increasing arsenic concentrations in runoff from 12 small forested catchments (czech republic,...

9
Increasing arsenic concentrations in runoff from 12 small forested catchments (Czech Republic, Central Europe): Patterns and controls Martin Novak a, , Lucie Erbanova a , Daniela Fottova a , Petra Voldrichova a , Eva Prechova a , Vladimir Blaha a , Frantisek Veselovsky a , Michael Krachler b a Department of Geochemistry, Czech Geological Survey, Geologicka 6, 152 00 Prague 5, Czech Republic b European Commission Joint Research Centre, Institute for Transuranium Elements (ITU), P.O. Box 2340, D-76125 Karlsruhe, Germany abstract article info Article history: Received 7 January 2010 Received in revised form 9 April 2010 Accepted 11 April 2010 Available online 21 May 2010 Keywords: Arsenic Catchment Runoff Acidication reversal Climate change The 40-year long period of heavy industrialization in Central Europe (19501990) was accompanied by burning of arsenic-rich lignite in thermal power plants, and accumulation of anthropogenic arsenic in forest soils. There are fears that retreating acidication may lead to arsenic mobilization into drinking water, caused by competitive ligand exchange. We present monthly arsenic concentrations in surface runoff from 12 headwater catchments in the Czech Republic for a period of 13 years (19962008). The studied area was characterized by a northsouth gradient of decreasing pollution. Acidication, caused mainly by SO x and NO x emissions from power plants, has been retreating since 1987. Between 1996 and 2003, maximum arsenic concentrations in runoff did not change, and were b 1 ppb in the rural south and b 2 ppb in the industrial north. During the subsequent two years, 20042005, maximum arsenic concentrations in runoff increased, reaching 60% of the drinking water limit (10 ppb). Starting in 2006, maximum arsenic concentrations returned to lower values at most sites. We discuss three possible causes of the recent arsenic concentration maximum in runoff. We rule out retreating acidication and a pulse of high industrial emission rates as possible controls. The pH of runoff has not changed since 1996, and is still too low (b 6.5) at most sites for an AsOH - ligand exchange to become signicant. Elevated arsenic concentrations in runoff in 20042005 may reect climate change through changing hydrological conditions at some, but not all sites. Dry conditions may result in elevated production of DOC and sulfur oxidation in the soil. Subsequent wet conditions may be accompanied by acidication leading to faster dissolution of arsenic-bearing suldes, dissolution of arsenic- bearing Fe-oxyhydroxides, and elevated transport of arsenic sorbed on organic matter. Anaerobic domains exist in normally well-aerated upland soils for hours-to-days following precipitation events. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Detrimental effects of bioavilable forms of arsenic on human health are well known (Haffert and Craw, 2008; Telford et al., 2009). Anthropogenic arsenic, originating from both industry and agricul- ture, has increasingly affected ecosystem health, especially after the beginning of the Industrial Revolution (Ozdilek et al., 2007; Nkongolo et al., 2008). Arsenic pollution peaked in the second half of the 20th century in most industrial countries (Clemente et al., 2008; Kim et al., 2009). Operation times of major arsenic polluters on the territory of today's Czech Republic varied (Novak et al., 2008). Chronologically, arsenic pollution was rst related to gold mining (since Middle Ages), then to coal burning (since 1860), to processing of anmal hides in tanneries (since ca. 1880), and most recently to pesticide application (since 1950). Local arsenic pollution resulted from gold mining in Roudne (SE of Prague; early 20th century), Jilove (S of Prague, ca. 19501980) and Zlate Hory (E of Prague, mid-20th century), and from base-metal mining in Kutna Hora (E of Prague, before 1980; Moravek et al., 1995). In the 1980s, new low-grade, arsenic-rich gold deposits were discovered S of Prague. Transport of ores to processing plants by heavy trucks was associated with dispersion of arsenic-containing dust in an area of about 50 km 2 . The small-scale mining operations were discontinued before 1990 (Drahota and Filippi, 2009; Drahota et al., 2006, 2009). Between 1950 and 1990, the centrally-planned Czech economy was characterized by environment-unfriendly heavy industries (Paces, 1985). Large production of lignite in northern Czech Republic, Saxony/Lusatia (Germany), and Silesia (Poland) caused much more widespread arsenic pollution than gold mining, industrial processing of animal hides, and farming (Bezacinsky, 1981; Benes et al., 1984; Erbanova et al., 2008). Emissions of arsenic were coupled with those of acidifying forms of sulfur (Novak et al., 2007). Sulfur emissions from power plants peaked in 1987. After 1990, both sulfur and arsenic emissions from 30 thermal power stations in the region dropped rapidly (Hejma, 1992; Minarik and Dousova, 1993). Heavy Science of the Total Environment 408 (2010) 36143622 Corresponding author. Tel.: + 420 251816540; fax: + 420 251818748. E-mail address: [email protected] (M. Novak). 0048-9697/$ see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2010.04.016 Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Upload: martin-novak

Post on 12-Sep-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Science of the Total Environment 408 (2010) 3614–3622

Contents lists available at ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r.com/ locate /sc i totenv

Increasing arsenic concentrations in runoff from 12 small forested catchments(Czech Republic, Central Europe): Patterns and controls

Martin Novak a,⁎, Lucie Erbanova a, Daniela Fottova a, Petra Voldrichova a, Eva Prechova a, Vladimir Blaha a,Frantisek Veselovsky a, Michael Krachler b

a Department of Geochemistry, Czech Geological Survey, Geologicka 6, 152 00 Prague 5, Czech Republicb European Commission Joint Research Centre, Institute for Transuranium Elements (ITU), P.O. Box 2340, D-76125 Karlsruhe, Germany

⁎ Corresponding author. Tel.: +420 251816540; fax:E-mail address: [email protected] (M. Novak

0048-9697/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.scitotenv.2010.04.016

a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 January 2010Received in revised form 9 April 2010Accepted 11 April 2010Available online 21 May 2010

Keywords:ArsenicCatchmentRunoffAcidification reversalClimate change

The 40-year long period of heavy industrialization in Central Europe (1950–1990) was accompanied byburning of arsenic-rich lignite in thermal power plants, and accumulation of anthropogenic arsenic in forestsoils. There are fears that retreating acidification may lead to arsenic mobilization into drinking water,caused by competitive ligand exchange. We present monthly arsenic concentrations in surface runoff from12 headwater catchments in the Czech Republic for a period of 13 years (1996–2008). The studied area wascharacterized by a north–south gradient of decreasing pollution. Acidification, caused mainly by SOx and NOx

emissions from power plants, has been retreating since 1987. Between 1996 and 2003, maximum arsenicconcentrations in runoff did not change, and were b1 ppb in the rural south and b2 ppb in the industrialnorth. During the subsequent two years, 2004–2005, maximum arsenic concentrations in runoff increased,reaching 60% of the drinking water limit (10 ppb). Starting in 2006, maximum arsenic concentrationsreturned to lower values at most sites. We discuss three possible causes of the recent arsenic concentrationmaximum in runoff. We rule out retreating acidification and a pulse of high industrial emission rates aspossible controls. The pH of runoff has not changed since 1996, and is still too low (b6.5) at most sites for anAs–OH− ligand exchange to become significant. Elevated arsenic concentrations in runoff in 2004–2005 mayreflect climate change through changing hydrological conditions at some, but not all sites. Dry conditionsmay result in elevated production of DOC and sulfur oxidation in the soil. Subsequent wet conditions may beaccompanied by acidification leading to faster dissolution of arsenic-bearing sulfides, dissolution of arsenic-bearing Fe-oxyhydroxides, and elevated transport of arsenic sorbed on organic matter. Anaerobic domainsexist in normally well-aerated upland soils for hours-to-days following precipitation events.

+420 251818748.).

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

Detrimental effects of bioavilable forms of arsenic on humanhealth are well known (Haffert and Craw, 2008; Telford et al., 2009).Anthropogenic arsenic, originating from both industry and agricul-ture, has increasingly affected ecosystem health, especially after thebeginning of the Industrial Revolution (Ozdilek et al., 2007; Nkongoloet al., 2008). Arsenic pollution peaked in the second half of the 20thcentury in most industrial countries (Clemente et al., 2008; Kim et al.,2009). Operation times of major arsenic polluters on the territory oftoday's Czech Republic varied (Novak et al., 2008). Chronologically,arsenic pollution was first related to gold mining (since Middle Ages),then to coal burning (since 1860), to processing of anmal hides intanneries (since ca. 1880), and most recently to pesticide application(since 1950). Local arsenic pollution resulted from gold mining in

Roudne (SE of Prague; early 20th century), Jilove (S of Prague, ca.1950–1980) and Zlate Hory (E of Prague, mid-20th century), and frombase-metal mining in Kutna Hora (E of Prague, before 1980; Moraveket al., 1995). In the 1980s, new low-grade, arsenic-rich gold depositswere discovered S of Prague. Transport of ores to processing plants byheavy trucks was associated with dispersion of arsenic-containingdust in an area of about 50 km2. The small-scale mining operationswere discontinued before 1990 (Drahota and Filippi, 2009; Drahotaet al., 2006, 2009). Between 1950 and 1990, the centrally-plannedCzech economy was characterized by environment-unfriendly heavyindustries (Paces, 1985). Large production of lignite in northern CzechRepublic, Saxony/Lusatia (Germany), and Silesia (Poland) causedmuch more widespread arsenic pollution than gold mining, industrialprocessing of animal hides, and farming (Bezacinsky, 1981; Beneset al., 1984; Erbanova et al., 2008). Emissions of arsenic were coupledwith those of acidifying forms of sulfur (Novak et al., 2007). Sulfuremissions from power plants peaked in 1987. After 1990, both sulfurand arsenic emissions from 30 thermal power stations in the regiondropped rapidly (Hejma, 1992; Minarik and Dousova, 1993). Heavy

3615M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

industries were not competitive any longer and were partly phasedout, while technological upgrades in power plants led to moreeffective dust removal from coal combustion products (Novak et al.,2005; Dousova et al., 2007). During peak industrial pollution, acid raincontributed to massive forest decline (1970–1996; Groscheova et al.,1998). Spruce die-back was recorded on a territory of more than1000 km2 (“Black Triangle”; Cerny, 1995). Tanneries caused only localpollution (Brtnice, Trutnov), and complied with strict environmentallegislation after 1990 (Benes, 1993; Dousova et al., 2007). The systemof state-controlled farming collapsed after 1990, and arsenic-basedpesticides were not in use after 1995 (Moldan and Hak, 2007).

Along with sulfur, arsenic accumulation in organic soils was aninevitable legacy of the late 20th century. Upland catchments are amajor source of drinking water in the densely populated CzechRepublic. More than one third of the territory of the Czech Republic iscovered by forests, with spruce monocultures prevailing over mixedand broadleaf stands. By analogy with less polluted sites in westernEurope and the US (Nimick et al., 2005), forested headwater catch-ments have been viewed as a net sink for pollutant arsenic (Huangand Matzner, 2007a,b). In theory, weathering of crystalline rocks,possibly enhanced by acid rain, should not significantly contributeto arsenic export by surface runoff because Fe-rich oxidation productsof pyrite and arsenopyrite adsorb arsenic (Lee et al., 2005). Thisadsorption may be temporary, depending on a number of abiotic andbiotic parameters (Jain et al., 1999; Smedley and Kinnburgh, 2002).Elevatedmobilization of arsenic in forest catchments is envisioned forthe period of retreating acidification (Huang and Matzner, 2007c).When pH of soil waters reaches circum-neutral values, OH− func-tional groups may start to replace arsenic at soil adsorption sites(Wenzel et al., 2001). Because a simultaneous combination of highaccumulation of anthropogenic arsenic in soil and decreasing anthro-pogenic acidity was never experienced by forest ecosystems before,there are fears that mobilized arsenic may contaminate sources ofdrinking water in these regions.

For a period of 13 hydrological years (1996–2008), we measuredarsenic concentrations in runoff from a network of small forestedcatchments in the Czech Republic. Some of the 12 sites were located ina heavily polluted, spruce die-back affected area of northern CzechRepublic, others were located in the less polluted south of the country.We describe temporal trends in arsenic concentrations in runoff, andevaluate mechanisms that might be responsible for these changes.

Fig. 1. Study sites. Sulfur deposition contours were taken from Fiala et al. (1997). Atmospheeach large circle) is from Novak et al. (2005).

2. Study sites

The 12 small forested catchments, situated throughout the CzechRepublic (Fig. 1), were similar in topography (V-shaped valley) andvegetation (mostly Norway spruce, with clearings after 1970). Bedrock,elevation, precipitation totals and pollution levels varied (Table 1;Novak et al., 2005). Arsenic concentrations in bedrock are given inTable 1. The largest arsenic concentration in bedrockwas found atMODand CER in the north. Soils were classified as dystric cambisols, only onesite, UHL, contained a smallwetlandnear its closing profile (80 m2), andthere were patches of peaty soils in the riparian zone of UDL (50 m2).Before the advent of acid rain in the early 20th century, soil pH (H2O)was close to 3.7 and 5.0 in organic and mineral horizons, respectively(Oulehle et al., 2010). In the mid-1990s, soil pH (H2O) was around 3.2and 4.3 in organic and mineral horizons, respectively (Novak et al.,2000). The annual hydrological minimum was around October 1, andmaximum between March and May (Kram and Fottova, 2007). In the1950s, eleven large lignite-burning power plants were put in operationin the north of the country. The Czech Republic became the third largestlignite producer in the world (Paces, 1985). Arsenic deposition ratespeaked in the 1980s (120 g As ha−1 yr−1; Santroch et al., 1989).Compared to thenorthernCzechRepublic, atmospheric deposition ratesin the south of the country were 4 times lower (Novak and Pacherova,2008). Erbanova et al. (2008) founddecreasing atmospheric depositionsof arsenic at four sites in the Czech Republic between 1996 and 2005.

Additionally, twomountain-top research plots were used to samplehorizontal and vertical atmospheric depositions (Fig. 1). Zajeci vrch(ZAJ; 1008 masl) was situated upwind from the coal-fired power plantsin the Krusne Mts. (north). Kamenec (NEC; 988 m) was located in theNovohradske Mts. in the least polluted area of the Czech Republic(south). Annual precipitation totals at both siteswere 1100 mm, and themean annual temperature was+3 °C. The sampling sites were situatedin an unforested area, remote from human dwellings. No local pointsources of arsenic pollution were present within 10 km.

3. Sampling

Starting on November 1, 1995 (water year 1996), runoff wassampled at the catchment outlet in a monthly interval. Samplingended on October 31, 2008 (water year 2008). 100 mL of runoff wascollected each month in each of the 12 catchments using acid-washed

ric sulfur deposition (kg ha−1 yr−1) for individual study sites (small numbers next to

Table 1Study site characteristics.

Site Location Elevation (masl) Catchmentarea (ha)

Bedrock As concentrationin bedrock (ppm)a

Mean annualtemperature (°C)

Annual precipitation(mm)

LIZ Na lizu 49°04′ N, 13°41′ E 828–1024 99 Sillimanite–biotite paragneiss b5 +4.9 905SPA Spalenec 48°55′ N, 13°59′ E 795–858 53 Granulite gneiss b5 +5.5 816POM Polomka 49°47′ N, 15°45′ E 512–640 69 Migmatite–orthogneiss, paragneiss 7 +6.3 695SAL Salacova Lhota 49°31′ N, 14°59′ E 557–744 168 Sillimanite–biotite paragneiss b5 +6.0 572LKV Loukov 49°38′ N, 15°21′ E 472–658 43 Biotite–muscovite granite b5 +6.9 715ANE Anenský potok 49°34′ E, 15°05′ E 480–540 27 Sillimanite–biotite paragneiss b5 +6.9 644MOD Modrý potok 50°42′ N, 15°42′ E 1010–1554 262 Muscovite mica–shist 60 to 850 +2.9 1666CER Cervík 49°27′ N, 18°23′ E 640–961 185 Flysh sandstone 13 +6.2 1155JEZ Jezeri 50°32′ N, 13°28′ E 475–924 261 Two-mica gneiss 8 +5.0 934LES Lesní potok 49°58′ N, 14°49′ E 400–495 70 Biotite granite 5 +7.0 613UHL Uhlirska 50°49′ N, 15°08′ E 780–870 187 Granite b5 +4.0 1231UDL U dvou loucek 50°13′ N, 16°29′ E 880–950 33 Two-mica gneiss b5 +5.0 1308

a Analyzed by RFA, detection limit 5 ppm.

3616 M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

PE vials. Unfiltered runoff samples were acidified in the field by HNO3

(1:1, 1 mL; Romil, As b10 ppt).A pilot study in four mountain-slope catchments showed that the

concentration of suspended matter in runoff was close to zero(b100 ppb; Jicinsky and Paces, 1980). Therefore, when the samplingstarted in the mid-1990s, we decided to use unfiltered water samplesfor analysis.

Throughout the 13 years, water runoff fluxes from each of the 12catchments were continuously recorded at a gauging station.

In winter 2008–2009, snow and ice accretion samples were takenat the two mountain-top research plots, ZAJ and NEC. The averageinterval between two samplings (exposure period) was 8 days. Snowrepresented vertical deposition, whereas ice accretions representedhorizontal deposition. Snow was collected from a rectangular plot(10×30 cm, 3 cm deep), in triplicate. The distance between the snowsampling plots was 30 to 50 m. Ice accretions were captured on acid-washed high-surface area polyethylene (PE) samplers, rectangular inshape (8×15 cm), placed 1.5 m above snow cover. Each replicate iceaccretion sample comprised four rectangular PE samplers. At the endof the exposure period, the four plastic samplers were placed in oneacid-washed polyethylene container and transported to the labora-tory. After the pooling, three replicate ice accretion samples wereavailable. The distance between three wooden stands, carrying fourice accretion samplers each, was 30 to 50 m.

4. Analysis

Runoff samples were analyzed for arsenic concentration by ETAAS(Perkin Elmer 4100; detection limit of 0.5 ppb) following thoroughshaking of the vial. The inner diameter of the tubing used to introducethe liquid sample into the AA spectrometer was 1 mm, effectivelyincluding small particles into the analysis. Due to acidification of thesamples in the field, colloids of Fe (hydr)oxides were not present inthe samples, and As3+ was oxidized to the more stable form As5+.Reproducibility of ETAAS analyses was determined using referencewater samples (Environment Canada, Burlington, Ontario) with acertified As concentration of 6.22 ppb (1 s=0.424 ppb). Our mean Asconcentration, based on 100 measurements over a 1–year period, was6.21 ppb (1 s=0.380 ppb).

Snow and ice accretion samples were processed in laminar flowboxes (ISO 5), using ultrapure acids and acid-washed labware. Toprevent adsorption of some trace elements on container surfaces,samples of melting snow and ice accretions were acidified withultrapure HNO3 (1:1) as soon as the first 20 mL of samples weremelted. Filtered meltwater samples (b0.45μm) were analyzed forarsenic concentration by ICP sector-field MS (ICP SF MS) Element 2(Thermo Finnigan), equipped with a guard electrode to eliminatesecondary discharge in the plasma to enhance overall sensitivity(detection limit of 0.3 ppt; Shotyk and Krachler, 2009).

Similar to runoff, the amount of suspended matter in meltwaterswas negligible (b40 ppb, n=200).

Arsenic in bedrock (Table 1) was determined by RoentgenFluorescence Analysis (RFA; Siemens SRS-200).

5. Results

5.1. Arsenic in catchment runoff

Arsenic concentrations in runoff are given in Fig. 2. LIZ and SPA,two sites situated in the southern Czech Republic, remote from largeindustrial sources of arsenic pollution, are depicted at the top of Fig. 2.UHL and UDL, two typical sites located in the industrial north of thecountry, are depicted at the bottom of Fig. 2.

Compared to polluted groundwaters in West Bengal, Cambodia,Argentina and some other countries (maximum of 5 ppmAs; Smedleyand Kinnburgh, 2002), arsenic concentrations in Czech headwatercatchments were low (maximum of 5 ppb As). In runoff from forestedcatchments in Germany, arsenic concentrations were slightly lower(maximum of 4.7 ppb; Matschullat et al., 2000), compared to theCzech study sites.

No seasonality was observed in arsenic concentrations in runoff(Fig. 2). However, an inter-annual trend emerged from comparison ofarsenic concentrations in runoff from the 12 catchments (Fig. 2). Afterthe beginning of the monitoring, arsenic concentrations showed littleinter-annual trend, and were lower (mainly b1 ppb in the south, andb2 ppb in the north), compared to the end of the observation period.For a period of two years toward the end of the observation period(2004–2005), maximum arsenic concentrations suddenly increasedmore than twice (up to 6 ppb at JEZ, LES and UDL), but decreasedagain as of 2006 (Fig. 2). Typical examples of the decrease in 2006 areLIZ, SPA, POM, MOD, and CER. At some other sites, maximum arsenicconcentrations in runoff also increased in 2004, but the decreasestarting in 2006 was less pronounced (SAL, LKV, JEZ, UHL and UDL). Inthe group of northern sites, situated close to coal-fired power plants,UDL and UHL were characterized by higher arsenic concentrations inrunoff in the early years of observation (1996–2003) compared toMOD and CER (Fig. 2).

5.2. Runoff pH

Fig. 3 gives monthly pH values of runoff from the 12 catchments.There was no inter-annual or decadal trend in runoff pH. Both at thebeginning and at the end of the observation period (1996 and 2008,respectively), pH values of runoff were mostly between 5 and 7.Decreasing acidity of nation-wide industrial emissions (decreasingSOx and NOx; Groscheova et al., 1998; Oulehle et al., 2008) did notresult in increasing pH of catchment runoff. Among the study sites,time series of the lowest and the highest runoff pH values were

Fig. 3. Lack of inter-annual trends in runoff pH at the Czech catchments. SAL in centralCzech Republic exhibits the highest pH, while the nearby site LKV exhibits the lowest pH.

Fig. 2. Time series of arsenic runoff concentrations. Relatively unpolluted southern sites are depicted at the top, extremely polluted northern sites are at the bottom of the figure.

3617M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

recorded at two nearby sites, LKV and SAL, located in central CzechRepublic (Fig. 1). LKV exhibited the lowest runoff pH (5.0), whereasSAL had the highest runoff pH (7.0; Fig. 3).

5.3. Water runoff fluxes

Water runoff fluxes, calculated from daily measurements, areshown in Fig. 4. Minimum andmaximum daily discharges are given inTable S1 (Electronic Annex). Some sites, one or two years before theonset of relatively high maximum arsenic concentrations in runoff(2004), experienced an extremely wet year followed by a dry year.Examples are MOD (2000 and 1000 mm, in 2002 and 2003,respectively), UDL (1500 and 700 mm in 2002 and 2003, respective-ly), and JEZ (900 and 200 mm in 2003 and 2004, respectively). Atthree sites, precipitation totals were 400 mm lower in 2003 comparedto 2002 (UHL, CER and SPA). At one site, LKV, there was a trendof decreasing water runoff fluxes throughout the observation period

Fig. 4. Water runoff fluxes for the Czech catchments. A relatively dry year 2003 folows a relatively wet year 2002 at 6 catchments (SPA, POM, MOD, CER, UHL, and UDL).

3618 M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

Table 2Average residence time of water in selected small catchmemts, based on δ18O-H2O.

Site Average water residence time (months) Reference

LIZ 6.7 Buzek (2009)SAL 11.5 Buzek et al. (2009)CER 15.9 Buzek (2009)JEZ 7.9 Buzek (2004)UHL 11.5 Buzek (2004)

3619M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

(5 times lower water runoff in 2007 compared to 1996). Fig. S1(Electronic Annex) shows that inter-annual minima in water runoffflux from LKV partly corresponded to lower annual precipitationtotals (1998, 2004).

Fig. S2 (Electronic Annex) shows that As concentration in runoffdid not correlate with water runoff flux (data for one polluted site,UHL, and one relatively unpolluted site, SPA). Table S1 (ElectronicAnnex) gives estimates of annual precipitation totals in an open-areafor all 12 catchments.

5.4. Arsenic concentrations in ice accretions and snow

As seen in Fig. 5, arsenic concentrations in mountain-top wintertime precipitation samples were about 15 times lower compared tocatchment runoff (Fig. 2). While arsenic concentrations in runoff wereabout 5 ppb (2004–2005), those in horizontal precipitation (iceaccretions) at NEC (2008) were 330 ppt.

In general, arsenic concentrations in vertical precipitation (snow)were less than one half of those in horizontal precipitation (iceaccretions). On average, arsenic concentrations at ZAJ represented ca.63–64% of arsenic concentrations at NEC, both in case of snow and iceaccretions. This difference was statistically significant at the 0.05 level.Mean arsenic concentration inwinter time precipitation at the southernsite NEC was surprisingly higher than at the northern site ZAJ, but thisdifference was not statistically significant at the 0.05 level.

6. Discussion

6.1. Contribution of bedrock arsenic to runoff

Using δ18O-H2O, Buzek (2004, 2009) showed that the meanresidence time of water in forested headwater catchments in theCzech Republic, underlain by crystalline bedrock, is 6 to 16 months(Table 2). The mean residence time of dissolved arsenic, which isexported from the catchments via runoff, may be different from thatof water (Grosbois et al., 2009). Weathering of accessory sulfidessupplies arsenic to groundwater, which, especially during baseflowperiods, may dominate runoff (Reedy et al., 2007). A directatmospheric component may leave the catchment by surface runoffhours-to-days after the beginning of the most recent precipitationevent, exporting atmogenic arsenic. A third mixing end member forrunoff arsenic is soil arsenic (Section 6.2).

As seen from Table 1, arsenic concentration in bedrock at 10 of 12catchments under study was close to the typical arsenic concentrationof the upper crust (1 to 8 ppm, Smedely and Kinniburgh, 2002).Bedrock at two sites in the north was relatively richer in arsenic(850 ppm at MOD, and 13 ppm at CER). These two sites, however, hadsome of the lowest arsenic concentrations in runoff in the entire data

Fig. 5. Comparison of dissolved arsenic concentrations in vertical deposition (snow)and horizontal deposition (ice accretions) in northern (ZAJ) and southern (NEC) CzechRepublic. Means and standard errors are given. Different letters denote significantlydifferent values (ANOVA, pN0.05).

set (Fig. 2). In the 1996–2003 period, MOD, for example, released1 ppb arsenic via runoff, less than UDL, whose bedrock arsenicconcentration was 170 times lower. Based on Table 1 and Fig. 2, weconclude that weathering of bedrock arsenic minerals was not theprimary control of arsenic concentrations in runoff.

Spatially averaged erosion rates have not been studied in ourcatchments. MOD and JEZ are situated on the steepest slope (Table 1)and may thus experience the highest erosion rates (Ferrier andKirchner, 2008). Bedrock-derived arsenic supply to runoff should behigh in such steep catchments, relative to mild-sloping catchmentswith a similar arsenic content in bedrock. A high erosion rate at MODdid not result in high As concentration in runoff.

6.2. Arsenic speciation in soils

Erbanova et al. (2008) studied arsenic speciation in the soil of fourcatchments (JEZ, UHL, UDL and LIZ). Arsenic associated with Fe–Aloxyhydroxides was always the most abundant arsenic species,making up ca. 60% of total soil arsenic. This form is immobile andunavailable to fauna and biota. Under anaerobic conditions, however,some arsenic can be released into solution due to microbiallymediated reductive dissolution of Fe-oxyhydroxides (Wenzel et al.,2001). Unpolluted LIZ had the lowest percentge of soluble and ion-exchangeable arsenic in organic soil (8%). Residual arsenic (rockdebris and a minor admixture of deposited coal ash particles)constituted ca. 20% of total arsenic.

For JEZ, UHL, UDL and LIZ, Erbanova et al. (2008) showed thatexport flux of total arsenic was directly proportional to water export.Consequently, some of the labile arsenic forms in soils, possibly olderpollutant arsenic, are increasingly flushed out of the catchments,whenever elevated water runoff fluxes are generated (cf., Huang andMatzner, 2007a).

6.3. Spatial variation in arsenic runoff concentrations

There is currently no geochemical tool to distinguish between“young” and “old” pollutant arsenic released from the catchments viarunoff. “Young” arsenic is defined as arsenic derived from recentprecipitation which did not undergo any (bio)geochemical cyclingwithin the ecosystem, while “old” arsenic was immobilized within theecosystem (mainly in soil) for a certain period of time betweendeposition and export.We assume that most of the exported arsenic isold pollutant arsenic because Dousova et al. (2007) and Erbanovaet al. (2008) have shown that atmospheric deposition of arsenic innorthern Czech Republic was up to 40 times higher in the 1980s,compared to 2006. Both studies also showed that the steepest declinein arsenic depositions occurred before 1996, by which year dustremoving equipment was installed into most Czech and East Germancoal-fired power plants.

At the begining of this study we hypothesized that (i) catchmentsat the northern border of the Czech Republic (i.e., JEZ, UHL, MOD, UDLand CER, see Fig. 1) would show high arsenic runoff concentrations,(ii) the group of catchments southeast of Prague (LES, ANE, SAL, LKVand POM) would produce lower arsenic concentrations in runoff thanthe northern catchments, but higher than LIZ and SPA in the south, and(iii) both vertical and horizontal precipitations at ZAJ in the north

Fig. 6. Comparison of three variables, As concentration in runoff (a), water flux (b), andAs concentration in atmospheric input (c), in the years 2004–2005 with the precedingand subsequent period. Data in the top panel were compiled from Fig. 2, data in themiddle panel are a compilation from Fig. 3, and data in the bottom panel were takenfrom Erbanova et al. (2008). Arsenic concentrations in runoff were available from 12catchments (Fig.1), whereas arsenic concentrations in atmospheric input wereavailable for JEZ, UHL, UDL and LIZ. Means and standard errors are given. Differentletters denote significantly different values (ANOVA, pN0.05).

3620 M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

would show higher winter-time arsenic concentrations than at thesouthern site NEC. These hypotheses were driven by the fact that thecluster of coal-fired power plants in the northern Czech Republic (butalso those near Leipzig and Cottbus in Germany) are situated upwindfrom most our northern catchments, and that arsenic air pollutionparallels sulfur air pollution and past acid rain damage to ecosystems.There were, for example, equally serious symptoms of forest decline15 years ago at UHL, MOD and UDL. In contrast, there were nosymptoms of forest decline in the southern Czech Republic. Data inFig. 2, however, show amore complicated pattern than that suggestedby our hypotheses. Assuming a similar level of pollution and a similarbuffering function of forest soils at two sites, the sitewith higherwaterrunoff fluxes would be expected to exhibit lower arsenic concentra-tions, due to dilution. This simple mechanism may explain why MOD,even though situated between two catchments with relatively higharsenic runoff concentrations, UHL and UDL, has low arsenic runoffconcentrations. MOD has by far the largest water runoff fluxes amongall catchments, reaching up to 2400 mm yr−1 (Fig. 4). Dilution doesnot, however, explain relative magnitudes of arsenic concentrationwhen data from other sites are compared. For example, about twicelowerwater export from JEZ did not lead to twice higher arsenic runoffconcentrations compared to UHL (cf., Figs. 2 and 4, bottom left).

As a result of the combination of site-specific factors contributingto runoff generation, the magnitude of the recent increase in arsenicconcentrations in runoff is similar in various parts of the country. Thepreviously highly polluted ecosystems in the Black Triangle region(north of the Czech Republic) do not produce the highest arsenicconcentrations in runoff. Sites in previously moderately pollutedregions, such as central Czech Republic, may produce arsenicconcentrations as high as sites in the spruce die-back ridden northof the Czech Republic (Fig. 2). An example are some of the highestarsenic runoff concentrations at LES southeast of Prague, a site 100 kmmore distant from coal-fired power plants than, e.g., UHL. LES and UHLhad the same arsenic concentration maxima in recent years (5 ppb).

6.4. Lack of north–south gradient of decreasing arsenic deposition in 2008

Data in Fig. 5 show that the spatial gradient of decreasing pollutionfrom the north to the south of the Czech Republic, known from anumber of previous studies (Fottova, 2003; Dousova et al., 2007;Novak and Pacherova, 2008) may no longer be valid. Counter-intuitively, NEC, the southern site close to nearly pristine nationalparks at the Czech Austrian border, captured a similar (or higher)amount of arsenic as the northern site ZAJ.We note that the lack of thenorth–south arsenic deposition trend (Fig. 5) may be a relativelyrecent phenomenon (2008), resulting from a major environmentalclean-up. Arsenic concentrations in ice accretions (350 ppt; Fig. 5) arenine times lower than those found in 2004 (Dousova et al. 2007).Nation-wide arsenic emissions decrease and may have reached such alow level that the previous spatial pollution gradient (polluted north)is not observed any longer.

In agreement with previous reports (Dousova et al., 2007),horizontal deposition (ice accretions) showed higher arsenic con-centrations than vertical deposition (snow). It is believed that theprovenance of arsenic in horizontal and vertical depositions differs.While snow receives arsenic from more local sources, the transportdistance arsenic captured by ice accretions may be hundreds tothousands kilometers.

6.5. Possible causes of elevated arsenic concentrations in runoff in2004–2005

So far, we have mainly discussed maximum arsenic runoffconcentrations, keeping in mind that it is instantaneous concentra-tions which matter when drinking water quality is of concern. Fig. 6agives mean arsenic concentrations in runoff across all 12 sites in the

years 2003, 2004–2005 and 2006–2007. The 2004–2005 increase inmean arsenic runoff concentrations is 2.5-fold, and rather robust.Fig. 6a also shows that, across the region, the mean arsenic con-centrations stopped increasing after 2005, at least temporarily.

The same trend of a sudden increase in maximum arsenic runoffconcentrations in a single year (2004), following 8 years of little-to-nochange, was found at a variety of sites. These sites were situated in asimilar climatic zone, on similar soils and with similar tree species(Norway spruce), but at different elevations, precipitation totals,water runoff fluxes, pollution levels, and on different bedrocks.Uniformity of the trend across most of the 12 sites (Fig. 2) allows usto search for a more general control mechanism. It is unlikely thatsuch synchronized change in arsenic behaviour in small forestedcatchments was initiated by site-specific factors. In the following

3621M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

paragraphs, we will review three potential mechanisms that might, inprinciple, explain trends depicted in Fig. 2.

We found no signs of increasing runoff pH during the 13 yearsof catchment observations (Fig. 3). This contrasts with slightly in-creasing pH of spruce canopy throughfall at some Central Europeansites (Oulehle et al., 2008). Forest soils effectively buffer chemicalproperties of atmospheric deposition and cause a delay in acidificationreversal. Based on sulfur isotope data (Novak et al., 2005) andmodelling results (Hruska and Kram, 2003), it may take more than50 years for soils and waters to recover from the pulse of acidityreceived in 1950–1990. We conclude that ligand exchange involvingAs and OH− in soils, typical of circum-neutral environments (Wenzelet al., 2001), did not cause the increase in arsenic runoff concentra-tions, seen in Fig. 2. A pH of around 6.5, common in Fig. 3, may, inprinciple, be sufficient for a limited arsenic mobilization, however, thelack of enhanced arsenic removal from the soil before 2004 under thesame pH would then remain unexplained.

The possible role of two other mechanisms of arsenic turnover inforest ecosystems is summarized in Fig. 6b,c. In Fig. 4 we have seenthat, at 6 of the 12 catchments, a wet year 2002 was followed by arelatively dry year 2003, just before the onset of higher arsenic runoffconcentrations (2004). To explore a more regional aspect of inter-annual trends in wetness/droughts (Fig. 4) in the specific years ofchanging arsenic runoff concentrations, in Fig. 6b we averaged waterrunoff fluxes from all 12 catchments. Water flux data for the year2003, just preceding the increase in arsenic runoff concentrations, arecompared with data for 2004–2005, and with data for the subsequenttwo years (2006–2007). It turns out that, on a regional basis, thewaterrunoff flux was only insignificantly lower in 2003 compared with2004–2005 and with 2006–2007. We conclude that, across the region,mobilization of arsenic under more reducing conditions and in thepresence of soil organic matter, which would be due to microbiallymediated reductive dissolution of FeOOH (Stuben et al., 2003), doesnot explain temporal trends in Fig. 2. The role of this mechanism atindividual catchments, experiencing large inter-annual changes inwetness/dryness, should be further explored. We note that ourprevious research in catchment situated in the polluted north of theCzech Republic (e.g., JEZ) showed that anaerobic domains exist inforest soils for one or two days following major precipitation events.Using sulfur isotopes, we were able to show that strictly anaerobicbacterial sulfate reduction is triggered off in normally well-aeratedsoils following a precipitation event (Novak et al., 2001). Intermittentanaerobic conditions in such soils may support disolution of Fe (hydr)oxides and release of adsorbed arsenic into solution. We also note thatRothwell et al. (2009) reported high arsenic mobility in organic soilsunder fluctuating water table conditions.

Fig. 6c rules out a pulse of high atmospheric arsenic deposition in2004–2005 as a possible cause for elevated arsenic concentrations inrunoff. Thismechanism, if viable, would require a short residence timeof atmospheric arsenic in the catchment. Data in Fig. 6c are acompilation from Erbanova et al. (2008) and include arsenicconcentrations in atmospheric deposition into four catchments, JEZ,UHL, UDL and LIZ. As seen from Fig. 6c, there was no pulse of higherarsenic concentration in atmospheric deposition in 2004–2005.Instead, there was a decrease in arsenic concentrations of the input.We also note that a very short residence time of anthropogenic arsenicin forest ecosystems would contradict observations from other partsof Europe, showing that catchments may be a net long-term sink fordeposited arsenic (Huang and Matzner, 2007c).

6.6. Absence of arsenic–TOC, arsenic–sulfate, and arsenic–ironcorrelations in runoff

Using data from one polluted and one unpolluted site (UHL and SPA,respectively), we illustrate that arsenic concentrations in runoffwere correlated with neither TOC, nor sulfate, nor Fe2+ concentrations

(Figs. S2 and S3, Electronic Annex). High levels of TOC in soils mayoutcompete solid-phase iron oxides for arsenic binding (Redman et al.,2002). Presence of TOC may inhibit formation of large FeOOH ag-gregates, keeping more arsenic in solution (Bauer and Blodau, 2009).Buschmann et al. (2006) reported that about 10% of total dissolvedarsenic may be bound to organic matter. If so, such low percentagemaynot be visible in total arsenic versus TOC graphs (cf., Fig. S2). A positivecorrelation between arsenic and dissolved organic matter was reportedalsobyBauer andBlodau (2009) in laboratorymesocosms, andRothwellet al. (2009) in wetlands. Elevated TOC production in forest soils mayresult from dry conditions. Dry conditions can also lead to soilacidification, due to oxidation of reduced sulfur in the soil, which, inturn, has the potential to dissolve arsenic-bearing Fe oxides (Rothwellet al., 2009; Smedley and Kinnburgh, 2002). Thus, a concurrent increasein sulfate and arsenic concentrations in runoff might help to explainhigher As mobility as a result of soil wetness. Again, we did not observesuch a correlation at our sites. Blodau et al. (2008) showed that releaseof arsenic from organic soil was linked to Fe2+. The whole process wascontrolled by root activity. Release of Fe may also be influenced bydecreasing acidity (Neal et al., 2008). No As–Fe relationship wasobserved at the Czech catchments.

7. Conclusions

Twelve forested headwater catchments, situated along a north–south pollution gradient in the Czech Republic (Central Europe), werestudied for a period of 13 years, following a maximum in arsenicpollution (1980s). The observation period (1996–2008) coincidedwith retreating acidity of atmospheric deposition. During the13 years, we were not able to observe an increase in pH of runofffrom any of the catchments. Recovery of soils and surface waters fromacidification, which had been caused by coal burning and heavyindustries mainly between 1950 and 1990, is delayed. Therefore itwas surprising to observe a sudden 2.5-fold increase in mean arsenicconcentration in runoff, which, in most cases, started only in the 9thyear of observations (2004). For the first 8 years, no systematic inter-annual trend in arsenic runoff concentrations was recorded. Maxi-mum arsenic runoff concentrations at most sites increased 4 to 5times in 2004. Two years later, starting in 2006, maximum arsenicrunoff concentrations decreased, and mean arsenic runoff concentra-tions stopped increasing. Based on previously published data onatmospheric arsenic depositions from four sites (JEZ, UHL, UDL, andLIZ; Erbanova et al., 2008) we were able to rule out a sudden increasein atmospheric input in 2004 as the cause of elevated arsenic runoffconcentrations: arsenic concentrations in atmospheric input de-creased over time. Future research will provide more insights intothe role of climate change in exporting arsenic from small forestcatchments in heavily polluted regions. One feasible mechanism forexport of arsenic-containing waters is microbial reduction of FeOOHin organic soil horizons in wet seasons, following a period of drought.We recorded a sequence of a wet and a dry year just before the onsetof high arsenic runoff concentrations at some, but not all sites.

Acknowledgements

This work was supported by the Financial Mechanism EEA/Norway,Brusells (Project no. CZ0136), the Scientific Centre “Advanced RemedialTechnologies and Processes” of the Ministry of Education of the CzechRepublic (Project no. IM0554), and the Czech Ministry of Environment(Project SP/1a6/151/07). We thank O. Myska and T. Paces for valuableinput concerning the role of suspended matter in runoff.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.scitotenv.2010.04.016.

3622 M. Novak et al. / Science of the Total Environment 408 (2010) 3614–3622

References

Bauer M, Blodau C. Arsenic distribution in the dissolved, colloidal and particulate sizefraction of experimental solutions rich in dissolved organic matter and ferric iron.Geochim Cosmochim Acta 2009;73:529–42.

Benes J, Mastalka J, Panyr J. Evaluation of trace elements in artificial fertilizers.Agrochemia 1984;24:108–13.

Benes S. Concentrations and mass budgets of chemical elements in the environment.Czech Ministry of Agriculture, Prague; 1993. 88 pp.

Bezacinsky M. Trace elements in soft coal and their emission into the atmosphere. OchrOvzdusi 1981;13:142.

Blodau C, Fulda B, Bauer M, Knorr K-H. Arsenic speciation and turnover in intact organicsoil mesocosms during experimental drought and rewetting. Geochim CosmochimActa 2008;72:3991–4007.

Buschmann J, Kappeler A, Lindauer U, Kistler D, Berg M, Sigg L. Arsenite and arsenatebinding to dissolved humic acids: influence of pH, type of humic acid, andaluminum. Environ Sci Technol 2006;40:6015–20.

Buzek F. Use of 15N and 18O isotopes to study runoff formation and nitrate dispersion inselected sub-catchments of the Labe catchment. Czech Geological Survey; 2004.Final Report Project no. VaV/6050/5/03.

Buzek F. Evaluation of the impact of climate change on hydrological budgets, anddevelopment of measures for their attenuation; 2009a. Annual Report Project no.VaV/1a6/151/07.

Buzek F, Paces T, Jackova I. Production of dissolved organic carbon in forest soils alongthe north–south European transect. Appl Geochem 2009;24:1686–701.

Cerny J. Recovery of acidified watersheds in the extremely polluted Krusne hory Mts.,Czech Republic. Water Air Soil Pollut 1995;85(21):51–63.

Clemente R, Dickinson NM, Lepp NW. Mobility of metals and metalloids in a multi-element contaminated soil 20 years after cessation of the pollution source activity.Environ Pollut 2008;155:254–61.

Dousova B, Erbanova L, Novak M. Arsenic in atmospheric deposition at the Czech–Polishborder: two sampling campaigns 20 years apart. Sci Total Environ 2007;387:185–93.

Drahota P, Paces T, Pertold Z, Mihaljevic M, Skrivan P. Weathering and erosion fluxes ofarsenic in watershed mass budgets. Sci Total Environ 2006;372:306–16.

Drahota P, Filippi M. Secondary arsenic minerals in the environment: a review. EnvironInt 2009;35:1243–55.

Drahota P, Rohovec J, Filippi M, Mihaljevic M, Rychlovky P, Cerveny V, Pertold Z.Mineralogical and geochemical controls of arsenic speciation and mobility underdifferent redox conditions in soil, sediment and water at the Mokrsko-West golddeposit, Czech Republic. Sci Total Environ 2009;407:3372–84.

Erbanova L, Novak M, Fottova D, Dousova B. Export of arsenic from forested catchmentsunder easing atmospheric pollution. Environ Sci Technol 2008;42:7187–92.

Ferrier KL, Kirchner JW. Effects of physical erosion on chemical denudation rates: anumerical modeling study of soil-mantled hilslopes. Earth Planet Sci Lett 2008;272:591–9.

Fiala J, Ostatnicka J, Hunova I, Novak V, Sladecek J. Air pollution in the Czech Republic in1997. Czech Hydrometeorological Institute, Prague; 1997. 190 pp.

Fottova D. Trends in sulfur and nitrogen deposition fluxes in the GEOMON network,Czech Republic, between 1994 and 2000. Water Air Soil Pollut 2003;150:73–87.

Grosbois C, Schafer J, Bril H, Blanc G, Bossy A. Deconvolution of trace element (As, Cr,Mo, Th, U) sources and pathways to surface waters of a gold mining-influencedwatershed. Sci Total Environ 2009;407:2063–76.

Groscheova H, Novak M, Havel M, Cerny J. Effect of altitude and tree species on δ34S ofdeposited sulfur (Jezeri catchment, Czech Republic). Water Air Soil Pollut1998;105:287–95.

Haffert L, Craw D. Processes of attenuation of dissolved arsenic downstream fromhistoric gold mine sites, New Zealand. Sci Total Environ 2008;405:286–300.

Hejma J. Air protection in the Czech Republic in the near future. Ochr Ovzdusi 1992;3:57–61.

Hruska J, Kram P. Modelling long-term changes in stream water and soil chemistry incatchments with contrasting vulnerability to acidification (Lysina and Pluhuv Bor,Czech Republic). Hydrol Earth Syst Sci 2003;7:525–39.

Huang JH, Matzner E. Fluxes of inorganic and organic arsenic species in a Norwayspruce forest floor. Environ Pollut 2007a;149:201–8.

Huang JH, Matzner E. Mobile arsenic species in unpolluted and polluted soils. Sci TotalEnviron 2007b;377:308–18.

Huang JH, Matzner E. Biogeochemistry of organic and inorganic arsenic species in aforested catchment in Germany. Environ Sci Technol 2007c;41:1564–9.

Jain A, Raven KP, Loeppert RH. Arsenite and arsenate adsorption on ferrihydrite: surfacecharge reduction and net OH− stoichiometry. Environ Sci Technol 1999;33:1179–84.

Jicinsky K, Paces T. The flux of elements in suspended matter from experimentaldrainage basins in Central Bohemia. IAHS–AISH Publications, 130. ; 1980. p. 271–6.

Kim YT, Yoon HO, Yoon C, Woo NC. Arsenic species in ecosystems affected by arsenic-rich spring water near an abandoned mine in Korea. Environ Pollut 2009;157:3495–501.

Kram P, Fottova D. Characteristics of daily runoff fluxes from 14 forested catchments ofthe GEOMON network (1994–2005). Final Project Report (VAV SP/1a6/151/07),Czech Geological Survey, Prague; 2007. 222 pp.

Lee P, Kang MJ, Choi SH, Touray JC. Sulfide oxidation and the natural attenuation ofarsenic and trace metals in the waste rocks of the abandoned Seobo tungsten mine,Korea. Appl Geochem 2005;20:1687–703.

Matschullat J, Maenhaut W, Zimmermann F, Fiebig J. Aerosol and bulk deposition in the1990's, Eastern Erzgebirge, Central Europe. Atmos Environ 2000;34:3213–21.

Minarik L, Dousova B. Characteristics of atmospheric dry deposition. Bull Czech EconomSoc 1993;1/2:55–60.

Moldan B, Hak T. Environment in the Czech Republic: a positive and rapid change.Environ Sci Technol 2007;41:358–62.

Moravek P, Pertold Z, Puncochar B, Studnicna B, Zacharias J. Gold deposits of the centraland SW part of the Bohemian Massif. Czech Geological Survey, Prague; 1995. 104 pp.

Nimick DA, Cleasby TE, McCleskey RB. Seasonality of diel cycles of dissolved trace-metalconcentrations in a Rocky Mountain stream. Environ Geol 2005;47:603–14.

Nkongolo KK, Vailancourt A, Dobrzenicka S, Mehes M, Beckett P. Metal content in soiland black spruce (Picea mariana) trees in the Sudbury region (Ontario, Canada):low concentration of arsenic, cadmium and nickel detected near smelter sources.Bull Environ Contam Toxicol 2008;80:107–11.

Novak M, Erel Y, Zemanova L, Bottrell SH, Adamova M. A comparison of lead pollutionrecord in Sphagnum peat with known historical Pb emission rates in the British Islesand the Czech Republic. Atmos Environ 2008;42:8997–9006.

Novak M, Jackova I, Prechova E. Temporal trends in the isotope signature of air-bornesulfur in Central Europe. Environ Sci Technol 2001;35:255–60.

Novak M, Kirchner JW, Fottova D, Prechova E, Jackova I, Kram P, Hruska J. Isotopicevidence for processes of sulfur retention/release in 13 forested catchmentsspanning a strong pollution gradient (Czech Republic, central Europe). GlobalBiogeochem Cycles 2005;19:GB4012.

Novak M, Kirchner JW, Groscheova H, Havel M, Cerny J, Krejci R, Buzek F. Sulfur isotopedynamics in two Central European watersheds affected by high atmosphericdeposition of SOx. Geochim Cosmochim Acta 2000;64:367–83.

Novak M, Mitchell MJ, Jackova I, Buzek F, Schweigstillova J, Erbanova L, Prikryl R,Fottova D. Processes affecting oxygen isotope ratios of atmospheric and ecosystemsulfate in two contrasting forest catchments in Central Europe. Environ Sci Technol2007;41:703–9.

Novak M, Pacherova P. Mobility of trace metals in pore waters of two Central Europeanpeat bogs. Sci Total Environ 2008;394:331–7.

Oulehle F, Hleb R, Hlouska J, Samonil P, Hofmeister J, Hruska J. Anthropogenicacidification effects in primeval forests in the Transcarpathian Mts., westernUkraine. Sci Total Environ 2010;408:856–64.

Oulehle F, McDowellWH, Aitkenhead-Peterson JA, Kram P, Hruska J, Navratil T, Buzek F,Fottova D. Long-term trends in stream nitrate concentrations and losses acrosswatersheds undergoing recovery from acidification in the Czech Republic.Ecosystems 2008;11:410–25.

Ozdilek HG, Mathisen PP, Pellegrino D. Distribution of heavy metals in vegetationsurrounding the Blackstone River, USA: considerations regarding sedimentcontamination and long term metals transport in freshwater riverine ecosystems.J Environ Biol 2007;28:493–502.

Redman AD, Macalady DL, Ahmann D. Natural organic matter affects arsenic speciationand sorption onto hematite. Environ Sci Technol 2002;36:2889–96.

Reedy RC, Scanion BR, Nicot JP, Tachovsky JA. Unsaturated zone arsenic distribution andimplications for groundwater contamination. Environ Sci Technol 2007;41:6914–9.

Rothwell JJ, Taylor KG, Ander EL, Evans MG, Daniels SM, Allott TEH. Arsenic retentionand release in ombrothropic peatlands. Sci Total Environ 2009;407:1405–17.

Santroch J, Honzak J, Mitosinkova M. Meteorologicke zpravy. Czech Hydrometeoro-logical Institute, Prague 1989;42:144–6 [in Czech].

Shotyk W, Krachler M. Determination of trace element concentrations in naturalfreshwaters: how low is “low”, and how low do we need to go? J Environ Monit2009;11:1747–53.

Smedley PL, Kinnburgh DG. A review of the source, behaviour and distribution ofarsenic in natural waters. Appl Geochem 2002;17:517–68.

Stuben D, Berner Z, Chandrasekharam D, Karmakar J. Arsenic enrichment ingroundwater of West Bengal, India: geochemical evidence for mobilization od Asunder reducing conditions. Appl Geochem 2003;18:1417–34.

Telford K, Maher W, Krikowa F, Foster S, Ellwood MJ, Ashley PM, Lockwood PV, WilsonSC. Bioaccumulation of antimony and arsenic in a highly contaminated streamadjacent to the Hillgrove Mine, NSW, Australia. Environ Chem 2009;6:133–43.

Wenzel WW, Kirchbaumer N, Prohaska T, Stingeder G, Lombi E, Adriano DC. Arsenicfractionation in soils using an improved sequential extraction procedure. AnalChim Acta 2001;436:309–23.