:. Minister of the Environment, Ottawa http://publications.gc.ca/pub?id¼396888&sl¼0. Environment Canada, 2014. National Pollutant Release Inventory. Database. Retrieved from: http://www.ec.gc.ca/inrp-npri. ESSA, Social Systems Analysis Ltd., Concord Scientific Corporation, 1987. Interim Target Loadings for Acidic Deposition in Western Canada: a Synthesis of Existing Information. Technical Report. Western and Northern Canada LRTAP Technical Committee. Friske, P.W.B., McCurdy, M.W., Day, S.J., Gross, H., Balma, R.G., Lynch, J.J., Durham, C.C., 1994a. National Geochemical Reconnaissance Lake Sediment and Water Data, Northeastern Saskatchewan (Parts of NTS 64L, 64M and 74P). Geological Survey of Canada, Open File 2857. Friske, P.W.B., McCurdy, M.W., Day, S.J., Gross, H., Balma, R.G., Lynch, J.J., Durham, C.C., 1994b. National Geochemical Reconnaissance Lake Sediment and Water Data, Northwestern Saskatchewan (Parts of NTS 74N and 74O). Geological Survey of Canada, Open File 2858. Geological Atlas of Saskatchewan, 2012a. Athabasca Lithogeochemistry. Digital Map: 1:250,000 Scale. Retrieved from: http://www.infomaps.gov.sk.ca/website/ SIR_Geological_Atlas/. Geological Atlas of Saskatchewan, 2012b. Lithogeochemistry Analysis. Digital Map: Scale 1:250,000. Retrieved from: http://www.infomaps.gov.sk.ca/website/SIR_ Geological_Atlas/. Geological Atlas of Saskatchewan, 2013. Bedrock Geology Layer Merged from 1: 250,000 Scale Compilation Bedrock Geology Map Series. Digital Map: 1: 250,000 Scale. Retrieved from: http://www.infomaps.gov.sk.ca/website/SIR_ Geological_Atlas/. Geological Survey of Canada, 1987. Regional Lake Sediment and Water Geochemical Reconnaissance Data, Saskatchewan. Geological Survey of Canada, Open File 2359. Gibson, J.J., Birks, S.J., Kumar, S., McEachern, P.M., Hazewinkel, R., 2010. Inter-annual variations in water yield to lakes in northeastern Alberta: implications for estimating critical loads of acidity. J. Limnol. 69 (1), 126e134. Government of Canada, 2000. Canadian Digital Elevation Data. Digital Map: Various Scales. Retrieved from: http://geobase.ca. Government of Canada, 2007. National Hydro Network, Canada. Digital Map. Retrieved from: http://geobase.ca/geobase/en/data/cded/index.html. Government of Saskatchewan, 2010. Sport Fishing in Saskatchewan: 2010 Summary Report. Retrieved online from: http://www.environment.gov.sk.ca. GRASS Development Team, 2012. Geographic Resources Analysis Support System (GRASS) Software, Version 6.4. Retrieved from:. Open Source Geospatial Foundation http://grass.osgeo.org. Hazewinkel, R.R., Wolfe, A.P., Pla, S., Curtis, C., Hadley, K., 2008. Have atmospheric emissions from the Athabasca Oil Sands impacted lakes in north eastern Alberta, Canada? Can. J. Fish. Aquatic Sci. 65 (8), 1554e1567. Henriksen, A., Posch, M., 2001. Steady-state models for calculating critical loads of acidity for surface waters. Water, Air Soil Pollut. Focus 1 (1e 2), 375e398. Henriksen, A., Dillon, P.J., Aherne, J., 2002. Critical loads of acidity for surface waters in south-central Ontario, Canada: regional application of the Steady-State Water Chemistry (SSWC) model. Can. J. Fish. Aquatic Sci. 59 (8), 1287e1295. Hengl, T., Heuvelink, G.B., Rossiter, D.G., 2007. About regression-kriging: from equations to case studies. Comput. Geosci. 33 (10), 1301e1315. Hornbrook, E.H.W., Friske, P.W.B., 1988a. Regional Lake Sediment and Water Geochemical Data, Northeastern Saskatchewan. Geological Survey of Canada, Open File 1643. Retrieved from: http://geochem.nrcan.gc.ca/cdogs/content/ pub/pub00126_e.htm. Hornbrook, E.H.W., Friske, P.W.B., 1988b. Regional Lake Sediment and Water Geochemical Data, Northern Saskatchewan. Geological Survey of Canada, Open File 1644. Retrieved from: http://geochem.nrcan.gc.ca/cdogs/content/pub/ pub00127_e.htm. Hornbrook, E.H.W., Friske, P.W.B., 1988c. Regional Lake Sediment and Water Geochemical Data, Northwestern Saskatchewan. Geological Survey of Canada, Open File 1642. Retrieved from: http://geochem.nrcan.gc.ca/cdogs/content/ pub/pub00125_e.htm. Jeffries, D.S., Clair, T.A., Dillon, P.J., Papineau, M., Stainton, M.P., 1995. Trends in surface water acidification at ecological monitoring sites in southeastern Canada (1981e1993). Water, Air Soil Pollut. 85, 577e582. Jeffries, D.S., Clair, T.A., Couture, S., Dillon, P.J., Dupont, J., Keller, W., McNicol, D.K., Turner, M.A., Vet, R., Weeber, R., 2003. Assessing the recovery of lakes in southeastern Canada from the effects of acidic deposition. Ambio 32 (3), 176e182. Jeffries, D.S., Semkin, R.G., Gibson, J.J., Wong, I., 2010. Recently surveyed lakes in northern Manitoba and Saskatchewan, Canada: characteristics and critical loads of acidity. J. Limnol. 69 (1s), 45e55. http://dx.doi.org/10.3274/JL10-69-S1-06. Johnston, C.M., Dewald, T.G., Bondelid, T.R., Worstell, B.B., McKay, L.D., Rea, A., Moore, R.B., Goodall, J.L., 2009. Evaluation of Catchment Delineation Methods for the Medium-resolution National Hydrography Dataset. Technical Report 2009-5233, U.S. Geological Survey Scientific Investigations. Retrieved from: http://pubs.usgs.gov/sir/2009/5233. Kurek, J., Kirk, J.L., Muir, D.C., Wang, X., Evans, M.S., Smol, J.P., 2013. Legacy of a half

9

century of Athabasca oil sands development recorded by lake ecosystems. Proc. Natl. Acad. Sci. 110 (5), 1761e1766. Kretser, W., Gallagher, J., Nicolette, J., 1989. Adirondack lakes Study, 1984e1987: an Evaluation of Fish Communities and Water Chemistry. Adirondack Lakes Survey Corporation. Laird, K.R., Das, B., Kingsbury, M., Moos, M.T., Pla-Rabes, S., Ahad, J.M., Cumming, B.F., 2013. Paleolimnological assessment of limnological change in 10 lakes from northwest Saskatchewan downwind of the Athabasca oils sands based on analysis of siliceous algae and trace metals in sediment cores. Hydrobiologia 720 (1), 55e73. Laxton, D.L., Watmough, S.A., Aherne, J., Straker, J., 2010. An assessment of nitrogen saturation in Pinus banksiana plots in the Athabasca oil sands region, Alberta. J. Limnol. 69 (1s), 171e180. Liaw, W., Atton, F., 1981. Acid rain on northern lakes: background studies in Saskatchewan. Musk-Ox 28, 26e42. Lien, L., Raddum, G.G., Fjellheim, A., Henriksen, A., 1996. A critical limit for acid neutralizing capacity in Norwegian surface waters, based on new analyses of fish and invertebrate responses. Sci. Total Environ. 177, 173e193. Lydersen, E., Larssen, T., Fjeld, E., 2004. The influence of total organic carbon (TOC) on the relationship between acid neutralizing capacity (ANC) and fish status in Norwegian lakes. Sci. Total Environ. 326 (1e3), 63e69. McKenney, D., Campbell, K., Lawrence, K., Valleau, J., Farr, K., 2006. Spatial Models of Canada- and North America-wide 1971/2000 Minimum and Maximum Temperature, Total Precipitation and Derived Bioclimatic Variables. Technical Report 106. Canadian Forestry Service. Retrieved from: http://cfs.nrcan.gc.ca/ publications?id¼26775. McMartin, I., Campbell, J., Dredge, L., 2008. Till Geochemistry from Archived Samples Collected over the La Ronge and Rottenstone Domains, Central Saskatchewan. Data file. Retrieved from: http://geochem.nrcan.gc.ca/cdogs/content/pub/ pub02238_e.htm. McMartin, I., Campbell, J., Dredge, L., Grunsky, E., 2007. Till Geochemistry from Archived Samples Collected over the Glennie, Kisseynew and Flin Flon Domains, East-central Saskatchewan. Data file. Retrieved from: http://geogratis.gc.ca. Moiseenko, T., 1994. Acidification and critical load in surface waters: Kola, Northern Russia. Ambio 23, 418e424. Moran, M., Dastoor, A., Gong, S., Gong, W., Makar, P., 1998. Conceptual Design for the AES Unified Regional Air Quality Modelling System (AURAMS) (Atmospheric Environment Service Air Quality Research Branch Internal Rep. Canada). Neary, B., Dillon, P.J., Munro, J.R., Clark, B.J., 1990. The Acidification of Ontario Lakes: an Assessment of Their Sensitivity and Current Status with Respect to Biological Damage. Ministry of the Environment Limnology Section, Dorset Research Centre, Queens Printer for Canada, Ottawa. New, M., Hulme, M., Jones, P.D., 2000. Global Monthly Climatology for the Twentieth Century. Data set. Oak Ridge National Laboratory Distributed Active Archive Center. Retrieved from: http://daac.ornl.gov. Nilsson, J., Grennfelt, P., 1988. Critical Loads for Sulphur and Nitrogen: Report from a Workshop Held at Stockloster, Sweden, 19e24 March 1988. Nordic Council of Ministers. Pebesma, E.J., 2004. Multivariable geostatistics in S: the gstat package. Comput. Geosci. 30, 683e691. Retrieved from: http://cran.r-project.org/package¼gstat. Posch, M., K€ am€ ari, J., Forsius, M., Henriksen, A., Wilander, A., 1997. Exceedance of critical loads for lakes in Finland, Norway, and Sweden: reduction requirements for acidifying nitrogen and sulfur deposition. Environ. Manag. 21 (2), 291e304. Posch, M., Aherne, J., Forsius, M., Rask, M., 2012. Past, present, and future exceedance of critical loads of acidity for surface waters in Finland. Environ. Sci. Technol. 46 (8), 4507e4514. Povak, N.A., Hessburg, P.F., McDonnell, T.C., Reynolds, K.M., Sullivan, T.J., Salter, R.B., Cosby, B.J., 2014. Machine learning and linear regression models to predict catchment-level base cation weathering rates across the southern Appalachian Mountain region, USA. Water Resour. Res. 50, 2798e2814. QGIS Development Team, 2014. QGIS Geographic Information System. Open Source Geospatial Foundation Project. Retrieved from: http://qgis.osgeo.org. Ramaekers, P., 1990. Geology of the Athabasca Group (Helikian) in Northern Saskatchewan. Technical Report 195. Saskatchewan Energy and Mines. Saffran, K.A., Trew, D.O., 1996. Sensitivity of Alberta Lakes to Acidifying Deposition: an Update of Sensitivity Maps with Emphasis on 109 Northern Lakes. Alberta Environmental Protection, Water Management Division, Water Sciences Branch. Schindler, D.W., Turner, M.A., Stainton, M.P., Linsey, G.A., 1986. Natural sources of acid neutralizing capacity in low alkalinity lakes of the Precambrian Shield. Science 232 (4752), 844e847. Schindler, D.W., 1988. Effects of acid rain on freshwater ecosystems. Science 239 (4836), 149e157. Scott, K.A., Wissel, B., Gibson, J.J., Birks, J.S., 2010. Chemical characteristics and acid sensitivity of boreal headwater lakes in northwest Saskatchewan. J. Limnol. 69 (1s), 33e44. http://dx.doi.org/10.4081/jlimnol.2010.s1.33. Shewchuk, S., 1982. An acid deposition perspective for northeastern Alberta and northern Saskatchewan. Water, Air, Soil Pollut. 18 (1e3), 413e419. Strang, D., Aherne, J., Shaw, D.P., 2010. The hydrochemistry of high-elevation lakes in the Georgia Basin, British Columbia. J. Limnol. 69 (1), 56e66. Sullivan, T.J., Webb, J.R., Snyder, K.U., Herlihy, A.T., Cosby, B.J., 2007. Spatial distribution of acid-sensitive and acid-impacted streams in relation to watershed features in the southern Appalachian Mountains. Water, Air, Soil Pollut. 182 (1e4), 57e71. The Canadian System of Soil Classification, 1995. Soil Landscapes of Canada, v.2.1. Digital Map and Database at 1:1 Million Scale. Retrieved from: http://sis.agr.gc.

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048

10

H. Cathcart et al. / Atmospheric Environment xxx (2016) 1e10

ca/cansis/nsdb/slc/index.html. United States Geological Survey, 1998. Global Land Cover Characteristics Database. Map: 1 Km Scale. Retrieved from: http://edc2.usgs.gov/glcc/glcc_version1.php. United States Geological Survey, 2000. LandDAAC MODIS Version 005 Canada, June 25eAugust 11, 2013. Retrieved from: http://iridl.ldeo.columbia.edu/SOURCES/. USGS/.LandDAAC/.MODIS/.version_005/. Vitt, D.H., Wieder, K., Halsey, L.A., Turetsky, M., 2003. Response of sphagnum fuscum to nitrogen deposition: a case study of ombrogenous Peatlands in Alberta, Canada. Bryologist 106 (2), 235e245.

Wolniewicz, M.B., Aherne, J., Dillon, P.J., 2011. Acid sensitivity of lakes in Nova Scotia, Canada: assessment of lakes at risk. Ecosystems 14 (8), 1249e1263. Wales, D.L., Beggs, G.L., 1986. Fish species distribution in relation to lake acidity in Ontario. Water, Air, Soil Pollut. 30 (3e4), 601e609. Whitfield, C.J., Mowat, A.C., Scott, K.A., Watmough, S.A., 2016. A modified approach for estimating the aquatic critical load of acid deposition in northern Saskatchewan, Canada. Atmos. Environ. (in this issue). Wiken, E., 1986. Terrestrial Ecozones of Canada. Report. Ecological land classification series No. 19. Environment Canada.

Please cite this article in press as: Cathcart, H., et al., Critical loads of acidity for 90,000 lakes in northern Saskatchewan: A novel approach for mapping regional sensitivity to acidic deposition, Atmospheric Environment (2016), http://dx.doi.org/10.1016/j.atmosenv.2016.08.048