at SciVerse ScienceDirect

Environmental Pollution 159 (2011) 3474e3483

Contents lists available

Environmental Pollution

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

Long-term mercury dynamics in UK soils

E. Tipping a,*, R.A. Wadsworth a, D.A. Norris b, J.R. Hall b, I. Ilyin c

aCentre for Ecology and Hydrology, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster LA1 4AP, United KingdombCentre for Ecology and Hydrology, Environment Centre Wales, Deiniol Road, Bangor, Gwynedd LL57 2UW, United KingdomcMeteorological Synthesizing Centre e East, Krasina pereulok, 16/1, 123056 Moscow, Russia

a r t i c l e i n f o

Article history:Received 26 May 2011Received in revised form9 August 2011Accepted 11 August 2011

Keywords:DynamicsEvasionLeachingMercurySoils

* Corresponding author.E-mail address: et@ceh.ac.uk (E. Tipping).

0269-7491/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.envpol.2011.08.019

a b s t r a c t

A model assuming first-order losses by evasion and leaching was used to evaluate Hg dynamics in UKsoils since 1850. Temporal deposition patterns of Hg were constructed from literature information.Inverse modelling indicated that 30% of 898 rural sites receive Hg only from the global circulation, whilein 51% of cases local deposition exceeds global. Average estimated deposition is 16 mg Hg m�2 a�1 to ruralsoils, 19 mg Hg m�2 a�1 to rural and non-rural soils combined. UK soils currently hold 2490 tonnes ofreactive Hg, of which 2140 tonnes are due to anthropogenic deposition, mostly local in origin. Topsoilcurrently releases 5.1 tonnes of Hg0 per annum to the atmosphere, about 50% more than the anthro-pogenic flux. Sorptive retention of Hg in the lower soil exerts a strong control on surface water Hgconcentrations. Following decreases in inputs, soil Hg concentrations are predicted to decline overhundreds of years.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Soil mercury accounts for 75% of the biogeochemically-activeelement (Mason and Sheu, 2002) and at sufficiently high levelscan exert toxic effects towards microbes, invertebrates and plants(Rundgren et al., 1992; Tipping et al., 2010a). The soil is both a sinkand a source for atmospheric mercury, accumulating the metal bywet and dry deposition, and releasing Hg0 by evasion (Grigal,2002). The latter is an important process in controlling atmo-spheric levels, and the evolution of soil Hg levels as inputs changeover time (Pirrone et al., 2008).

A significant issue at national or regional scales is the source ofdeposited Hg (Pai et al., 1999). Globally, with the pollution ofremote locations in mind, long-range transport of Hg0 in theatmosphere before oxidation and deposition is the main focus, butat smaller scales local emissions and their speciation are important.Anthropogenic emissions are roughly equally divided between Hg0,which can directly join the global circulation, and other forms of Hg(reactive gaseous mercury, RGM, and particulate mercury Hgp)which deposit closer to the source (Mason et al., 1994). Theimportance of short-distance transport and deposition is evidentfrom the spatial distribution of soil Hg pools. Thus, Nater and Grigal(1992) reported systematic declines in soil Hg with distance fromindustry in the Great Lakes states. Observations for the UK from

All rights reserved.

national scale monitoring programmes show substantial spatiallyauto-correlated variability in soil Hg contents (Tipping et al., 2011),apparently dependent upon proximity to urban and industrialareas.

Here we address the dynamics of soil Hg, in an attempt toaccount for the observed distribution of UK soil pools. Our approachwas to attribute these to past deposition by accounting for the keyloss processes of leaching (Hg bound to DOM) and evasion. Twosources of Hg deposition were distinguished, the global atmo-spheric pool of the metal, and local anthropogenic emissions ofRGM and Hgp. At each soil site, the amount of Hg required to bedeposited over time to achieve observed Hg pools in topsoils wascalculated, using plausible historical variations in Hg inputs. Modeloutputs were compared with measured Hg deposition, accumula-tion in deeper soil, and surface water concentrations, and withmodel predictions of deposition to the UK. The model was used toestimate future soil Hg dynamics under different scenarios.

The findings were expected to provide information about UK Hgin soils, its likely turnover rates and its contribution to Hg0 in theglobal circulation by re-emission. This would help in assessinglikely toxic effects and the suitability of emission controls, espe-cially with respect to the Critical Loads approach (UBA, 2004).

2. The atmospheric deposition of mercury

Knowing how much Hg is being deposited and the history ofdeposition are key to understanding how Hg in soil has developed.

0

2

4

6

8

10

1800 1850 1900 1950 2000

depo

sitio

n re

lativ

e to

200

0

global local

Fig. 1. Assumed patterns of deposition, relative to a contemporary value of unity.

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e3483 3475

Here we review information relevant to the UK, and proposehistorical deposition patterns.

2.1. Forms of atmospheric mercury

Mercury has a complex atmospheric chemistry, unlike othermetals that are transported as aerosols (Schroeder and Munthe,1998). Three principal forms of atmospheric Hg are distinguished;elemental mercury Hg0, reactive gaseous mercury (RGM) whichcomprises HgCl2 and HgBr2 and possibly organic forms, andparticulate Hg (Hgp). The term total gaseous mercury (TGM) refersto the sum of gaseous elemental Hg0 (GEM) and RGM. Inputs to theglobal circulation of Hg0 come in about equal parts from naturalsources (volcanoes, geothermally active areas, the mineral matterof uncontaminated soils; Gustin et al., 2008), direct anthropogenicinputs, and re-emission of previously-deposited Hg (Mason et al.,1994; Schroeder and Munthe, 1998). Because it is relativelyunreactive, Hg0 has an atmospheric residence time of the order ofone year and therefore it can spread globally. It is lost from theatmosphere after conversion to RGM or Hgp both of which aredeposited rapidly as wet or dry deposition. Details of atmosphericoxidation processes are still actively researched (Hynes et al., 2009).The RGM and Hgp forms can also enter the atmosphere directly, asanthropogenic emissions. Pacyna and Pacyna (2002) estimated that53% of total global emissions are as Hg0, and data quoted byTravnikov and Ilyin (2005) indicate a similar value for Europeancountries, with a value of 52% for the UK. In 2005, UK emissionscomprised 53% Hg0 (Dore et al., 2007).

2.2. Contemporary Hg deposition to the UK

Rowland et al. (2010a) reported a volume-weighted meanconcentration of 3.3 ng Hg l�1 inwet deposition to 10 rural UK sitesover the period 2005e2009. An overall mean deposition of 2.8 mgm�2 a�1 was derived, with ranges of 2.5e3.5 mg m�2 a�1 amongyears and 2.1e4.3 mg m�2 a�1 among sites. The authors acknowl-edged that these concentrations and deposition rates are surpris-ingly low, offering as explanations the choice of remote siteswithoutlocal Hg sources, a possible decline in global emissions of Hg, andanalytical or technical factors. Due to the lack of monitoring of RGM,this UK rural network did not permit estimates of Hg dry deposition.Miller et al. (2005) estimated wet deposition fluxes of 3e7 mgm�2 a�1 to rural sites in the north-east of North America, andsuggested that an equal amount might come from dry deposition.A higher average concentration of 23 ng Hg l�1 in bulk depositionwas reported by Yang et al. (2002) for a single remote rural sitein Scotland, for one year (1997e1998), with a deposition fluxof 35.9 mg m�2 a�1. Yang et al. (2009) reported annual averageconcentrations for a site in central London to fall from 76.0 ng l�1 in1999 to 43.8 ng l�1 in 2005 (fluxes 45.3e15.0 mg m�2 a�1).

Lee et al. (2001) conducted a modelling study using reportedanthropogenic emissions from UK and other European countries,combined with knowledge and assumptions about long-distancetransport, and the atmospheric generation of RGM. They esti-mated wet deposition of Hg to be 9.9 tonnes for 1998, whichcorresponds to an average deposition rate of 41 mgm�2 a�1 to theUKland surface, but this would include all sites, not just rural remoteones. Ilyin et al. (2008) modelled deposition to Europe for 2006,based on inputs from both anthropogenic and “natural” emissions,and reported an average rate for the UK of 14 mg Hg m�2 a�1.

2.3. Past deposition

The amount of circulating Hg in the atmosphere has increasedby about a factor of three over the last 100e200 years (Mason et al.,

1994), consistent with sediment records of lakes and peat bogs inremote locations (Swain et al., 1992; Biester et al., 2007). However,larger increases are shown by the sediments of lakes closer toemission sources (Landers et al., 1998; Lockhart et al., 2000). Forfour remote undisturbed UK lakes, Yang and Rose (2003) found thatratios of present to past deposition rates were between two andfour, but there were substantial variations in the estimated pre-industrial fluxes of Hg. In two cases the inputs were 4.5 and6 mg Hg m�2 a�1, while in the other two they were much greater,with values of 22 and 40 mg Hgm�2 a�1. Farmer et al. (2009) studiedpeat cores from different parts of Scotland and reported depositionrates of w4 mg Hg m�2 a�1 in the pristine past, rising to maximumrates of 50e200 mg Hgm�2 a�1 in the 20th century. The estimates ofhistorical deposition for the UK are clearly not fully consistent, butseveral of the sites give pre-industrial rates reasonably in accordwith the general historical rate of 3e3.5 mg Hg m�2 a�1 given byBiester et al. (2007), while rates estimated for the 20th century arehigher.

2.4. Historical Hg deposition scenarios for modelling UK soils

Based on the above information, we propose a simple pattern ofannual Hg deposition to the UK land surface over time (Fig. 1).Mercury deposition is considered to come from “global” and “local”sources. From modelling with STOCHEM (Yang et al., 2010) weassume global deposition to be dominated by wet deposition,characterised by a single average concentration of Hg, so the globalflux at any point in the UK is proportional to annual rainfall.Present-day values are assumed to be three times those in thepristine past. We assumed that the increase in rainfall Hg concen-tration began in 1850 and was linear until 1950, after which itremained constant at 5 ng l�1.

The main local UK sources are given by UK National Atmo-spheric Emissions Inventory (NAEI: http://www.naei.org.uk) ascombustion (energy and transformation), iron and steel productionand waste, although considerable amounts of mercury may havebeen added to agricultural land with fertilisers, lime, and manure(Fergusson, 1990). Our historical trend was constructed byassuming no locally-sourced deposition before 1850, then a linearrise to 1960e1970, then a fall to the present-day value. The ratio ofthe 1960e1970 value to that in 2000 is set to 60:7, based on thetonnage of UK anthropogenic emissions for 1970 and 2000 pub-lished by NAEI, and the approximately linear decline between thoseyears. The same pattern is assumed to apply to non-UK localsources.

The same global and local patterns are assumed for all sites. Asalready stated, spatial variation in the global inputs is assumed tobe due only to variations in rainfall. Local inputs are considered to

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e34833476

be specific to each soil sampling point in terms of absolutemagnitude, but are scaled to the common pattern of Fig. 1.

3. Mercury in soil

Herewe describe the data available to characterise UK soils withrespect to mercury, review the key loss processes, and presenta simple dynamic model to simulate Hg in soils at different loca-tions over time. For orientation, Fig. 2 shows the system used forthe modelling analysis. Mercury enters the topsoil in deposition,and can be lost by leaching, bound to DOM, to the lower soil or byevasion. Loss from the lower soil is by DOM-bound transport tosurface water or even deeper soil. As the net result of theseprocesses the soil has a burden of biogeochemically reactive Hg dueto deposition. It also contains unreactive Hg within the parent soilminerals, which is measured as part of the total soil Hg, but theweathering of Hg from this source and its subsequent evasion orleaching are assumed negligible.

3.1. Soil data

Soil data used for this work were described in detail by Tippinget al. (2011). Values of loss-on-ignition (LOI), equivalent to soilorganic matter (SOM), bulk density (BD), and total Hg concentra-tion, are available for rural and urban soils, and soils in the vicinitiesof various industrial activities, for Great Britain and NorthernIreland. All the Hg determinations were made by the sameextraction (hot aqua regia) and measurement (ICP-MS) procedures.The results for Hg were in general agreement with the smallerFOREGS data set for the UK, and with FOREGS data for Europe asawhole (Salminen et al., 2005). The Hg soil contents are regarded as“total”, incorporating both “reactive” metal, able to participate inbiogeochemical reactions and potentially bioavailable, and“unreactive”metal, present in mineral matrices. Values of the latterwere on average 30% of the total, and were estimated by assumingmineral matter to contain 0.05 mg Hg g�1 (Tipping et al., 2011).

The data are from four different surveys, for 108 and 404 ruralsites respectively from the Countryside Surveys of Great Britain for1998 (Black et al., 2002) and 2007 (Spurgeon et al., 2008), 647 sites

RGM

HgP

Hg0

Upper soil Hg-SOM

Hg-DOM

global

local

RGM

HgP

Lower soil Hg-SOM

Hg-DOM

Fig. 2. Schematic of the dynamic model, showing pools and fluxes of reactive Hg.

(366 rural, 87 urban, 194 industrial) from the UK Soil and HerbagePollutant Survey (Environment Agency of England and Wales),conducted in 2001 and 2002 (Barraclough, 2007), and a survey of20 sites in NW England conducted by Tipping et al. (2011). In theCountryside Surveys, soils were sampled to a depth of 15 cm and inthe Soil and Herbage Pollutant Survey to 5 cm, but we previouslyshowed that the 5 cm data could be assumed to apply to 15 cm(Tipping et al., 2011). For the 20 rural soils of NW England, sampledtopsoil depths varied from 10 to 17 cm, and soil immediately belowthe topsoils (layer thicknesses of 7e14 cm) was also sampled.

The locations of the rural soils in Great Britain were knownthrough grid references, and provided a reasonably even distribu-tion. However, grid references were not available for NorthernIreland (which accounts for 6% of the total UK land area) and so aneven spatial occurrence coverage was assumed.

3.2. Leaching of Hg

Mercury is lost from soils by leaching, leading to temperateand boreal zone surface water fluxes mostly in the range1e3 mg m�2 a�1 (Grigal, 2002). Nearly all Hg in soil solution isbound to dissolved organic matter (DOM), while soil organicmatter(SOM) is the dominant sorbent of Hg in the solid phase (Meili, 1991;Schuster, 1991). Thus the leaching behaviour of Hg depends prin-cipally on the partitioning of OM between solid and solution, andthe relative binding of Hg. Åkerblom et al. (2008) demonstratedthat the Hg:OM ratio is approximately the same in each phase, thiswas adopted in the UNECE Mapping Manual (UBA, 2004), andwe make the same assumption. Therefore the Hg leaching flux(Hgleach, g m�2 a�1) can be estimated from the equation;

Hgleach ¼ HgpoolDOMleach=SOMpool (1)

where DOMleach is the leaching flux of DOM (g m�2 a�1), and Hgpool(g m�2), and SOMpool (g m�2) are the amounts of Hg and SOM in thesoil layer of interest. It is convenient to express the Hg leaching fluxin mg m�2 a�1.

For non-peat UK topsoils, an average DOM leaching flux of10 gC m�2 a�1, and for peats a flux of 25 gC m�2 a�1, can beadopted (Buckingham et al., 2008). We made the approximationthat DOC fluxes from lower soil were 25% of those from the uppersoil, because of DOM removal by sorption and subsequentmineralization (Michalzik et al., 2003; Buurman and Jongmans,2005). For non-peat soils, the resulting fluxes correspond to DOCconcentrations of a few mg l�1, as generally observed for UK ruralsurface waters at low discharge when runoff is dominated by flowfrom deeper soil horizons, while for peats baseflow [DOC] issomewhat higher (see e.g. Monteith and Evans, 2000). For thepurposes of this study, we assumed that soils with LOI > 75% werepeats. Organic matter, both solid and dissolved, was assumed to be50% C.

3.3. Evasion from the terrestrial surface

Mercury is lost from soil by reduction to Hg0 followed byevasion, also referred to as emission or volatilisation (Rundgrenet al., 1992; Zhang and Lindberg, 1999; Schlüter, 2000). Factorscontrolling the rate of evasion include temperature, moisture,turbulent mixing, light, interactions with vegetation, sorption tosoil surfaces and microbial activity (Zhang and Lindberg, 1999;Johnson et al., 2003; Choi and Holsen, 2009; Bahlmann et al.,2006; Fritsche et al., 2008). Modelling of mercury exchange at theplot level over short timescales has been performed with some ofthese factors as controlling variables (Scholtz et al., 2003; Bash

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e3483 3477

et al., 2007), but this is currently not possible for application ata national scale, as required here.

For our modelling, we assumed that the primary control on theevasion flux is the amount of reactive mercury in the soil, and thatthe evasion is simply proportional to this pool. Thus the evasionflux (g m�2 a�1) is given by

Hgevasion ¼ kevHgpool (2)

where kev (a�1) is the first-order rate constant, and Hgpool is definedas for equation (1). We compiled published field observations orestimates of evasion fluxes and corresponding measured or esti-mated Hg soil pools, obtaining data for 30 different sites (Table S1).The range of estimated soil Hg pools (1e2000 mg m�2) in thecompilation easily encompassed the observed UK range reportedby Tipping et al. (2011). For each site, a value of kev was derived,a median value of 0.0028 a�1 being obtained with a range of valuesfrom 0.0003 to 0.016 a�1. As a rounded default value of kev weadopted 0.0025 a�1. For sensitivity analysis, we also performedcalculations with values of kev of 0.001 and 0.006 a�1, covering 60%of the derived values. The fairly strong relationship (r2 ¼ 0.76)between the logarithms of the evasion flux and the soil pool(Fig. S1) supports the first-order assumption.

Because equations (1) and (2) have similar forms, an averagefirst-order leaching constant (kle) can be derived from the ratioDOMleach/SOMpool. Assuming organic matter to be 50% C, then themean values of DOCleach (10 g m�2 a�1) and SOC (7000 gC m�2)produce kle ¼ 0.0014 a�1, which is about half of the evasion rateconstant kev. Thus Hg losses from soil by the evasion and leachingroutes are of comparable magnitude.

Table 1Properties of three contrasting topsoils, and model outputs for three values of kev.

A B C

ObservationsSOM % 5.9 14.7 13.8BD g cm�3 1.1 0.7 1.2Total Hg mg g�1 0.035a 0.184 0.499Average annual rainfall mm a�1 1423 762 640Model outputs kev ¼ 0.001 a�1

Global Hg deposition 2000 mg m�2 a�1 7 4 3Local Hg deposition 1960e1970 mg m�2 a�1 0 197 1073Local Hg deposition 2000 mg m�2 a�1 0 23 125Hg evasion 2000 mg m�2 a�1 2 15 80Hg leaching flux 2000 mg m�2 a�1 1 19 61Default model outputs kev ¼ 0.0025 a�1

Global Hg deposition 2000 mg m�2 a�1 7 4 3Local Hg deposition 1960e1970 mg m�2 a�1 0 214 1176Local Hg deposition 2000 mg m�2 a�1 0 25 137Hg evasion 2000 mg m�2 a�1 3 37 201Hg leaching flux 2000 mg m�2 a�1 1 19 61Model outputs kev ¼ 0.006 a�1

Global Hg deposition 2000 mg m�2 a�1 7 4 3Local Hg deposition 1960e1970 mg m�2 a�1 0 260 1437local Hg deposition 2000 mg m�2 a�1 0 30 168Hg evasion 2000 mg m�2 a�1 4 87 486Hg leaching flux 2000 mg m�2 a�1 1 19 62

a 50% of the analytical detection limit.

3.4. The dynamic soil model

The model is run on an annual time-step. To obtain a depositionscenario at a given site, present-day global deposition is calculatedfrom rainfall, and a trial value for the present-day local depositionchosen. These are then scaled as in Fig. 1 to generate total deposi-tional inputs for each year of calculation, including pre-1850 wheninput is assumed to have been due only to natural global Hg cycling.The DOM fluxes (from topsoil and lower soil) are assumed constantover time. The soil is assumed to have contained no reactive Hg10,000 years ago, i.e. soon after the retreat of glaciers that coveredmuch of the UK. The model then simulates the accumulation ofreactive Hg until the year in which the soil was sampled, and theestimated unreactive Hg added to obtain the calculated total soil Hgconcentration. Comparison with the observed value yields a newtrial value of present-day deposition, and the process is repeateduntil calculated and observed topsoil total Hg concentrationsmatch, or the local input falls to zero, at which point smalldisagreements between observed and calculated soil Hg concen-tration must be accepted.

This model is similar to the one employed to estimate re-emission of Hg within the European Monitoring and EvaluationProgramme (EMEP) MSCE-HM atmospheric model of heavy metaltransboundary air pollution in Europe (Travnikov and Ilyin, 2005;Ryaboshapko and Ilyin, 2001). The differences are (a) our use oftwo soil layers and two leaching fluxes, and (b) the use of a shortertimescale and temperature-dependence of evasion in the MSCE-HM model. Our model uses measured soil Hg to estimate deposi-tion, the inverse of the mode of application of theMSCE-HMmodel.

For mapping, model outputs were spatially interpolated usingthe gstat package (http://www.gstat.org/) in R (http://www.r-project.org/). After inspecting the variograms, we chose toperform kriging by fitting a spherical data model with a range of30 km.

4. Results

4.1. Model applications to specific sites

Table 1 summarizes information about three contrastingtopsoils, one in a remote location unlikely to be affected by localdeposition (A), the second rurally located but with appreciable localHg inputs (B), and the third near an industrial facility with highlocal deposition (C). As described in Section 3.4, in each case thecurrent local depositionwas adjusted to force a match between theobserved and calculated topsoil total Hg concentration. Results areshown for three values of kev, the default (0.0025 a�1) and theselected lower and higher values of 0.001 and 0.006 a�1 (Section3.3). We did not perform an uncertainty analysis for leaching lossesbecause although DOC flux varies among UK rural soils (relativestandard deviation c. 50%), the average is reasonably well defined(Buckingham et al., 2008). In contrast, the estimates of kev are fromdisparate observations in a range of soils, none of which are in theUK (see Supplementary information).

As shown by the time-series results of total soil Hg concentra-tions for the default kev, plotted in Fig. 3, the simulated topsoil Hgconcentrations in the polluted soils B and C reached a peak in about1980, and then fell slightly to the present-day values. The back-ground soil continued to gain Hg but at a low level, not evident inthe Figure. Similar patterns are seen for the other two kev values.With kev ¼ 0.001 a�1 the peak soil Hg concentration occurs earlierand is lower, and the decline is smaller, while the opposite trendsapply with kev ¼ 0.006 a�1, but the differences are slight. Thedifferent choices of kev results in different evasion fluxes for 2000(Table 1), and this means that the contribution of leaching to theloss of Hg from the topsoil falls from 40e50% for kev ¼ 0.001 a�1 to10e20% for kev ¼ 0.006 a�1. Another significant feature is thata higher evasion rate constant demands greater deposition rates toattain a given soil Hg concentration.

4.2. Spatially resolved Hg dynamics at rural sites

The model was used to calculate present-day local Hg deposi-tion for each of the 898 rural sites. For the default kev, 30% of the

0

0.1

0.2

0.3

0.4

0.5

0.6

1750 1800 1850 1900 1950 2000

tota

l soi

l Hg µ

g g-1

A

B

C

Fig. 3. Time-dependence of total topsoil Hg concentrations in the three soils of Table 1.

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e34833478

sites are calculated to receive no local deposition, and only in 51%does the local exceed the global. In 10% of cases the local depositionis 5 times or more that of the global. Thus there is considerablevariation in calculated local Hg deposition.

Inspection of the outputs showed that the results were spatiallyauto-correlated and so mapping by interpolation (kriging) wasjustified. Fig. 4 shows maps of deposition from the global pool andlocal sources separately, and their combination. The areas ofgreatest local deposition are in the north midlands and north-westof England, and the area around London. Not surprisingly, thepattern is similar to that of soil Hg pools mapped by Tipping et al.(2011). Evasion is greatest where deposition and reactive soil Hgare greatest, and so the evasion map looks like the deposition mapin terms of spatial pattern.

The mapped deposition fields were used to compare calculateddeposition rates with measured wet deposition values at the 10sites of the CEH monitoring network (Section 2.2; Rowland et al.,2010a). If only the calculated globally-sourced Hg deposition ratesare compared with the measured ones, quite good agreement isobtained, as would be expected in view of our choice of a fixedconcentration of 5 ng l�1 for rainfall due to global Hg (Section 2.4).However, at all sites some local deposition was estimated byinverse modelling and so all calculated total Hg deposition valueswere considerably higher than the observations. In 8 cases, totalcalculated Hg was 3e5 times the measured value, while in twocases the ratio was 40e50 times.

The deposition of Hg in Europe is modelled with the EMEPMSCE-HM atmospheric model of heavy metal transboundary airpollution in Europe (Travnikov and Ilyin, 2005; Ryaboshapko andIlyin, 2001), the results being used to evaluate soil ecologicaleffects of Hg through the Critical Loads approach (Hettelingh andSliggers, 2006; Slootweg et al., 2007). The MSCE-HM model takesaccount of primary emissions from anthropogenic activities, andalso both natural emission and re-emission of pollutant Hg.Modelled results for the year 2000 gave an average total depositionof 13.5 mg Hg m�2 a�1 to the UK, only slightly lower than our valueof 16.3 mg Hg m�2 a�1 for rural sites. Furthermore the patterns ofdeposition calculated by the two approaches show broad similarity(Fig. 4), the highest deposition rates being predicted for central andnorthern England. However there is a difference with respect toevasion, even though we use the same rate constant. Whereas weestimate a total evasion rate of 16.1 mg Hg m�2 a�1 for rural soils in2000, the EMEP value is only 4.8 mg Hg m�2 a�1. This differencemust reflect the continual greater local inputs of Hg calculated byinverse modelling, which have led to larger soil Hg pools andtherefore greater evasion rates.

4.3. Non-rural sites

Soil data for sites in urban and industrial areas came from theUK Soil and Herbage Pollutant Survey (Barraclough, 2007), forwhich site selection was not made with the aim of obtaininga representative spatial coverage. However, we assume that thecombined data (281 points) for the sampled soils providesa reasonable measure of Hg in non-rural soils, which account forapproximately 6% of the UK land area. Measured data for these sitesare summarised in Table 2, and compared with model results,assuming all Hg in non-rural soils to be due to atmospheric depo-sition. The present-day total depositional load to these sites iscalculated to be 1.0 tonnes, i.e. 27% of the total load to rural sites.Note that if this input is added to the rural load, the averagedepositional flux to the UK increases from 16.3 to 19.3 mg Hgm�2 a�1 (see Section 4.2).

4.4. The overall UK budget of mercury over time

Table 3 shows the pools and loads for 1800, 1900, 1970 and2000, calculated with the default kev of 0.0025 a�1. The results referto a simple average of rural data, combined with averaged resultsfor non-rural sites and weighted according to rural and non-ruralareas to produce values for the UK as a whole. Very similarresults are obtained if the rural data are aggregated using thegridded maps of the land area of the UK at 5 km resolution, whichindicates a good representative spatial coverage. Fig. S2 shows theresults expressed as average amounts per unit area, based on thescheme of Fig. 2.

With the default kev we find that in 2000, for the UK as a whole,Hg deposition from global sources was about 30% of the total, therest coming from local sources. This represents a major changefrom the situation in 1970, when global deposition comprised onlyabout 5% of the total, due to the much greater local deposition. Themean evasion rate in 2000 exceeded the mean deposition and thiswill not have changed significantly since, so the soils are currentlya source of Hg0 to the atmosphere. This output of 5.1 tonnes is about70% of the total anthropogenic emissions from the UK.

These overall UK estimates depend upon the choice of kev (seeTable S2). If a lower value is assumed, then the calculated total soilHg pools tend to be higher, although not greatly so. The maindifference is in the evasion flux, which for the year 2000 is esti-mated at 2.1 tonnes with kev ¼ 0.001 a�1, but 12.4 tonnes withkev ¼ 0.006 a�1. The local deposition is calculated to increase withkev; for the year 2000 it is 3.0 tonnes with kev ¼ 0.001 a�1,4.2 tonnes with kev ¼ 0.006 a�1. Mapped calculated total Hgdeposition for the years 1900, 1970 and 2000 are shown in Fig. S3.

4.5. Mercury concentrations in the lower soil and stream drainage

For 20 sites in NW England we determined Hg soil concentra-tions in both the topsoil and in the soil immediately beneath it. Themodel forces agreement between observed and simulated Hgconcentration in the topsoil, but predicts the lower soil concen-tration. The observed lower soil Hg concentrations ranged from0.14 to 0.44 mg g�1 with a mean of 0.22 mg g�1 (standard deviation0.09 mg g�1). With kev ¼ 0.0025 a�1 the predicted range was0.07e0.30 mg Hg g�1 with a mean of 0.16 mg g�1 (standard deviation0.09 mg g�1). The calculated mean concentration is significantlylower (t-test, p < 0.05) but is nonetheless in reasonable agreementwith the observed mean. The measured and calculated values werenot correlated, which must partly reflect their small ranges. The fullresults are given in Table S3, which also shows results forkev ¼ 0.001 and 0.006 a�1. Closer agreement between observed andcalculated values is found with kev ¼ 0.001 a�1 since the slower

Fig. 4. Modelled Hg deposition (mg m�2 a�1) to rural sites in 2000. The top two maps and left-hand lower map are derived from inverse modelling from soil Hg, the right-handlower map uses output from the EMEP MSCE-HM model (see Section 4.2).

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e3483 3479

Table 2Summary of results for 281 non-rural (urban and industrial) topsoils. Modelledresults refer to a 15 cm soil depth with kev ¼ 0.0025 a�1.

Observed total Hg concentration mg g�1 0.33Observed total Hg pool mg m�2 42

Modelled global deposition 1970 mg m�2 4.5Modelled global deposition 2000 mg m�2 4.5

Modelled local deposition 1970 mg m�2 554Modelled local deposition 2000 mg m�2 65

0

0.1

0.2

0.3

0.4

0.5

0.6

1800 1900 2000 2100 2200

tota

l soi

l Hg µ

g g-1

C

B

Fig. 5. Predicted total topsoil Hg concentrations for soils B and C of Table 1. Full lines;Hg deposition held constant at current values. Dashed lines; local Hg deposition falls tozero over the period 2010e2030.

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e34833480

evasion leads to a greater calculated leaching flux to the lower soilsand hence greater lower soil Hg concentrations.

We compared measured filtered stream Hg concentrations inthree upland catchments, with model predictions of Hg concen-tration in water leaving the lower soil. The mean measured Hgconcentration over the three sites was 2.9 ng l�1, and the calculatedone 1.9 ng l�1, while the corresponding mean Hg:DOM ratios were0.30 and 0.38 mg g�1 (details for the individual sites are given inTable S4). This can be considered reasonable agreement. Anothercomparison that can be made is with the results for unfiltered Hgfor a larger river system in NW England studied by Rowland et al.(2010b) which yield a mean Hg:DOM ratio of 0.41 mg g�1. The NWof England has relatively high levels of Hg in rural sites; for theentire rural data set the median and mean Hg:DOM ratios calcu-lated for water leaching from the lower soil are 0.18 and 0.22 mg g�1

respectively.The leaching fluxes of Hg from the lower soil are small,

compared to both deposition and evasion (Table 3, Fig. S2).Considering just the rural sites, the calculated 5-percentile, medianand 95-percentile leaching fluxes of Hg from the topsoil were 0.8,7.4 and 31.9 mg m�2 a�1 in 2000, whereas the corresponding fluxesfrom the lower soil were 0.5, 1.0 and 3.5 mg m�2 a�1, i.e. anappreciably narrower range. Thus the model calculates consider-able increases over time in the leaching flux of Hg from the topsoil,but sorptive accumulation of the metal has buffered leaching fromthe lower soil.

4.6. Future soil Hg concentrations

We ran the default model for the two polluted soils (B and C) ofTable 1 and Fig. 3 under two simple future scenarios. Firstly bothglobal and local deposition of Hg were maintained constant atcurrent levels. Secondly the local deposition was reduced to zeroover 20 years from 2010, but global deposition was maintainedconstant. In the first case, soil Hg falls by about 20% from the peakvalue by 2200, in the second by about 40% (Fig. 5). To approach towithin 10% of the final steady state, corresponding to backgrounddeposition after removal of all local inputs, would take about 1500years. If the model is run with kev ¼ 0.006 a�1 for the scenario inwhich local Hg deposition falls to zero by 2030, about half of the soil

Table 3Estimated Hg budget over time for the United Kingdom, values in tonnes. Calculatedwith kev ¼ 0.0025 a�1.

1800 1900 1970 2000

Annual deposition global 0.5 0.9 1.4 1.4Annual deposition local 0 13.2 29.0 3.4Annual evasion from soils 0.3 1.0 4.4 5.1Annual leaching from topsoils 0.2 0.7 2.9 3.4Annual leaching from lower soils 0.2 0.2 0.3 0.3

Annual anthropogenic emissions 0 27 60 7

Reactive Hg soil poola 350 690 2140 2490

a Topsoil and lower soil combined.

Hg pool is predicted to have been lost by 2020, whereas if kev is setto 0.001 a�1 then the fractional loss is only about 20% (Fig. S3).

5. Discussion

Themodel used here for Hg is simpler than the CHemistry of theUplands Model, which we used previously to simulate the long-term dynamics of atmospherically-deposited Ni, Cu, Zn, Cd andPb in UK catchments (Tipping et al., 2006, 2007, 2010b). Simplifi-cation is possible because the very strong binding of Hg to organicmatter means that inorganic fluxes of Hg can be ignored. However,our underlying picture of Hg(II) chemistry is still based on itsphysico-chemical partitioning between SOM and DOM; the inor-ganic forms are still chemically active, and the key to biologicaluptake and toxicity, but quantitatively negligible in terms of outputfluxes from soil. This accords with the conceptual model of Zhangand Lindberg (1999), who also described the soil behaviour ofHg(II) in terms of strong but reversible chemical reactions involvingsorption to SOM and soil surfaces, but added transformation to Hg0

and consequent evasion. A different conceptual approach has beentaken by other researchers, who propose that Hg is essentiallyirreversibly associated with SOM (and DOM), so the metal dependsfor release on the decomposition of the organic matter (Meili, 1991;Smith-Downey et al., 2010).

Our analysis highlights evasion of Hg0 as the principal processgoverning Hg dynamics in soils. Even with the lowest value of kevconsidered (0.001 a�1) the loss of Hg by the evasion route consid-erably exceeds the leaching of Hg from the lower soil (Table S2),although in this case the leaching flux from the upper soil slightlyexceeds the evasion flux. Our representation of the evasion processis simple, the flux being directly proportional to the soil pool ofreactive Hg, with a proportionality constant estimated from pub-lished field data (Table S1). The default evasion constant derived inthis way is equivalent to (the reciprocal of) the residence time of400 years adopted by Ryaboshapko and Ilyin (2001). Similarly, Selinet al. (2008), in a global modelling effort, quoted a turnover time ofHg in soil of 100e1000 years, which includes the reciprocal of ourdefault kev and approximately coincides with the values used toassess sensitivity (Table 1). Despite this apparent agreementhowever, it is clear from the literature reviewed briefly in Section3.3 that the exchange of Hg0 with terrestrial ecosystems (plant andsoil) is complex, and considerable uncertainties exist (Gustin et al.,2008), including the possibility that soil Hg is heterogeneous, e.g.partitioned between “old” and “new” Hg (Hintelmann et al., 2002).

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e3483 3481

Thus while the simple representation of evasion used here can beconsidered reasonable, appropriate caution should be exercisedwhen considering its predictions. Model outputs based on a rangeof kev from 0.001 to 0.006 a�1 provide guidance as to the sensitivityof soils to the choice of evasion flux (Table 1, Table S2, Fig. S3).

The other loss process for Hg from soil is by leaching, complexedwith DOM. According to the calculations, substantial transport ofHg from the topsoil to the lower soil takes place. Currently this fluxis about two-thirds of the evasion flux for kev ¼ 0.0025 a�1

(Table 3). However, the loss of Hg from the lower soil is small andstrongly regulated by sorption (Section 4.5; see also Grigal, 2002).Therefore, natural Hg, accumulated in soil before anthropogenicemissions took place, may still be playing a significant role in theleaching of Hg to surface waters. This might in turn explain whysurface water Hg:DOM ratios and Hg fluxes vary relatively little.Thus Grigal (2002) collated results showing Hg:DOM ratios of c.0.1e0.15 mg g�1 for relatively unimpacted sites in North Americaand Sweden, values not much smaller than our calculated valuesfor UK rural sites, which have a median of 0.18 (Section 4.5). Grigalalso highlighted the relatively small range of surface water Hgfluxes, with 75% of reported values falling within the range1e3 mg m�2 a�1.

Some other modelling simplifications should be admitted. Wehave not included the loss of Hg due to fire (Schroeder andMunthe,1998; Cinnirella and Pirrone, 2006; Wiedinmyer and Friedli, 2007),which may be especially important for UK moorlands wherecontrolled burning of heather is a land management tool. Removalof mercury in harvested crops or trees is neglected. The historicaldeposition pattern is highly simple and assumed to be the same ateach site. The sequential passage of water through the two layersprobably does not apply to the organic-rich blanket peats thatcharacterise appreciable areas of the UK uplands, for which surfacerunoff is common, and therefore the computations for the lowersoils in such areas are inaccurate. No account is taken of likelytemporal changes in soil DOM concentrations and fluxes (see e.g.Monteith et al., 2007). However, elaboration of the model to dealwith these shortcomings would be unlikely to change the mainresults or conclusions significantly.

As noted in Section 4.2, our deposition rates estimated byinverse modelling are all greater than Hg wet deposition reportedby Rowland et al. (2010a) for the 10 sites of the UK metals moni-toring network. In the eight cases for which the discrepancy isa factor five or less, this may partly be explained by the absence ofdry deposition estimates in the measured values, which can besignificant (Miller et al., 2005), but this would not explain theresults for the two sites where deposition is 50 times too high.Another comparison that can be made is for Lochnagar in thenorth-east Scotland, for which we estimate a total deposition rateof c. 15 mg m�2 a�1 in 1998, appreciably lower than the value of35.9 mg m�2 a�1 for the period 1997e1998 reported by Yang et al.(2002). Less direct estimates of rural Hg deposition rates frompeat bog records in Scotland (Farmer et al., 2009) gave quite highvalues (Section 2.3), and again here we estimate lower rates. Wecalculate a mean deposition rate of 70 mg m�2 a�1 for non-ruralsites (Table 2), in fair agreement with the results of Yang et al.(2009) for a single site in London (Section 2.2). As described inSection 4.2, our estimates of deposition for 2000 are in fair agree-ment with results from the EMEP MSCE-HM atmospheric model.Lee et al. (2001) carried out atmospheric modelling at higherresolution, specifically for the UK, and estimated a load of9.9 tonnes of Hg to the UK in 1998, made up of 2.5 tonnes from theglobal circulation and 7.4 tonnes fromUK and European sources, i.e.local deposition in our terms. Our calculated values for 1998 are1.4 tonnes of global deposition and 5.1 tonnes of local deposition(somewhat greater than the value for 2000 given in Table 3 because

of the decline in emissions and therefore deposition), makinga total of 6.5 tonnes, i.e. about two-thirds of the Lee et al. estimate.Overall, our estimates of Hg deposition from inverse modelling fallin the middle of the range of values from measurement andatmospheric modelling, while our dynamic model makes reason-able predictions of lower soil Hg concentrations and fluxes tosurface waters (Section 4.5). Therefore the modelling approachappears justified and its predictions should be useful.

Calculations of the UK mercury budget summarised in Table 3show soils (rural and non-rural combined) of the UK to holda large amount of reactive Hg, with 2490 tonnes of the metal intotal. Since the soils were close to steady state in 1850 (i.e. wouldnot have accumulated more non-anthropogenic Hg), this repre-sents an accumulation of 2140 tonnes, a 7-fold enrichment, due toanthropogenic emissions. The total anthropogenic deposition since1850 is calculated to have been 2460 tonnes, of which 2370 tonnes(96%) was from local sources, and 90 tonnes from the globalcirculation. Therefore 87% of the total anthropogenic depositionremains in the soil, and will continue to be a source of Hg0 to theatmosphere. We estimate the present evasion flux to be5.1 tonnes a�1, which is c. 70% of current anthropogenic emissions.However, since about half of the latter are in forms other than Hg0

(Section 2.1) then our calculations suggest that emissions to theglobal atmosphere from UK soils are currently c. 50% greater thanall combined UK anthropogenic sources.

Because local Hg deposition has fallen markedly over recentdecades (Fig. 1, Section 2.4), soil Hg concentrations in the UK arepredicted to decline (Fig. 5). Although hundreds of years will beneeded to approach background levels, further reductions in localanthropogenic emissions are likely to hasten the process. Since themain loss process is evasion, this recovery of UK soils will be at theexpense of a long-term contribution of Hg0 to the global circulation.

6. Conclusions

- Topsoil mercury concentrations in the UK can be approxi-mately accounted for in terms of Hg deposition from the globalcirculation and local sources, taking account of likely historicaldeposition patterns, and losses by evasion of Hg0 and leachingof Hg complexed by dissolved organic matter (DOM).

- Inverse modelling of Hg deposition indicates that 30% of ruralsites receive Hg only from the global circulation. In 51% of caseslocal deposition exceeds global, while in 10% it is 5 times ormore that of the global. The greatest deposition rates occur incentral and northern England.

- The average total rate of Hg deposition to rural sites in the UKin 2000 is estimated by inverse modelling to have been16 mg m�2 a�1, close to the value of 13.5 mg m�2 a�1 estimatedby atmospheric transport modelling with the EMEP MSCE-HMmodel. Deposition to non-rural sites was 70 mgm�2 a�1 if all Hgis assumed to be atmospheric in origin. The overall depositionrate to the UK was 19 mg m�2 a�1.

- UK soils currently hold 2490 tonnes of reactive Hg, of which2140 tonnes are due to anthropogenic deposition since 1850,representing a 7-fold enrichment overall. Of the 2460 tonnes oftotal anthropogenic deposition since 1850, 96% is local in originand 87% has been retained by soils.

- Soil Hg is currently a source of Hg0 to the atmosphere, with anestimated evasion flux of 5.1 tonnes a�1, corresponding to c.70% of current anthropogenic emissions. Since only about halfof the latter are in the form of Hg0 then emissions to the globalatmosphere from UK soils are currently c. 50% greater than allcombined UK anthropogenic sources.

- The overall leaching flux of DOM-bound Hg from topsoil tolower soil is slightly less than the evasion flux, but sorptive

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e34833482

retention exerts a strong control on leaching in the lower soiland therefore on surface water Hg concentrations.

- If both global and local deposition of Hg were maintainedconstant at current levels, the Hg contents of contaminatedsoils would fall by about 20% from their 1980 peak value by2200.Were local deposition reduced to zero over 20 years from2010, but global deposition maintained constant, a decrease ofabout 40%would be achieved. To approach to within 10% of thefinal steady state, corresponding to background depositionafter removal of all local inputs, would take about 1500 years.

- The key role of Hg0 evasion in controlling soil Hg levels meansthat research is required to improve its representation inmodels operating at large spatial scales.

Acknowledgements

We thank D. Barraclough (Environment Agency of England andWales) for making the UK Soil and Herbage Survey data available,C.M. Wood (CEH) for help with accessing Countryside Survey data,and J. Poskitt, A.J. Lawlor and J. Antelo for carrying out fieldworkand analysis. We are grateful to R.G. Derwent, P.J. Coleman (Defra)and J.N. Cape (CEH) for their helpful comments on the draftmanuscript. This work was financially supported under contractAQ0805, by the UK Department for Environment, Food and RuralAffairs, the Scottish Executive, the National Assembly of Wales andthe Department of the Environment (in Northern Ireland), and byNERC.

Appendix. Supplementary information

Supplementary information associated with this article can befound, in the online version, at doi:10.1016/j.envpol.2011.08.019.

References

Åkerblom, S., Meili, M., Bringmark, L., Johansson, K., Kleja, D.B., Bergkvist, B., 2008.Partitioning of Hg between solid and dissolved organic matter in the humuslayer of boreal forests. Water, Air, and Soil Pollution 189, 239e252.

Bahlmann, E., Ebinghaus, R., Ruck, W., 2006. Development and application ofa laboratory flux measurement system (LFMS) for the investigation of thekinetics of mercury emissions from soils. Journal of Environmental Manage-ment 81, 114e125.

Barraclough, D., 2007. UK Soil and Herbage Pollutant Survey. UKSHS Report No. 1.Environment Agency, Bristol.

Bash, J.O., Bresnahan, P., Miller, D.R., 2007. Dynamic surface interface exchanges ofmercury: a review and compartmentalized modeling framework. Journal ofApplied Meteorology and Climatology 46, 1606e1618.

Biester, H., Bindler, R., Martinez-Cortizas, A., Engstrom, D., 2007. Modeling the pastatmospheric deposition of mercury using natural archives. EnvironmentalScience & Technology 41, 4851e4860.

Black, H.I.J., Garnett, J.S., Ainsworth, G., Coward, P.A., Creamer, R., Ellwood, S.,Horne, J., Hornung, M., Kennedy, V.H., Monson, F., Raine, L., Osborn, D.,Parekh, N.R., Parrington, J., Poskitt, J.M., Potter, E., Reeves, N., Rowland, A.P.,Self, P., Turner, S., Watkins, J., Woods, C., Wright, J., 2002. MASQ: Monitoring andAssessing Soil Quality in Great Britain. Survey Model 6: Soils and Pollution.Environment Agency, Bristol.

Buckingham, S., Tipping, E., Hamilton-Taylor, J., 2008. Concentrations and fluxes ofdissolved organic carbon in UK topsoils. Science of the Total Environment 407,460e470.

Buurman, P., Jongmans, A.G., 2005. Podzolisation and soil organic matter dynamics.Geoderma 125, 71e83.

Choi, H.D., Holsen, T.M., 2009. Gaseous mercury emissions from unsterilized andsterilized soils: the effect of temperature and UV radiation. EnvironmentalPollution, 1673e1678.

Cinnirella, S., Pirrone, N., 2006. Spatial and temporal distributions of mercuryemissions from forest fires in Mediterranean region and Russian Federation.Atmospheric Environment 40, 7346e7361.

Dore, C.J., Watterson, J.D., Murrells, T.P., Passant, N.R., Hobson, M.M., Choudrie, S.L.,Thistlethwaite, G., Wagner, A., Jackson, J., Li, Y., Bush, T., King, K.R., Norris, J.,Coleman, P.J., Walker, C., Stewart, R.A., Goodwin, J.W.L., Tsagatakis, I., Conolly, C.,Downes, M.K., Brophy, N., Hann, M.R., 2007. UK Emissions of Air Pollutants 1970to 2005. National Atmospheric Emission Inventory (NAEI) Annual Report. AEAEnergy and Environment, Harwell. http://www.airquality.co.uk/reports/cat07/0801140937_2005_Report_FINAL.pdf.

Farmer, J.G., Anderson, P., Cloy, J.M., Graham, M.C., MacKenzie, A.B., Cook, G.T., 2009.Historical accumulation rates ofmercury in four Scottish ombrotrophic peat bogsover the past 2000 years. Science of the Total Environment 407, 5578e5588.

Fergusson, J., 1990. The Heavy Elements: Chemistry, Environmental Impact andHealth Effects. Pergamon, Oxford.

Fritsche, J., Obrist, D., Alewell, C., 2008. Evidence of microbial control of Hg0

emissions from uncontaminated terrestrial soils. Journal of Plant Nutrition andSoil Science 171, 200e209.

Grigal, D.F., 2002. Inputs and outputs of mercury from terrestrial watersheds:a review. Environmental Reviews 10, 1e39.

Gustin, M.S., Lindberg, S.E., Weisberg, P.J., 2008. An update on the natural sourcesand sinks of atmospheric mercury. Applied Geochemistry 23, 482e493.

Hettelingh, J.P., Sliggers, J. (Eds.), 2006. Heavy Metal Emissions, Depositions, CriticalLoads and Exceedances in Europe. Ministry of the Environment VROM DGM,Netherlands.

Hintelmann, H., Harris, R., Heyes, A., Hurley, J.P., Kelly, C., Krabbenhoft, D.P.,Lindberg, S., Rudd, J.W.M., Scott, K.J., St Louis, V.L., 2002. Reactivity and mobilityof new and old mercury deposition in a boreal forest ecosystem during the firstyear of the METAALICUS study. Environmental Science and Technology 36,5034e5040.

Hynes, A.J., Donohoue, D.L., Goodsite, M.E., Hedgecock, I.M., 2009. Our currentunderstanding of major chemical and physical processes affecting mercurydynamics in the atmosphere and at the air-water/terrestrial interfaces. In:Pirrone, N., Mason, R. (Eds.), Mercury Fate and Transport in the Global Atmo-sphere. Springer, Dordrecht, pp. 427e457.

Ilyin, I., Rozovskaya, O., Travnikov, O., Hettelingh, J.P., Reinds, G.J., 2008. HeavyMetals: Transboundary Pollution of the Environment. EMEP/MSC-E StatusReport 2/2008.

Johnson, D.W., Benesch, J.A., Gustin, M.S., Schorran, D.S., Lindberg, S.E., Coleman, J.S.,2003. Experimental evidence against diffusion control of Hg evasion from soils.Science of the Total Environment 304, 175e184.

Landers, D.H., Gubala, C., Verta, M., Lucotte, M., Johansson, K., Vlasova, T.,Lockhart, W.L., 1998. Using lake sediment mercury flux ratios to evaluate theregional and continental dimensions of mercury deposition in arctic and borealecosystems. Atmospheric Environment 32, 919e928.

Lee, D.S., Nemitz, E., Fowler, D., Kingdon, R.D., 2001. Modelling atmosphericmercury transport and deposition across Europe and the UK. AtmosphericEnvironment 35, 5455e5466.

Lockhart, W.L., Macdonald, R.W., Outridge, P.M., Wilkinson, P., DeLaronde, J.B.,Rudd, J.W.M., 2000. Tests of the fidelity of lake sediment core records ofmercury deposition to known histories of mercury contamination. Science ofthe Total Environment 260, 171e180.

Mason, R.P., Sheu, G.R., 2002. Role of the ocean in the global mercury cycle. GlobalBiogeochemical Cycles 16, 1093. doi:10.1029/2001GB001440.

Mason, R.P., Fitzgerald, W.F., Morel, F.M.M., 1994. The biogeochemical cycling ofelemental mercury: anthropogenic influences. Geochimica et CosmochimicaActa 58, 3191e3198.

Meili, M., 1991. The coupling of mercury and organic matter in the biogeochemicalcycle e towards a mechanistic model for the boreal forest zone. Water, Air, andSoil Pollution 56, 333e347.

Michalzik, B., Tipping, E., Mulder, J., Gallardo Lancho, J.F., Matzner, E., Bryant, C.L.,Clarke, N., Lofts, S., Vicente Esteban, M.A., 2003. Modelling the production andtransport of dissolved organic carbon in forest soils. Biogeochemistry 66,241e264.

Miller, E.K., Vanarsdale, A., Keeler, G.J., Chalmers, A., Poissant, L., Kamman, N.C.,Brulotte, R., 2005. Estimation and mapping of wet and dry mercury depositionacross northeastern North America. Ecotoxicology 14, 53e70.

Monteith, D.T., Evans, C.D., 2000. UK Acid Waters Monitoring Network: 10 YearReport. Ensis Ltd, London.

Monteith, D.T., Stoddard, J.L., Evans, C.D., de Wit, H.A., Forsius, M., Høgåsen, T.,Wilander, A., Skjelkvåle, B.L., Jeffries, D.S., Vuorenmaa, J., Keller, B., Kopácek, J.,Vesely, J., 2007. Dissolved organic carbon trends resulting from changes inatmospheric deposition chemistry. Nature 450, 537e541.

Nater, E.A., Grigal, D.F., 1992. Regional trends in mercury distribution across theGreat Lakes states, north central USA. Nature (London) 358, 139e141.

Pacyna, E.G., Pacyna, J.M., 2002. Global emission of mercury from anthropogenicsources in 1995. Water, Air, and Soil Pollution 137, 149e165.

Pai, P., Karamchandani, P., Seigneur, C., Allan, M.A., 1999. Sensitivity of simulatedatmospheric mercury concentrations and deposition to model input parame-ters. Journal of Geophysical Research e Atmospheres 104, 13,855e13,868.

Pirrone, N., Hedgecock, I.M., Sprovieri, F., 2008. New directions: atmosphericmercury, easy to spot and hard to pin down: impasse? Atmospheric Environ-ment 42, 8549e8551.

Rowland, A.P., Lawlor, A.J., Guyatt, H.J., Wadsworth, R.A., 2010a. Background wetdeposition of mercury in Great Britain. Journal of Environmental Monitoring 12,1747e1755.

Rowland, A.P., Neal, C., Scholefield, P., Halford, A.P., Vincent, C.D., Hockenhull, K.,2010b. Mercury in UK rivers: from rural headwaters to the heartlands of thehistoric industrial base. Journal of Environmental Monitoring 12, 2299e2306.

Rundgren, S., Rühling, A., Schlüter, A., Tyler, G., 1992. Mercury in Soil e Distribution,Speciation and Biological Effects. Nordic Council of Ministers, Copenhagen.

Ryaboshapko, A., Ilyin, I., 2001. Mercury re-emission to the atmosphere in Europe.In: Midgley, P.M., Reuther, M.J., Williams, M. (Eds.), Proceedings of EUROTRACSymposium 2000. Transport and Chemical Transformation in the Troposphere.Springer-Verlag, Berlin Heidelberg.

E. Tipping et al. / Environmental Pollution 159 (2011) 3474e3483 3483

Salminen, R., Batista, M.J., Bidovec, M., Demetriades, A., De Vivo, B., De Vos, W.,Duris, M., Gilucis, A., Gregorauskiene, V., Halamic, J., Heitzmann, P., Lima, A.,Jordan, G., Klaver, G., Klein, P., Lis, J., Locutura, J., Marsina, K., Mazreku, A.,O’Connor, P.J., Olsson, S.Ǻ., Ottesen, R.-T., Petersell, V., Plant, J.A., Reeder, S.,Salpeteur, I., Sandström, H., Siewers, U., Steenfelt, A., Tarvainen, T., 2005.Geochemical Atlas of Europe. Part 1 e Background Information, Methodologyand Maps. Geological Survey of Finland, Espoo.

Schlüter, K., 2000. Review: evaporation of mercury from soils. An integration andsynthesis of current knowledge. Environmental Geology 39, 249e271.

Scholtz, M.T., Van Heyst, B.J., Schroeder, W.H., 2003. Modelling of mercury emis-sions from background soils. Science of the Total Environment 304, 185e207.

Schroeder, W.H., Munthe, J., 1998. Atmospheric mercury e an overview. Atmo-spheric Environment 32, 809e822.

Schuster, E., 1991. The behaviour of mercury in the soil with specific emphasis oncomplexation and adsorption processes e a review of the literature. Water, Air,and Soil Pollution 56, 667e680.

Selin, N.E., Jacob, D.J., Yantosca, R.M., Strode, S., Jaegle, L., Sunderland, E.M., 2008.Global 3-D landeoceaneatmosphere model for mercury: present-day versuspre-industrial cycles and anthropogenic enrichment factors for deposition.Global Biogeochemical Cycles 22, GB2011. doi:10.1029/2007GB003040.

Slootweg, J., Hettelingh, J.-P., Posch, M., Schütze, G., Spranger, T., De Vries, W.,Reinds, G.J., Van’t Zelfde, M., Dutchak, S., Ilyin, I., 2007. European critical loads ofcadmium, lead and mercury and their exceedances. Water, Air, and Soil Pollu-tion Forum 7, 371e377.

Smith-Downey, N.V., Sunderland, E.M., Jacob, D.J., 2010. Anthropogenic impacts onglobal storage and emissions of mercury from terrestrial soils: insights froma new global model. Journal of Geophysical Research e Biogeosciences 115.doi:10.1029/2009JG001124 Article No. G03008.

Spurgeon, D.J., Rowland, P., Ainsworth, G., Rothery, P., Long, S., Black, H.I.J., 2008.Geographical and pedological drivers of distribution and risks to soil fauna ofseven metals (Cd, Cu, Cr, Ni, Pb, V and Zn) in British soils. EnvironmentalPollution 153, 273e283.

Swain, E.B., Engstrom, D.R., Brigham, M.E., Henning, T.A., Brezonik, P.L., 1992.Increasing rates of atmospheric mercury deposition in midcontinental NorthAmerica. Science 257, 784e787.

Tipping, E., Lawlor, A.J., Lofts, S., Shotbolt, L., 2006. Simulating the long-term chemistryof an upland UK catchment: heavymetals. Environmental Pollution 141,139e150.

Tipping, E., Yang, H., Lawlor, A.J., Rose, N.L., Shotbolt, L., 2007. Trace metals in thecatchment, loch and sediments of Lochnagar: measurements and modelling. In:Rose, N.L. (Ed.), Lochnagar: The Natural History of a Mountain Lake. Springer,Dordrecht, pp. 345e373.

Tipping, E., Lofts, S., Hooper, H., Frey, B., Spurgeon, D.J., Svendsen, C., 2010a. Criticallimits for Hg(II) in soils, derived from chronic toxicity data. EnvironmentalPollution 158, 2465e2471.

Tipping, E., Rothwell, J.J., Shotbolt, L., Lawlor, A.J., 2010b. Dynamic modelling ofatmospherically-deposited Ni, Cu, Zn, Cd and Pb in Pennine catchments(northern England). Environmental Pollution 158, 1521e1529.

Tipping, E., Poskitt, J.M., Lawlor, A.J., Wadsworth, R.A., Norris, D.A., Hall, J.R., 2011.Mercury in United Kingdom topsoils; concentrations, pools, and critical limitexceedances. Environmental Pollution 159, 3721e3729.

Travnikov, O., Ilyin, I., 2005. Regional Model MSCE-HM of Heavy Metal Trans-boundary Air Pollution in Europe. EMEP/MSC-E Technical Report 6/2005.

UBA, 2004. Manual on Methodologies and Criteria for Modelling and Mapping ofCritical Loads and Levels and Air Pollution Effects, Risks and Trends. Umwelt-bundesamt, Berlin. www.icpmapping.org.

Wiedinmyer, C., Friedli, H., 2007. Mercury emission estimates from fires: an initialinventory for the United States. Environmental Science and Technology 41,8092e8098.

Yang, H., Rose, N.L., Battarbee, R.W., Boyle, J.F., 2002. Mercury and lead budgets forLochnagar, a Scottish mountain lake and its catchment. Environmental Scienceand Technology 36, 1383e1388.

Yang, H., Rose, N.L., 2003. Distribution of mercury in six lake sediment cores acrossthe UK. Science of the Total Environment 304, 391e404.

Yang, H., Berry, A., Rose, N., Berg, T., 2009. Decline in atmospheric mercury depo-sition in London. Journal of Environmental Monitoring 11, 1518e1522.

Yang, H., Battarbee, R.W., Turner, S.D., Rose, N.L., Derwent, R.G., Wu, G., Yang, R.,2010. Historical reconstruction of mercury pollution across the Tibetan Plateauusing lake sediments. Environmental Science and Technology 44, 2918e2924.

Zhang, H., Lindberg, S.E., 1999. Processes influencing the emission of mercury fromsoils: a conceptual model. Journal of Geophysical Research 104, 21,889e21,896.