concentrations and fluxes of dissolved organic carbon in uk topsoils

S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 4 6 0 – 4 7 0

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te / sc i to tenv

Concentrations and fluxes of dissolved organic carbon inUK topsoils

S. Buckinghama,b, E. Tippinga,⁎, J. Hamilton-Taylorb

aCentre for Ecology and Hydrology, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster, LA1 4AP, United KingdombDepartment of Environmental Sciences, Lancaster University, Lancaster, LA1 4YQ, United Kingdom

A R T I C L E D A T A

⁎ Corresponding author. Tel.: +44 0 1524 59586E-mail address: [email protected] (E. Tipping).

0048-9697/$ – see front matter © 2008 Elsevidoi:10.1016/j.scitotenv.2008.08.020

A B S T R A C T

Article history:Received 10 April 2008Received in revised form7August 2008Accepted 13 August 2008Available online 24 September 2008

Dissolved organic carbon (DOC) concentrations in soil water samples collected from depthsof 5 to 20 cm at 10 moorland and 11 forest sites during the period 2000–2006 were obtainedfrom new measurements and from the monitoring programmes of the UK EnvironmentalChange Network and the International Cooperative Programme (ICP) on Forests. Data on soilproperties and vegetation type were also assembled. Considering data from Prenart tensioncollectors, which were used at nearly all the sites, mean annual concentrations ranged from1.3 to 97.5 g m−3 with means of 19.5 (standard deviation 15.2) and 27.6 (SD 23.3) g m−3 formoorland and forest sites respectively. Interannual mean DOC concentration at anindividual site varied by only 1.5-fold, averaged over all sites with at least three years'data. Concentrations during summer months (April to September) were on average 17%greater than those in winter (October to March). If data from two sites (the single peatlandand an unusual forest site) were ignored, DOC concentrations were strongly inverselyrelated to water flux, estimated from rainfall and evaporation data. Fluxes of DOC,calculated by combining concentration with water flux, ranged from 2.2 to 71.9 gC m−2 yr−1

over all sites and years, with overall means of 19.2 (SD 13.6) and 12.2 (SD 13.9) gCm−2 yr−1 forthe moorland and forest sites respectively. However, if the two exceptional sites wereomitted, the overall mean was 9.1 gC m−2 yr−1 with a standard deviation of only 4.9 gC m−2

yr−1. Annual DOC flux was strongly dependent upon annual water flux, varying by 3.5-foldbetween years when averaged over all sites. On average, 75.5% of the DOC was exportedduring the winter period (October to March).

© 2008 Elsevier B.V. All rights reserved.

Keywords:Dissolved organic carbonConcentrationFluxSoilWater flux

1. Introduction

Dissolved organic matter (DOM), usually measured as dis-solved organic carbon (DOC) which comprises about half of itsmass, is a product of organic matter decomposition in soils,and consists of a wide range of molecules, ranging fromsimple acids and sugars to complex humic substances (Moore,1998). Transport of DOM from the surface soil is a significantcomponent of the carbon cycle, contributing to C storage indeeper horizons (e.g. Buurman and Jongmans, 2005), and beingtransported to surface waters where it has a number of

6.

er B.V. All rights reserved

chemical and biological functions (Thacker et al., 2005). Recentreports have highlighted widespread increases in DOC con-centrations in surface waters of the UK (Worrall et al., 2004;Evans et al., 2005) and more widely (Monteith et al., 2007).Because of its binding properties, DOM also plays a role in thetransport, geochemical reactivity and bioavailability of metalsand organic compounds. For example, metal transport fromtopsoils by DOM is a significant flux in the calculation ofCritical Loads of Heavy Metals (Hall et al., 2006). Dissolvedorganic matter also contributes significantly to soil nutrientcycling (Corey et al., 2004; Park and Matzner, 2003).

.

mailto:[email protected]

http://dx.doi.org/10.1016/j.scitotenv.2008.08.020

461S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 4 6 0 – 4 7 0

Most work on the concentrations and fluxes of soildissolved organic carbon has been performed on forestecosystems (see Michalzik et al., 2000 for a review). Less isknown about other kinds of natural or “semi-natural” plant–soil systems, including the moorlands that are prevalent inthe United Kingdom. Although peatlands have receivedattention, most studies refer to net streamwater exportsrather than internal soil fluxes (e.g. Moore et al., 2003; Pastoret al., 2003; Worrall et al., 2003; Moore and Clarkson, 2007).However, recent years have seen increased DOC monitoringactivity in UK soils. The purpose of the present work was tocollate and extend the available data, and use them to explorespatial and temporal variations in both concentration andflux, and possible relationships to plant and soil type.

The DOC and other data come from our own work(Buckingham et al., 2008), and from the monitoring pro-grammes of the United Kingdom Environmental ChangeNetwork (ECN) and the International Cooperative Programme(ICP) United Kingdom Level II Intensive Forest MonitoringNetwork, operated by Forest Research. The data set covers 21sites, with approximately equal numbers of moorlands andforests, over the period 2000–2006. The data are obtained fromthe analysis of soil water samples obtained with collectorslocated at depths of up to 20 cm below the soil surface. Thisapproximately covers the rooting zone, where the majority ofDOM generation is thought to occur. Thus they provide anoverall picture of DOC transport from the most active part ofthe soil with respect to carbon cycling. However, in most casesthe collectors are installed at specific depths rather than at soilhorizon boundaries, and this limits detailed interpretation ofthe results. Nonetheless, the results provide valuable informa-

Table 1 – Site descriptions

Site name Abbr. Location Soil type

MoorlandsCowdale Slack C Cumbria Typical brown earthDoe House Gill D Cumbria Acid rankerDrayton Dr Warwickshire Heavy clay, calcareousGlensaugh G Aberdeenshire Humus iron podzolMask Hill M Cumbria Cambic stagnohumic gleyMoor House 1 Mo1 Cumbria PeatMoor House 2 Mo2 Cumbria PeatRavenstonedale R Cumbria Ferric stagnopodzolSnowdon S Gwynedd Brown podzol, gleysSourhope So Roxburghshire Peaty gley podzol

ForestsCoalburn Co Northumberland Cambic stagnohumic gleyGrizedale Gr Cumbria Typical brown podzolLlynn Brianne L Carmarthen Cambic stagnohumic gleyMeathop Wood 1 Me1 Cumbria Acidic brown earthMeathop Wood 2 Me2 Cumbria Brown earthRannoch Ra Perthshire Humo ferric gley podzolSavernake Sa Wiltshire Argillic pelosolSherwood Sh Nottinghamshire Brown podzolic soilThetford T Norfolk Brown calcareous sandsTummel Tu Perthshire Ferric podzolWytham W Oxfordshire Heavy clay soil

a Mean annual temperature °C.b Value estimated from regression of C:N on % C for the other sites.

tion about DOC concentrations and fluxes, which can then belinked to other processes, for example contaminant transport.

Most of the DOC concentration data refer to samplescollected with tension lysimeters, although for some of thesites comparisons with other sampler types are made (Buck-ingham et al., 2008). Fluxes of DOC were estimated bycombining the concentration data with simple estimates ofwater flux, mostly made from rainfall and evaporation, but insome cases using stream discharge as a surrogate for the soilwater flux (McDowell and Likens, 1988). As well as compilinginformation on DOC, we collated complementary informationon soils and vegetation. The field sites (Table 1, Fig. 1) cover arange of soil and vegetation types located at various altitudes.

2. Methods

We monitored topsoil water at five moorland and two forestsites in north–west England, using tension-free collectors(TFC), and two types of tension collector, Prenart collectors(PC) and Rhizon collectors (RC). Three replicate samplers weredeployed at each site, and monitored fortnightly for one year,April 2005 to April 2006. Details are given by Buckingham et al.(2008). Samples were passed through glass fibre filters (What-man GF/F, nominal pore size 0.7 µm), and DOC concentrationsdetermined using a TOC-VCPH Total Organic CarbonAnalyzer.Six Prenart samplers are used at each ECN site, withcollections being made weekly. Triplicate soil water samplesare collected with Prenart samplers every twoweeks at the ICPsites and filtered using 0.45 µm membrane filters. Details ofthe ECN sampling and analytical procedures are given by

Vegetation Altitudem

% C C:N pH MATa Datasource

Grassland 350 3.0 10.9 4.8 7.0 This studyGrassland 357 23.9 13.6 4.7 6.9 This studyGrazed pasture 60 6.4 12.8 6.9 9.9 ECNHeather 312 12.8 11.5 4.0 6.2 ECNHeather 301 3.5 12.0 5.0 7.3 This studyHeather 577 49.1 43.3 4.0 5.5 This studyHeather 577 49.1 43.3 4.0 5.5 ECNGrassland 406 31.4 18.4 4.7 6.6 This studyAcidic grassland 692 7.4 14.8 4.6 5.3 ECNCoarse grassland 400 7.6 13.6 4.2 6.2 ECN

Sitka spruce (1974) 300 43.2 25.7 4.2 6.8 ICPOak (1920) 90 33.3 30.5 4.1 8.7 ICPSitka spruce (1973) 460 41.1 18.4 3.9 7.3 ICPMixed deciduous 45 4.2 12.7 6.4 9.0 This studyMixed deciduous 50 4.7 11.3 6.7 8.9 This studyScots pine (1965) 460 50.8 39.0 4.9 5.3 ICPOak (1950) 100 9.4 23.9 4.9 9.6 ICPScots pine (1952) 250 11.0 26.7 4.9 8.1 ICPScots pine (1967) 30 3.0 6.1 5.0 9.6 ICPSitka spruce (1969) 375 7.7 ~13b 5.5 5.8 ICPAsh deciduous 113 6.1 12.2 5.7 9.5 ECN

Fig. 1 –Map showing site positions. Open symbols represent moorlands, closed symbols forests.

462 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 4 6 0 – 4 7 0

Sykes and Lane (1996), those for the ICP sites are in manualsavailable at http://www.icp-forests.org/Manual.htm.

For the sites that we studied ourselves, soils were sampledto the depths (10–20 cm) at which the tension-free collectorswere located. The soil samples were thoroughly mixed, sieved

(4 mm mesh) and their water contents determined by dryingat 110 °C. Carbon and nitrogen contents were determinedwitha Universal CHNS-O Vario EL elemental analyzer, and pH wasmeasured with a glass electrode on suspensions prepared bymixing the wet soil with deionized water (1:2 ratio by weight).

http://www.icp-forests.org/Manual.htm

Fig. 2 –DOC flux results from methods B (closed squares) andC (open squares) plotted against results from method A. Thedata refer to all three collector types. The 1:1 line is shown.


Similar information was obtained from sampling and analysisat the ECN and ICP sites. For the ICP forest sites, soil variablesare reported by horizon, but the soil water collectors arelocated at a standard depth of 10 cm. Therefore we calculatedthe compositions that would have been obtained by mixingthe top 10 cm of the soils. To perform this calculation weestimated bulk density from soil organic matter content usingthe equation of Tipping et al. (2003).

For our own sites and the ICP sites, we obtained dailyrainfall data from the nearest meteorological monitoringstation of the British Atmospheric Data Centre (BADC). Forthe ECN sites, rainfall data specific to the site were used.Monthly actual evaporation data were obtained from Meteor-ological Office MORECS database, which provides values for40×40 km grids throughout Great Britain, and used to obtain adaily average for each month, and yearly totals.

In flux calculation method A, the annual average DOCconcentration was multiplied by the annual water flux (rain-fall minus actual evaporation), to obtain the flux for anindividual yearly period. In method B the annual flux wasobtained by adding the weekly or fortnightly fluxes, obtainedbymultiplying the DOC concentration for the collection periodby the daily water flux (daily rainfall minus average dailyevaporation). This method was only applied when the DOCconcentration data covered at least 75% of the time. For someof the ECN data, missing values were estimated by linearinterpolation between preceding and following data. Fluxcalculation method C utilised stream discharge data toestimate the water flux through the soil (McDowell andLikens, 1988). This was applied to Glensaugh, Moor House,Snowdon and Sourhope using 15 min discharge data andcatchment areas from ECN, and to Ravenstonedale Commonusing data provided by The University of Newcastle Depart-ment of Civil Engineering. Use of this method carries theassumption that each catchment is effectively watertight, i.e.there are no losses to deep groundwater.

We report DOC concentrations in g m− 3 and fluxes ing m−2 yr− 1 for consistency. The g m−3 unit is equivalent tothe conventional mg l−3 more commonly used for DOCconcentrations. The numerical value of the flux expressed ingC m−2 yr−1 can be multiplied by 10 to obtain kgC ha−1 yr−1.

3. Results and discussion

3.1. Estimation methods

Fluxes of DOCwere calculated for individual one-year periods,although these were not always calendar years. Theoretically,calculation method B should be superior to method A becauseit takes better account of temporal variation in both DOCconcentration andwater flux. However, the twomethods gaveresults in good agreement (Fig. 2), and therefore fluxesestimated by the simpler, more widely-applicable method A(multiplication of the annual mean DOC concentration by theannual water flux) can be considered reliable. At the five siteswhere it could be applied, method C also gave results inagreement with the other two methods (Fig. 2). We note thatthe estimation of fluxes by these methods entails theassumption that soil water storage is equal at the start and

end of the period for which a flux is calculated. Theoretically,this could lead to amaximum error of 25% for a 10 cm depth ofsoil of 50% porosity with water flux of 200 mm, but this wouldfall to 5% for water flux of 1000 mm.

With regard to collector type, we have already reported thattension-free samplers tend to collect water with higher DOCconcentrations (Buckingham et al., 2008), and this is evidentfrom the data in Tables 2 and 3. On average the tension-freeconcentrations are 1.27 times those obtained with the Prenartsamplers, and this is also reflected in flux estimates. However,given that the ECN and ICP monitoring programmes areconducted exclusively with Prenart tension samplers, com-parisons of concentrations and fluxes are most usefullyperformed using data obtained by this common method.Therefore we examined spatial and temporal variations onlyfor Prenart-derived data, andwith fluxes estimated bymethodA. However if the results presented here were to be used forother purposes, e.g. estimating the contribution of DOC to co-transport of other components such as heavy metals, slightlyhigher values might be appropriate.

With regard to collector depths, these have been chosen tobe standard in the ECN and ICP monitoring schemes,whereas in our own work we took horizon properties intoaccount in some cases, and used different depths fordifferent collectors in order to take into account likelyupward movement of water when using tension collectors.Therefore if there is a dependence of DOC concentration withdepth in topsoils, variations in the assembled data will reflectthis. The only case where the same collector was used atdifferent depths at the same location (although not the sameplot, nor over the same time period) is in the peat at MoorHouse, where the Prenart collectors that we placed at 5 cmgave an average concentration about 50% higher than theECN collectors, placed at 10 cm.

3.2. Overall means and ranges of concentrations and fluxes

The mean annual DOC concentration at the forest sites issomewhat greater than the value for the moorland sites, butthe reverse is true for the mean fluxes (Table 4). Moreover, the

Table 2 – Data for moorland sites

Site Method Depth n Period P E [DOC] DOC flux

Min Max Mean Wt. mean A B C

Cowdale Slack TFC 15 317 2005–6 1466 490 4.1 23.2 12.9 (5.1) 11.0 17.9 (10.7) 12.5 (8.6)PC 7.5 263 2005–6 1466 490 4.1 10.0 6.0 (1.4) 6.3 6.0 (1.4) 6.1RC 7.5 297 2005–6 1466 490 3.5 6.2 4.8 (0.7) 4.8 4.7 4.7

Doe House Gill TFC 20 322 2005–6 2251 535 1.0 4.2 2.0 (0.9) 1.8 3.5 (1.6) 3.1 (0.9)PC 7.8 323 2005–6 2251 535 0.7 2.4 1.3 (0.5) 1.2 2.2 (0.1) 2.1 (0.2)RC 9 253 2005–6 2251 535 1.6 8.8 3.2 (1.4) 3.0 5.6 5.1

Drayton PC 15 364 2001 700 513 20.0 58.0 45.3 (16.5) 32.3 8.5 (2.8) 6.0 (1.9)PC 15 364 2002 712 531 36.7 96.0 65.1 (8.4) 59.5 11.8 (3.3) 10.8 (2.9)

Glensaugh PC 10 361 2001 912 492 12.0 27.4 18.5 (4.2) 17.8 7.8 7.5 10.3PC 10 362 2002 1490 478 13.2 46.4 23.0 (8.2) 8.2 22.2 23.3 19.3

Mask Hill TFC 15 310 2005–6 1466 490 7.0 15.8 11.6 (2.3) 10.9 11.4 (2.7) 10.6 (2.2)PC 7.5 325 2005–6 1466 490 3.4 32.0 9.2 (7.1) 7.3 9.5 7.1RC 7.5 339 2005–6 1466 490 4.9 16.3 7.3 (2.4) 6.4 7.0 6.2

Moor House 1 TFC 10 254 2005–6 1586 515 18.3 46.1 34.3 (7.9) 31.6 37.2 (6.0) 33.3 (6.4)PC 5 80 2005–6 1586 515 12.8 35.6 30.0 (3.7) 30.0 33.3 32.6RC 5 271 2005–6 1586 515 18.4 36.3 25.7 (4.6) 25.1 27.8 27.3

Moor House 2 PC 10 365 2000 2755 531 14.7 34.5 23.2 (6.0) 10.4 51.6 48.2 47.1PC 10 365 2001 1731 511 17.1 38.6 23.0 (6.0) 18.8 28.1 28.4PC 10 365 2002 2078 490 15.6 38.1 23.4 (6.0) 14.7 37.2 35.1 38.7PC 10 365 2003 1561 547 18.0 32.3 22.3 (3.7) 22.0 22.6 20.8PC 10 365 2004 2318 498 15.6 34.2 23.9 (5.1) 13.1 43.5 42.7PC 10 365 2005 1814 516 14.6 39.7 25.7 (8.1) 19.8 33.3 31.7

Ravenstonedale TFC 15 353 2005–6 1714 491 16.9 41.7 25.1 (6.2) 22.8 30.8 (2.7) 27.9 (2.5) 38.5PC 7.5 297 2005–6 1714 491 11.8 23.3 17.6 (4.0) 11.8 21.6 14.5 25.2RC 7.2 332 2005–6 1714 491 8.4 27.5 17.1 (4.7) 13.9 21.0 17.0 14.5

Snowdon PC 10 361 2000 4736 532 1.3 5.4 3.1 (1.1) 2.7 13.2 (1.7) 11.2 (1.9)PC 10 364 2001 3234 509 2.2 4.2 3.0 (0.5) 2.9 8.2 (0.9) 7.8 (1.0)PC 10 358 2002 3324 490 1.2 12.3 3.3 (2.0) 2.9 9.5 (1.2) 8.1 (0.8) 10.1PC 10 357 2003 1828 558 2.2 11.2 4.2 (2.3) 3.8 5.3 (1.0) 4.9 (0.9)

Sourhope PC 10 364 2001 869 470 18.6 18.6 18.6 (0.0) 18.6 7.4 (0.9) 7.4 (0.9) 16.8PC 10 363 2002 1002 433 10.4 35.0 20.3 (6.0) 17.1 11.5 (1.0) 9.7 (2.4) 14.9

Key: TFC tension-free collector, PC Prenart (tension) collector, RC Rhizon (tension) collector, depths are in cm, n = number of samples, P =precipitation, E = evaporation, [DOC] concentration of dissolved organic carbon in g m−3, DOC flux is in gC m−2 yr−1. Uncertainties in [DOC],shown in parentheses, are standard deviations calculated from all measurements, those in DOC flux are from comparisons of individualcollectors at the same site.


ranges of values overlap considerably and there are too fewdata to draw a general conclusion about any systematicdifference. The highest individual concentrations (i.e. for aone- or two-week periods) were c. 100 g m−3, at the two driestsites (DraytonandThetford). Table 4 also shows summary dataomitting results for two “outlier” sites, namelyMoorHouseandLoch Tummel. As will be seen, the results from these do not fitwith those from the other sites. To our knowledge this is thefirst survey to cover a range of soil and vegetation types,although the geographical locations are limited to sites in theUK. Most other studies have been conducted in forestedecosystems, especially coniferous forests. Michalzik et al.(2000) reviewed results for forest soils, and reported DOCconcentrations for the Oa layer to range from 20 to 90 g m−3,and fluxes from 10 to 40 g m−2 yr−1, similar to the values inTables 3 and 4. Neff and Asner (2001) reviewed further data,and quoted a range of 2 to 84 g m−2 yr−1 for DOC fluxes in soilsup to 20 cmdeep, under a variety of vegetation cover, althoughmostly eucalyptus, coniferous and deciduous forest.

To put the DOC fluxes into context within the carbon cycle,the range of observed topsoil DOC fluxes is substantially lowerthan carbon inputs in plant litter, which are of the order of

several hundred gCm−2 yr−1. As exampleswe canmention thedata reported by Perkins et al. (1978), which correspond tolitter inputs between 200 and 700 gCm−2 yr−1 for grasslands ofdifferent nutrient status in NorthWales, an estimate of above-ground litter input of ca. 350 gCm−2 yr−1 for heather moorlandatMoor House (Smith and Forrest, 1978), and litterfall values inthe range 200 to 700 gCm−2 yr−1 for the ICP Level II sites used inthe present work (http://www.forestresearch.gov.uk/website/forestresearch.nsf). Thus the DOC fluxes given in Table 4represent no more than 20%, and more typically c. 5%, of thelitter C input, most of which leaves the system as CO2.

In their review of carbon loss from UK soil, Dawson andSmith (2007) reported DOC fluxes in the range 0.8 to 26 g m−2

yr−1 for 24 upland streams and rivers, with a mean value of c.10 g m−2 yr−1, quite similar to the soil fluxes in Table 4.However, many of the locations in the compilation of Dawsonand Smith contain substantial areas of peatland, and so thequoted surface water fluxes are probably at the high end of therange. More generally, streamDOC fluxes are somewhat lowerthan the corresponding topsoil values, due to the sorption andmineralization of DOC during transport through deeper soil(Kalbitz et al., 2000).

http://www.forestresearch.gov.uk/website/forestresearch.nsf

http://www.forestresearch.gov.uk/website/forestresearch.nsf

Table 3 – Data for forest sites

Site Method Depth n Period P E [DOC] DOC flux

Min Max Mean Wt. mean A B

Coalburn PC 10 225 2002 1379 475 6.0 9.7 7.8 (0.9) 7.8 7.1 7.0PC 10 295 2003 925 512 5.9 7.6 6.7 (0.4) 6.8 2.8 2.8PC 10 365 2004 1518 499 6.4 15.3 9.9 (8.7) 10.0 10.1 10.2PC 10 309 2005 1223 522 5.9 16.0 10.2 (2.2) 10.9 7.1 7.6

Grizedale PC 10 351 2003 1478 494 4.7 11.2 8.0 (2.1) 7.6 7.9 7.5PC 10 365 2004 2163 501 3.0 15.5 7.9 (3.2) 8.0 13.1 13.3PC 10 337 2005 1370 497 2.3 18.3 10.9 (4.0) 8.9 9.5 7.8

Llyn Brianne PC 10 365 2002 1255 470 7.1 23.3 14.8 (5.0) 13.9 11.7 10.9PC 10 309 2003 768 509 11.1 27.0 18.6 (4.0) 20.0 1.3 5.2PC 10 351 2004 1168 487 10.7 30.5 22.9 (4.2) 24.6 15.6 16.7PC 10 365 2005 1190 504 12.3 41.4 23.6 (7.6) 26.2 16.2 18.0

Meathop Wood 1 TFC 15 303 2005–6 1385 562 9.7 26.7 15.2 (3.9) 14.3 12.5 11.8Meathop Wood 2 TFC 15 335 2005–6 1385 562 5.4 29.8 21.1 (6.8) 23.0 17.3 18.9Rannoch PC 10 365 2002 1284 432 3.3 7.4 5.6 (0.9) 5.5 4.8 4.7

PC 10 281 2003 657 466 2.2 5.4 28.9 (2.6) 28.9 2.6 2.2Savernake PC 10 323 2002 1153 573 12.0 31.3 21.1 (18.6) 19.4 12.2 11.2

PC 10 281 2003 689 485 14.6 30.6 19.3 (18.5) 18.6 3.4 3.8PC 10 117 2004 846 589 8.0 36.2 28.8 (8.0) 7.4PC 10 236 2005 643 553 23.9 37.5 28.9 (2.7) 24.7 2.6 2.2

Sherwood PC 10 365 2002 1311 580 10.7 29.4 19.0 (5.6) 18.4 13.9 13.4PC 10 365 2003 859 627 10.6 34.0 20.3 (7.8) 25.6 4.7 3.8PC 10 365 2004 1164 614 14.5 41.3 24.0 (7.3) 26.2 13.5 14.9

Thetford PC 10 99 2003 613 542 69.0 92.6 81.5 (10.9) 5.8PC 10 46 2004 992 638 27.6 110.7 71.6 (33.7) 16.3PC 10 67 2005 672 584 92.4 103.5 97.5 (4.5) 8.6

Tummel PC 10 211 2002 1728 408 12.5 50.5 31.1 (6.9) 33.7 41.1 44.6PC 10 167 2003 831 465 20.9 27.4 24.4 (1.8) 8.9PC 10 109 2004 1519 451 41.6 75.9 55.7 (11.1) 59.5PC 10 141 2005 1884 473 21.1 82.0 50.9 (22.1) 71.9

Wytham PC 16.5 263 2000 730 444 19.5 23.5 21.5 (0.5) 21.7 6.2 6.2

Key: see Table 2. Results in italics refer to periods during which fewer than 50% of the samplers provided water samples.


3.3. Interannual variations

Fig. 3 shows annual mean DOC concentrations (not flow-weighted) and fluxes for the nine sites providing three or moreyears' data over the period 2000 to 2005. The concentrations at agiven site vary rather little fromyear to year, the greatest relativerange (2.3-fold) being for Tummel, while the average range isonly 1.5-fold. All the sites show a slight increase in DOCconcentration with time, the significance of which was tested

Table 4 – Summary of DOC concentration (g m−3) and flux(g m−2 yr−1) data obtained using Prenart samplers

All data Omitting Mo1, Mo2,Tu

Moorland Forest Moorland Forest

ConcentrationsRange 1.3–65.1 5.6–97.5 1.3–65.1 5.6–97.5Mean 19.5 27.6 17.0 25.4Standard deviation 15.2 23.3 18.3 23.9

FluxesRange 2.2–51.6 1.3–71.9 2.2–22.2 1.3–16.3Mean 19.2 12.2 11.1 8.5Standard deviation 13.6 13.9 5.7 4.7

The last two columns refer to data from all sites except Moor Houseand Tummel.

by linear regression of the weekly or fortnightly data, for the 7sites with sampling success rates N50%. In five cases, Coalburn,Grizedale, Llyn Brianne, Savernake and Sherwood (all forests),the increase in DOC concentration was highly significant(pb0.001). The time period is too short for these yet to beconsideredsustainedchanges, but the increases are reminiscentof those in surface waters seen over much longer time periods(see Introduction), andcontinuedmonitoring is clearlydesirable.

The annual DOC fluxes vary much more from year to yearthan the concentrations (Fig. 3), the greatest range being 8.1-fold at Tummel and the average range 3.5 fold. This reflectsthe strong dependence of flux on water flux, which explainswhy there are no temporal flux trends, and why the lowestannual flux occurred at 8 of the 9 sites of Fig. 3 in the especiallydry year of 2003 (Tables 2 and 3).

3.4. Seasonality

Fig. 4 illustrates annual seasonality of DOC concentration andflux for four representative sites. Concentrations are higher inthe summer period, whereas fluxes are greater during wintermonths. Fig. 5 summarises seasonality for all the sites, interms of winter (October to March) and summer (April toSeptember) periods. For 13 of the 16 sites, the ratio of summerDOC concentration to the annual mean value exceeds one, theoverall average ratio being 1.08. Thus the concentrationsduring the summer period are typically 17% greater than those

Fig. 3 –Mean annual DOC concentrations and fluxes at sites with three or more years' data.


Fig. 4 –Seasonal variations in DOC concentration and flux, and water flux, for four representative sites; monthly valuesaveraged over all available years.


in the winter. At all sites the DOC flux in winter accounted forsubstantially more than half the total flux, the average being75.5%. In the extreme cases (Glensaugh, Rannoch, LlynnBrianne) 90% or more of the annual flux occurs in winter. Ina modelling study of one of the sites, Doe House Gill, Tippinget al. (2007) concluded that the higher summer DOC concen-trations were due to evaporative concentration rather thanhigher rates of DOC production.

3.5. Prediction of DOC concentrations and fluxes

We attempted to correlate DOC concentrations and fluxes withsite variables including soil C content, C:N ratio, soil pH, meanannual temperature and altitude (Table 1), but no strongrelationships emerged, either singly or in multiple regression.In particular, our data did not show any convincing dependence

of flux upon soil C:N ratio, which has been shown to apply tostream DOC fluxes by Aitkenhead and McDowell (2000).

Plots of concentration and flux against water flux (Fig. 6)are helpful in understanding similarities and differencesamong sites, although one must be aware that the plot ofDOC flux against water flux involves dependent axes andmust not be overinterpreted. The concentration-water fluxplot suggests an inverse relationship, but two sites, MoorHouse, a peatland, and the forest at Tummel, deviateappreciably. If these sites are ignored the shape of the plot isthat expected for a common flux (9.1 gC m−2 yr−1) at all sites,with DOC concentration varying simply due to dilution. Thetwo “outliers” also strongly affect the general relationshipbetween flux and water flux; without them the DOC fluxes fallwithin a relatively narrow range, much of the variation beingattributable to interannual variation inwater flux (Section 3.3).

Fig. 5 –Seasonality in DOC concentration and flux at all sites for which more than 50% of the soil water collections weresuccessful. The results refer to all collector types and are ordered in terms of average concentration or flux. Closed squares referto Prenart collectors, open squares to tensionless lysimeters and open circles to Rhizon collectors. The closed triangles in theflux plot indicate negative calculated summer fluxes, due to an excess of evaporation over precipitation.


If it is accepted that the long-term mean DOC flux from alltopsoils other than peatlands can be approximated by themean value of 9.1 (standard deviation 4.9) gC m−2 yr−1 derivedfrom the 19 sites not including Tummel, then the long-term

Fig. 6 –DOC concentrations and fluxes plotted against waterflux. The lines are drawn for a DOC flux of 9.1 gC m−2 yr−1.

average DOC concentration for sites without DOC measure-ments can possibly be estimated by dividing this flux by thelong-term annual water flux of the site of interest. Becausevariations in flux between years depend so strongly on waterflux (Fig. 3), the flux in a particular year might be estimated asthe mean flux (9.1 gC m−2 yr−1) multiplied by the water fluxdivided by the long-term average water flux. Given the widerange of DOC concentration but the relatively narrow range of“long-term” average fluxes, it is preferable to use the averageDOC flux to characterise thesoil transportofDOCandassociatedcomponents, such as nutrients, metals and organic contami-nants.Asnotedabove, a slightlyhigher value, say10 gCm−2 yr−1,would be preferred in view of the lower DOC concentrationsthat tend to be obtained with Prenart samplers.

While the value of 10 gC m−2 yr−1 may be a reasonabletypical choice, it should be stressed that it applies to a soilthickness of the order of 15 cm (see above). Certainly, higherfluxes appear to operate at shallower depths, especially inforest soils, where DOC fluxes at the base of the O-horizontend to be higher, of the order of 20 gC m−2 yr−1 or more(Michalzik et al., 2000). This reflects the tendency of the flux todecline with depth, as DOC is sorbed and mineralized.

3.6. “Outlier” sites

Why are the DOC concentrations and fluxes at Moor Houseand Tummel different? In fact the DOC concentrations at MoorHouse are quite typical of blanket bog and wetlands in general(e.g. Glatzel et al., 2003; Moore and Clarkson, 2007). But fromFig. 6 we see that, for the water flux amount at Moor House,the DOC concentration lies well above the illustrative trendline. This difference can be attributed to the plant system in a


peat bog and the continually high moisture content, whichrestricts mineralization and promotes the formation of DOC,although that is not to suggest that the DOC flux is by anymeans the major output of C from such systems. Anotherlikely factor is the lack of mineral sorbents in peats, whichencourages the rapid escape of DOC. It seems likely thatMoor House is quite typical of peats and that a higher char-acteristic DOC flux than the 10 gm−2 yr−1 suggested above, say25 g m−2 yr−1, might be an appropriate typical value for suchsites. In all probability Moor House is not a true outlier, butrather belongs to a different population of soils.

It is much more difficult to explain the high DOCconcentrations and fluxes at Tummel. The soil at this site isrelatively low in carbon, compared to the other forest sites(Table 1) and so it is even more surprising to find DOC fluxesthat are several times those of the other forest values. Thelitterfall at Tummel (338 gCm−2 yr−1) is not exceptionally high,which would appear to rule out the possibility of an unusuallyhigh rate of C input to the soil. Thereforewe cannot yet find anexplanation for the high DOC values. As more data arecollected, the status of Tummel as an “outlier” may becomeclearer.

It is worth noting that the high DOC fluxes at these twosites make the data points conspicuous in Fig. 6, but themarked positive relationship between DOC flux andwater fluxis in fact seen for all the sites (Section 3.3).

3.7. Mechanisms of DOC production and transport intopsoils

Kalbitz et al. (2000) reviewed the numerous factors that governthe production and transport of DOC in soil, which includeplant type, soil C pool sizes, substrate quality, microbialactivity, sorption, pH, and metal reactions with organicmatter. The authors drew attention to the often conflictingconclusions that come from laboratory and field studies,highlighted the importance of hydrology in determining DOCflux, and stressed the need for field research. With regard tothe last point, the present study is one contribution.

If the two “outlier” sites are discounted, the soils studiedhere release DOC at rates that fall within a fairly narrow range,and fluxes averaged over a number of years will be still closer.This behaviour generates the strong inverse relationshipbetween DOC concentration and water flux (Fig. 6). However,if individual sites are considered, the DOC flux is essentiallyproportional to water flux, because the DOC concentrationtends not to vary. Resolution of these two apparently contra-dictory trends would go some way towards understandingDOC concentrations and fluxes in soils.

The DyDOC model (Michalzik et al., 2003; Tipping et al.,2005, 2007) uses the concept of “potential DOC”, i.e. a pool ofmaterial that can be transferred to percolating soil water, inwhich it is measured as actual DOC. The potential DOC isassumed to be generated bymicrobially-driven decompositionprocesses, and to be lost from the soil either as DOC inleachate or after mineralization to CO2. Depending upon theproperties of the soil solids and on the solution chemistry,some ormost of the potential DOCmay be sorbed. Therefore, anumber of factors determine observed DOC concentrationsand fluxes. For example, the dependence of flux upon water

flux at a given site may arise because during dry years lesspotential DOC is generated, less diffusional exchange canoccur, a larger proportion is lost via mineralization, or the soilsorption properties change. The similarity of fluxes amongmost of our study sites (Fig. 6) may be wholly or partlycoincidental, reflecting different combinations of differentprocesses, and should therefore not be taken to mean that thedifferent soils have similar input rates of potential DOC. Ourfailure to find predictive relationships between soil propertiesand either DOC concentrations or fluxes (Section 3.5) suggeststhat disentangling interactions among the controlling pro-cesses is not straightforward.

4. Conclusions

(a) Annual average concentrations of DOC in topsoils at 21moorland and forest sites in the UK, derived from dataobtained with tension samplers between 2000 and 2006,ranged from 1.3 to 97.5 g m−3. The mean concentrationfor the 10 moorland sites was 19.5 (standard deviation15.2) g m−3 while that for the forest sites was 27.6 (SD23.3) g m−3.

(b) For sites with data for three or more years, interannualdifferences in average DOC concentration were small(average 1.5-fold). The concentrations during summer(April to September) were on average 17% greater thanthose in winter (October to March).

(c) If data from two of the sites are ignored, the DOCconcentrations are strongly inversely related to waterflux, approximately consistent with a constant averageDOC flux at all sites. One of the two exceptionalsites is the only peatland of the study, while the otheris a forest on mineral soil with unusually high DOCconcentrations.

(d) Fluxes of DOC could reliably be estimated by combiningthe DOC concentrations with rainfall and evaporationdata, either by summing results for one- or two-weekperiods, or simply as the product of mean annual DOCconcentration and annual water flux.

(e) The DOC fluxes ranged from 2.2 to 71.9 gC m−2 yr−1 overall sites and years, with overall means of 19.2 (SD 13.6)and 12.2 (SD 13.9) gC m−2 yr−1 for the moorland andforest sites respectively. If the two exceptional siteswere omitted, the overall meanwas 9.1 gCm−2 yr−1 witha relatively small standard deviation of 4.9 gC m−2 yr−1.

(f) For all the sites where data for three or more years wereavailable, DOC flux was strongly dependent upon waterflux, varying by 3.5 times between years when averagedover all sites. On average, 75.5% of the DOC wasexported during the winter period (October to March).

Acknowledgements

We thank our CEH colleagues (S. Thacker, C.Woods, D. Howard,C. Wood, P. Chamberlain and F. Sanderson) for help withfieldwork, analysis, data acquisition and statistics, and staff ofthe UK Environmental Change Network (L. Sherrin), ForestResearch (N. Barsoum, E.Vanguelova) and the University of


Newcastle upon Tyne Department of Civil Engineering(G. Parkin, M. Wilkinson, C. Mills) for the provision of data andadvice.Weare grateful toB.Williams,M.Bowstead, theNationalTrust and Natural England for granting access to the samplingsites. J. Quinton (Lancaster University) provided helpful com-ments on the manuscript. This work was supported by a Ph.D.studentship award to S. Buckingham by the UK NaturalEnvironment Research Council.

R E F E R E N C E S

Aitkenhead JA, McDowell WH. Soil C:N ratio as a predictor ofannual riverine DOC flux at local and global scales. Glob.Biogeochem. Cycles 2000;14:127–38.

Buckingham S, Tipping E, Hamilton-Taylor J. Dissolved organiccarbon in soil solutions: a comparison of collection methods.Soil Use Manage. 2008;24:29–36.

Buurman P, Jongmans AJ. Podzolisation and soil organic matterdynamics. Geoderma 2005;125:71–83.

Cory RM, Green SA, Pregitzer KS. Dissolved organic matterconcentration and composition in the forests and streams ofOlympic National Park, WA. Biogeochem 2004;67:269–88.

Dawson JJC, Smith P. Carbon losses from soil and its consequencesfor land-use management. Sci. Tot. Environ. 2007;382:165–90.

Evans CD, Monteith DT, Cooper DM. Long-term increases insurface water dissolved organic carbon: observations, possiblecauses and environmental impacts. Environ. Pollut.2005;137:55–71.

Glatzel S, Kalbitz K, Dalva M, Moore T. Dissolved organic matterproperties and their relationship to carbon dioxide efflux fromrestored peat bogs. Geoderma 2003;113:397–411.

Hall JR, Ashmore MR, Fawehinmi J, Jordan C, Lofts S, Shotbolt L,Spurgeon DJ, Svendsen C, Tipping E. Developing a critical loadapproach for national risk assessments of atmospheric metaldeposition. Environ. Tox. Chem. 2006;25:883–90.

Kalbitz K, Solinger S, Park JH, Michalzik B, Matzner E. Controls onthe dynamics of dissolved organic matter in soils: a review. SoilSci. 2000;165:277–304.

McDowell WH, Likens GE. Origin, composition, and flux ofdissolved organic carbon in the Hubbard Brook valley. Ecol.Monogr. 1988;58:177–95.

Michalzik B, Kalbitz K, Park J-H, Solinger S, Matzner E. Fluxes andconcentrations of dissolved organic carbon and nitrogen — asynthesis for temperate forests. Biogeochem 2000;52:1–33.

Michalzik B, Tipping E, Mulder J, Lancho JFG, Matzner E, Bryant CL,Clarke N, Lofts S, Esteban MA. Modelling the production andtransport of dissolved organic carbon in forest soils. Biogeo-chem 2003;66:241–64.

Monteith DT, Stoddard JL, Evans CD, de Wit HA, Forsius M,Høgåsen T,Wilander A, Skjelkvåle BL, Jeffries DS, Vuorenmaa J,

Keller B, Kopácek J, Vesely J. Dissolved organic carbon trendsresulting from changes in atmospheric deposition chemistry.Nature 2007;450:537–40.

Moore TR. Dissolved organic carbon: sources, sinks, and fluxes androle in the soil carbon cycle. In: Lal R, Kimble KM, Follett RF,Stewart BA, editors. Soil Processes and the Carbon Cycle. CRCPress LLC; 1998. p. 281–92.

Moore TR, Clarkson BR. Dissolved organic carbon in New Zealandpeatlands. NZ. J. Mar. Freshwat. Res. 2007;41:137–41.

Moore TR, Matos L, Roulet NT. Dynamics and chemistry ofdissolved organic carbon in Precambrian catchments and animpounded wetland. Can. J. Fish. Aquat. Sci. 2003;60:612–23.

Neff JC, AsnerGP. Dissolvedorganic carbon in terrestrial ecosystems:synthesis and a model. Ecosystems 2001;4:29–48.

Park J-H, Matzner E. Controls on the release of dissolved organiccarbon and nitrogen from a deciduous forest floor investigatedby manipulations of aboveground litter inputs and water flux.Biogeochem 2003;66:265–86.

Pastor J, Solin J, Bridgham SD, Updegraff K, Harth C,Weishampel P,Dewey B. Global warming and the export of dissolved organiccarbon from boreal peatlands. Oikos 2003;100:380–6.

Perkins DF, Jones V, Millar RO, Neep P. Primary production,mineral nutrients and litter decomposition in the grasslandecosystem. In: Heal OW, Perkins DF, editors. ProductionEcology of British Moors and Montane Grasslands. Berlin:Springer; 1978. p. 304–31.

Smith RAH, Forrest GI. Field estimates of primary production. In:HealOW, PerkinsDF, editors. Production Ecologyof BritishMoorsand Montane Grasslands. Berlin: Springer; 1978. p. 17–37.

Sykes JM, Lane AMJ, editors. The United Kingdom EnvironmentalChange Network: Protocols for Standard Measurements atTerrestrial Sites. London: Stationery Office; 1996.

Thacker S, Tipping E, Baker A, Gondar D. Development andapplication of functional assays for freshwater dissolvedorganic matter. Water Res. 2005;39:4559–73.

Tipping E, Rieuwerts J, Pan G, Ashmore MR, Lofts S, Hill MTR,FaragoME, Thornton I. The solid-solution partitioning of heavymetals (Cu, Zn, Cd, Pb) in upland soils of England and Wales.Environ. Pollut. 2003;125:213–25.

Tipping E, Fröberg M, Berggren D, Mulder J, Bergkvist B. DOCleaching from a coniferous forest floor: modelling a manip-ulation experiment. J. Plant Nutr. Soil Sci. 2005;168:316–24.

Tipping E, Smith EJ, Bryant CL, Adamson JK. The organic carbondynamics of a moorland catchment in N.W England. Biogeo-chem 2007;84:171–89.

Worrall F, Reed M, Warburton J, Burt T. Carbon budget for a Britishupland peat catchment. Sci. Tot. Environ. 2003;312:133–46.

Worrall F, Harriman R, Evans CD, Watts CD, Adamson JK, Neal C,Tipping E, Burt T, Grieve IC, Montieth D, Naden PS, Nisbet T,Reynolds B, Stevens P. Trends in dissolved organic carbon inUK rivers and lakes. Biogeochem 2004;70:369–402.

concentrations and fluxes of dissolved organic carbon in uk topsoils

Documents