Science of the Total Environment 470–471 (2014) 543–550

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Sulphate leaching from diffuse agricultural and forest sources in a large central European catchment during 1900–2010 Jiří Kopáček a,⁎, Josef Hejzlar a, Petr Porcal a, Maximilian Posch b a b

Biology Centre AS CR, Institute of Hydrobiology, Na Sádkách 7, 37005 České Budějovice, Czech Republic Coordination Centre for Effects, RIVM, P.O. Box 1, NL-3720 BA Bilthoven, The Netherlands

H I G H L I G H T S • • • • •

Study is based on 50-year monitoring of SO4–S export from the upper Vltava catchment. SO4–S export primarily reflects hydrology and S inputs in fertilisers and deposition. But, mineralization and desorption contribute to SO4–S leaching from soils. Leaching of accumulated SO4–S delays recovery of surface waters from acidification. S losses from farmland increase a risk of S deficiency for S-demanding crops.

a r t i c l e

i n f o

Article history: Received 30 July 2013 Received in revised form 6 October 2013 Accepted 7 October 2013 Available online 26 October 2013 Editor: Christian EW Steinberg Keywords: Modelling Sulphate leaching Sulphur mineralization Diffuse sources

a b s t r a c t Using dynamic, mass budget, and empirical models, we quantified sulphate–sulphur (SO4–S) leaching from soils in a large central European catchment (upper Vltava river, Czech Republic) over a 110-year period (1900–2010). SO4–S inputs to soils with synthetic fertilisers and atmospheric deposition increased in the 1950s–1980s, then rapidly decreased (~80%), and remained low since the middle 1990s. The proportion of drained agricultural land rapidly increased from 4 to 43% between the 1950s and 1990s; then the draining ability of the system slowly decreased due to its ageing. Sulphate concentrations in the Vltava exhibited similar trends as the external SO4–S inputs, suggesting that they could be explained by changes in atmospheric and fertiliser S inputs. The available data and modelling, however, showed that (i) internal SO4–S sources (mineralization of soil organic S in the drained agricultural land), (ii) a hysteresis in SO4–S leaching from forest soils (a net S retention at the high S inputs and then a net release at the lowered inputs), and (iii) hydrology must be taken into account. An empirical model was then employed, based on parameters representing hydrology (discharge), external SO4–S sources (inputs by synthetic fertilisers and atmospheric deposition), and internal SO4–S sources (mineralization related to soil drainage). The model explained 84% of the observed variability in annual SO4–S concentrations in the Vltava river during 1900–2010 and showed that forest soils were a net sink (105 kg ha−1) while agricultural land was a net source (55 kg ha−1) of SO4–S during 1960–2010. In the late 1980s, forest soils changed from a sink to a source of S, and the present release of SO4–S accumulated in forest soils thus delays recovery of surface waters from acidification, while S losses from agricultural soils increase the risk of future S deficiency in S-demanding crops. © 2013 Published by Elsevier B.V.

1. Introduction Since the mid-1900s, when anthropogenic acidification was recognized as a wide-spread phenomenon in many parts of Europe and North America, great progress has been made in the documentation, understanding, and modelling of sulphur (S) deposition effects on terrestrial and aquatic ecosystems (e.g. Psenner and Catalan, 1994; Norton and Veselý, 2004). Long-range atmospheric transport of S has contributed to the acidification of sensitive areas and resulted in ⁎ Corresponding author. Tel.: +420 38 7775878; fax: +420 385 310 248. E-mail address: [email protected] (J. Kopáček). 0048-9697/$ – see front matter © 2013 Published by Elsevier B.V. http://dx.doi.org/10.1016/j.scitotenv.2013.10.013

elevated sulphate–sulphur (SO4–S) concentrations in receiving fresh waters. In contrast to unmanaged (forest and alpine) areas, where atmospheric deposition represents the major S input, agricultural land has also received SO4–S as a part of S-bearing synthetic fertilisers such as ammonium sulphate, potassium sulphate, superphosphate and complex fertilisers since the early 20th century (Eriksen, 2009). Sulphur is not only an acidifying pollutant, but also an essential nutrient required for plant growth, and plays an important role in many plant processes such as synthesis of essential amino acids, chlorophyll, and fixation of nitrogen (N) by leguminous plants (Blair, 2002; Eriksen, 2009). Consequently, significant reductions in S emissions since the 1980s (Smith et al., 2011), decreasing concentrations of

544

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

sulphate in superphosphate since the 1960s, and reduced consumption of synthetic fertilisers in many European countries since the early 1990s (FAO, 2010) have resulted in a ‘surprising’ phenomenon — sulphur deficiency in crops (Schnug, 1991; Zhao et al., 2002; Zelený and Zelená, 2002). Sulphur deficiency was first recognized in S-demanding crops such as canola, and since the mid-1990s also in cereals (Pedersen et al., 1998). Large heterogeneous catchments may thus exhibit the paradoxical situation that the continuing mineralization of organic S stored in unmanaged soils may further contribute to their acidification, while S may become a limiting nutrient in some agricultural areas. Sulphate leaching from (and S accumulation in) unmanaged soils can be successfully predicted using dynamic models, e.g. MAGIC (Model of Acidification of Groundwater In Catchments; Cosby et al., 1985). The MAGIC model has been widely used in a variety of applications to simulate acidification of soils and surface waters; it simulates sulphate dynamics in terrestrial and aquatic ecosystems, based on sulphate retention/release kinetics in soils. Stable S isotope research and mass budget studies have indicated that biological S turnover is (besides adsorption–desorption) an important process in the soil S cycle (Alewell, 2001; Prechtel et al., 2003; Novák et al., 2005). Simple input– output (adsorption–desorption or immobilisation–mineralization) models provide reasonable estimates of sulphate fluxes, if they are considered as a carrier of cations from soils in acidification studies. If sulphate leaching is considered as a loss of an essential nutrient from cropland, however, additional processes affecting S cycling and pools in soils should be considered (Schoenau and Germida, 1992). Sulphate applied to agricultural land may be (i) adsorbed in the soil, (ii) reduced and stored in vegetation, soil microbial biomass and as poorly soluble sulphides in soils and/or released to the atmosphere as H2S, and (iii) leached (e.g., Reuss and Johnson, 1986; Novák et al., 2004, 2005; Ercoli et al., 2012). In contrast, SO4–S may be produced in soils by mineralization of soil organic S (Ghani et al., 1993; Clark et al., 2006). Mineralization and immobilisation of S occur concurrently, and the leaching or incorporation of SO4–S into soil organic matter is thus a net result of several processes, reflecting soil physico-chemical properties (Keer et al., 1986; Eriksen et al., 1995; Eriksen, 2009; Ercoli et al., 2012). Soil processes responsible for sulphur transformations such as oxidation and reduction are mainly microbially mediated and are therefore affected by soil permeability, aeration, moisture, pH, and substrate availability (Ghani et al., 1993; Eriksen, 1997). The same land use changes that affect mineralization of soil organic N (e.g., drainage, tilling, water table management, conversion of arable land to meadows and vice versa; Kopáček et al., 2013a;b) may thus be important factors in the mobilisation of soil S pools in managed agricultural soils. In fact, Singh et al. (2004) reported significant reduction in soil S pools after long-term cultivation of pastures. Another important mechanism controlling sulphate retention in soils via adsorption and mineralization is soil pH. Sulphate adsorption onto Al and Fe oxyhydroxides increases as pH declines to ~3–3.5, because the total positive charge of their surface increases (Stumm, 1992). In contrast, hydroxyl anions effectively replace sulphate adsorbed in soils at higher pH (Tisdale et al., 1984). Curtin and Syers (1990) found that most of soil sulphate was in solution at pH N 6. Elevated soil pH after liming also temporarily increases mineralization of soil organic matter (Nyborg and Hoyt, 1978) and the increased S mineralization then contributes (besides sulphate desorption) to elevated SO4–S leaching (Bolan et al., 1988). Changes in land use and agricultural practices usually occur in parallel (e.g., increased fertilisation rate, soil drainage and liming), prohibiting proper disentangling of their individual effects on SO4–S leaching. An exception to this pattern is the present development of agriculture in the Czech Republic (and other post-communist European countries) due to the abruptly decreased S inputs to agricultural land, resulting from the reduced consumption of synthetic fertilisers and a ~90% reduction in S emission/deposition rates at an otherwise stable

proportion of drained soils (Kopáček et al., 2012, 2013a,b). This largescale ‘experiment’ enables better understanding of the contribution sources (fertilisation plus atmospheric of external and internal SO2− 4 leaching deposition versus mineralization plus desorption) to SO2− 4 from agricultural land. The aims of this study are (i) the quantification of S sources for forest and agricultural areas of a large heterogeneous catchment in central Europe during 1900–2010, (ii) the application of mass budget and dynamic models to estimate the role of diffuse forest and agricultural sources in the surface water pollution with sulphate, and (iii) to evaluate effects of hydrology, external S sources (atmospheric deposition and mineral fertilisers), and internal S sources (elevated mineralization after drainage and tilling of waterlogged agricultural land and reduced soil ability to adsorb sulphate after liming) on SO4–S leaching from agricultural land. To fulfil these aims we first estimated the major SO4–S inputs into the catchment and measured and modelled the major export fluxes from the catchment sources to surface waters. Then we developed and calibrated an empirical model for SO4–S leaching from agricultural land, and finally compared the export fluxes to the model estimates.

2. Materials and methods 2.1. Site description The upper Vltava catchment (12,968 km2, elevation 271–1378 m; Fig. 1) stretches from the mountain range between the Czech Republic, Austria, and Germany to the Slapy Reservoir, built in 1954 ~40 km upstream of Prague. The bedrock of the catchment is mostly formed by crystalline rocks, and soils are dominated by cambisols, with depths usually b1 m in steep mountain areas and N 1 m elsewhere. At present, agricultural land, forests (mostly plantations of Norway spruce; Picea abies), surface waters, and urban areas cover 52%, 42%, 3%, and 3% of the catchment, respectively, but their proportions changed during the 20th century. The forest proportion increased from 32% to 42%, whereas agricultural land decreased from 63% to 52% between 1900 and 2010, with the major changes occurring during 1945–1947 (Kopáček et al., 2013b). The area and volume of surface waters (mostly shallow polymictic ponds) increased from ~360 to 460 km2 and from ~0.3 to 1.66 km3, respectively, in the study catchment between 1900 and 1991 due to the construction of eight deep valley reservoirs. Water residence times in the surface waters thus increased from 0.1–0.2 years during 1900–1960 to a wide range of 0.3–0.9 years during 1980–2010 (Kopáček et al., 2013b).

Fig. 1. Catchment of the upper Vltava river (grey area) and boundary of the administrative South Bohemian District and their location in the Czech Republic. Points with abbreviations indicate locations of the sampled forest lakes and streams (full names and other details are in Supplementary Information, Table SI-2).

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

The area of the upper Vltava catchment is almost identical to the area of the administrative South Bohemian District (11,347 km2; Fig. 1). Annual statistics on agricultural activities (application of synthetic fertilisers and drainage of farmland) are available for this district from 1957 onwards, and population data (for individual settlements) since 1900 (yearbooks by the Czech Statistical Office). Annual agricultural statistics for the district prior to 1957 were derived from the available statistics for the whole area of the Czech Republic (Kopáček et al., 2013b).

545

using a regional regression model developed for the upper Vltava catchment (Kopáček et al., 2012), using central European SO2 emission trends for 1900–2010. The model was based on measured atmospheric deposition of SO4–S at ten bulk precipitation and nine throughfall stations during 1978–2009. The stations were situated in (or near) the catchment and distributed along an elevation gradient from 275 to 1334 m. The modelled deposition trends were related to the mean catchment elevation of 579 m (for more details see SI, Part 2 and Fig. SI-2). 2.3. Sulphur export to surface waters

2.2. Sulphur input to the catchment Total S input to the catchment (IS) was the sum of SO4–S inputs to agricultural land (IAL), forests (IFO), and imported for industrial purposes (IIN), i.e. IS =IAL +IFO +IIN; for a list of abbreviations see Table 1. The total annual input of SO4–S to agricultural land was calculated as the sum of SO4–S in synthetic fertilisers and bulk deposition. The annual SO4–S input to forests was set equal to throughfall deposition. The annual amount of S from industry was estimated from wood pulp production as the most important S consuming industrial process in the catchment (see below). The SO4–S input associated with application of synthetic fertilisers was based on the annual consumption of MgSO4, CaSO4 and individual S-bearing N, P, and K fertilisers in the upper Vltava catchment and their respective S concentrations. The historical development of the percent S contributions to N, P, and K fertilisers was reconstructed from a literature survey and linear interpolation between available values. For more details see Supplementary Information (SI, Part 1, Fig. SI-1). Sulphur input in organic fertilisers was not included, assuming that this flux roughly compensated for S removal from soils in feed and straw. Annual fluxes of bulk and throughfall depositions of SO4–S in the open area (without trees) and forests, respectively, were calculated

Table 1 List of symbols used in the modelling. Symbol

Description

AW CS, CS*

Total area of all surface waters in the catchment Discharge-weighted mean concentration of SO4–S in the Slapy Reservoir: CS is measured; CS* is calculated (Eq. (6)) from the modelled ES* values Concentration of SO4–S in the water draining agricultural land (CAL = EAL/QAL) Percent proportion of drained agricultural land Total SO4–S export from the catchment to surface waters: ES is calculated from Eq. (4); ES* = EAL* + EFO + EWW + EIN + EAD SO4–S exports from agricultural land: EAL is calculated from Eq. (5); EAL* is modelled as an empirical function of Q, IAL, and %D (Eq. (7)) SO4–S exports from forests (FO), waste waters (WW), and industry (IN) Total S input to the catchment (IS = IAL + IFO + IIN) S input to agricultural land in synthetic fertilisers and bulk deposition S input to forests in throughfall deposition S input to the catchment for industrial purposes Annual change in storage of SO4–S in the Slapy Reservoir SO4–S output from the Slapy Reservoir: OS is measured; OS* is calculated from the modelled CS* (OS* = Q · CS*) Annual average discharge of the Vltava river at the Slapy Reservoir Discharges originating from agricultural land (AL) and forests (FO) Water load per unit area of surface waters in the catchment (qW = Q/AW) Net removal of SO4–S in the surface waters Proportion of SO4–S removed from waters by internal processes (rS = RS/ES); Eqs. (3) and (4) Mass transfer coefficient for SO4–S

CAL %D ES, ES* EAL, EAL*

EFO, EWW, EIN IS IAL IFO IIN ΔMS OS, OS* Q QAL, QFO qW RS rS sS

The total annual SO4–S export (ES, g yr−1) from terrestrial and atmospheric sources to the surface waters of the upper Vltava catchment was estimated from the mass balance: ES ¼ OS þ ΔMS þ RS

ð1Þ

where OS (g yr−1) is the total annual SO4–S output from the Slapy Reservoir, ΔMS (g yr−1) is the annual change in storage of SO4–S in the reservoir, and RS (g yr−1) is the net removal of SO4–S in the water bodies. The OS flux was calculated as the product of the annual Vltava discharge (Q; m3 yr−1) and the discharge-weighted mean concentration of SO4–S (CS, g m−3) in the Slapy Reservoir. The change in SO4–S storage was calculated as the difference between the SO4–S concentration in the reservoir in January of the year of interest (n) and the previous year (n − 1), multiplied by its water volume (V, m3), i.e., ΔMS = V · (Cn − Cn − 1). Data on annual average discharges of the Vltava at the Slapy Reservoir (Q, m3 yr−1) for 1900–2010 come from Kopáček et al. (2013b). Concentrations of SO4–S were measured at the Slapy Reservoir 5–17 times per year during 1960–1969 and then regularly at three-week intervals by a colorimetric method (barium chloranilate; Procházková, 1961; 1960–1996) and by ion chromatography (since 1996). For details on the comparability of analytical methods, data quality control, and the reconstruction of unreliable or missing data in 1960–1995 see SI (Part 3). Annual average SO4–S concentrations were calculated as discharge-weighted means. Historical data (1875–1959) on sulphate concentrations in the Vltava river were reviewed in the literature (for sources and reliability see SI, Part 3). An average of the sulphate concentrations in 1875, 1885, and 1886 was used as a background value in 1900 in this study. The term RS in Eq. (1) represents the sum of SO4–S losses from surface waters due to reduction and the net burial in sediments. The internal SO4–S removal was not measured, but can be estimated as RS = ES · rS, where rS is the relative proportion of the SO4–S export from catchment and atmospheric sources to surface waters that is removed from the waters by internal processes (reduction, assimilation and burial in sediments). It can be obtained from either the mass balances, provided all fluxes are known: rS ¼

RS O þ ΔM S ¼ 1− S ES ES

ð2Þ

or from an empirical relationship widely used for modelling SO4–S removal in lakes (e.g., Kelly et al., 1987; Baker and Brezonik, 1988): r S ¼ sS =ðsS þ qW Þ

ð3Þ

where sS is the net mass transfer coefficient and qW (m yr−1) is the water load per unit area of waters. A value of sS = 0.5 m yr−1, recommended as default for the First-order Acidity Balance (FAB) model (Posch et al., 2012), was used for the whole study period. The qW values were calculated as qW = Q / AW, where AW (m2) is the total area of all surface waters in the catchment. Combining Eqs. (2) and (3) allows the estimation of the total annual S inputs to surface waters from external sources on the basis of

546

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

measured values of SO4–S concentrations and output from the Slapy Reservoir: ES ¼

OS þ ΔMS 1−sS =ðsS þ qW Þ

ð4Þ

The resulting total ES equals the sum of SO4–S exports from agricultural land (EAL) and forests (EFO), which are the dominant diffuse sources, waste waters from sanitary systems (EWW), and industry (EIN), which are the dominant point sources, and from atmospheric deposition (EAD) onto the water surface: ES = EAL + EFO + EWW + EIN + EAD. The individual EFO, EWW, EIN, EAD, and EAL fluxes of SO4–S from external sources to surface waters of the upper Vltava catchment were estimated as follows. Sulphate export from forested catchments was measured at three lakes and at seven first- or second-order streams in the upper Vltava catchment or its surroundings, varying in elevation from 662 to 1090 m (Fig. 1, Table SI-2). The lakes were sampled in the early 1960s and then regularly at 2-week to 6-month intervals since 1984. The streams were sampled at 3-week intervals during several years in the 2000s (Table SI-2). Annual EFO fluxes were the products of the average SO4–S concentrations in the forest lakes and streams and the respective discharges originating from forested areas (QFO). The QFO values were calculated from Q as QFO = 0.54 · Q, where the 54% proportion in the total Q was estimated by a hydrological model (Kopáček et al., 2013a). The time series of SO4–S concentrations in forest surface waters was reconstructed using measured and modelled SO4–S concentrations. Sulphate concentrations prior to 1984 were based on trends reconstructed for the Bohemian Forest lakes (Majer et al., 2003), using the MAGIC model, version 7 (Cosby et al., 2001). Historical SO4–S concentrations used for model calibration were taken from a literature review (see SI, Part 4). For the 1984–2010 period, the time series of EFO fluxes was obtained as the geometric mean of all available data measured at all forest catchments (Fig. SI-3). Annual EWW fluxes were estimated on the basis of population in the study catchment and the per capita S production of 2.7 g capita−1 day−1 in urine (Pitter, 2009), assuming that all this S sooner or later enters the surface waters as SO4–S. Wood pulp production was the most important industrial process contributing to water pollution with S compounds. Annual EIN flux was estimated as a product of annual wood pulp and paper production in the upper Vltava catchment and an average consumption of 25–30 g of Na2SO3 (~7.5 g S) kg−1 of the product (Sjöström, 1981). For details see SI (Part 5). Annual EAD fluxes of SO4–S onto the water surfaces were set equal to bulk deposition (Fig. SI-2A). Annual EAL fluxes (i.e., S export to surface waters from agricultural land) were calculated for the 1960–2010 period as: EAL ¼ ES –EFO –EWW –EIN –EAD :

ð5Þ

The annual average concentration of SO4–S in the water draining from agricultural land (CAL) was then calculated as CAL = EAL / QAL for individual years, where QAL represents discharge originating from agricultural land, and was calculated as QAL = 0.41 · Q on the basis of the hydrological model by Kopáček et al. (2013a). A model computing EAL values from variables characterizing sulphate leaching from soils in the 1960–2010 period (see next section) was then used to estimate annual EAL fluxes during 1900–1959. 2.4. Modelling sulphur export from agricultural land A two-step regression model between the element export from agricultural land and variables characterizing hydrology and its external

and internal sources (Kopáček et al., 2013b) was used to estimate the EAL fluxes of SO4–S during 1960–2010. The model was based on empirical relationships between the EAL fluxes and discharge, SO4–S input by atmospheric deposition and synthetic fertilisers (IAL) as a major external S input to agricultural land, and internal SO4–S sources (mineralization of soil organic S and sulphate desorption). We used the proportion of drained agricultural land (%D) as a proxy for mineralization, and soil liming (application rate of Ca; kg ha−1 yr−1) as a proxy for soil pH change and its sorption/desorption ability. In the model, the element amount exported from terrestrial systems with a unit volume of water depends primarily on its leachable amount in the soil (Kopáček et al., 2013b). The first step was based on linear regressions (with zero intercepts) calculated between EAL (g yr−1) and QAL (m3 yr−1); the slopes of the EAL vs. QAL relationships have a unit of mg l−1 and represent the average SO4–S concentrations in the water outflow from agricultural land for the given periods. These relationships were calculated for five 6-yr (1959–1989) and three 7-yr (1990–2010) periods. These increments were chosen to equally (and statistically significantly) cover periods of the progressive drainage of agricultural land and the decreasing draining ability of the draining system, respectively (Fig. 2C). The second step was based on regressions between the slopes and the IAL, %D, and Ca-dose values averaged for the same periods as the respective slopes. Thus, the modelled EAL flux (EAL*) was defined as EAL* = f(Q, IAL, %D, Ca). The model results were checked by comparison of the calculated annual average concentrations of SO4–S (CS*, mg l−1) with those measured in the Vltava (CS). The CS* concentrations were calculated from the annual EAL* fluxes and Q at the Slapy Reservoir according to: 

CS ¼

ð1−r S Þ  ES −ΔM S Q

ð6Þ

where ES* (g yr−1) is the calculated total SO4–S export from terrestrial and atmospheric sources to the surface waters in the catchment (ES* = EAL* + EFO + EWW + EIN + EAD). The best regression model for computing the EAL* fluxes (i.e., the model best explaining the observed CS variability in the upper Vltava during 1960–2010) was then used to reconstruct the EAL* fluxes and CS* concentrations throughout 1900–2010. Note: A list of all symbols used in the modelling is provided in Table 1. 2.5. Land use changes important for the model calibration The proportion of drained agricultural land in the upper Vltava catchment rapidly increased from ~4% in 1960 to 43% in 1990. The construction of new drainage systems declined in the early 1990s, and completely ceased (including maintenance of the existing system) in 1994. The proportion of drained agricultural land prior to 1960 was based on linear extrapolation between 0%, 1.3%, 2.8%, and 4% in 1920, 1935, 1945, and 1960, respectively. Most of the drainage was done with intensive subsurface drainage systems with no water level control. This system continuously loses capacity due to the damage of tubes or clogging with roots and silt, and its average lifetime is ~40 years (Kulhavý et al., 2007). Consequently, we assumed that the draining ability of the ageing system decreased annually by 2.5% (=100 / 40) from its maximum efficiency in 1994 (Kopáček et al., 2013b). The average areal consumption of Ca for liming (as limestone, dolomite, and calcium oxide) of agricultural land comes from Vaněk and Penk (1991) and databases of the Ministry of Agriculture of the Czech Republic. Intensive liming in the 1970s and 1980s resulted in the alkalization of agricultural land. The proportion of acidic arable soils (pH b 5.5) decreased from ~55% to 22% in the study catchment between the 1960s and the early 1990s. After a ~90% reduction in the liming doses since the early 1990s, soils began to acidify again and the proportion of acidic arable soils increased to the present ~36% (Kopáček et al., 2013b).

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

250

25

A 15

150

10

100

5

50

0

0 70

70

S input (kg ha-1 yr-1)

60

Q (m3 s -1) .

200

CS Q

B

60

IS ES

50

50

40

40

30

30

20

20

10

10

0

0

S export (kg ha-1 yr-1) .

Cs (mg l-1)

20

300

30

250

%D Ca

200 150

20 100

Ca (kg ha-1 yr-1)

Drainage (%)

40

C

10 50 0 1900

0 1920

1940

1960

1980

2000

Year Fig. 2. Historical development of SO4–S concentrations (CS) and discharge (Q) in the upper Vltava river (A) and major variables responsible for this trend: (B) total SO4–S input to the upper Vltava catchment (IS) in synthetic fertilisers, atmospheric deposition and for industrial purposes and total SO4–S export to surface waters calculated from Eq. (4), both expressed on a catchment area basis, and (C) percent proportion of drained agricultural area (%D; points are based on statistics, lines are estimated) and Ca dose to agricultural land associated with liming of fields.

3. Results 3.1. Sulphur fluxes in the catchment Concentrations of SO4–S exhibited a pronounced trend in the Vltava during 1960–2010, with maxima of 15–21 mg l−1 in 1975–1995 (Fig. 2A). The total SO4–S export from external sources, calculated with Eq. (4), brought 16–87GgSyr−1 to the surface waters in the catchment (Fig. SI-4). This ES flux was for the whole study period dominated by sulphate exports from agricultural land (EAL = 9–59 Gg S yr−1) and forest areas (EFO = 4–28 Gg S yr−1), while other sources (EIN, EWW, and EAD) were 1–2 orders of magnitude lower (Fig. SI-5). The cumulative ES value was 1591 kg ha−1 on a whole catchment area basis and was almost equal to the total S input to the catchment (IS = 1586 kg ha−1) in synthetic fertilisers, atmospheric deposition and for industrial purposes (i.e., 728, 816, and 42 kg ha−1, respectively) during 1960–2010 (Fig. 2B). The maximum values of SO4–S inputs (52 kg ha−1 yr−1)

547

occurred in 1975–1985, and then decreased sharply to the present 11–13 kg ha−1 yr−1 (Fig. 2B). Sulphate concentrations in the Vltava were discharge independent, when tested on an annual basis for the whole 1960–2010 period, but variability in the discharge explained 20% of the observed CS variability (positive correlation, p b 0.05) during the period of high CS values (1975–1995; Fig. 2A). In this period, large pools of leachable SO4–S occurred in agricultural and forest soils due to high S inputs. The S input in synthetic fertilisers brought 40–62 kg ha−1 yr−1 to the agricultural land (Fig. SI-1), and together with atmospheric deposition (Fig. SI-2) they reached 52–72kgha−1 yr−1 during 1975–1985 (Fig. 3A). Throughfall deposition brought 41–51 kgha−1 yr−1 of SO4–S to the forest areas in the same period (Fig. 3A). The SO4–S exports from agricultural land were on average 6 kg ha−1 yr−1 higher than those from forest areas in 1960–2010, but both fluxes exhibited similar trends (Fig. 3B). The respective cumulative inputs and export fluxes of SO4–S were 1772 and 1827 kg ha−1 in agricultural land and 1397 and 1292 kg ha−1 in forests during 1960–2010. These results suggest that S accumulated in unmanaged forest soils (on average 105 kg ha−1) while it was depleted in agricultural land (on average 55 kg ha−1) during the last 50 years. The net sulphate losses from agricultural land probably resulted from mineralization and desorption (or reduced soil sorption capacity) of drained, cultivated, and limed soils as suggested by the following correlations between SO4–S export and concentrations in runoff water from agricultural land. The trend in CAL positively correlated with IAL and Ca dose, which individually explained 39% and 21%, respectively, of its variability (Table 2). Linear regression between CAL and %D was not significant during the whole 1960–2010 period, but variability in the %D explained 75% and 55% of the CAL variability in 1960–1983 and 1990–2010, respectively. As expected, discharge explained most (52%) of the EAL variability, because EAL is a product of QAL and CAL. SO4–S input to agricultural land and liming explained 24% and 15% of the EAL variability, respectively, while drainage of soils did not correlate (in contrast to nitrate leaching; Kopáček et al., 2013b) with the EAL values (Table 2). However, adding %D to the independent variables Q and IAL explained more variability of EAL (75% vs. 68%), as well as CAL (54% vs. 42%). Similarly, adding of Ca dose to the regression further improved explanatory power of both EAL and CAL values (Table 2). These results suggested that a model for computing EAL fluxes and CAL in the upper Vltava catchment should incorporate all four variables, representing hydrology, as well as external sulphate sources (IAL) and proxies for the internal sulphate sources (%D and Ca dose). 3.2. Empirical model of sulphate leaching from agricultural land All eight EAL vs. QAL relationships were significant (p b 0.05) and their slopes, representing the average SO4–S concentrations in the water outflow from agricultural land for the given periods, increased from 16 to 32 mg l−1 between 1960 and 1983, and then declined to 13 mg l−1 in 2003–2010 (Fig. 4A). IAL, %D, and Ca individually explained 72%, 9%, and 57% of the slope variability over the whole 1960–2010 period, respectively, and the slope vs. IAL and slope vs. %D relationships exhibited nonlinear trends (Fig. 4B, C). The slopes also positively correlated with the Ca input to agricultural land (Fig. 4D). The combination of IAL and %D values explained 95% of the slope variability. The adding of liming, however, improved the model only slightly (to 97%), due probably to its autocorrelation with other variables: 64% with IAL during the whole study period, and 88% with %D during 1960–1990 (Fig. 2C). Consequently, we modelled the EAL* fluxes as the product of QAL and the slope, which was an empirical function of IAL (kg ha−1 yr−1) and %D, whereas the small additional effect of Ca dose was neglected in this study: EAL  ¼ 0:41  Q  Slope ¼ 0:41  Q  ð3:8 þ 0:32  IAL þ 0:22  %DÞ

ð7Þ

(the empirical relationship of slope vs. IAL and %D is in the brackets).

548

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

A

80

AL FO

60

S export (kg ha-1 yr-1)

S input (kg ha-1 yr-1)

80

40

20

0 1900

1920

1940

1960

1980

2000

Year

B

60

AL (Eq. 5) AL (Eq. 7) FO

40

20

0 1900

1920

1940

1960

1980

2000

Year

Fig. 3. Major SO4–S fluxes in the upper Vltava catchment expressed on an area basis of the respective land categories: (A) SO4–S input to agricultural land (AL) and forests (FO), and (B) SO4–S export from these areas to surface waters. Lines AL (Eq. 5) and AL (Eq. 7) denote S exports from agricultural land calculated with Eqs. (5) and (7), respectively.

The CS* concentrations calculated using this model and Eq. (6) explained 84% of the observed CS variability in the upper Vltava during 1900–2010 (Fig. 5) and provided similar average SO4–S concentrations of 13.4 and 13.3 mg l−1, respectively, over 1960–2010. The calculated SO4–S output from the Slapy Reservoir (OS* = Q · CS*) explained 93% of the observed OS values (Fig. SI-6). The EAL* fluxes calculated using Eq. (7) varied within 5–25 (average of 11) kg ha− 1 yr− 1 during 1900–1959 and were on average 2.5 kg ha− 1 yr− 1 higher than SO4–S export from forests in this period (Fig. 3B).

4. Discussion Our model for estimating the sulphate export from agricultural sources is based on both directly measured data and calculated fluxes, which entail different degrees of uncertainty (see SI, Part 6). Among them, the EFO flux had the relatively highest effect on the EAL flux (computed from Eq. (5)) because both fluxes were of similar magnitude (Fig. 3). The cumulative EFO flux was, however, only slightly lower (due to S retention in soils) than the cumulative S input to forests, suggesting that the EFO fluxes were estimated reasonably well for the purpose of this study. Despite a pronounced hysteresis in SO4–S input–export fluxes in forest soils (a net S retention at the high S inputs until the late 1980s, and then a net release at the lower inputs), the modelled EFO fluxes fit well the observed exports from forests throughout the whole study period and especially during the 1980s–2000s (Fig. 6). The latter period was crucial for the development of mass budget and regression models used for computing the sulphate export from agricultural land, because most changes in atmospheric deposition and consumption of synthetic fertilisers occurred during this time,

Table 2 Results of linear regressions (R2) between calculated annual SO4–S export from agricultural land (EAL) or annual average SO4–S concentrations in the outlet from agricultural land (CAL) in the upper Vltava catchment in 1959–2010 and 4 independent variables or their combinations (n = 51). Dependent variables EAL CAL

Independent variablesa Q

IAL

%D

Ca

Q, IAL

Q, IAL, %D

Q, IAL, %D, Ca

0.52*** 0.05

0.24*** 0.39***

0.01 0.06

0.12* 0.21***

0.68*** 0.42***

0.75*** 0.54***

0.77*** 0.59***

a Independent variables: Discharge (Q), S input to agricultural land (IAL; synthetic fertilisers and atmospheric deposition), proportion of drained agricultural area (%D), and Ca dose to agricultural land (a proxy for soil pH), associated with liming of fields in the catchment. Asterisks indicate significances: * b0.05; ** b0.01; *** b0.001.

while the percent proportion of drained farmland changed less dramatically (Fig. 2). The different rates of changes in agricultural practice enabled partial disentangling of the effects of external and internal sources for S leaching (Fig. 4) and indicated that drainage and liming mobilised S pools in soils and contributed to sulphate leaching from agricultural land. The slopes of EAL vs. QAL relationships increased almost linearly with the increasing IAL in 1960–1983, but declined at a slower rate than could be expected from the decline in IAL values in the 1990s (Fig. 4B). A possible explanation for this pattern is an important contribution of internal S sources to the SO4–S leaching. Mineralization of soil organic S commonly varies between 0.5 and 3% of its soil pool (Keer et al., 1986; Eriksen et al., 1995) and is intensified after soil drainage and cultivation (Singh et al., 2004). The maximum proportion of drained soils occurred in the early 1990s (Fig. 2C). In addition, the proportion of acidic agricultural soils was lowest in the catchment at that time (Kopáček et al., 2013b) due to intensive liming in the 1980s (Fig. 2C). Our data suggest that the ability of agricultural soils to adsorb sulphate thus probably reached its minimum during the study period, diminishing the adsorption of sulphate added to soils in synthetic fertilisers, atmospheric deposition and liberated by mineralization of soil S pools. The continuing mineralization and low adsorption (and possibly also partial desorption of previously adsorbed sulphate in more acidic soils) maintained high concentrations of soil SO4–S available for leaching, and resulted in high SO4–S exports from soils even in the 1990s, when external S inputs to agricultural land decreased abruptly (Fig. 4B). The rapid decline in external sulphate input to agricultural land (Fig. 3A) caused a nonlinear trend of the slope vs. %D relationship (Fig. 4C). The slopes increased linearly with the increasing percentage of drained agricultural land in 1960–1983, and then decreased at similar rate (but with a lower intercept) during the period of drainage system ageing in the last two decades (Fig. 4C). Our empirical model reasonably explained the observed long-term trend in SO4–S concentrations in the Vltava during 1900–2010, with the exception of the high inter-annual variability in the middle 1980s, which clearly followed hydrological conditions (Fig. 5). Similar to nitrate (Kopáček et al., 2013b), sulphate concentrations were low during dry years in the 1980s (Fig. 5), probably because SO4–S could not be leached from soils due to low hydraulic conductivity in the soil profile. We assume that due to longer residence time in soils during dry periods, more SO4–S was microbially reduced in anoxic soil microsites and temporally stored as organic S or sulphides in the upper soil horizons, similar to forest soils (Mörth et al., 2005; Novák et al., 2005; Kopáček et al., 2013c). The effect of temperature, soil moisture and microbial composition on SO4–S mineralization and immobilisation are thus additional variables, which should be considered in modelling S leaching from soils, especially in

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

40

A

**

30

*

*

**

** **

20

Slope (mg SO4-S l-1)

Slope (mg SO4-S l-1)

40

***

**

10 0 1960

B

30

1972-77

20

1997-03 10

1966-71 1960-65

2003-10

0 1970

1980

1990

2000

2010

0

10

20

1972-77

Slope (mg SO4-S l-1)

Slope (mg SO4-S l-1)

1978-83

30

1984-89 1990-96

1966-71

1997-03 1960-65

10

30

40

50

60

70

IAL (kg S ha-1 yr-1) 40

C

20

1978-83 1984-89

1990-96

Year 40

549

2003-10

D

1978-83 1972-77

30

1984-89

1990-96

20 1997-03 10

1966-71 1960-65

2003-10

0

0 0

10

20

30

40

0

50

50

100

150

200

250

Liming (kg Ca ha-1 yr-1)

Drainage (% of AL)

Fig. 4. Slopes (mg l−1) of linear regression between SO4–S export from agricultural land (EAL, g yr−1, Fig. 3B) and annual average discharge (Q, m3 yr−1, Fig. 2A) in the upper Vltava catchment calculated for eight 6–7-year intervals between 1960 and 2010: (A) Time series of the slopes. Asterisks above line indicate significance of individual regressions (*, p b 0.05; **, p b 0.01; ***, p b 0.001). Relationships between slopes and (B) net SO4–S input to agricultural land (IAL, in synthetic fertilisers and atmospheric deposition), (C) proportion of the drained agricultural land, and (D) calcium (Ca) dose associated with liming of agricultural land. Time periods for which slopes were calculated are indicated at the points.

models aiming to explain its inter-annual and seasonal variability (e.g., Schoenau and Germida, 1992). While the MAGIC model is a reasonable tool for reconstruction and prediction of sulphate leaching from unmanaged soils on the basis of soil properties and atmospheric S deposition (Fig. 6), similar models for cultivated soils should include more explanatory variables. Our results show that both external and internal SO4–S sources significantly contribute to SO4–S leaching from agricultural land. The changing proportion of drained agricultural land was a good proxy for mineralization of soil S (as well as organic N; Kopáček et al., 2013b) pools. Even though adding of soil liming did not significantly improve our model on sulphate leaching from farmland in the upper Vltava catchment (due to its correlations with other independent variables), it probably plays an important role in soil SO4–S immobilisation, as observed elsewhere (Tisdale et al., 1984; Bolan et al., 1988). Consequently, variables like drainage, cultivation and either liming or changes in soil pH should

250

25

20

200

20

15

150

10

100

5

50

5

0 1900

0

0

25

CS

B

Q

CS* (mg l-1)

CS*

Q (m3 s-1)

A SO4-S concnetration (mg l-1)

not be omitted in similar modelling of long-term trends in sulphate leaching from agricultural land. Our results imply two important consequences for forest and agricultural land in the upper Vltava catchment. Due to the net SO4–S accumulation in forest soils during the most of the 20th century (Fig. 6), these ecosystems are enriched with S by 105 kg ha−1 compared to 1960 and by N200 kg ha−1 compared to 1850 (Kopáček et al., 2001). The present leaching of SO4–S exceeds its throughfall deposition and delays recovery of surface waters from acidification, which remains slow despite the ~90% reduction in SO2 emissions in the region (Kopáček et al., 2012). A new steady state in the input–output fluxes of SO4–S in the forest soils is expected to occur in the study area as late as in the middle of the 21st century (Majer et al., 2003). In contrast to forests, agricultural soils lost 55 kg ha−1 of S compared to 1960, despite very high SO4–S inputs in synthetic fertilisers and atmospheric deposition (Fig. 3A). This loss will probably continue, especially in the drained areas, increasing thus the risk of S deficiency for S-demanding crops in the upper Vltava catchment.

15

10

y = 1.00x R2 = 0.84

1920

1940

1960

Year

1980

2000

0

5

10

15

20

25

CS (mg l-1)

Fig. 5. Modelled (CS*) and observed (CS) SO4–S concentrations and discharge (Q) in the Vltava (Slapy Reservoir) in 1900–2010 (A), and linear regression between CS* and CS (solid line) (B).

550

J. Kopáček et al. / Science of the Total Environment 470–471 (2014) 543–550

Terrestrial SO4-S export (kg ha-1 yr -1)

50

1985

Observed Series1 Series3 MAGIC 40

1990 30

2000

1:1

20

1960 2010

10

2020 1900 0 0

10

20

30

40

50

SO4-S deposition (kg ha-1 yr-1) Fig. 6. Input and output fluxes of SO4–S in Čertovo Lake catchment (an unmanaged mountain forest area). Details on observed and modelled (MAGIC) exports can be found in Kopáček et al. (2006) and Majer et al. (2003), respectively. Numbers at filled points indicate years.

Acknowledgements We thank our colleagues who participated in the chemical monitoring of the Slapy Reservoir and forest streams. We also thank two anonymous reviewers for their constructive criticism of an earlier version of the paper. This study was partially supported by the Grant Agency of the Czech Republic (Project Nos. P504/12/ 1218 and 526/09/0567). Appendix A. Supplementary data Supplementary Information associated with this article can be found in the online version, at http://dx.doi.org/10.1016/j.scitotenv.2013.10. 013. These data include Google map of the most important areas described in this article. References Alewell C. Predicting reversibility of acidification: the European sulfur story. Water Air Soil Pollut 2001;130:1271–6. Baker LA, Brezonik PL. Dynamic model of in-lake alkalinity generation. Water Resour Res 1988;24(1):65–74. Blair GJ. Sulphur fertilisers: a global perspective, proceedings no 498. York, United Kingdom: International Fertiliser Society; 2002. Bolan NS, Syers JK, Tillman RW, Scotter DR. Effect of liming and phosphate additions on sulphate leaching in soils. J Soil Sci 1988;39:493–504. Clark JM, Chapman PJ, Heathwaite AL, Adamson JK. Suppression of dissolved organic carbon by sulphate induced acidification during simulated droughts. Environ Sci Technol 2006;40:1776–83. Cosby BJ, Hornberger GM, Galloway JN, Wright RF. Modelling the effects of acid deposition: assessment of a lumped-parameter model of soil water and stream water chemistry. Water Resour Res 1985;21:51–63. Cosby BJ, Ferrier RC, Jenkins A, Wright RF. Modelling the effects of acid deposition: refinements, adjustments and inclusion of nitrogen dynamics in the MAGIC model. Hydrol Earth Syst Sci 2001;5:499–518. Curtin D, Syers JK. Extractability and adsorption of sulphate in soils. J Soil Sci 1990;41: 295–304. Ercoli L, Arduini I, Mariotti M, Lulli L, Masoni A. Management of sulphur fertiliser to improve durum wheat production and minimise S leaching. Eur J Agron 2012;38: 74–82. Eriksen J. Sulphur cycling in Danish agricultural soils: inorganic sulphate dynamics and plant uptake. Soil Biol Biochem 1997;29:1379–85. Eriksen J. Soil sulphur cycling in temperate agricultural systems. Adv Agron 2009;102: 55–89. Eriksen J, Mortensen JV, Dissing Nielsen J, Nielsen NE. Sulphur mineralization in five Danish soils as measured by plant uptake in a pot experiment. Agric Ecosyst Environ 1995;56:43–51.

FAO. Statistical Databases, Food and Agriculture Organization of the United Nations, Rome. http://faostat.fao.org, 2010. Ghani A, McLaren RG, Swift RS. Mobilization of recently-formed soil organic sulphur. Soil Biol Biochem 1993;25:1739–44. Keer JI, McLaren RG, Swift RS. The sulphur status of intensive grassland sites in southern Scotland. Grass Forage Sci 1986;41:183–90. Kelly CA, Rudd JWM, Hesslein RH, Schindler DW, Dillon PJ, Driscoll CT, et al. Prediction of biological acid neutralisation in acid-sensitive lakes. Biogeochemistry 1987;3: 129–40. Kopáček J, Veselý J, Stuchlík E. Sulphur and nitrogen fluxes and budgets in the Bohemian Forest and Tatra Mountains during the industrial revolution (1850–2000). Hydrol Earth Syst Sci 2001;5(3):391–405. Kopáček J, Turek J, Hejzlar J, Kaňa J, Porcal P. Element fluxes in watershed–lake ecosystems recovering from acidification: Čertovo Lake, the Bohemian Forest, 2001–2005. Biologia 2006;61(Suppl. 20):S413–26. Kopáček J, Posch M, Hejzlar J, Oulehle F, Volková A. An elevation-based regional model for interpolating sulphur and nitrogen deposition. Atmos Environ 2012;50:287–96. Kopáček J, Hejzlar J, Posch M. Factors controlling the export of nitrogen from agricultural land in a large central European catchment during 1900–2010. Environ Sci Technol 2013a;47:6400–7. Kopáček J, Hejzlar J, Posch M. Quantifying nitrogen leaching from diffuse agricultural and forest sources in a large heterogeneous catchment. Biogeochemistry 2013b. http://dx.doi.org/10.1007/s10533-013-9825-5. [online]. Kopáček J, Cosby BJ, Evans CD, Hruška J, Moldan F, Oulehle F, et al. Nitrogen, organic carbon and sulphur cycling in terrestrial ecosystems: linking nitrogen saturation to carbon limitation of soil microbial processes. Biogeochemistry 2013c. http://dx.doi.org/10.1007/s10533-013-9892-7. [online]. Kulhavý Z, Doležal F, Fučík P, Kulhavý F, Kvítek T, Muzikář R, et al. Management of agricultural drainage systems in the Czech Republic. Irrig Drain 2007;56:141–9. Majer V, Cosby BJ, Kopáček J, Veselý J. Modelling reversibility of central European mountain lakes from acidification: Part I — the Bohemian Forest. Hydrol Earth Syst Sci 2003;7(4):494–509. Mörth C-M, Torssander P, Kjønaas JO, Stuanes AO, Moldan F, Giesler R. Mineralization of organic sulfur delays recovery from anthropogenic acidification. Environ Sci Technol 2005;39:5234–40. Norton SA, Veselý J. Acidification and acid rain. In: Lollar BS, editor. Environmental Geochemistry. Treatise on GeochemistryOxford: Elsevier-Pergamon; 2004. p. 367–406. Novák M, Michel RL, Přechová E, Štěpánová M. The missing flux in a 35S budget for the soils of a small polluted catchment. Water Air Soil Pollut Focus 2004;4:517–29. Novák M, Kirchner JW, Fottová D, Prechová E, Jačková I, Krám P, et al. Isotopic evidence for processes of sulfur retention/release in 13 forested catchments spanning a strong pollution gradient (Czech Republic, central Europe). Global Biogeochem Cycles 2005;19:GB4012. Nyborg M, Hoyt PB. Effects of soil acidity and liming on mineralization of soil nitrogen. Can J Soil Sci 1978;58:331–8. Pedersen CA, Knudsen L, Schnug E. Sulphur fertilization. In: Schnug E, editor. Sulphur in agroecosystems. Dordrecht: Kluwer; 1998. p. 115–34. Pitter P. Hydrochemistry. Prague: VŠCHT; 2009 [in Czech]. Posch M, Aherne J, Forsius M, Rask M. Past, present, and future exceedance of critical loads of acidity for surface waters in Finland. Environ Sci Technol 2012;46:4507–14. Prechtel A, Armbruster M, Matzner E. Modelling sulphate stream concentrations in the Black Forest catchments Schluchsee and Villingen. Hydrol Earth Syst Sci 2003;7(4): 552–60. Procházková L. On the application of the chloranilic acid method for the colorimetric determination of sulphate in water (Über die Anwendung des Chloranilsäureverfahrens zur colorimetrischen Sulfatbestimmung im Wasser). Fresen Z Anal Chem 1961;182:103–7. [in German]. Psenner R, Catalan J. Chemical composition of lakes in crystalline basins. In: Margalef R, editor. Limnology now: a paradigm of planetary problems. Amsterdam: Elsevier; 1994. p. 255–314. Reuss JO, Johnson DW. Acid deposition, soil and waters. Ecological studies vol. 50. New York: Springer; 1986. Schnug E. Sulphur nutritional status of European crops and consequences for agriculture. Sulphur Agric 1991;15:7–12. Schoenau JJ, Germida JJ. Sulphur cycling in upland agricultural systems. In: Howarth RW, Stewart JWB, Ivanov MV, editors. Sulphur cycling on the continents, SCOPE 48. Scientific Committee on Problems of the Environment (SCOPE); 1992 [http://www. scopenvironment.org/downloadpubs/scope48/contents.html]. Singh B, Hedley MJ, Saggar S, Francis GS. Chemical fractionation to characterize changes in sulphur and carbon in soil caused by management. Eur J Soil Sci 2004;55:79–90. Sjöström E. Wood chemistry: fundamentals and applications. Academic Press; 1981. Smith SJ, Van Aardenne J, Klimont Z, Andres RJ, Volke A, Delgado Arias S. Anthropogenic sulfur dioxide emissions: 1850–2005. Atmos Chem Phys 2011;11:1101–16. Stumm W. Chemistry of the solid–water interface: processes at the mineral–water and particle–water interface in natural systems. New York: Wiley; 1992. Tisdale SL, Nelson W, Beaton JD. Soil fertility and fertilizers. New York: Macmillan; 1984. Vaněk V, Penk J. Liming of agricultural soils, Ministerstvo zemědělství České Republiky, Prague; 1991 [in Czech]. Zelený F, Zelená E. Changes in sulfur need for oilseed rape in the Czech Republic. Phyton 2002;42(3):265–70. Zhao FJ, McGrath SP, Blake-Kalff MMA, Link A, Tucker M. Crop responses to sulphur fertilisation in Europe, proceedings no 504. York, United Kingdom: International Fertiliser Society; 2002.