lable at ScienceDirect

Atmospheric Environment 146 (2016) 311e323

Contents lists avai

Atmospheric Environment

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

Critical levels and loads and the regulation of industrial emissions innorthwest British Columbia, Canada

P. Williston a, *, J. Aherne b, S. Watmough b, D. Marmorek c, A. Hall c, P. de la Cueva Bueno c,C. Murray c, A. Henolson d, J.A. Laurence e

a British Columbia Ministry of Environment, 3726 Alfred Ave, Smithers, BC, V0J 2N0, Canadab Environmental and Resource Studies, Trent University, Peterborough, ON, K9J 7B8, Canadac ESSA Technologies Ltd., 600-2695 Granville St., Vancouver, BC, V6H 3H4, Canadad Trinity Consultants Inc., 20819 72nd Avenue S, Kent, WA, 98103, USAe Consulting Plant Pathologist, 6201 NE 22nd Avenue, Portland, OR, 97211, USA

h i g h l i g h t s

� LNG terminals proposed for NW British Columbia may double emissions of NOx and SO2.� Soil critical loads of acidity averaged 181e219 meq m�2 yr�1.� Exceedance in small areas was predicted under worst-case emissions scenarios.� Up to 20% of lakes were predicted to exceed critical loads of acidity.� Critical levels and loads can inform LNG impact assessment and emissions regulation.

a r t i c l e i n f o

Article history:Received 7 April 2016Received in revised form18 August 2016Accepted 19 August 2016Available online 21 August 2016

Keywords:smelter emissionsliquefied natural gas (LNG)SO2

NO2

acidificationeutrophicationregulationaquatic and terrestrial ecosystemsKitimatPrince Rupertnorthwest British Columbia

* Corresponding author.E-mail address: patrick.williston@gov.bc.ca (P. Wil

http://dx.doi.org/10.1016/j.atmosenv.2016.08.0581352-2310/© 2016 The Authors. Published by Elsevier

a b s t r a c t

Northwest British Columbia, Canada, a sparsely populated and largely pristine region, is targeted forrapid industrial growth owing to the modernization of an aluminum smelter and multiple proposedliquefied natural gas (LNG) facilities. Consequently, air quality in this region is expected to undergoconsiderable changes within the next decade. In concert, the increase in LNG capacity driven by gasproduction from shale resources across North America has prompted environmental concerns andhighlighted the need for science-based management decisions regarding the permitting of air emissions.In this study, an effects-based approach widely-used to support transboundary emissions policy nego-tiations was used to assess industrial air emissions in the Kitimat and Prince Rupert airsheds underpermitted and future potential industrial emissions. Critical levels for vegetation of SO2 and NO2 andcritical loads of acidity and nutrient nitrogen for terrestrial and aquatic ecosystems were estimated forboth regions and compared with modelled concentration and deposition estimates to identify the po-tential extent and magnitude of ecosystem impacts. The critical level for SO2 was predicted to beexceeded in an area ranging from 81 to 251 km2 in the Kitimat airshed owing to emissions from anexisting smelter, compared with <1 km2 in Prince Rupert under the lowest to highest emissions sce-narios. In contrast, the NO2 critical level was not exceeded in Kitimat, and ranged from 4.5 to 6 km2 inPrince Rupert owing to proposed LNG related emissions. Predicted areal exceedance of the critical load ofacidity for soil ranged from 1 to 28 km2 in Kitimat and 4e10 km2 in Prince Rupert, while the arealexceedance of empirical critical load for nutrient N was predicted to be greater in the Prince Rupertairshed (20e94 km2) than in the Kitimat airshed (1e31 km2). The number of lakes that exceeded thecritical load of acidity did not vary greatly across emissions scenarios in the Kitimat (21e23 out of 80sampled lakes) and Prince Rupert (0 out of 35 sampled lakes) airsheds. While critical loads have beenwidely used to underpin international emissions reductions of transboundary pollutants, it is clear thatthey can also play an important role in managing regional air emissions. In the current study, exceedanceof critical levels and loads suggests that industrial emissions from the nascent LNG export sector mayrequire careful regulation to avoid environmental impacts. Emissions management from LNG exportfacilities in other regions should consider critical levels and loads analyses to ensure industrial

liston).

Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323312

development is synergistic with ecosystem protection. While recognizing uncertainties in dispersionmodelling, critical load estimates, and subsequent effects, the critical levels and loads approach is beingused to inform regulatory decisions in British Columbia to prevent impacts that have been well docu-mented in other regions.© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

British Columbia seeks to join the global trade in liquefiednatural gas (LNG); a total export capacity of approximately236 Mt yr�1 has been proposed for the region (British ColumbiaMinistry of Natural Gas Development, 2015). Many of the pro-posed LNG export facilities plan to use gas-fuelled turbines to drivecompressors that liquefy natural gas, and this, combined withincreased shipping, may substantially elevate local nitrogen oxide(NOx) and sulphur dioxide (SO2) emissions. Furthermore, globalLNG capacity is increasing, driven in part by gas production fromshale resources, and there is growing concern that emissions fromliquefaction facilities may lead to environmental impacts if notadequately managed. The potential for increased industrial airemissions from LNG production, particularly in regions withexisting sources of NOx and SO2, suggests a need for science-basedinformation to guide policy decisions, permitting, and airshedmanagement at the regional level that is protective of theenvironment.

Sulphur dioxide and NOx are recognized as having direct andindirect effects on vegetation, soils and surface waters (WHO ,2000; EEA, 2013; US EPA, 2011). Direct exposure to SO2 atelevated concentration and for sufficient time has been shown tocause both visible injury to sensitive vegetation and growth andyield loss in agricultural crops and forest trees (Cape, 1993; Wulffet al., 1996; Likens et al., 1998; Horsely et al., 2000; US EPA,2008). Elevated concentrations of NO2 cause similar effects; how-ever, concentrations reported to cause such injury are usuallyhigher than for SO2 (Mansfield et al., 1985; US EPA, 2008). Lichens,especially thosewith cyanobacteria symbionts, are among themostsensitive to these air pollutants, and are among the first species tobe lost in regions with elevated industrial emissions (Richardsonand Cameron, 2004).

Sulphur dioxide and NOx are also precursors of ‘acid rain’ andcan be deposited in wet deposition or fall as dry deposition leadingto the acidification of soils and surface waters located in regionswith shallow, base poor soils with low mineral weathering rates(Reuss and Johnson, 1986). Consequences of acidification includethe loss of base cations from soils and an increase in aluminum (Al)concentration in soil water and surface waters, which can adverselyimpact forest ecosystems and lead to the loss of aquatic biota (Reussand Johnson, 1986). In addition, nitrogen (N) is considered to be alimiting nutrient for many ecosystems and increased levels of Ndeposition can lead to eutrophication, which can impact commu-nity composition, cause the loss of sensitive species, alter forestproductivity, and increase leaching of NO3

� (nitrate) to surfacewaters (Aber et al., 1998). However, adverse effects caused byacidification or eutrophication are usually not immediate due to theinherent buffering capacity of soils and time-lags in ecosystemresponse (Cosby et al., 1985).

The concepts of both critical levels and critical loads have beenadopted worldwide, perhaps most forcefully in Europewhere theseeffects-based approaches have been key components in policy de-cisions that have led to dramatic decreases in SO2 and, morerecently, NOx emissions (Rafaj et al., 2014; EEA, 2013). As defined by

the Convention on Long-Range Transboundary Air Pollution(CLRTAP), critical levels are “concentrations of pollutants in the at-mosphere above which direct adverse effects on receptors, such ashuman beings, plants, ecosystems or materials, may occur according topresent knowledge”; whereas critical loads are “a quantitative esti-mate of exposure to one or more pollutants below which significantharmful effects on specified elements of the environment do not occuraccording to present knowledge” (UNECE , 2004). Critical levels arepredominately based on empirical data where impacts to sensitivebiota (often lichens) have been observed, whereas critical loads arecommonly based on steady-statemodels (for terrestrial and aquaticecosystems) and are useful for determining the potential area ornumber of surface water bodies at risk from acidification during thelonger term, but do not provide an estimate of when chemical orbiological effects occur (UNECE , 2004). The critical load concepthas also been embraced in Canada, both within the federal gov-ernment where the objective is to reduce acidic deposition to meetcritical loads (CCME, 2013), as well as regionally, such as in theAlberta Oil Sands, where the critical load concept has been incor-porated into management decisions (Foster et al., 2001). Criticalloads and exceedance have been previously determined for specificregions within British Columbia (Mongeon et al., 2010; Strang et al.,2010; Nasr et al., 2010; Krzyzanowski and Innes, 2010) and are nowbeing used to inform provincial government decisions with respectto proposed industrial development (BCMOE, 2014b; 2014c, 2015b;ESSA et al., 2014b).

The objective of this study was to assess the risk of direct andindirect impacts of SO2 and NOx emissions on terrestrial andaquatic ecosystems in the Kitimat and Prince Rupert airsheds,British Columbia. We used air dispersion modelling with criticallevels and critical loads to estimate the magnitude and areal extentof exceedances across a range of emissions scenarios. The assess-ment provided science-based estimates of potential impacts, whichhave been used to evaluate risk and inform the regulation ofpermitted air emissions. While the assessment focused on north-west British Columbia, the growth in LNG capacity worldwide maysuggest a greater need for critical levels and loads assessments tosupport the management of regional air quality.

2. Methods

2.1. Study area

Kitimat (54� 00 N, 128� 41’ W) is at the northern end of a 90 kmlong fjord with a 12 kmwide valley confined by 1500e1800 m highmountains to the east and west (Fig. 1). The climate in Kitimat istemperate: the July mean high is 22 �C and the January mean lowis �4 �C, while the annual precipitation is 2211 mm (EnvironmentCanada, 2015). The Kitimat Valley is mainly forested with westernhemlock (Tsuga heterophylla), Sitka spruce (Picea sitchensis), Pacificsilver fir (Abies amabilis), lodgepole pine (Pinus contorta var. lat-ifolia), and western redcedar (Thuja plicata). The valley bottom isdominated by stands of 30e50 year old regenerating westernhelmlock. This is bisected by the Kitimat River and floodplainswhich support stands of deciduous black cottonwood (Populus

Fig. 1. Northwest coast of British Columbia showing the two study areas, Prince Rupert (west) and Kitimat (east), shaded by opaque white. The locations of existing and proposedemissions sources and study lakes with observations of water chemistry are also shown. The dashed lines delineate focused study areas used to display deposition and critical loads(see Fig. 2 and 3) based upon the results of deposition modelling. The inset shows the location of the study area in relation to the west coast of North America.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 313

trichocarpa) and red alder (Alnus rubra). Higher elevations areoriginal old-growth conifer forests that intergrade to alpine heath.The region supports limited commercial agriculture: a few smallfarms in the vicinity of Terrace, at the northern edge of the studyarea (Fig. 1). Soils are typically podzols (Soil Classification WorkingGroup, 1998), and in places are strongly influenced by deep gla-ciofluvial deposits and alluvial processes (Clague, 1985).

Prince Rupert (54� 180N, 130� 260 W) is also coastal temperatebut has considerably more rain, with an average annual precipita-tion of 3060 mm. High rainfall promotes the regional developmentof peatlands in most low undulating terrain, principally sphagnumbogs and bog-forest complexes with western redcedar, yellow-cedar (Xanthocyparis nootkatensis), and shore pine (Pinus contortavar. contorta). Productive forests grow on steeper terrain of thenearby Coast Mountains, where Sitka spruce, Pacific silver fir,western hemlock, and western redcedar can grow to 60 m inheight. Mineral soils are typically podzols and folisols (SoilClassification Working Group, 1998), though they tend to be over-lain by organic veneers and blankets in areas of gentle topography(Banner et al., 2005). The climate is strongly influenced by the Pa-cific Ocean; the July mean high is 16.2 �C, and the January low

is�0.8 �C (Environment Canada, 2015). The Prince Rupert airshed islargely unconfined to the south, west, and north; and is restrictedto the east by the Coast Mountains (Fig. 1).

The Kitimat airshed has received air emissions from analuminum smelter in operation since 1954 (Fig. 1). The smelter,which recently modernized, is currently permitted to emit 42 t SO2d�1. Kitimat is the preferred location of at least three proposed LNGexport terminal facilities with a planned capacity of approximately34 Mt yr�1 (British Columbia Ministry of Natural Gas Development,2015). The current industrial emissions in the Prince Rupert regionare from shipping activities at the port of Prince Rupert, which atpresent includes grain, coal and wood pellets, in addition tocontainer traffic with rapidly increasing capacity (Prince RupertPort Authority, 2015). The Prince Rupert airshed has at least sixproposed LNG facilities (Fig. 1) with an export capacity of approx-imately 108 Mt yr�1 (British Columbia Ministry of Natural GasDevelopment, 2015).

2.2. Air dispersion modelling

Air dispersion modelling techniques are used worldwide to

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323314

predict or estimate ambient air concentrations and deposition ratesdue to industrial or other sources of air emissions. In the currentstudy, air quality under a range of emissions scenarios was pre-dicted using the CALPUFF modelling system (version 6.42, Level110325). The assessment followed the methods and proceduresdescribed in Guidelines for Air Quality Dispersion Modelling in BritishColumbia (BC MOE, 2015a). CALPUFF has been adopted by the USEnvironmental Protection Agency in its Guideline on Air QualityModels as the preferredmodel for assessing long-range transport ofpollutants and their impacts on Federal Class I areas, and as suchhas been widely used for cumulative effects assessments (e.g., Vittet al., 2003; Johnson et al., 2011). The study area domain for theKitimat airshed was 6772 km2, while the domain for Prince Rupertwas 10,956 km2 (Fig. 1). The meteorological and CALPUFF compu-tational grid spacing was 4 km � 4 km in the Kitimat airshed, andwas subsequently refined to 1 km � 1 km in the Prince Rupertairshed (the second area studied), with a 1 km � 1 km receptor(ecosystem) grid in both airsheds. MESOPUFF II was used to predictsecondary atmospheric chemistry in the Kitimat airshed, whileRIVAD-ISORROPIA was used in Prince Rupert as a refinement toimprove the secondary chemical model predictions. Lackingambient ozone data for the Kitimat airshed, the MESOPUFF IIchemistry model used an assumed ozone concentration of 80 ppb,which is considered conservatively high, resulting in a probableoverestimation of oxidation reaction rates. The RIVAD-ISORROPIAchemistry model used monthly background ozone concentrationsbased onmonthly averages for the Prince Rupert area. One scenarioin Prince Rupert was also run with MESOPUFF II and 80 ppb ozonefor comparison. The CALMET meteorological processor, part of theCALPUFF model system, was used to generate one year of meteo-rological data for Kitimat (2008) using 5th generation mesoscalemodel output (MM5); while in Prince Rupert, CALMET was used togenerate one year of weather data (2012) using Weather Researchand Forecasting (WRF) output. Twelve air emissions scenarios wereexamined in the Kitimat airshed and seven scenarios wereconsidered in Prince Rupert. The scenarios included a range ofemission reduction technologies, and numbers and locations ofproposed developments. Emissions sources included estimates forexisting industries, shipping, rail transport, LNG facilities, a pro-posed refinery, and power generating facilities, with scenariosranging from 3.3 to 55.8 t d�1 SO2 and 11.9e90.0 t d�1 NO2 (Fig. 1,Table 1). For a full description of emissions scenarios and proced-ures used to generate air concentration and deposition estimatessee ESSA et al. (2014a, 2016).

2.3. Critical levels for vegetation

Modelled air concentrations were compared to North Americanand European critical levels (reported in CCME, 1999; BC MOE,2014a; US EPA, 2012; WHO , 2000; and EU , 2008); however,given the proximity of pristine ecosystems and the presence of

Table 1Range in sulphur dioxide (SO2) and nitrogen dioxide (NO2) emissions (from lowest to higscenarios) airsheds, and ranges in average [and maximum] modelled air concentrationsdomains (see Fig. 1).

Kitimat airshed l

Sulphur dioxide (SO2) emissions (t d�1) 16.3e55.8Nitrogen dioxide (NO2) emissions (t d�1) 11.9e30.3Annual average SO2 concentration [maximum] (mg m�3) a 1.10e2.25 [35.2eAnnual average NO2 concentration [maximum] (mg m�3) a 0.43e0.74 [12.2eAnnual sulphur deposition [maximum] (meq m�2 yr�1) 8.17e21.91 [224Annual nitrogen deposition [maximum] (meq m�2 yr�1) 1.77e3.72 [34.7e

a Modelled air concentrations do not include background concentrations (Kitimat SNO2 ¼ 5.64 mg m�3).

nationally endangered and regionally endemic lichen species(COSEWIC, 2006, 2010), only the most conservative critical levelsare presented here to assess the area of modelled exceedance, i.e.,annual average SO2 limit of 10 mg m�3 (to protect sensitive lichens)and NO2 limit of 30 mg m�3 (to protect natural vegetation) (WHO ,2000; EU , 2008).

2.4. Terrestrial critical loads

Steady-state mass balance and nutrient mass balance models(UNECE , 2004) were used to determine critical loads for Kitimatand Prince Rupert airsheds (see Supporting Information Tables SI-1and SI-2). In addition, an empirical assessment for nutrient Ndeposition was performed using critical load values reported byBobbink and Hettelingh (2011) (see Supporting InformationTable SI-3). In both regions, a base cation to aluminum ratio(Bc:Al) of 1 was used as the critical limit for coniferous forestsgrowing on mineral soil, which is protective of the region's domi-nant tree species including western hemlock and western red-cedar; while a ratio of 6 was applied to deciduous forests asprotective of the dominant genus Populus (Sverdrup andWarfvinge,1993). For forests growing on organic soils, a critical Bc:H ratio wasused. An empirical critical load of 4 or 5 kg N ha�1 yr�1 for nutrientNwas used based on reported impacts to lichen communities in thePacific Northwest (see Table SI-3, Geiser et al., 2010).

The study domains were delineated into 1 km � 1 km receptorgrids aligned with the modelled deposition grid, and the propor-tion of receptor ecosystems, e.g., coniferous and deciduous(including mixed) forest, recorded for each grid. Critical loads foracidity and nutrient N were estimated for each 1 km � 1 km gridcell and compared with modelled sulphur (S) and N deposition topredict exceedances. An ‘effects domain’ was developed to facilitatethe assessment of risk (i.e., proportion of area with exceedance),and was defined as the receptor ecosystem area enclosed by the 15meq m�2 yr�1 modelled S and N deposition isopleth under thehighest emissions scenario. A deposition threshold of 15 meq m�2

yr�1 was selected as the minimum deposition expected to causenegative impacts to acid sensitive ecosystems based on the lowestcritical load category in the Skokloser classification (UNECE , 2004).The area of predicted exceedance was compared to the effectsdomain rather than an arbitrary study area boundary (such as arectangle drawn on a map) to determine the relative risk of eachemissions scenario. The effects domain approach allows for com-parisons across study areas of different sizes.

Kitimat: Soil base cation weathering is an important determi-nant of critical loads of acidity; soil data from 80 forest plots in theKitimat airshed were used to generate estimates of base cationweathering using the A2M solver (Posch and Kurz, 2007) and thePROFILE model (Sverdrup and Warfvinge, 1988; Warfvinge andSverdrup, 1992). Spatial prediction or regionalisation of soilweathering rates were carried out using established geostatistical

hest scenarios) examined in the Kitimat (n ¼ 12 scenarios) and Prince Rupert (n ¼ 7and deposition under the lowest and highest emissions scenario across the study

owestehighest scenario Prince rupert airshed lowestehighest scenario

3.3e7.834.0e90.0

41.4] 0.40e0.41 [22.6e22.7]13.4] 1.87e2.94 [98.5e101.4].4e1141.6] 2.59e2.68 [98.9e100.8]42.4] 3.14e4.55 [127.7e134.7]

O2 ¼ 1.07 mg m�3, and NO2 ¼ 4.3 mg m�3; Prince Rupert SO2 ¼ 4.0 mg m�3, and

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 315

mapping techniques (McBratney et al., 2003), i.e., regression-kriging following Hengl et al. (2004). Forests on mineral soilswere assessed for critical loads of acidity in the Kitimat airshed,covering approximately 65% of the terrestrial study area; while allnatural terrestrial ecosystems were assessed for nutrient N criticalloads (85% of the study area).

Prince Rupert: Detailed digital soil maps and soil geochemistrydata were not available for the Prince Rupert airshed; as such,bedrock geology was used as an indicator of soil parent material todefine acid sensitivity and weathering rates following theSkokloster classification (UNECE , 2004). This represented a so-called ‘Level 0’ semi-quantitative critical loads analysis asdescribed in the ICP Mapping Manual (UNECE , 2004) and theBritish Columbia Ministry of the Environment Critical LoadScreening Guidance for Acidification and Eutrophication of TerrestrialEcosystems (BC MOE, 2014c). Furthermore, peatlands (organic soils)were included in the Prince Rupert airshed assessment, contrib-uting ~17% to the 82% of the terrestrial study area assigned criticalloads of acidity.

2.5. Aquatic critical loads

Aquatic critical loads were calculated using two steady-statemodels: the Steady-State Water Chemistry (SSWC) model(Henriksen et al., 2002); and the First-order Acidity Balance (FAB)model (Henriksen and Posch, 2001, Aherne et al., 2004; UNECE ,2004; Posch et al., 2012). The calculated critical loads werecompared to S and N deposition estimates (only S for the SSWC) toassess the capacity of the aquatic ecosystems in the study area towithstand (buffer) the effects of acidic deposition. A critical con-centration limit of Acid Neutralising Capacity (ANC) equal to 26 meqL�1 was specified to protect biota based on a critical pH of 6, whichwas empirically derived from a titration curve using the best fit ofthe Small and Sutton (1986) equation to laboratory pH and GranANC data from 41 lakes in the Kitimat airshed. For the Prince Rupertstudy, this equation was modified to include a dissolved organiccarbon (DOC) term, since lakes in the Prince Rupert area had higherconcentrations of DOC (ESSA et al., 2016). The critical ANC for thePrince Rupert area was 23 meq L�1 at average DOC (10 mg L�1).

The exceedance of critical loads of acidity was estimated foreach lake under the lowest and highest emissions scenarios. Inaddition, the expected change in pH was estimated for eachsampled lake using a modification of the steady-state ESSA-DFOmodel (Marmorek et al., 1990; ESSA et al., 2013). Lake chemistrywas sampled for 80 lakes greater than 1 ha in the Kitimat airshedand for 35 lakes in the Prince Rupert airshed (Fig. 1); aquatic criticalloads were estimated only for these sampled lakes. The study lakesin the Kitimat area were a representative sample of all lakesreceiving moderate to high levels of predicted acidic depositionunder one or more scenarios (ESSA et al., 2014a), and also includedlakes receiving low levels of acidic deposition in areas with bedrocksensitive to acidification. In contrast, data for only 35 lakes wereavailable for the Prince Rupert airshed, many of which were rela-tively large in area, and located in areas receiving relatively lowlevels of acidic deposition (see Figs. 1 and 2; ESSA et al., 2016). ThePrince Rupert lake data set is not considered to be regionallyrepresentative, and may underestimate the risk of acidification.

In Prince Rupert the risk of eutrophication was assessed usingempirical critical loads for nutrient N compared to predicted Ndeposition for the 35 study lakes and all 859 lakes >1 ha in theentire study area. Two critical limits were examined for lakeeutrophication: 3 kg N ha�1 yr�1 (protective of dystrophic lakes)and 5 kg N ha�1 yr�1 (protective of oligotrophic lakes) (De Wit andLindholm, 2010). Dystrophic lakes were defined as those with apH < 6 and with organic anions constituting more than 50% of their

anion composition (ESSA et al., 2016). All other lakes were classifiedas oligotrophic. An equivalent assessment of lake eutrophicationwas not completed in the Kitimat airshed (owing to the proposedlow NO� emissions; Table 1). The lake acidification risk assessmentincluded a secondmetric (in addition to critical load exceedance) ofpredicted DpH � 0.3 pH units from baseline pH to inform riskcategorization based upon reported effects thresholds (Marmoreket al., 1990; F€olster et al., 2007; ESSA et al., 2013).

3. Results

3.1. Air dispersion modelling

Air dispersion modelling showed two distinctly different pat-terns in deposition between the study areas (Fig. 2). In the confinedKitimat airshed, the modelled plume under the highest emissionsscenario was a long and relatively narrow north-south ellipse withthe highest deposition in the central and western portions of thevalley (Fig. 2). In Prince Rupert, the modelled plume under thehighest emissions scenario was somewhat broader, with thegreatest deposition predicted in an isopleth oriented from south-east to northwest (Fig. 2). Both plumes reflect the prevalence ofsoutheasterly winds that result from the regular progression of lowpressure systems that migrate toward the Gulf of Alaska, which arecharacteristic of regional weather patterns (Lange, 2003). Thesewinds deflect both plumes in a northerly or northwesterly direc-tion. High pressure systems in association with winter outflowwinds can produce seasonal northerly and easterly winds, thoughgenerally for shorter duration. These seasonal winds result inplume dispersion to the southwest in Kitimat and west in PrinceRupert (Fig. 2). Modelled maximum S deposition ranged from 224to 1142 meq m�2 yr�1 in Kitimat, compared with 99e101 meq m�2

yr�1 in Prince Rupert under the lowest to highest emissions sce-narios (see Table 1), reflecting the higher SO2 emissions in Kitimatassociated with a smelter. In contrast, modelled maximum Ndeposition was higher in Prince Rupert (128e135 meq m�2 yr�1)compared with Kitimat (35e42 meq m�2 yr�1) due to the greaternumber of proposed LNG facilities. The effects domain wasconsiderably larger in Kitimat (1388 km2) than in Prince Rupert(586 km2; Table 2), owing to the greater magnitude of SO2 emis-sions in Kitimat (Table 1), and the spatial configuration of emissionssources and stack heights (closer sources and taller stacks in Kiti-mat than in Prince Rupert), in addition to the influence of localmeteorology and physical geography. In both airsheds, modelledconcentrations and deposition were very high close to existing andproposed sources, falling away with distance (see Fig. 2).

3.2. Exceedance of critical levels

In the Kitimat airshed, the critical level of SO2 at which sensitivelichens may be impacted was predicted to be exceeded in an arearanging from 81 to 251 km2 under the lowest to highest emissionsscenarios (Fig. 2, Table 2). In contrast, the NO2 critical level for theprotection of vegetation was not predicted to be exceeded underthe any of the twelve emissions scenarios in the Kitimat airshed.The SO2 critical level was predicted to be exceeded in an arearanging from 0 to 1 km2 in Prince Rupert, while the area predictedto exceed the NO2 critical level was 4.5e6 km2 (Fig. 2, Table 2).

3.3. Exceedance of terrestrial critical loads

Despite differences in the data sources for critical load modelinputs (see Tables SI-1 and SI-2), the average estimated criticalloads of acidity in Kitimat and Prince Rupert were similar (181 meqm�2 yr�1 and 219 meq m�2 yr�1 respectively; Table 2). However,

Fig. 2. Modelled total sulphur (top panels) and nitrogen (bottom panels) deposition (meq m�2 yr�1) in the Prince Rupert airshed (left panels) and Kitimat airshed (right panels)under the worst-case (highest) emissions scenarios (see Table 1). Annual concentration isopleths of sulphur dioxide (SO2 ¼ 5 and 10 mg m�3: top panels) and nitrogen dioxide(NO2 ¼ 20 and 30 mg m�3: bottom panels) are also shown. Note: SO2 and NO2 isopleths include a value for background air concentration (see footnote Table 1); in contrast, totalsulphur and nitrogen deposition do not include background values. See the dashed line in Fig. 1 for the location of mapped study areas.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323316

the ranges in values differed: in the Kitimat airshed, the criticalloads of acidity across the study area ranged from 59 to 305 meqm�2 yr�1, while in Prince Rupert they ranged from 42 to 964 meqm�2 yr�1 (Fig. 3, Table 2). Overall, much of the study regions had

moderate to high critical loads of acidity, and consequently mod-erate to low sensitivity to acidic deposition. In Kitimat, the area ofthe receptor ecosystem (forests) exceeded under twelve emissionsscenarios was low to moderate, ranging from 1 to 28 km2 (Table 2).

Table 2Statistical summaries for terrestrial and aquatic critical loads (see Figs. 3 and 4); and predicted exceedance of air concentration limits, and critical loads of acidity and nutrientnitrogen for the Kitimat and Prince Rupert airsheds. Range represents impacts (e.g., areal exceedance or number of lakes exceeded) under the lowest and highest emissionsscenarios (see Table 1).

Kitimat airshed Prince rupert airshed

Deposition ‘effects domain’ (km2) 1388 586Average base cation weathering ratea (meq m�2 yr�1) [regional range] 56.6 [18.7e113.5] 76.2 [19.4e393.1]Average critical loads of acidityb (CLmaxS; meq m�2 yr�1) [regional range] 181.1 [59.3e305.1] 218.7 [42.2e963.7]Average critical loads of nutrient nitrogen (CLnutN; meq m�2 yr�1) [regional range] 57.9 [12.3e167.4] 75.5 [38.1e252.7]Empirical critical loads of nutrient nitrogen (CLempN; meq m�2 yr�1) [kg ha�1 yr�1] 28.6e35.7 [4e5] 35.7 [5]Baseline pH range of sampled lakes (pH units) 4.50e7.98 4.84e7.03Average critical loads of acidity for lakes (CLmaxS; meq m�2 yr�1) [regional range] 263.3 [0e1696.5] 188.7 [25.4e807.5]Range in area of SO2 critical level (10 mg m�3) exceedance (km2) 80.6e250.6 0.2e0.8Range in area of NO2 critical level (30 mg m�3) exceedance (km2) 0e0 4.5e6.0Range in area of critical loads of acidity exceedance (km2) [%]c 0.56e27.94 [0.04e2.01] 4.04e10.27 [0.69e1.75]Range in area of critical loads of N(nut) exceedance (km2) [%]c 0.56e2.17 [0.04e0.16] 4.97e23.54 [0.97e4.02]Range in area of critical loads of N(emp) exceedance (km2) [%]c 0.76e31.14 [0.05e2.24] 19.57e93.82 [3.22e15.44]Range in number of lakes predicted to exceed the critical load of acidityd [% of number of lakes] 21e23 [26.3e28.8] 0e0Range in number of lakes predicted to exceed the critical load of acidityd excluding lakes with

high organic acids [% of number of lakes]5e7 [6.3e8.8] 0e0

Range in number of lakes predicted to exceed D pH 0.3 (%) 0e7 (0.0e8.8) 0e0Range in number of lakese predicted to exceed the dystrophic empirical critical load (3 kg N ha�1 yr�1)

for N [% of number of lakes]NA 17e25 [2e3]

Range in number of lakes predicted to exceed the oligotrophic empirical critical load (5 kg N ha�1 yr�1)for N [% of number of lakes]

NA 6e10 [1e1]

a The base cation weathering rates were calculated using soil chemistry data and the A2M solver with the PROFILE model for the Kitimat airshed and were derived from theSkokloster classes for the Prince Rupert airshed.

b The critical limits for ecosystem protection were set at Bc:Al ¼ 1 coniferous, and Bc:Al ¼ 6 deciduous forests.c The percent is expressed as a proportion of the effects domain, which is defined as the area enclosed by a 15 meqm�2 yr�1 modelled S and N deposition isopleth under the

maximum emissions scenario.d Excluding lakes with a pH below 6 in the absence of acidic deposition from industrial emissions.e This analysis included all 859 lakes in the study area. An equivalent analysis of Kitimat area lakes was not completed.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 317

The greatest areal exceedance represented <2.5% of the mappedreceptor ecosystem within the effects domain. Even though arelatively small area was predicted to be exceeded, the averageexceedance was high, ~50 meq m�2 yr�1 under all emissions sce-narios, indicating that a relatively small area of forested ecosystemson mineral soil received acidic deposition greatly in excess of thecritical load. With respect to N deposition, the exceeded area wasgreater using empirical critical loads (CLempN) compared with thenutrient mass balance model under all emissions scenarios(Table 2). In the Kitimat airshed, the largest mass balance N arealexceedancewas predicted to be 2 km2, while in contrast, the largestempirical critical load areal exceedance for nutrient N wasapproximately 31 km2 (Fig. 3, Table 2). In Prince Rupert, the area ofthe receptor ecosystems (forest, shrub and wetland) receivingacidic deposition in excess of the critical load ranged from 4 to10 km2 and even under the highest emissions, exceedance was <2%of the effects domain (Table 2). Similarly under all scenarios, theexceeded area was much greater (>4 times) for empirical N criticalloads compared with the nutrient N mass balance (Table 2). TheCLnutN exceeded area ranged from 5 to 24 km2 compared to20e94 km2 for empirically derived nutrient N exceedance (Table 2).

3.4. Exceedance of aquatic critical loads

The distribution of critical loads of acidity (CLmaxS) was lower(more sensitive) in the 80 lakes sampled in the Kitimat airshed thanin the 35 lakes sampled in the Prince Rupert area (Table 2, Fig. 4),which may partly reflect the differences in lake selection (as notedabove). One of the challenges with assessing the impacts of acidi-fication in both study areas was the prevalence of lakes with lownatural pH, typically owing to low base cation concentrations andhigh dissolved organic acids. Thirty of the 80 sampled lakes in theKitimat airshed had a current pH� 6.0. A range of 21e23 lakes werepredicted to exceed the critical load of acidity (Fig. 4) across thelowest to highest emissions scenarios (Table 1). When lakes with

naturally low pH owing to high concentrations of organic acidswere excluded, this dropped to 5e7 lakes (Table 2). Furthermore, atotal of 7 lakes in Kitimat were predicted to experience a change inpH of �0.3 pH units under the maximum emissions scenario.

No lakes in the Prince Rupert airshed were predicted to exceedthe critical load of acidity under the highest emissions scenario, andonly one lake had a predicted pH change as high as 0.29 pH units.Across the entire study domain, 17e25 of 859 lakes (up to 3%) werepredicted to exceed the dystrophic empirical critical load fornutrient N, while 6e10 lakes (up to 1%) were predicted to exceedthe oligotrophic empirical critical load (Table 2; see Fig. 4 for areareceiving � 3 kg N ha�1 yr�1; note that water chemistry was notavailable for these lakes). Of the 35 lakes sampled for waterchemistry, one was predicted to exceed the nutrient N critical loadunder the highest emissions scenario (3% of the sampled lakes);however, as discussed above, relatively few of the sampled lakeswere within the area of highest predicted deposition (Fig. 2).

4. Discussion

The predicted air concentrations in the Kitimat and PrinceRupert airsheds were comparable to other regions with industrialemissions in North America. The maximum annual concentrationspredicted for the Kitimat airshed (35e41 mg m�3 SO2; Table 1) werewithin the range of observations at backgroundmonitoring stationsin the United States (14e69 mg m�3 SO2; Bytnerowicz et al., 2013),but generally higher than ambient concentrations reported na-tionally and regionally across Canada (28 mg m�3 SO2; EnvironmentCanada, 2016). In contrast, predictedmaximumNO2 concentrationsin the Kitimat airshed (12e13 mg m�3 NO2) were lower thanambient concentrations from urban Canadian monitoring stations(20e38 mg m�3 NO2; Environment Canada, 2016; Reid and Aherne,2016). Predicted average NO2 concentrations in the Prince Rupertairshed (1.9e2.9 mg m�3 NO2; Table 1) were lower than most ruralambient air monitoring stations in Canada (3e10 mg m�3;

Fig. 3. Critical loads of acidity (CLmaxS: top panels) and nutrient nitrogen (CLnutN: bottom panels) for terrestrial ecosystems in the Prince Rupert airshed (left panels) and Kitimatairshed (right panels) mapped on a 1 km � 1 km grid resolution. White-filled squares denote grids with exceedance of their respective critical loads under the highest modelledtotal sulphur and nitrogen deposition (see Fig. 2). The ‘effects domain’ (top panels) and regions receiving greater than the nitrogen deposition limit (4 kg N ha�1 yr�1 for Kitimat and5 kg N ha�1 yr�1 for Prince Rupert; bottom panels) are also shown. The bottom panels delineate regions with exceedance of empirical critical loads of nutrient nitrogen (CLempN)under the worst-case scenario. See the dashed line in Fig. 1 for the location of mapped study areas.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323318

Environment Canada, 2016); however, the maximum values wereconsiderably higher (99e101 mgm�3 NO2; Table 1), more than threetimes greater than the critical level for protecting sensitive vege-tation albeit for a small area (�6 km2; Table 2).

Impacts from air emissions on vegetation in the Kitimat airshedhave been a concern for more than four decades, as monitoringstudies have shown evidence of effects to sensitive ecosystemcomponents (Bunce, 1978, 1984; Richards, 1986; Weinstein and

Fig. 4. Critical loads of acidity (CLmaxS) for lakes in the Prince Rupert airshed (west) and Kitimat airshed (east). White-filled circles denote lakes with exceedance of critical loadsunder the highest modelled sulphur and nitrogen deposition (see Fig. 2). The dashed line denotes regions (lakes) receiving � 3 kg N ha�1 yr�1, the empirical nutrient nitrogendeposition limit for dystrophic lakes, under the worst-case scenario.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 319

Davison, 2004). Visual injury to plants and tissue concentrationanalysis of fluoride and S have been monitored on at least a bian-nual basis since 1970 due to concerns related to hydrogen-fluoride(HF) and SO2 emissions from the existing smelter (Cambria Gordon,2011). The trend to date has been one of decreasing visible injury toplants in response to changes in operations and permitting(Cambria Gordon, 2011). Lichen studies in the 1970s and 1980s(Bunce, 1978; Richards, 1986) showed an area of impact consistentwith the area described by the highest emissions concentrationspredicted under our air dispersion modelling (Fig. 2). However, thechemical profile of emissions sources has changed substantiallyover time. During the past decade, operations ceased at a pulp andpaper mill and a gas plant, both of which emitted S in the Kitimatairshed; furthermore, the recent smelter modernization, presentlyin the commissioning phase, has decreased HF and increased SO2emissions. As such, it is likely that the observed patterns of impactsmay change to reflect the new emissions profile, supporting theneed for ongoing vegetation monitoring.

In contrast, visual surveys of vegetation, plant tissue analysis,and lichen and plant community monitoring studies have notoccurred in the Prince Rupert airshed, and little is known about

potential impacts from historic or current emissions. This repre-sents an important information gap for emissions management,especially considering the high concentrations of NO2 that may begenerated by proposed LNG developments.

Overall, the potential risk to terrestrial ecosystems broughtabout by elevated acidic (S and N) deposition in the regionappeared low to moderate, even under the highest emissions sce-narios. Base cation weathering rates ranged between 19 and 393meqm�2 yr�1 (average: 76meqm�2 yr�1) in the top 50 cm of soil inPrince Rupert and between 24 and 118 meq m�2 yr�1 (average: 57meq m�2 yr�1) in Kitimat. In general, weathering rates in the twoairsheds were similar to those in other forested regions in Canada:3e13 meq m�2 yr�1 in Nova Scotia (Whitfield et al., 2006); 58e446meq m�2 yr�1 in Quebec (Houle et al., 2012); 21e79 meq m�2 yr�1

in Ontario (Koseva et al., 2010); and 19e351meqm�2 yr�1 in BritishColumbia (Mongeon et al., 2010). High rainfall and associatedrunoff in the region (see Table SI-2) increased critical loads owing tohigh estimated critical ANC leaching. The estimated average criticalloads of acidity for forest soils in Kitimat (181 meq m�2 yr�1) andPrince Rupert (219 meq m�2 yr�1) were higher than the Canadianaverage (82 meq m�2 yr�1; median: 67 meq m�2 yr�1), but were

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323320

within the range of values reported from a Canada-wide compari-son (18e640 meq m�2 yr�1; Carou et al., 2008). The choice of anappropriate critical chemical limit is much debated (Løkke et al.,1996); in other regions of Canada a critical Bc:Al ratio of 10 hasbeen used, which is suggested to maintain soil base saturationabove 20%, but has not been linked with declines in tree health(Ouimet et al., 2006).

The areal exceedance of the empirical critical load for N waspredicted to be higher than potential acidification impacts in bothairsheds, especially in Prince Rupert where almost 100 km2 waspredicted to receive N deposition in excess of the critical load underthe highest emissions scenario. In contrast to many parts of Europeand eastern North America, much of northwest British Columbiahas not been impacted by air pollution and there is emerging evi-dence that even at low N deposition values, sensitive biota may beaffected (Emmett, 2007; Fenn et al., 2008; Geiser et al., 2010;Wilkins and Aherne, 2016). Empirical critical loads in this assess-ment were selected to recognize the uncertainty in the sensitivityof pristine environments. These low empirical critical loads werebased upon studies in the western US that suggest N deposition>3 kg N ha�1 yr�1 may drive community changes among epiphyticmacro-lichens in forests in low precipitation (<500 mm) areas(Geiser et al., 2010). However in areas with median precipitation(1860 mm), a CLempN of 5 kg N ha�1 yr�1 was proposed, and in highprecipitation (4510 mm) areas, a CLempN of 9.2 kg N ha�1 yr�1 wassuggested (Geiser et al., 2010). As such, to protect lichen commu-nities in coniferous and mixed forests in the study area, a CLempN of5 kg N ha�1 yr�1 was chosen for Prince Rupert and a CLempN of4 kg N ha�1 yr�1 was chosen for Kitimat based on a synthesis ofmultiple studies (Bobbink et al., 2015; Pardo et al., 2011; Blett et al.,2014). Given the historic dominance of S deposition across Canada,few studies have evaluated the impacts of N deposition; nonethe-less, the potential impacts of N deposition on forest ecosystemservices have been suggested (Aherne and Posch, 2013). In contrast,N deposition is viewed as one of the most important drivers ofbiodiversity change in natural ecosystems across Europe (reviewedby Bobbink and Hicks, 2014).

The Kitimat and Prince Rupert areas bear broad similarities towestern Atlantic Europe, where peatlands are common, precipita-tion is comparable (1500e3000 mm yr�1), and the climate iscoastal temperate. Aherne and Curtis (2003) found that 6.9% of 277sampled lakes in Ireland exceeded the critical load of acidity, whichis higher than predicted in the Prince Rupert airshed (0%) and lowerthan in the Kitimat airshed (26e29%; 18e20% across both studies).While S deposition in Ireland was higher than the highest emis-sions scenario in Kitimat and Prince Rupert, the average criticalloads of acidity for Irish lakes (478 meq m�2 yr�1) was approxi-mately two times higher than in Prince Rupert (189 meq m�2 yr�1)and Kitimat (263 meq m�2 yr�1). Studies in the United Kingdom(Curtis et al., 2000) and Nordic countries (Norway, Sweden andFinland; Henriksen et al., 1992), show levels of exceedance > 25% atthe national scale (proportionally similar to Kitimat) but underconsiderably higher S deposition.

Under the Canada-Wide Acid Rain Strategy for Post-2000(CCME, 2013), provinces have committed to restoring 95% ofimpacted lakes to below the critical load of acidity. The estimatedcritical loads of acidity to protect 95% of the lakes sampled underprovincial surveys (as summarised by Aherne and Jeffries, 2015)generally ranged from <5 meq m�2 yr�1 in northern Saskatchewanand Manitoba, 10 meq m�2 yr�1 in coastal southern BritishColumbia, 15e30 meq m�2 yr�1 in Eastern Canada, and up to 60meq m�2 yr�1 in northern Alberta, reflecting the range in regionalacid sensitivity, associated water chemistry and specified levels ofprotection. In comparison, critical loads of acidity to protect 95% ofthe lakes in the current study ranged from approximately 20 meq

m�2 yr�1 (Kitimat) to 40 meq m�2 yr�1 (Prince Rupert); however,restoring ecosystems impacted by acidification is not equivalent toprotecting pristine ecosystems, and allowing up to 5% of lakes to beimpacted may not be acceptable nor consistent with the Strategy(CCME, 2013).

Empirical critical loads for nutrient N reported by Bobbink andHettelingh (2011) for a wide range of ecosystems show that oligo-trophic and dystrophic lakes (3e10 kg N ha�1 yr�1), and blanketbogs (5e10 kg N ha�1 yr�1) have low critical loads. It is possible that2000e3000 mm (or greater) of annual precipitation in northwestBritish Columbia may result in regional critical loads that are in themiddle to high end of the reported range for these very sensitiveecosystems; however, lacking regionally specific empirical data, weused the lowest values as a conservative approach to account foruncertainties. It is notable that lower critical loads (~1.5 kg N ha�1

yr�1) have been proposed for mountain lakes in the western US(Saros et al., 2011; Nanus et al., 2012), and there are likely ecologicalanalogues for these extremely sensitive lakes in British Columbia,but not on the northwest coast with its high rainfall.

Bobbink et al. (2015) list six factors that determine the severityof impacts from atmospheric N deposition. They are, in brief, (i) theduration and total amount of inputs, (ii) the chemical and physicalform of N inputs, (iii) the sensitivities of the receptors, (iv) theabiotic conditions, including climate, (v) land use, and (vi) soil (andwater) biogeochemistry. Relatively little quantitative data exists tocharacterize these factors in northwest British Columbia. Oneexception is the NADP (2015) wet deposition chemistry data;annual deposition observed during 2014 was 0.5e1.3 kg N ha�1

yr�1 for the Kitimat airshed and 1.0 kg N ha�1 yr�1 for the PrinceRupert airshed, suggesting that the present risk of eutrophicationunder existing emissions may be low.

We approached the task of linking our analyses to emissionsregulation by developing decision support tools: a risk frameworkfor each receptor. The framework used predicted critical loadexceedances to define risk categories (Table 3) that could be used asthresholds for escalating management actions such as increasedmonitoring, or limiting and reducing emissions. For example, thesoils framework (Table 3) included a comparison of predicted ex-ceedance to the effects domain. One disadvantage of the effectsdomain approach is that as total emissions increase, the effectsdomain also increases, thereby expanding the area of acceptablepotential exceedance. Furthermore, the risk framework did notdifferentiate risk based upon where exceedances were predicted,for example, places with special land values such as parks andprotected areas.

While the critical levels and loads approach has been used toassess emissions across a range of industriesde.g., oil sandsdevelopment (Foster et al., 2001), coal fired power plants (Josipovicet al., 2011), and agriculture (SNIFFER, 2011), British Columbia is thefirst jurisdiction to formalize this approach to the environmentalimpact assessment of emissions from major LNG projects (BC MOE,2014b; 2014c, 2015b). Furthermore, critical loads analyses havebeen recently used in British Columbia in the development ofenforceable thresholds for monitoring and emissions reduction in arecent smelter air permit (ESSA et al., 2014b), which represents aninnovation in smelter permitting. These and other efforts haveestablished critical loads as an important science-based tool for airquality management, not only at the national and internationalscale, but also for ensuring protection of the environment at thescale of individual airsheds through the regulation of industrial airpermits.

The emissions scenarios examined in this study included a rangeof potential control technologies and future development out-comes. Full build-out of all projects proposed for the Kitimat andPrince Rupert airsheds (represented by the highest emissions

Table 3Risk categories and definitions for terrestrial ecosystems.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 321

scenarios) may overestimate the risk as industry analysts suggestthat British Columbia could support no more than two to threelarge LNG export terminal facilities (Hughes, 2015; Stastny, 2016).Furthermore, existing permitted emissions were modelled at fullpermit limit levels rather than at actual operational discharges,which are typically lower. Nonetheless, to ensure long-termecosystem protection it is important that critical level and loadassessments include conservative assumptions that representworst-case scenarios to accommodate uncertainties in modelapplication and ecosystem response.

5. Conclusions

The determination of critical levels and loads provided ascreening level assessment of the risks associated with emissionsfrom proposed industrial developments in northwest BritishColumbia. Exceedances were predicted, indicating that airshedmanagement, emissions regulation, and air quality monitoring inthe region need to consider the risks of direct effects from exposureto SO2 and NO2 and indirect effects from acidification and eutro-phication. To reduce uncertainties in exceedance estimates, re-finements of critical loads estimates need to be supported by thecollection of additional soils and surface water data to quantifyregional sensitivity. Similarly, continued monitoring of wet depo-sition and air concentrations is required to evaluate regional airdispersion models. Moreover, empirical studies to determineregionally specific critical thresholds for pristine ecosystems wouldimprove model assumptions. Into the future, it may be necessary toassess the time for predicted impacts to occur in the Kitimat andPrince Rupert airsheds, which will require process-based dynamicmodelling. Irrespective, under proposed emissions scenarios it isclear that current airshed management in northwest BritishColumbia, Canada should use an effects-based approach to informthe regulation of industrial emissions including those from pro-posed LNG export facilities. Indeed, the environmental assessmentof LNG facilities worldwide would benefit from critical levels andloads analyses to ensure long-term protection of adjacentecosystems.

The Canada-Wide Acid Rain Strategy for Post-2000 set a long-term goal of achieving the threshold of critical loads for acidicdeposition across Canada (CCME, 2013). An important principle ofthis strategy is ‘keeping clean areas clean’, whichmeans preventingpristine areas from exceeding critical levels and loads. The criticallevels and loads analyses presented here are science-based toolsthat support the management of emissions in the Kitimat andPrince Rupert airsheds to meet the intent of this principle.

Acknowledgements

Frazer McKenzie, from the British Columbia Ministry of Envi-ronment, encouraged the first use of the critical loads approach innorthwest British Columbia. Ian Sharpe, recognized the value ofthis approach and initiated the first government-led critical loadsassessment in the regiondthe results are presented in this study.Support for this work was provided by Ed Hoffman, Ann Godon,Dennis Fudge, BenWeinstein andWarrenMcCormick. The researchalso benefited from review comments provided by Chris Perrin,Uwe Gramann, Rock Ouimet, Tim Sullivan, Dean Jeffries, AllanLegge, Sirkku Manninen, Doug Maynard, Pat Shaw, Max Posch, andtwo anonymous reviewers. Environment Canada, British ColumbiaMinistry of Environment, Prince Rupert LNG, and Rio TintoAluminum provided lake chemistry data used in this assessment.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2016.08.058.

References

Aber, J., McDowell, W., Nadelhoffer, K., Magill, A., Bertson, G., Kamakea, M.,McNulty, S., Currie, W., Rustad, L., Fernandez, I., 1998. Nitrogen saturation intemperate forest ecosystems. BioScience 48, 921e934.

Aherne, J., Curtis, C.J., 2003. Critical loads of acidity for Irish lakes. Aquat. Sci. 65,21e35.

Aherne, J., Posch, M., Dillon, P.J., Henriksen, A., 2004. Critical loads of acidity forsurface waters in south-central Ontario, Canada: regional application of thefirst-order acidity balance (FAB) model. Water, Air Soil Pollut. Focus 4, 25e36.

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323322

Aherne, J., Posch, M., 2013. Impacts of nitrogen and sulphur deposition on forestecosystem services in Canada. Curr. Opin. Environ. Sustain. 5, 108e115.

Aherne, J., Jeffries, D., 2015. Critical load assessments and dynamic model appli-cations for lakes in North America (Chapter 19). 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, 25. Springer, Dor-drecht, Netherlands, pp. 485e504. Environmental Pollution.

Banner, A., LePage, P., Moran, J., de Groot, A. (Eds.), 2005. The HyP3 Project: Pattern,Process, and Productivity in Hypermaritime Forests of Coastal British Columbiae a Synthesis of 7-year Results. British Columbia Ministry of Forests. ResearchBranch, Victoria, p. 142. Special Report 10.

BC MOE (British Columbia Ministry of Environment), 2014a. British ColumbiaInterim Air Quality Objectives. Updated October 21, 2014. http://www.bcairquality.ca/reports/pdfs/aqotable.pdf.

BC MOE (British Columbia Ministry of Environment), 2014b. Air Emissions ImpactAssessment for Liquefied Natural Gas Export Terminal Facilities: GeneralGuidance for Conducting Environmental and Human Health Effects Assess-ment. November 21, 2014.

BC MOE (British Columbia Ministry of Environment), 2014c. Air Emissions ImpactAssessment for Liquefied Natural Gas Export Terminal Facilities: Critical LoadScreening Guidance for Acidification and Eutrophication of Terrestrial Ecosys-tems. November 21, 2014.

BC MOE (British Columbia Ministry of Environment), 2015a. British Columbia AirQuality Dispersion Modelling Guideline, p. 104. November 2015.

BC MOE (British Columbia Ministry of Environment), 2015b. Air emissions impactassessment for liquefied natural gas export terminal facilities: guidance for theassessment of acidification and eutrophication of aquatic ecosystems. June 25,2015.

Blett, T.F., Lynch, J.A., Pardo, L.H., Huber, C., Haeuber, R., Pouyat, R., 2014. FOCUS: apilot study for national-scale critical loads development in the United States.Environ. Sci. Policy 38, 225e236.

British Columbia Ministry of Natural Gas Development, 2015. http://engage.gov.bc.ca/lnginbc/.

Bunce, H.W.F., 1978. Lichen Surveys 1975-1977, p. 11. Kitimat Valley, B.C. A report tothe Aluminum Company of Canada Limited by Reid Collins and Associates.

Bunce, H.W.F., 1984. Fluoride emissions and forest survival, growth and regenera-tion. Environ. Pollut. Ser. A) 35, 169e188.

Bobbink, R., Hettelingh, J.-P. (Eds.), 2011. Review and Revision of Empirical CriticalLoads and Dose-response Relationships: Proceedings of an Expert Workshop,Noordwijkerhout, 23e25 June 2010. National Institute for Public Health and theEnvironment, Bilthoven, The Netherlands, p. 243. RIVM Report 680359002.http://www.rivm.nl/en/Documents_and_publications/Scientific/Reports/2011/mei/Review_and_revision_of_empirical_critical_loads_and_dose_response_relationships_Proceedings_of_an_expert_workshop_Noordwijkerhout_23_25_June_2010.

Bobbink, R., Hicks, W.K., 2014. Factors affecting nitrogen deposition impacts onbiodiversity: an overview (Chapter 14). In: Sutton, M.A., Mason, K.E.,Sheppard, L.J., Sverdrup, H., Haeuber, R., Hicks, W.K. (Eds.), Nitrogen Deposition,Critical Loads and Biodiversity. Springer, Dordrecht, The Netherlands,pp. 127e138.

Bobbink, R., Tomassen, H., Weijters, M., van den Berg, L., Strengbom, J., Braun, S.,Nordin, A., Schütz, K., Hettelingh, J.-P., 2015. Effects of empirical critical loads ofnitrogen for Europe (Chapter 4). In: de Vries, W., Hettelingh, J.-P., Posch, M.(Eds.), Critical Loads and Dynamic Risk Assessments: Nitrogen, Acidity andMetals in Terrestrial and Aquatic Ecosystems. Springer, Dordrecht, TheNetherlands, pp. 85e128. Environmental Pollution 25.

Bytnerowicz, A., Fenn, M., McNulty, S., Yuan, F., Pourmokhtarian, A., Driscoll, C.,Meixner, T., 2013. Interactive effects of air pollution and climate change onforest ecosystems in the United States: current understanding and future sce-narios (Chapter 16). In: Global Dimension of Air Pollution as Part of ClimateChange. Developments in Environmental Science, vol. 13. Elsevier, Amsterdam,The Netherlands, pp. 333e369.

Cape, J.N., 1993. Direct damage to vegetation caused by acid rain and polluted cloud:definition of critical levels for forest trees. Environ. Pollut. 82, 167e180.

Cambria Gordon, 2011. 2010 Vegetation Report: Annual Vegetation Inspection,Monitoring, and Assessment Program. A report to: Rio Tinto Alcan. February2011.

Carou, S., Dennis, I., Aherne, J., Ouimet, R., Arp, P.A., Watmough, S.A., DeMerchant, I.,Shaw, M., Vet, B., Bouchet, V., Moran, M., 2008. A National Picture of AcidDeposition Critical Loads for Forest Soils in Canada. Canadian Council of Min-isters of the Environment, p. 6. Report PN 1412. http://www.ccme.ca/files/Resources/air/acid_rain/national_picture_acid_deposition_pn1412.pdf.

CCME (Canadian Council of Ministers of the Environment), 1999. Canadian nationalambient air quality objectives: process and status. In: Canadian environmentalQuality Guidelines. Canadian Council of Ministers of the Environment, Winni-peg. http://ceqg-rcqe.ccme.ca/download/en/133/.

CCME (Canadian Council of Ministers of the Environment), 2013. 2010e2011Progress Report on the Canada-Wide Acid Rain Strategy for Post-2000, p. 35.http://www.ccme.ca/files/Resources/air/acid_rain/pn_1490_ar_2010_2011_prgrs_rpt_1.0_e.pdf.

Clague, J.J., 1985. Deglaciation of the Prince Rupert e Kitimat area, british Columbia.Can. J. Earth Sci. 22, 256e265.

Cosby, B.J., Hornberger, G.M., Galloway, J.N., Wright, R.F., 1985. Time scales ofcatchment acidification. Environ. Sci. Technol. 19, 1144e1149.

COSEWIC, 2006. COSEWIC Assessment and Update Status Report on the Cryptic

Paw Nephroma Occultum in Canada. Committee on the Status of EndangeredWildlife in Canada, Ottawa, p. 28 vi þ. http://www.sararegistry.gc.ca/virtual_sara/files/cosewic/sr_cryptic_paw_lichen_e.pdf.

COSEWIC, 2010. COSEWIC Assessment and Update Status Report on the OldgrowthSpecklebelly Pseudocyphellaria Rainierensis in Canada. Committee on the Statusof Endangered Wildlife in Canada, Ottawa, p. 38 vi þ. http://www.sararegistry.gc.ca/virtual_sara/files/cosewic/sr_Oldgrowth%20Specklebelly_0810_e.pdf.

Curtis, C., Allott, T., Hall, J., Harriman, R., Helliwell, R., Hughes, M., Kernan, M.,Reynolds, B., Ullyett, J., 2000. Critical loads of sul- phur and nitrogen for freshwaters in Great Britain and assessment of deposition reduction requirementswith the First-order Acidity Balance (FAB) model. Hydrology Earth Syst. Sci. 4,125e140.

De Wit, H.A., Lindholm, M., 2010. Nutrient Enrichment Effects of Atmospheric NDeposition on Biology in Oligotrophic Surface Waters e a Review. NorwegianInstitute for Water Research, p. 39. ICP Waters report 101/2010.

Emmett, B.A., 2007. Nitrogen saturation of terrestrial ecosystems: some recentfindings and their implications for our conceptual framework. Water, Air, SoilPollut. Focus 7, 99e109.

European Environment Agency (EEA), 2013. Air Quality in Europed2013 Report,p. 107. http://www.eea.europa.eu/publications/air-quality-in-europe-2013.

Environment Canada, 2015. Canadian Climate Normals: 1981e2010. AccessedNovember 23, 2015. climate.weather.gc.ca/climate_normals.

Environment Canada, 2016. Canadian Environmental Sustainability Indicators.Accessed January 15, 2016. http://www.ec.gc.ca/indicateurs-indicators/.

ESSA Technologies, Laurence, J., Limnotek, Risk Sciences International, Rio TintoAlcan, Trent University, Trinity Consultants, and University of Illinois, 2013.Sulphur Dioxide Technical Assessment Report in Support of the 2013 Applica-tion to Amend the P2-0001 Multimedia Permit for the Kitimat ModernizationProject, vol. 2. Final Technical Report. Prepared for Rio Tinto Alcan, Kitimat, B.C.450. http://www.riotintobcoperations.com/wp-content/uploads/STAR_Volume2_Apr-10-2013.pdf.

ESSA Technologies, Laurence, J., Limnotek, Risk Sciences International, Trent Uni-versity, and Trinity Consultants, 2014a. Kitimat Airshed Emissions EffectsAssessment. Report prepared for. British Columbia Ministry of Environment,Smithers, B.C, p. 205. appendices. http://www.bcairquality.ca/airsheds/docs/ESSA-Kitimat-Airshed-Report_20140425.pdf.

ESSA Technologies, Laurence, J., Limnotek, Risk Sciences International, Trent Uni-versity, and Trinity Consultants, 2014b. Kitimat Modernization Project SulphurDioxide Environmental Effects Monitoring Program. Rio Tinto Alcan, Vancouver,B.C, p. 102. Report prepared for. http://www.riotintobcoperations.com/wp-content/uploads/EEM_Program_Plan_141007.pdf.

ESSA Technologies, Laurence, J., Limnotek, Risk Sciences International, Trent Uni-versity, and Trinity Consultants, 2016. Prince Rupert Airshed Study. Reportprepared for. British Columbia Ministry of Environment, Smithers, B.C, p. 235(appendices).

EU (European Union). 2008. Directive 2008/50/EC of the European Parliament andof the Council of 21 May 2008 on ambient air quality and cleaner air for Europe(OJ L 152, 11.6.2008, pp. 1e44). http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri¼CELEX:32008L0050&from¼EN.

Fenn, M.E., Jovan, S., Yuan, F., Geiser, L., Meixner, T., Gimeno, B.S., 2008. Empiricaland simulated critical loads for nitrogen deposition in California mixed coniferforests. Environ. Pollut. 155, 492e511.

Foster, K.R., McDonald, K., Eastlick, K., 2001. Development and application of critical,target and monitoring loads for the management of acid deposition in Alberta,Canada. Water, Air, Soil Pollut. Focus 1, 135e151.

F€olster, J., Andr�en, C., Bishop, K., Buffam, I., Corey, N., Goedkoop, W., Holmgren, K.,Johnson, R., Laudon, H., Wilander, A., 2007. A novel environmental quality cri-terion for acidification in Swedish lakes e an application of studies on therelationship between biota and water chemistry. Water, Air, Soil Pollut. 7,331e338.

Geiser, L.H., Jovan, S.E., Glavich, D.A., Porter, M., 2010. Lichen-based critical loads fornitrogen deposition in western Oregon and Washington forests, USA. Environ.Pollut. 158, 2412e2421.

Hengl, T., Heuvelink, G., Stein, A., 2004. A generic framework for spatial predictionof soil variables based on regression-kriging. Geoderma 122 (1e2), 75e93.

Houle, D., Lamoureux, P., B�elanger, N., Bouchard, M., Gagnon, C., Couture, S.,Bouffard, A., 2012. Soil weathering rates in 21 catchments of the CanadianShield. Hydrology Earth Syst. Sci. 16, 685e697.

Henriksen, A., K€am€ari, J., Posch, M., Wilander, A., 1992. Critical loads of acidity:Nordic surface waters. Ambio 21 (5), 356e363.

Henriksen, A., Posch, M., 2001. Steady-state models for calculating critical loads ofacidity for surface waters. Water, Air Soil Pollut. Focus 1 (1e2), 375e398.

Henriksen, A., Dillon, P.J., Aherne, J., 2002. Critical loads of acidity for surface watersin south-central Ontario, Canada: regional application of the Steady-StateWater Chemistry (SSWC) model. Can. J. Fish. Aquat. Sci. 59, 1287e1295.

Horsley, S.B., Long, R.P., Bailey, S.W., Hall, T.J., 2000. Factors associated with thedecline of sugar maple on the Allegheny Plateau. Can. J. For. Res. 30, 1365e1378.

Hughes, D.J., 2015. A Clear Look at BC LNG: Energy Security, Environmental Impli-cations and Economic Potential. Canadian Centre for Policy Alternative. May2015.

Johnson, D., Lalonde, K., McEachern, M., Kenney, J., Mendoza, G., Buffin, A., Rich, K.,2011. Improving cumulative effects assessment in Alberta: regional strategicassessment. Environ. Impact Assess. Rev. 31 (5), 481e483.

Josipovic, J., Annegarn, H.J., Kneen, M.A., Pienaar, J.J., Piketh, S.J., 2011. Atmophericdry and wet deposition of sulphur and nitrogen species and assessment of

P. Williston et al. / Atmospheric Environment 146 (2016) 311e323 323

critical loads of acidic deposition exceedance in South Africa. South Afr. J. Sci.107 (3/4), 10.

Koseva, I.S., Watmough, S.A., Aherne, J., 2010. Estimating base cation weatheringrates in Canadian forest soils using a simple texture-based model. Biogeo-chemistry 101 (1e3), 183e196.

Krzyzanowski, J., Innes, J., 2010. Back to the basics e estimating the sensitivity offreshwater to acidification using traditional approaches. J. Environ. Manag. 91,1227e1236.

Lange, O.S., 2003. The Veil of Chaos: Living withWeather along the British ColumbiaCoast. Environment Canada, p. 198.

Likens, G.E., Driscoll, C.T., Buso, D.C., Siccama, T.G., Johnson, C.E., Lovett, G.M.,Fahey, T.J., Reiners, W.A., Ryan, D.F., Martin, C.W., Bailey, S.W., 1998. Thebiochemistry of calcium at Hubbard. Biogeochemistry 41, 89e173.

Løkke, H., Bak, J., Falkengren-Grerup, U., Finlay, R.D., Ilvesniemi, H., Nygaard, P.H.,Starr, M., 1996. Critical loads of acidic deposition for forest soils: is the currentapproach adequate? Ambio 25, 510e516.

Mansfield, T.A., Whitmore, M.E., Pande, P.C., Freer-Smith, P.H., 1985. Responses ofherbaceous and woody plants to the dry deposition of SO2 and NO2. In:Hutchinson, T.C., Meema, K.M. (Eds.), Effects of Atmospheric Pollutants onForests, Wetlands and Agricultural Ecosystems. Series G, Ecological Sciences,vol. 16, pp. 131e144.

Marmorek, D.R., Jones, M.L., Minns, C.K., Elder, F.C., 1990. Assessing the potentialextent of damage to inland lakes in Eastern Canada due to acidic deposition:development and evaluation of a simple “site” model. Can. J. Fish. Aquat. Sci. 47,55e66.

McBratney, A.B., Mendonça Santos, M.L., Minasny, B., 2003. On digital soil mapping.Geoderma 117, 3e52.

Mongeon, A., Aherne, J., Watmough, S.A., 2010. Steady-state critical loads of acidityfor forest soils in the Georgia Basin, British Columbia. J. Limnol. 69 (1), 193e200.

NADP (National Atmospheric Deposition Program), 2015. National AtmosphericDeposition Program 2014 Annual Summary. Illinois State Water Survey, Uni-versity of Illinois at Urbana-Champaign, IL. NADP Data Report 2015-01.

Nanus, L., Clow, D.W., Saros, J.E., Stephens, V.C., Campbell, D.H., 2012. Mappingcritical loads of nitrogen deposition for aquatic ecosystems in the RockyMountains, USA. Environ. Pollut. 166, 125e135.

Nasr, M., Castonguay, M., Ogilvie, J., Raymond, B.A., Arp, P.A., 2010. Modelling andmapping critical loads and exceedances for the Georgia Basin, British Columbia,using a zero base-cation depletion criterion. J. Limnol. 69 (1), 181e192.

Ouimet, R., Arp, P.A., Watmough, S.A., Aherne, J., Demerchant, I., 2006. Determi-nation and mapping critical loads of acidity and exceedances for upland forestsoils in eastern Canada. Water, Air, Soil Pollut. 172, 57e66.

Pardo, L.H., Fenn, M., Goodale, C.L., Geiser, L.H., Driscoll, C.T., Allen, E., Baron, J.,Bobbink, R., Bowman, W.D., Clark, C., Emmett, B., Gilliam, F.S., Greaver, T.,Hall, S.J., Lilleskov, E.A., Liu, L., Lynch, J., Nadelhoffer, K., Perakis, S., Robin-Abbott, M.J., Stoddard, J., Weathers, K., Dennis, R.L., 2011. Effects of nitrogendeposition and empirical nitrogen critical loads for ecoregions of the UnitedStates. Ecol. Appl. 21, 3049e3082.

Posch, M., Kurz, D., 2007. A2MdA program to compute all possible mineral modesfrom geochemical analyses. Comput. Geosci. 33 (4), 563e572.

Posch, M., Aherne, J., Forsius, M., Rask, M., 2012. Past, present and future exceedanceof critical loads of acidity for surface waters in Finland. Environ. Sci. Technol. 46,4507e4514.

Prince Rupert Port Authority, 2015. Port of Prince Rupert Annual Report 2014.Prince Rupert Port Authority, Prince Rupert, Canada, p. 8. http://www.rupertport.com/documents/2014-annual-report/pdf.

Rafaj, P., Amann, M., Siri, J., Wuester, H., 2014. Changes in European greenhouse gasand air pollutant emissions 1960e2010: decomposition of determining factors.Clim. Change 124, 477e568.

Reid, H., Aherne, J., 2016. Staggering reductions in atmospheric nitrogen dioxideacross Canada in response to legislated transportation emissions reductions.Atmos. Environ. 146, 243e251.

Reuss, J.O., Johnson, D.W., 1986. Acid Deposition and the Acidification of Soils andWaters. Ecological Studies, vol. 59. Springer, Dordrecht, The Netherlands, p. 118.

Richards, J., 1986. Lichens Surveys 1979-1984, Kitimat Valley, B.C, p. 40. A report tothe Aluminum Company of Canada Limited by Reid Collins and Associates.

Richardson, D.H.S., Cameron, R.P., 2004. Cyanolichens: their response to pollutionand possible management strategies for their conservation in northeasternNorth America. Northeast. Nat. 11, 1e22.

Saros, J.E., Clow, D.W., Blett, T., Wolfe, A.P., 2011. Critical nitrogen deposition loads inhigh-elevation lakes of the western US inferred from paleolimnological records.Water, Air, Soil Pollut. 216 (1), 193e202.

Small, M.J., Sutton, M.C., 1986. A regional pH-alkalinity relationship. Water Res. 20,335e343.

SNIFFER (Scotland and Northern Ireland Forum for Environmental Research), 2011.Development of Site Relevant Critical Loads for APIS. SNIFFER Project ER04,p. 35.

Soil Classification Working Group, 1998. The Canadian System of Soil Classification,third ed. Agriculture and Agri-Food Canada. Publication 1646, p. 187 http://sis.agr.gc.ca/cansis/publications/manuals/1998-cssc-ed3/cssc3_manual.pdf.

Stastny, R.P., 2016. Oilweek. Canada's LNGWindow Is Closing. Published January 14,2016. http://www.oilweek.com/index.php/701-canada-s-lng-window-is-closing.

Strang, D., Aherne, J., Shaw, D.P., 2010. The hydrochemistry of high-elevation lakesin the Georgia Basin, British Columbia. J. Limnol. 69 (1), 56e66.

Sverdrup, H., Warfvinge, P., 1988. Weathering of primary silicate minerals in thenatural soil environment in relation to a chemical weathering model. Water,Air, Soil Pollut. 38, 387e408.

Sverdrup, H., Warfvinge, P., 1993. The Effects of Soil Acidification on the Growth ofTrees, Grass and Herbs as Expressed by the (CaþMgþK)/Al Ratio. Department ofChemical Engineering II, Lund University, Lund, Sweden. Reports in Ecology andEnvironmental Engineering 1993:2.

UNECE (United Nations Economic Commission for Europe), 2004. Manual onMethodologies and Criteria for Modelling and Mapping Critical Loads andLevels and Air Pollution Effects, Risks and Trends. UNECE Convention on Long-range Transboundary Air Pollution, ICP Modelling and Mapping, p. 254. www.icpmapping.org/Mapping_Manual.

US EPA (United States Environmental Protection Agency), 2008. Final Report: In-tegrated Science Assessment (ISA) for Oxides of Nitrogen and Sulfur e

Ecological Criteria. United States Environmental Protection Agency, Washing-ton, DC. EPA/600/R-08/082F. December 2008.

US EPA (United States Environmental Protection Agency), 2011. Policy Assessmentfor the Review of the Secondary National Ambient Air Quality Standards forOxides of Nitrogen and Oxides of Sulfur. EPA-452/R-11e005a. February 2011.

US EPA (United States Environmental Protection Agency), 2012. National AmbientAir Quality Standards (NAAQS). http://www3.epa.gov/ttn/naaqs/criteria.html.

Vitt, D.H., Wieder, K., Halsey, L.A., Turetsky, M., 2003. Response of Sphagnum fuscumto nitrogen deposition: a case study of ombrogenous peatlands in Alberta,Canada. Bryologist 106 (2), 235e245.

Warfvinge, P., Sverdrup, H., 1992. Calculating critical loads of acid deposition withPROFILE - a steady-state soil chemistry model. Water, Air Soil Pollut. 63,119e143.

Weinstein, L.H., Davison, A., 2004. Fluorides in the Environment: Effects on Plantsand Animals. CABI Publishing, Wallingford, UK, p. 287.

Whitfield, C.J., Watmough, S.A., Aherne, J., Dillon, P.J., 2006. A comparison ofweathering rates for acid-sensitive catchments in Nova Scotia, Canada and theirimpact on critical load calculations. Geoderma 136, 899e911.

WHO (World Health Organization), 2000. Air Quality Guidelines for Europe, seconded (CD ROM version). (Chapter 10). Effects of sulfur dioxide on vegetation:critical levels. Available at: http://www.euro.who.int/data/assets/pdf_file/0016/123091/AQG2ndEd_10effso2.pdf.

Wilkins, K., Aherne, J., 2016. Vegetation community change points suggest thatcritical loads for nutrient nitrogen deposition are too high. Atmos. Environ. 146,315e322.

Wulff, A., Crossley, A., Sheppard, L.J., 1996. Fine structure of acid mist treated Sitkaspruce needles: open-top chamber and field experiments. Ann. Bot. 77, 1e10.