recovery of sulfate saturated soils in the plynlimon catchments, mid-wales following reductions in...

Dynamic Article LinksC<Journal ofEnvironmentalMonitoringCite this: J. Environ. Monit., 2012, 14, 1531

www.rsc.org/jem PAPER

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online / Journal Homepage / Table of Contents for this issue

Recovery of sulfate saturated soils in the Plynlimon catchments, mid-Walesfollowing reductions in atmospheric S inputs from the 1980s to 2011

Steven Hughes,*a Brian Reynolds,a David A. Norris,a Sarah A. Brittain,a Ashlee L. Dere,b Clive Woods,c

Linda K. Armstrong,d Sarah A. Harmand and Heather D. Wickhamd

Received 26th January 2012, Accepted 7th March 2012

DOI: 10.1039/c2em30070b

Sulfate adsorption capacity of B-horizons of base-poor, predominantly stagnopodzol, soils from the

Plynlimon catchments, mid-Wales was determined by combination of laboratory adsorption and

desorption isotherms. Results show that sulfate adsorption capacity of a range of stagnopodzol (Histic-

stagno-podzol (Leptic), WRB), brown podzolic soil (Histic-umbrisol (Leptic), WRB) and stagnohumic

gley (Histic-stagno-gleysol, WRB) B-horizons was positively related to the amounts of extractable

(pyrophosphate and oxalate) Fe + Al, with the stagnopodzol and brown podzolic soil Bs horizon

having the largest adsorption capacity and stagnohumic gley Bg horizon the smallest adsorption

capacity. Results show that dissolved organic carbon (DOC) has a negative but limited effect on sulfate

adsorption in these soils. Results obtained from a set of historical soil samples revealed that the

grassland brown podzolic soil Bs horizon and afforested stagnopodzol Bs horizon were highly

saturated with sulfate in the 1980s, at 63% and 89% respectively, whereas data from some recently

sampled soil from two sites revisited in 2010–11 indicates that percentage sulfate adsorption saturation

has since fallen substantially, to 41% and 50% respectively. Between 1984 and 2009 the annual rainfall-

weighted mean excess SO4-S concentration in bulk precipitation declined linearly from 0.37 mg S l�1 to

0.17 mg S l�1. Over the same period, flow weighted annual mean stream water SO4-S concentrations

decreased approximately linearly from 1.47 mg S l�1 to 0.97 mg S l�1 in the plantation afforested Hafren

catchment compared to a drop from 1.25 to 0.69 mg S l�1 in the adjacent moorland catchment of the

Afon Gwy. In flux terms, the mean decrease in annual stream water SO4-S flux has been approximately

0.4 kg S ha�1 yr�1, whilst the recovery in stream water quality in the Afon Cyff grassland catchment

has been partly offset by loss of SO4-S by desorption from the soil sulfur pool of approximately

0.2 kg S ha�1 yr�1.

aCentre for Ecology and Hydrology, Environment Centre Wales, DeiniolRoad, Bangor, Gwynedd LL57 2UW, United Kingdom. E-mail: [email protected]; [email protected]; [email protected]; [email protected]; Fax: +44(0) 1248 362133; Tel: +44 (0) 1248 374500bDepartment of Geosciences, Pennsylvania State University, UniversityPark, Pennsylvania PA16802, USA. E-mail: [email protected]; Fax: +01814 863 8094; Tel: +01 858 243 2644cCentre for Ecology and Hydrology, Lancaster Environment Centre,Library Avenue, Bailrigg, Lancaster LA1 4AP, United Kingdom.

E-mail: [email protected]; Fax: +44 (0) 1524 61536; Tel: +44 (0) 1524595800dCentre for Ecology and Hydrology, Maclean Building, Benson Lane,Crowmarsh Gifford, Wallingford, Oxfordshire OX10 8BB, UnitedKingdom. E-mail: [email protected]; [email protected]; [email protected];Fax: +44 (0) 1491 692424; Tel: +44 (0) 1491 838800

Environmental impact

This study shows that Plynlimon grassland brown podzolic soil and afforested stagnopodzol Bs horizons have large sulfate

adsorption capacities and were highly saturated with sulfate in the 1980s amid historically high inputs of atmospheric S deposition.

Data obtained from some recently sampled soil from two sites revisited in 2010–11 shows that percentage sulfate adsorption

saturation of the soils has since fallen substantially, in line with on-going reductions in anthropogenic S deposition in the UK over

the past two decades. At the catchment scale, the flux of SO4-S by desorption from the soil S pool has been large enough to partly

offset the recovery in stream water quality in at least one of the four main sub-catchments.

This journal is ª The Royal Society of Chemistry 2012 J. Environ. Monit., 2012, 14, 1531–1541 | 1531

http://dx.doi.org/10.1039/c2em30070b






http://pubs.rsc.org/en/journals/journal/EM

http://pubs.rsc.org/en/journals/journal/EM?issueid=EM014006

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

Introduction

Sulfate retention in soils has been a major focus of research over

the last three decades, and particularly so during the 1980s, since

it has long been established that anthropogenic sulfur from acid

rain plays a major role in soil acidification. Retention of SO4 by

physico-chemical adsorption by the soil can mitigate some of the

effects of acid deposition by removing the mobile anion from the

soil solution, and therefore decrease the risk of leaching of base

cations from the soil. However, SO4 adsorption is a delaying

mechanism, rather than a permanently ameliorative process,

owing to the reversibility of the process in soils.1 For predictions

of acidification reversal, soil characteristics, sulfur pools and

their dynamics have to be evaluated.2

The consensus view is that sulfate adsorption increases with

depth in the soil, usually reaching a maximum in the soil B

horizon,3,4 and is known to occur on the surfaces of Fe & Al

hydroxy-oxides.5,6 Other factors known to influence sulfate

retention in soils are pH,5,7–9 texture1 and soil organic matter

content.10 The latter has received more recent attention since it is

known that organic anions in soil leachates can compete with

sulfate for adsorption sites on the soil.8

Whereas research has been carried out on the ability of soils to

retain sulfate, there is still a pressing need for more quantitative

data on sulfate adsorption and desorption in soils susceptible to

acid deposition for inclusion into dynamic acidification models

e.g. MAGIC.11 This has become more urgent with the need for

these models to accurately predict the likely recovery rates of

surface water quality draining soils affected by acidification in

the light of on-going reductions in anthropogenic sulfur depo-

sition. However, there is a common problem associated with

a direct use of measured soil SO4 adsorption data in MAGIC. In

some studies, e.g.modelling the reversibility of Central European

mountain lakes from acidification, it has been found that

measured soil SO4 adsorption maxima were too low for the

model to accurately predict the observed delay between SO4

deposition and in-lake SO4 concentrations during both the

acidification advance and recovery. Indeed, it is necessary to use

several times higher values of soil SO4 adsorption maxima in the

model calibration either for forest12 or alpine soils13 to achieve

a good fit. This ‘‘overestimation’’ of soil SO4 adsorption capacity

probably compensates for biological S-turnover in soils, which is

an important process indicated by stable S isotope studies.14,15

Notwithstanding the above limitations, the aim of the present

study is to provide quantitative data on sulfate adsorption

capacity and relative saturation of susceptible base-poor,

predominantly stagnopodzol, soils from the Plynlimon catch-

ments, mid-Wales. It is hoped that these data will help improve

the predictive capabilities of the current acidification models at

the regional scale.

We report here detailed laboratory investigations of sulfate

adsorption and desorption isotherms of stagnopodzol, brown

podzolic soil, gleyed podzol and stagnohumic gley B-horizon

soils on a batch of historical (air-dried, ground & sieved <2 mm)

soil samples collected from a number of sites within the Plynli-

mon catchments in the 1980s. The study extended to include

some more recently sampled soil (2002 & 2010–11) from

a limited number of selected sites revisited in the catchment, for

comparison. Additional studies on some of the soils sampled in

1532 | J. Environ. Monit., 2012, 14, 1531–1541

2002 were carried out to investigate the effects of increasing

concentrations of dissolved organic carbon on sulfate adsorp-

tion. The soil results are set in the context of long-term

measurements of the concentrations and fluxes of wet deposited

sulfate at Plynlimon and the corresponding response in four

headwater streams.

Materials and methods

Site description

Summary features of the Plynlimon catchments (location: a box-

bounded by latitude 52� 260 2100 N and longitude 3� 450 320 0 W for

the lower left hand corner, and latitude 52� 290 3600 N and

longitude 3� 420 050 0 W for the upper right hand corner) are

shown in Table 1 and a land cover map and location of soil

sampling sites are shown in Fig. 1. Two main catchments were

chosen for soil studies at Plynlimon. The Afon Cyff, which is

a mixture of semi-natural acid heathland and grassland subject to

low density sheep grazing, with areas of agriculturally improved

grassland which has received additions of lime and fertiliser at

irregular intervals over the last 80 years. The improved grassland

is grazed by sheep and cattle. The other chosen catchment, the

Afon Hore, is predominantly covered by plantation forestry

consisting mainly of Sitka spruce (Picea sitchensis), and planted

in several phases since the 1940s. The Lower Hore sub-catchment

was clear felled in 1985–86 and replanted with Sitka spruce. The

remainder of the catchment has been subject to patch clear felling

and replanting since that time. A few soil samples were also taken

in 2002 from another adjacent grassland catchment, the Afon

Gwy. The Afon Cyff forms part of the headwaters of the River

Wye, whilst the Afon Hore is located in the headwaters of the

River Severn. Geology consists of Lower Palaeozoic mudstones

or shales. Locally derived glacial and post-glacial drift occurs in

all the catchments. Soil cover consists of a mosaic of acid upland

types including peats, brown earths, stagnohumic gleys, brown

podzolic soil and stagnopodzols. Hill-top peats (Histosols, in the

World Reference Base for Soil Resources and USDA nomen-

clature) and stagnopodzols16 (Histic-stagno-podzol (Leptic),

WRB; Aquods, USDA) predominate at Plynlimon. Whereas

brown podzolic soil (Histic-umbrisol (Leptic), WRB; Orthods,

USDA) and stagnohumic gley soil (Histic-stagno-gleysol, WRB;

Aquepts, USDA) together comprise a much smaller percentage

of the soil cover at Plynlimon; they are important soil types

locally in two of the four main sub-catchments i.e.Afon Cyff and

Afon Hore. Detailed soil descriptions and soil chemistry of

stagnopodzols from the Afon Cyff and Afon Hore catchments

are provided by Reynolds et al.17 The brown podzolic soil16 at site

1 in the Cyff has an orange-brown ‘‘rusty and pellety’’ Bs horizon

which is a distinctive characteristic of brown podzolic soil. The

brown podzolic soil Bs horizon and stagnopodzol Bs horizon at

Plynlimon averages 25 cm in thickness and have a low base

saturation (Na, K, and Ca & Mg) of between 6 to 8% (1980s

data). Soil cover information for Plynlimon was obtained (under

licence) from LandIS, NSRI.

Unfortunately, the sulfur chemistry of the streams draining the

Cyff and Hore catchments is complicated by geogenic sulfur

sources within the bedrock. The upper reaches of the Hore

contain derelict workings of the Snowbrook mine from the 19th

This journal is ª The Royal Society of Chemistry 2012


Table 1 Summary features of the Plynlimon catchments

Afon Gwy Afon Cyff Afon Hafren Afon Hore

Catchment area (ha) 390 308 347 335Altitude range (m) 380–730 355–690 350–690 339–738Hill-top peat (%) 37 14 52 33Improved Grassland (%) 7 39 0 0Forest cover 1980s (%) 1 0 48 77Mean annual rainfall 1983–2002 (mm) 2665 2595 2545 2530Mean annual runoff 1983–2002 (mm) 2275 2145 2054 2059Bedrock mineralization None CaCO3 veins None Pb/Zn sulfides + CaCO3

Arithmetic mean SO4 concentration (mgSO4 l�1) for lowest 5%ile flows (1984–2009) 2.67 5.10 3.47 4.71

Arithmetic mean Ca concentration (mg l�1) for lowest 5%ile flows (1984–2009) 1.01 2.83 1.10 3.44

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

and early 20th centuries. The mine worked Pb and Zn sulfide ore

bodies associated with calcite secondary mineralisation which

results in high pH drainage water enriched in calcium, bicar-

bonate and sulfate. Apart from a small trial pit, there was no

mining in the Cyff, but at low flows, the comparatively high

stream water SO4 concentrations (Table 1) along with elevated

pH and calcium are strongly indicative of mineralisation asso-

ciated with the faults running across the catchment. The bedrock

mineralisation does not affect soil chemistry except in the

immediate vicinity of the mine but to provide a catchment

context for the soil study, stream water chemistry data from the

Afon Gwy and Afon Hafren are also reported in this paper (see

Table 1). There is no evidence of bedrock sulfide mineralisation

Fig. 1 Land cover map and soil samplin


in either of these catchments. The Afon Gwy, which is adjacent

to the Cyff, is dominated by acid grass and dwarf shrub heath

vegetation grazed at low intensity by sheep. Less than 10% of the

catchment has been agriculturally improved and the land

management has been unchanged since the start of the water

quality sampling in 1979. Approximately 50% of the Afon Haf-

ren catchment is covered by plantation Sitka spruce forest which

has been progressively thinned, felled and replanted since the

early 1990s. The upper part of the Hafren is semi-natural dwarf

shrub heath and bog habitat supporting low intensity sheep

grazing. The areas, altitude ranges, geology of the Gwy and

Hafren are very similar to the Cyff and Hore and a similar range

of acid soils occurs across all the catchments (Table 1).

g sites for the Plynlimon catchments.

J. Environ. Monit., 2012, 14, 1531–1541 | 1533


Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

Soil sulfate desorption and adsorption isotherms

All analysis was necessarily done on air-dried, ground and sieved

(<2 mm) soil. Duplicate samples of 5 g soil were shaken over-

night at 5 �C in 100 ml 0.01 M NaCl solution, filtered through

Whatman 0.45 mm membrane cellulose nitrate filters, and the

solutions kept and analysed for SO4. The filtered soil was washed

back into the same flasks with fresh 0.01 M NaCl solution and

the procedure repeated for between a minimum of five, and

a maximum of eight sequential extractions. All soil extractions

were carried out at approximately the native soil pH values (�0.1

units). This was done for each soil extraction by selecting one

solution from a choice of a number of 0.01 M NaCl solutions,

each with a different pre-set pH value from a range set in

increments of approximately 0.3 units from pH 4 to 6 (HCl or

NaOH was used to set the desired pH of the 0.01 M NaCl

solutions). The effects of 0.01 M NaCl in depressing the soil

solution pH was measured and taken into account in the selec-

tion process, but was found to be relatively minor for the Cyff

grassland soils. However, the effect of 0.01MNaCl was observed

to depress the solution pH by as much as 0.2 units for the Hore

forest soils. Cumulative desorption curves were obtained for each

soil by plotting cumulative desorption against the reciprocals of

each sequential extraction number (e.g. Fig. 2). The fitted curve

extrapolated to the intercept on the y-axis yields the native

amount of SO4 adsorbed on the soil i.e. extrapolated to an

infinite number of soil extractions.18 There are a number of

potential errors in this procedure; namely the inclusion of

entrapped soluble but non-adsorbed SO4 in the initial extraction,

and carry over between extractions. Not all of the historically

adsorbed SO4 will be reversible. The cumulative desorbed SO4 at

each extraction point, when subtracted from the native adsorbed

SO4 content of the soil, yields the estimated amounts of SO4

adsorbed at various equilibrium SO4 concentrations. Additional

points were obtained by spiking additional 5 g soil suspensions

with various amounts of 0.005 M Na2SO4 solution. For the

additional data points, the amount of 0.01 M NaCl solution was

adjusted so that the final solution volume was 100 ml. Sulfate was

analysed by ion chromatography.

Experimentally derived adsorption data were fitted to the

Langmuir monolayer adsorption model by way of a common

Fig. 2 Cumulative sulfate desorption curves for two selected Plynlimon

soils sampled in the 1980s.

1534 | J. Environ. Monit., 2012, 14, 1531–1541

linear transformation of the Langmuir equation i.e. by plotting 1/

Es against 1/[SO4] (where Es ¼ amount of SO4 adsorbed,

expressed in this study as mg SO4 adsorbed kg�1 air-dried soil;

[SO4] concentration expressed in mg l�1).

Langmuir model:

Es ¼ Emx

�SO2�

4

�

C þ �SO2�

4

� (1)

(Emx ¼ SO4 adsorption maximum; C ¼ equilibrium concentra-

tion of SO4 in soil solution at ½ Emx).

Inverting eqn (1) and rearranging:

1

Es

¼ C

Emx

� 1�SO2�

4

�þ 1

Emx

(2)

From y ¼ mx + c:

Plot of 1/Es against 1/[SO4] gives an intercept on the y-axis of

1/Emx and slope C/Emx (e.g. Fig. 3).

Dissolved organic carbon (DOC) studies

The effects of increasing concentrations of DOC on sulfate

adsorption and desorption were carried out as above at the

native pH of the soil solution, by diluting a natural peat water

leachate to two pre-selected DOC concentrations (4 mg l�1 &

8 mg l�1 respectively). The diluted natural peat water was

adjusted to the native soil solution pH with NaOH and, likewise,

to an approximate ionic strength of 0.01 M with the addition of

NaCl. The small amounts of SO4 in the 100 ml peat water

leachate at the two selected DOC concentrations (0.016 mg SO4

& 0.032 mg SO4 respectively) was subtracted from the total

amounts of SO4 in the solutions at each equilibrium point to give

an estimate of the total amounts desorbed or adsorbed.

Extractable soil Fe and Al

Acid ammonium oxalate (0.2 M) extractable & sodium pyro-

phosphate (0.1 M) extractable soil Fe and Al were determined at

room temperature on sieved (<2 mm) air-dried soil according to

the methods of McKeague and Day,19 and Bascomb.20

Fig. 3 Langmuir sulfate adsorption isotherm plots of 1/Es against

1/[SO4] for selected Plynlimon soils sampled in the 1980s.



Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

Ammonium oxalate extracts inorganic amorphous & organically

bound Fe and Al; pyrophosphate extracts organically bound Fe

and Al.21 Aluminium was analysed in the soil extracts by flame

atomic absorption spectrophotometry (AAS) using an N2O/

C2H2 flame and with deuterium background correction. Iron was

also analysed by flame AAS using an air/C2H2 flame and with

deuterium background correction.

Soil pH

Soil pH was determined in a 1 : 2.5 soil : water suspension using

the standard method of the Soil Survey of England and Wales.22

Wet deposition and stream water chemistry

Rainfall samples have been collected since May 1983 using

a continuously open collector located at the top of the Afon

Hafren catchment at Carreg Wen at an altitude of 575 m above

sea level. The rain collector funnel is 1.5 m above the ground

surface and is protected by a stainless steel coronet to deter birds

from perching. Two rain collectors have been deployed at the site

to reduce sample losses from contamination by bird droppings.

Samples are rejected where there is obvious contamination of the

collector funnels and further screening of the data for contami-

nation is undertaken using the concentrations of phosphorus and

sample pH as indicators.

Samples of cloud water have been collected using a passive

‘‘harp type’’ collector23 located near the rain collector since

September 1990. The harp is mounted approximately 1.8 m

above the ground surface and consists of a powder coated wire

frame supporting a cone of Teflon filaments which direct inter-

cepted cloud droplets via a glass funnel and into a collecting

bottle. Although the cloud collector has a horizontal lid

approximately 1.5 m in diameter, wind driven rain and aerosols,

as well as cloud droplets, are intercepted by the filaments. The

cloud water sample therefore provides a qualitative estimate of

the enhancement of wet deposition to forest and woodland

canopies.24

Rain and cloud water samples have been collected weekly,

providing a composite sample over the measurement period.

Snowfall occurs periodically over the winter at Plynlimon but

rarely remains on the ground for longer than a week. The wet

deposition samples therefore include any snow that has fallen

during the collection period. Weekly rainfall volume is measured

using a ground level storage rain-gauge. All rain and cloud water

samples are filtered in the laboratory within 24 h of collection.

Weekly ‘‘grab’’ samples of stream water have been collected

from the Hore and Hafren since May 1983 and at intervals

varying between weekly and monthly from the Cyff and Gwy

since September 1979. At all sites samples are collected from

a point approximately 50 m upstream of the permanent steep

stream flume gauging structures which provide runoff data at

15 min intervals. Samples for sulfate analysis are filtered through

0.45 mm cellulose nitrate membrane filters. Initially, samples

collected from the Cyff andGwy were filtered in the laboratory at

CEH Bangor, but since 2007 they have been filtered in the field.

Samples collected from the Hore and Hafren have been filtered in

the field over the entire monitoring period.


A variety of analytical techniques have been used to measure

sulfate concentrations in water samples since 1979. Between 1979

and 1984, sulfate concentrations in samples from the Cyff and

Gwy were determined by a BaCl2 turbidimetric technique using

a continuous flow auto-analyser. This method was insufficiently

sensitive for the sulfate concentrations encountered in the water

samples and was replaced by ion chromatography in 1984. Since

1983, concentrations of total dissolved sulfur have been

measured in samples from the Hore, Hafren and in rain and

cloud water using inductively coupled plasma emission spec-

trometry (ICPOES). Detailed speciation and comparison with

ion chromatography showed that concentrations of sulfide,

sulfite and organic sulfur in the samples were negligible and that

total dissolved sulfur equated to sulfate. The quality of the

ICPOES data was evaluated against standard water samples

provided by the Aquacheck LGC Inter-laboratory Proficiency

Testing Scheme whilst the ion chromatography analysis is

accredited to ISO 17025:2005.

The data for SO4 in rain and stream waters are reported as

annual rainfall volume-weighted mean or annual flow-weighted

mean SO4 concentrations. The cloud water collector data are

reported as annual geometric mean concentrations. The annual

rainfall wet deposition flux of sulfur to the catchments has been

calculated as the product of the annual rainfall weighted mean

concentration and the catchment annual total rainfall. Annual

stream water sulfur fluxes are the product of the annual flow

weighted mean concentration and the catchment annual total

runoff.25 Sulfate in rainfall and cloud water from non sea-salt

sources has been denoted as XsSO4. The sea-salt contribution to

total SO4 has been calculated using the ratio of Na to SO4 in

seawater.

Results and discussion

It can be seen in Table 2, soils possessing a recognisable Bs

horizon, i.e. a soil horizon containing significant amounts of Fe

and Al sesquioxides (Sites 1 & 8), generally had the highest SO4

adsorption capacities. By contrast, the stagnohumic gley forest

soil from the Hore catchment (Site 9), gleyed podzols (Site 3),

and stagnopodzols from the Cyff (grassland) catchment that

had previously been limed (Sites 4 & 5), had the lowest

adsorption capacities. For the majority of cases, the historical

soils used in the study were more than 60% saturated with SO4

at the time of sampling in the 1980s. It can also be seen in Table

2 that the soil with the lowest SO4 adsorption capacity, i.e. the

Hore stagnohumic gley Bg horizon of the Ynys soil series

(Site 9), was also the most highly saturated, with the calculated

Emx very close to the estimated native SO4 content. The forest

stagnopodzol Bs horizon (Site 8) was also found to have very

high percentage SO4 saturation. This might be as expected since

these forest soils received larger S inputs in throughfall, from

capture by the forest canopy, than is the case for the grassland

soils (29 kg SO4-S ha�1 and 22 kg SO4-S ha�1for the forest and

grassland, respectively),24 and inherently larger soil water SO4

concentrations than the latter. Indeed, the mean soil water SO4

concentration in the forest stagnopodzol Bs horizon in the mid

1980s was 10 mg l�1, compared with about 3 mg l�1 for the

unimproved grassland Bs horizon.17 As expected, the soils with

the largest SO4 adsorption capacities usually gave the best

J. Environ. Monit., 2012, 14, 1531–1541 | 1535


Table 2 Native SO4 contents, calculated adsorption capacities (Emx) and concentrations of sulfate in solution at equilibriumwith the soil at half Emx (C)of some Plynlimon catchment soils sampled in the 1980sa

Site Soil sample (1980s) Soil type pHSO4 content/mg kg�1

Emx/mg kg�1 C/mg l�l % saturation

1 Bs/1 Cyff Brown podzolic soil 3.83 138 222 1.49 62Bs/2 Cyff Brown podzolic soil 4.2 141 217 1.33 65Bs/3 Cyff Brown podzolic soil 4.84 186 294 2.79 63Bs/4 Cyff Brown podzolic soil 4.47 232 357 1.79 65Bs/5 Cyff Brown podzolic soil 4.27 146 263 1.74 56Bs/6 Cyff Brown podzolic soil 4.61 165 303 4.03 55Bs/7 Cyff Brown podzolic soil 4.2 155 250 1.30 62Bs/8 Cyff Brown podzolic soil 4.28 141 208 1.25 68Bs/9 Cyff Brown podzolic soil 4.45 250 357 2.18 70

6 B/10 Cyff Stagnopodzol 4.7 99 125 0.41 79B/11 Cyff Stagnopodzol 4.6 115 149 0.52 77

4 B/12 Cyff Stagnopodzol 4.87 69 100 0.70 69B/13 Cyff Stagnopodzol 4.84 59 77 0.60 77B/14 Cyff Stagnopodzol 5.38 54 66 0.35 82

5 B/15 Cyff Stagnopodzol 4.38 69 100 0.70 69B/16 Cyff Stagnopodzol 4.9 69 99 0.50 69B/17 Cyff Stagnopodzol 4.48 67 75 0.35 89

1 Fe pan/18 Cyff Brown podzolic soil 3.87 329 417 1.92 792 Bsg/19 Cyff Gleyed podzol 4.55 179 213 1.06 843 Bsg/20 Cyff Gleyed podzol 5.46 70 83 0.44 84

Bsg/21 Cyff Gleyed podzol 4.43 73 99 0.53 748 Bs/22 Hore Stagnopodzol 4.16 384 455 2.50 84

Bs/23 Hore Stagnopodzol 3.95 312 333 1.97 949 Bg/24 Hore Stagnohumic gley 4.35 82 86 0.36 96

Bg/25 Hore Stagnohumic gley 4.18 91 79 0.40 116

a Soils are numbered sequentially for ease of presentation. See Fig.1 for site locations.

Fig. 4 Experimentally derived & fitted (Langmuir) sulfate adsorption

isotherms for selected Plynlimon soils sampled in the 1980s.

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

adsorption curves, and the best fit of the experimental data to

the Langmuir isotherm (Fig. 3 & 4).

No significant differences were found in the SO4 adsorption

capacity (Emx) between the historical (1980s) Bs horizon soil

(mean ¼ 275 mg kg�1 air-dried soil) and the recently sampled

(2010) Bs horizon soil in the Cyff grassland catchment at Site 1

(mean ¼ 259 mg kg�1) (two-sample t-test, unpaired, equal vari-

ances: T ¼ 0.49 P ¼ 0.630 DF ¼ 13). However, it can be seen in

Fig. 5 that the percentage SO4 saturation of the high SO4

adsorption capacity brown podzolic soil Bs horizon has fallen

quite substantially since the 1980s. The percentage SO4 satura-

tion of the Cyff brown podzolic soil Bs horizon at Site 1 has

fallen from an average of 63% saturation in the 1980s to under

41% in 2010 (two-sample t-test, unpaired, no assumption of

equal variances was made: T ¼ 3.75 P ¼ 0.013 DF ¼ 5).

Unsurprisingly, the native adsorbed SO4 content of the recently

sampled (2010) Bs horizon soil was also found to be significantly

smaller (mean ¼ 110 mg kg�1) than for the historical (1980s) Bs

horizon soil (mean ¼ 173 mg kg�1) (two-sample t-test, unpaired,

equal variances: T ¼ 2.35 P ¼ 0.036 DF ¼ 13). The measured

decline in percentage sulfate adsorption saturation in the soils

follows the characteristic ‘‘hockey stick’’ pattern predicted by

modelling in other European countries, e.g. Lysina and Pluhuv

Bor, Czech Republic,26 wherein their models predicted a rapid

decline in the soil pool of adsorbed SO4 from the 1980s to the

year 2010, following which the decrease in the adsorbed SO4 pool

will be small from the year 2010 to 2030. To date, the models

have proved very successful in matching the observed trends e.g.

at Lysina.27

These findings are consistent with the trend in the data for wet

deposited sulfate at Plynlimon collected over the last 26 years.

1536 | J. Environ. Monit., 2012, 14, 1531–1541

Wet deposition accounts for most of the sulfur deposited at

Plynlimon24 and between 1984 and 2009, there was an approxi-

mately linear decrease from 1.7 mg SO4 l�1 to 1.2 mg SO4 l

�1 in

annual rainfall weighted mean SO4 concentration (Fig. 6A). This

reflected the decline in annual rainfall weighted mean concen-

tration of XsSO4 from 1.1 mg SO4 l�1 to 0.6 mg SO4 l�1.

Correspondingly, annual wet deposited sulfur in rainfall

decreased from 13 to 10 kg S ha�1 yr�1 and annual excess wet

sulfur deposition from 8 to 5 kg ha�1 yr�1 (Fig. 6B). The decline

in XsSO4 has led to an increase in the proportion of sea-salt SO4



Fig. 5 Time series plots of percentage sulfate adsorption saturation in

Plynlimon grassland and afforested soil Bs horizons from the 1980s to

2010–11. Data shown are from site 1 and site 8, respectively. Error bars

are �1 s.d.

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

in rainfall from ca. 36% in 1984 to 50% in 2009. The SO4

concentrations in the cloud water samples are an order of

magnitude greater than in rainwater but the time series data

show similar trends (Fig. 6C). Due to the windiness of the site

and the limited effectiveness of the lid, samples from the cloud

water collector comprise a mixture of cloud droplets, rainwater

and aerosols and these components can be isolated by more

sophisticated sampling methods.23 However the Plynlimon data

provide a semi-quantitative description of the enhanced inputs

Fig. 6 Time series plots of: (A) annual volume-weighted mean SO4

concentrations in rainfall; (B) annual wet deposited sulfur in the Gwy

catchment; (C) annual geometric mean SO4 concentrations in cloud water

at Plynlimon from 1984 to 2009.


potentially experienced by forest canopies. The trends in rainfall

SO4 chemistry measured at Plynlimon are consistent with those

reported for the UK as a whole28,31 and reflects the major decline

in UK and European SO2 emissions over the period.

In the Hore forest catchment, the percentage SO4 saturation

for the stagnopodzol Bs horizon at Site 8 was observed to have

fallen from a high of 89% in the 1980s to 55% in 2002 (two-

sample t-test, unpaired, equal variances: T ¼ 7.53 P ¼ 0.017

DF ¼ 2), and since fallen further to around 50% in 2011 (two-

sample t-test, unpaired, equal variances: T ¼ 7.59 P ¼ 0.001

DF ¼ 5). However, the calculated Emx of the Bs horizon soil

sampled in the 1980s was significantly larger (mean ¼ 394 mg

kg�1) than the Emx for the Bs horizon soil sampled in 2011

(mean ¼ 208 mg kg�1) (two-sample t-test, unpaired, equal vari-

ances: T ¼ 4.02 P ¼ 0.010 DF ¼ 5). We can report that the

precise location in the forest from which the stagnopodzols were

sampled during the 1980s is uncertain, but the soils re-sampled in

2002 and 2011 are all believed to lie within a 100 m radius of the

original plots. This suggests a large spatial variability in soil

quality at Site 8 as the most likely explanation for the large

differences in Emx between the soils sampled in the 1980s and

those sampled in 2002 and in 2011. Regardless of the differences

in Emx, part of the large reduction in percentage SO4 saturation

in the forest soil at Site 8 between the 1980s and 2011 is probably

a result of reduced SO4 inputs due to clear felling and subsequent

re-growth of a 2nd generation crop. In fact, estimates of the

annual deposition loading of SO4-S in throughfall in the forest in

the mid 1980s was 29 kg ha�1 which compares with only 18 kg

ha�1 for felling debris throughfall in the two years immediately

post-felling (1988–89), and with 14 kg ha�1 for bulk precipitation

over the same time. Therefore, we infer there was about a 35%

reduction in S reaching the ground under forest debris compared

to the mature standing crop. The increased water flux post-felling

would also further reduce soil water SO4 concentrations, leading

to SO4 desorption. In contrast to the stagnopodzol Bs horizon,

no significant differences in percentage SO4 saturation were

found between the historical (1980s) soil and the recently

sampled soil (2010) for the Hore stagnohumic gley Bg horizon at

Site 9 (two-sample t-test, unpaired, equal variances: T¼ 2.01 P¼0.138 DF ¼ 3). The Hore stagnohumic gley Bg horizon remains

highly saturated with sulfate. This result is as expected, given that

the calculated SO4 adsorption capacity (Emx) of the stagnohumic

gley Bg horizon is much lower than that for the stagnopodzol Bs

horizon, with concomitant lower calculated values for the equi-

librium concentration of sulfate at half the Emx i.e. C. The mean

value ofC for the stagnohumic gley Bg horizon was 0.4 mg l�1 (all

data i.e. including both historical and recently sampled soil),

which is much lower than the mean measured sulfate concen-

trations in precipitation (1.2 mg l�1) at Plynlimon in 2009.

The results of the DOC competition experiments are shown in

Table 3. The data shows that increasing concentrations of DOC

reduced the predicted (Langmuir) sulfate adsorption maxima in

the grassland brown podzolic soil Bs horizon tested from Site 1.

This is indicative of some occlusion by organic molecules, which

cannot easily be displaced from adsorption sites even at high SO4

concentrations. The cumulative effect was non-linear, which

suggests that organic molecules in these soils, even at relatively

high DOC concentrations, can occlude only a proportion of the

available SO4 adsorption sites.

J. Environ. Monit., 2012, 14, 1531–1541 | 1537


Fig. 7 Plots of SO4 adsorption maxima (Emx) against total acid

ammonium oxalate and pyrophosphate extractable soil Fe and Al for the

Plynlimon stagnopodzol, stagnohumic gley and brown podzolic soils

sampled in the 1980s. The plots for the Bs horizons include data from

both brown podzolic soil and stagnopodzols. The plots for the B horizons

are the other data from stagnopodzols, and the plots for the Bg horizons

are the data from stagnohumic gleys.

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

Plots of Emx against total ammonium oxalate extractable and

total pyrophosphate extractable soil Fe and Al for the stagno-

podzol, brown podzolic soil and stagnohumic gley B-horizons

(Fig. 7) gave positive linear relationships (R2 ¼ 0.51, p < 0.001

and R2 ¼ 0.55, p < 0.001 respectively), which is consistent with

adsorption increasing with increasing amounts of soil Fe and Al.

However, the relatively low R2 values indicate that other factors

are involved in determining the adsorption capacity of the

different soils. Indeed, it is well known that sulfate adsorption

increases with decreasing pH in the soil, with adsorption maxima

usually occurring around pH 4,5,9 although adding pH as a factor

did not result in a significant improvement in the regression

models in our study.

The storage capacity and SO4 pool of the (high Emx) soils may

be expressed per hectare of sub-catchment with same soil type

and soil series cover, and can be derived from the average bulk

densities of the mineral soil horizon and the average soil horizon

depth. This may then be compared to known annual inputs of

SO4 to provide an estimate of the soils ability to act as a long-

term source/sink of SO4. From Tables 2–4, the average adsorp-

tion capacity for the unimproved grassland (Cyff) brown

podzolic soil Bs horizon at Site 1 is 282 mg SO4 kg�1 air-dried soil

(9 historical soils + 3 soils sampled in 2002 + 6 soils sampled in

2010). Assuming an average bulk density of 0.86 g air-dried soil

cm�3 for the mineral soil horizon and an average Bs horizon

depth of 25 cm, the total adsorption capacity on a per unit area

basis is estimated as 0.06 kg SO4 m�2 (625 mmol SO4 m�2), or

201 kg SO4-S ha�1. The estimated annual S wet deposition

loading during the 1980s was 22 kg ha�1,24 therefore the total

adsorption capacity of the Plynlimon brown podzolic soil Bs

horizon at Site 1 represented about 9 years’ S deposition.

Moreover, the average percentage SO4 saturation of the Bs

horizon, at the time of sampling during the 1980s, was 63%.

Therefore, the total amount of adsorbed SO4 in the Bs soil

horizon was equivalent to about 6 years’ of annual S deposition

during the 1980s.

Calculations for the Hore forest stagnopodzol at Site 8 (SO4

adsorption capacity of 247 mg kg�1 air-dried soil, an estimated

annual S wet deposition loading of 29 kg ha�1 and a Bs horizon

Table 3 Native SO4 contents, calculated adsorption capacities (Emx) and concof some Plynlimon catchment soils sampled in 2002a

Site Soil sample (2002) Soil type pH

1 Bs/26 Cyff Brown podzolic soil 4.16Bs/27 Cyff Brown podzolic soil 4.20Bs/28 Cyff Brown podzolic soil 4.35Bs/26 + 4 mg l�1 DOC Cyff* Brown podzolic soil 4.16Bs/27 + 4 mg l�1 DOC Cyff* Brown podzolic soil 4.20Bs/28 + 4 mg l�1 DOC Cyff* Brown podzolic soil 4.35Bs/26 + 8 mg l�1 DOC Cyff* Brown podzolic soil 4.16Bs/27 + 8 mg l�1 DOC Cyff* Brown podzolic soil 4.20Bs/28 + 8 mg l�1 DOC Cyff* Brown podzolic soil 4.35

7 B/Cg/29 Gwy Gleyed podzol 4.27Bsg/30 Gwy Gleyed podzol 4.24Bsg/31 Gwy Gleyed podzol 4.34

9 Bg/32 Hore Stagnohumic gley 4.288 Bs/33 Hore Stagnopodzol 4.22

Bs/34 Hore Stagnopodzol 4.32

a * Denotes adsorption/desorption isotherm experiments amended with natu

1538 | J. Environ. Monit., 2012, 14, 1531–1541

depth of 25 cm), indicates that the SO4 adsorption capacity of the

Bs horizon (553mmol SO4 m�2 or 178 kg SO4-S ha�1) represented

about 6 years’ S deposition. At 89% saturation, the native

amount of SO4 adsorbed was equivalent to about five and a half

years of annual S deposition during the 1980s. In view of this, it is

expected that desorption of SO4 would only have an influence as

a significant control on the flux of SO4 through the Plynlimon

catchments in the medium term. Moreover, from a comparison

of the average percentage sulfate saturation of the historical

(1980s) Bs horizon soil in the Cyff grassland catchment at Site 1

(63%) with the most recently sampled (2010) Bs horizon soil

(41%) from the same site, it is estimated that the latter figure of

41% sulfate adsorption saturation now represents under 4 years

of S deposition loading at the 1980s rate of annual S deposition.

Therefore, the Cyff grassland catchment brown podzolic soil Bs

entrations of sulfate in solution at equilibriumwith the soil at half Emx (C)

SO4 content/mg kg�1 Emx/mg kg�1 C/mg l�1 % saturation

123 370 1.96 3395 250 1.30 38184 435 2.17 42121 313 1.53 3982 189 0.98 43183 400 1.92 46120 286 1.34 4277 172 0.81 44185 400 2.04 4628 24 0.08 11749 68 0.36 7255 78 0.41 7189 89 0.34 100106 192 0.90 55112 208 1.00 54

ral peat water leachates. See Fig.1 for site locations.



Table 4 Native SO4 contents, calculated adsorption capacities (Emx) and concentrations of sulfate in solution at equilibriumwith the soil at half Emx (C)of some Plynlimon catchment soils sampled in late 2010 or early 2011a

Site Soil sample (2010) Soil type pHSO4 content/mg kg�1

Emx/mg kg�1 C/mg l�1 % saturation

1 Bs/35 Cyff Brown podzolic soil 4.36 54 167 2.03 32Bs/36 Cyff Brown podzolic soil 4.30 91 213 1.55 43Bs/37 Cyff Brown podzolic soil 4.37 231 345 1.66 67Bs/38 Cyff Brown podzolic soil 4.34 96 250 2.10 38Bs/39 Cyff Brown podzolic soil 4.56 115 323 3.13 36Bs/40 Cyff Brown podzolic soil 4.36 71 256 1.97 28

9 Bg/41 Hore Stagnohumic gley 4.23 78 87 0.28 90Bg/42 Hore Stagnohumic gley 4.41 107 147 0.78 73Bg/43 Hore Stagnohumic gley 4.26 60 66 0.32 91

8 Bs/44 Hore* Stagnopodzol 4.20 90 161 1.00 56Bs/45 Hore* Stagnopodzol 4.08 96 167 0.94 57Bs/46 Hore* Stagnopodzol 4.28 113 250 1.30 45Bs/47 Hore* Stagnopodzol 4.23 116 256 1.38 45Bs/48 Hore* Stagnopodzol 4.21 96 204 0.98 47

a * Denotes soil sampled in early 2011. See Fig.1 for site locations.

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

horizon at Site 1 is estimated to have cumulatively desorbed

about two years worth of the 1980s rates of annual S deposition

loading over the last two decades, representing about 46 kg SO4-

S ha�1. Similarly, calculations on the forest Bs horizon soil at Site

8 reveal that at 50% saturation the Bs horizon (as of January

2011) represents just over 3 years’ annual S loading at the 1980s

rate of annual S deposition. Therefore, the forest Bs horizon soil

at Site 8 is estimated to have cumulatively desorbed about two

and a half years worth of the 1980s rates of annual S deposition

loading over the previous two decades, or about 70 kg SO4-S

ha�1. However, owing to the small number of samples of soil

taken in the forest site both during the 1980s and more recently in

2002 & 2011, we would caution that the figures for the forest site

should be treated as rough estimates only.

In Sweden, modelling (SAFE) of the adsorbed sulfate pool in

the forest B horizon from the Lake Gardsjon roof project shows

that the adsorbed SO4 pool closely follows the S-deposition rate.

The adsorbed SO4 pool is estimated to have peaked in the late

1980s at 700 mmol m�2, thereafter falling to about 550 mmol m�2

by the year 2000; a decrease of 21%.9 In our laboratory study, the

native adsorbed SO4 pool in the Hore forest stagnopodzol Bs

horizon at Site 8 is estimated to have reached 492 mmol m�2 in

the mid-1980s, falling to about 276 mmol m�2 by 2011; a decrease

of 44%. To reiterate, it should be noted that the Hore forest site

was clear felled in 1986, and thereafter a 2nd generation crop has

since been grown.

Stagnopodzols (50%) cover a large proportion of the Plynli-

mon total catchment area, and a further significant proportion is

also covered by hilltop peats (35%) and brown podzolic soil

(4%). The areas covered by brown podzolic soil and stagno-

podzols are considered to be dominated by two soil series;

namely the Manod (Mj) and Hiraethog (Hi) series. The former

generally possess a distinctive orange-brown ‘‘rusty’’ Bs horizon

and (predicted) larger sulfate adsorption capacity than the latter.

However, the soil development in the large areas dominated by

stagnopodzols also comprises of sporadic patches of mixed Hi/

Mj and Hi/peat soils, making it difficult to ascribe sulfate

adsorption capacity at this scale. The estimated cumulative

release of SO4 from the soil Bs horizon over the last 25 years is


46 kg SO4-S ha�1 at Site 1 and 70 kg SO4-S ha�1 at Site 8. Site 1

in the Cyff catchment is covered by brown podzolic soil.

Moreover, the proportion of the total Cyff catchment area

covered by brown podzolic soils is calculated to be about 11.6%.

Therefore we estimate that over the last 25 years, the cumulative

flux of desorbed sulfate from the soil (Mj) pool is equivalent to

about 5 kg SO4-S ha�1 spread across the total Cyff catchment

area or about 0.2 kg SO4-S ha�1 yr�1. Results from our labora-

tory studies indicates that the SO4-S flux from the soil pool of

soils of low Emx in the Cyff catchment would be negligible, with

these soils likely remaining at high percentage SO4 saturation to

the present date, since they typically have C values well below

1 mg l�1 (Table 2).

The estimates of the release of sulfur from the soils can be

contextualised by considering the changes in the annual flow-

weighted mean stream water SO4 concentrations and SO4-S

fluxes for the Plynlimon sub-catchments over the 26 year period

since 1984. Time series plots of standardised annual flow-

weighted mean SO4 concentrations show strong linear down-

ward trends that are consistent across the four monitored

sub-catchments at Plynlimon (Fig. 8). For individual sites, linear

regression of annual flow-weighted mean concentration against

year was highly significant (r2 values ranging between 0.55 for the

Cyff and 0.78 for the Hafren), with a mean slope across the sites

of �0.065 mg SO4 l�1 yr�1 (Table 5). This compares with the

annual rate of change in volume-weighted mean SO4 concen-

tration in rainfall of ��0.03 mg SO4 l�1 yr�1 (Fig. 6A).

Annual stream SO4-S fluxes are consistently greater than

estimated SO4-S (rainfall) inputs (Fig. 9A & 9B) and these

differences could be the result of:

(i) An underestimation of S inputs from unmeasured aerosol,

cloud water and dry deposition sources to the catchments.

(ii) Leaching of SO4-S from geogenic and to a lesser extent

agricultural sulfur sources (SO4 is a common impurity in rock

phosphate fertiliser), in catchments affected by bedrock miner-

alisation and agricultural improvement, contributing to stream

water SO4-S fluxes.

(iii) Enhanced capture of aerosols and cloud droplets by forest

canopies compared to low moorland vegetation.

J. Environ. Monit., 2012, 14, 1531–1541 | 1539


Fig. 8 Time series plots of: (A) standardised annual flow-weighted mean

stream water SO4 concentrations and (B) standardised annual dissolved

stream water S fluxes for the Cyff, Gwy, Hafren and Hore at Plynlimon

between 1984 and 2009. For each site in (A), the data were standardised

by calculating for each year, the difference between the annual flow-

weighted mean and the 26 year average of flow-weighted means and

dividing this by the standard deviation of the 26 year average of flow

weighted means. For each site in (B) the difference between each annual

flux and the average of the 26 annual fluxes was calculated and divided by

the standard deviation of the 26 annual fluxes. Effectively this re-scales

the y-axis in both plots to number of standard deviations from the long-

term annual mean, and was done in order to facilitate presentation of the

data for all four sub-catchment streams on only two plots.

Table 5 Coefficients and r2 values for linear regression relationshipsbetween annual flow-weighted mean stream water SO4 concentrationsand annual stream water S fluxes against year for the Cyff, Gwy, Hafrenand Hore catchments at Plynlimon for 1984 to 2009

Site Slope r2 Site Slope r2

aCyff �0.07 0.55 bCyff �0.43 0.27aGwy �0.06 0.77 bGwy �0.38 0.57aHafren �0.06 0.78 bHafren �0.28 0.26aHore �0.07 0.72 bHore �0.34 0.24

a Annual flow weighted mean stream water SO4 concentration (mg l�1)against year. b Annual stream water S fluxes (kg S ha�1 yr�1) against year.

Fig. 9 Scatter plots of annual stream water dissolved S flux against

annual wet S deposition for: (A) the Cyff and Gwy and (B) the Hafren

and Hore catchments at Plynlimon from 1984 to 2009.

Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

The long term changes in stream water SO4-S fluxes are much

more variable year to year and between the sites (Fig. 8B)

compared to flow weighted mean concentrations (Fig. 8A). This

presumably results from combining the annual variability in

catchment runoff with variations in SO4 concentration. For

individual sites, linear regression of annual SO4-S flux against

year was significant, with a mean slope across the sites of �0.36

kg S ha�1 yr�1 (Table 5). This compares with a slower annual

decline in rainfall SO4-S flux of �0.21 kg S ha�1 yr�1, and with

a calculated offset loss of SO4-S by desorption from the soil S

pool in the Afon Cyff catchment of about 0.2 kg S ha�1 yr�1.

1540 | J. Environ. Monit., 2012, 14, 1531–1541

Thus there should have been little change in the stream water

SO4-S flux, whereas in fact it has declined by about �0.4 kg S

ha�1 yr�1. This discrepancy can most likely be explained by

a large decrease in the unmeasured dry S deposition and/or

occult S deposition to the catchments. Although not measured

directly in this study, concentration based estimated deposition

(CBED) data28–30 for the 5 � 5 km square that covers most of the

Plynlimon catchment indicates that dry deposition comprised

a significant proportion (37%) of the total non-marine S depo-

sition to the catchments during the 1980s, whilst the CBED total

non-marine S (35 kg S ha�1 yr�1 (1986–88)) decreased by over

82% in the following 20 years to just over 6 kg S ha�1 yr�1 (2006–

08). This compares with a 50% drop in the measured (non-

marine) XsSO4-S flux in rainfall over the same period (Fig. 6B).

This implies a very large fall in the dry and/or occult S deposition

to the catchments has occurred over the 20 years prior to 2006.

Indeed, this supposition is supported by a very large decline

(70%) in the measured annual geometric mean XsSO4 concen-

trations in cloud water at Plynlimon over the years 1991–2009

(Fig. 6C). Other studies also lend support to this, e.g. Fowler

et al.31 reported that the contribution to total sulfur deposition

from dry deposition is significant, even in the uplands and in the

west in Britain, and the change in dry S deposition is the largest

contributor to the changes in the UK deposition budget.

However, the partitioning of the deposition between wet and dry

deposition is changing quite rapidly with time.

For sulfur, the acidity response to reductions in anthropogenic

sulfur deposition predicted by current models ignores the

dynamics of the large organic (+ sulfides) sulfur pool. Mobi-

lisation of this store in response to climate change could provide

a source of sulfur at present inadequately accounted for in

acidification models. At Plynlimon, significantly increased



Publ

ishe

d on

16

Mar

ch 2

012.

Dow

nloa

ded

by S

tate

Uni

vers

ity o

f N

ew Y

ork

at S

tony

Bro

ok o

n 23

/10/

2014

04:

43:2

2.

View Article Online

pore-water SO4 concentrations have been observed in a flush

wetland in the Afon Gwy catchment in the Autumn–Winter

following consecutive simulated and/or natural summer

drought.32 Given the significant percentage cover of hilltop peats

in three of the Plynlimon catchments, this scenario might be of

some concern in the future, and might warrant further

investigation.

Conclusions

The results from this study show that the Bs horizons of two of

the predominant soil types (stagnopodzols and brown podzolic

soil) at Plynlimon have a relatively high sulfate adsorption

capacity of about 180 to 200 kg SO4-S ha�1. Results also show

that the percentage sulfate adsorption saturation of the grassland

brown podzolic soil Bs horizon and afforested stagnopodzol Bs

horizon was high during the 1980s, at 63% and 89% respectively,

amid historically high inputs of atmospheric S deposition. Data

on some recently sampled soil from two sites revisited in 2010–11

shows that the percentage sulfate saturation of the soils has since

fallen substantially, to 41% and 50% respectively, in line with the

on-going reductions in anthropogenic S deposition in the UK

over the past two decades.

Acknowledgements

The authors would like to thank the Journal’s reviewers who

provided helpful and constructive comments on the original

manuscript. The authors would also like to thank Jane Hall for

compiling the data for Fig. 1. The Natural Environment

Research Council funded the research. Land cover data is

derived from the Countryside Council for Wales Phase I Habitat

Survey. Soil cover information was obtained (under licence) from

Land Information System (LandIS), The National Soil

Resources Institute (NSRI), Cranfield University, UK.

Notes and references

1 D. Barton, D. Hope, M. F. Billett and M. S. Cresser, Appl. Geochem.,1994, 9, 127–139.

2 A. Prechtel, C. Alwell, M. Armbruster, J. Bittersohl, J. M. Cullen,C. D. Evans, R. Helliwell, J. Kopacek, A. Marchetto, E. Matzner,H. Meesenburg, F. Moldan, K. Moritz, J. Vesely and R. F. Wright,Hydrol. Earth Syst. Sci., 2001, 5(Sp. Iss. SI), 311–325.

3 D. W. Johnson, Biogeochemistry, 1984, 1, 29–43.


4 R. B. Harrison, D. W. Johnson and D. E. Todd, J. Environ. Qual.,1989, 18, 419–426.

5 T. T. Chao, M. E. Harward and S. C. Fang, Soil Sci. Soc. Am. Proc.,1964, 28, 632–635.

6 R. L. Parfitt and R. S. C. Smart, Soil Sci. Soc. Am. J., 1978, 42, 48–50.7 B. R. Singh, Soil Sci., 1984, 138, 346–353.8 G. R. Gobran, F. Courchesne and A. Dufresne, in Experimentalreversal of acid rain effects: The Gardsjon Roof Project, ed. H.Hultberg and R. Skeffington, John Wiley & Sons Ltd, 1998, pp.206–229.

9 L. Martinson, M. Alveteg and P. Warfvinge, Environ. Pollut., 2003,124, 119–125.

10 D. W. Johnson and D. E. Todd, Soil Sci. Soc. Am. J., 1983, 47, 792–799.

11 B. J. Cosby, P. G. Whitehead and R. Neale, J. Hydrol., 1986, 84, 381–401.

12 V. Majer, B. J. Cosby, J. Kopacek and J. Vesely, Hydrol. Earth Syst.Sci., 2003, 7, 494–509.

13 J. Kopacek, B. J. Cosby, V. Majer, E. Stuchlik and J. Vesely, Hydrol.Earth Syst. Sci., 2003, 7, 510–524.

14 M. Novak, J. W. Kirchner, H. Groscheova, M. Havel, J. Cerny,R. Krejci and F. Buzek, Geochim. Cosmochim. Acta, 2000, 64, 367–383.

15 C. Alewell, Water, Air, Soil Pollut., 2001, 130, 1271–1276.16 B. W. Avery, Soil Survey Technical Monograph No. 14, Harpenden,

1980.17 B. Reynolds, C. Neal, M. Hornung, S. Hughes and P. A. Stevens,

Water, Air, Soil Pollut., 1988, 38, 55–70.18 J. Westerberg, Acidification and sulphate adsorption by Lake District

soils, Master’s Thesis, Royal Institute of Technology, Stockholm,Sweden, 1998.

19 J. A. McKeague and J. H. Day, Can. J. Soil Sci., 1966, 46, 13–22.20 C. L. Bascomb, J. Soil Sci., 1968, 19, 251–268.21 L. I. Evans and W. G. Wilson, Can. J. Soil Sci., 1985, 65, 489–496.22 B. W. Avery and C. L. Bascomb, Soil Survey Technical Monograph

No.6, Harpenden, 1974.23 B. Reynolds, D. Fowler and S. Thomas, Sci. Total Environ., 1996,

188, 115–125.24 B. Reynolds, D. Fowler, R. I. Smith and J. R. Hall, J. Hydrol., 1997,

194, 305–329.25 B. Reynolds, M. Hornung and P. A. Stevens,Catena, 1987, 14, 13–23.26 J. Hruska and P. Kram, Hydrol. Earth Syst. Sci., 2003, 7, 525–539.27 F. Oulehle, B. J. Cosby, R. F. Wright, J. Hruska, J. Kopacek,

P. Kram, C. D. Evans and F. Moldan, Environ. Pollut., 2012, 165,158–166.

28 RoTAP, Review of Transboundary Air Pollution: Acidification,Eutrophication, Ground Level Ozone and Heavy Metals in the UK,Defra, 2011, http://www.rotap.ceh.ac.uk.

29 R. I. Smith, D. Fowler, M. A. Sutton, C. Flechard and M. Coyle,Atmos. Environ., 2000, 34, 3757–3777.

30 R. I. Smith and D. Fowler, Water, Air, Soil Pollut., 2001, 1, 341–354.

31 D. Fowler, R. I. Smith, J. B. A. Muller and K. J. Vincent, Environ.Pollut., 2005, 137, 15–25.

32 S. Hughes, B. Reynolds, J. A. Hudson and C. Freeman,Hydrol. EarthSyst. Sci., 1997, 1, 661–669.

J. Environ. Monit., 2012, 14, 1531–1541 | 1541


recovery of sulfate saturated soils in the plynlimon catchments, mid-wales following reductions in...

Documents