nitrogen in rainfall, cloud water, throughfall, stemflow, stream water and groundwater for the...

The Science of the Total Environment 314–316(2003) 121–151

0048-9697/03/$ - see front matter� 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0048-9697(03)00100-1

Nitrogen in rainfall, cloud water, throughfall, stemflow, streamwater and groundwater for the Plynlimon catchments of mid-Wales

Colin Neal *, Brian Reynolds , Margaret Neal , Linda Hill , Heather Wickham ,a, b a a a

Bronwen Pughb

Centre for Ecology and Hydrology, Crowmarsh Gifford, Wallingford, Oxon, OX10 8BB, UKa

Centre for Ecology and Hydrology Bangor, University of Wales Bangor, Deiniol Road, Bangor, Gwynedd, LL57 2UP, UKb

Accepted 2 January 2003

Abstract

An extensive study of acidic and acid sensitive moorland and forested catchments in mid-Wales is used to showthe water quality functioning with respect to nitrate and ammonium. For this, long-term records of rainfall, cloudwater, throughfall, stemflow and stream water(up to 18 years of weekly data) are combined with shorter durationinformation on stream water associated with small tributary sources and drainage ditches, ground water from a net-work of exploratory boreholes and paired control and felled catchments. The ratio of nitrate to ammonium is aboutone in rainfall, cloud water, throughfall and stemflow but the concentrations are much lower in rainfall(;25mM l ) than in cloud water(;300 mM l ) while throughfall and stemflow are intermediate(;80 mM l ).y1 y1 y1

Within the streams draining moorland and forested areas, nitrate concentrations are close to the mean value in rainfallwhile ammonium concentrations are often over an order of magnitude lower in the stream than in rainfall and aretypically only about a fifth that of nitrate. With felling, stream water nitrate concentrations increase for podzolic soilsbut show a variable response for gley soils. For the streams draining forested podzols, the concentrations of nitratecan be up to an order of magnitude higher for the first few years after felling compared to than pre-fell values but inlater years, concentrations decline to pre-fell and even lower levels. Felling for the podzolic soils barely leads to anychanges in ammonium concentration. For the gley soils, felling results in an order of magnitude increase in nitrateand ammonium for a small drainage ditch, but the pulse barely reaches the main stream channel. Rather, within-catchment and within-stream processes not only take up the nitrate and ammonium fluxes generated, but in the caseof nitrate, concentrations with- and post-felling are lower than pre-felling concentrations. Groundwater concentrationsof nitrate for the moorland and forested catchments are slightly lower than for the streams while for ammonium thereverse is the case: ammonium concentrations in groundwater are about a tenth those of nitrate. With felling, ground-water nitrate concentrations show sporadic increases. For two boreholes, these increases occur during wet periodswhen groundwater levels are at their shallowest; for one other borehole, there is a gradual and sustained increaseover several years. The results are explained in relation to the dominant hydrogeochemical processes operative.� 2003 Elsevier Science B.V. All rights reserved.

Keywords: Plynlimon; Hafren; Hore; Rainfall; Stemflow; Throughfall; Stream; Nitrate; Ammonium; Spruce; Forestry; Felling;Podzol; Gley

*Corresponding author. Tel.:q44-1491-838800; fax:q44-1491-692424.E-mail address: [email protected](C. Neal).

122 C. Neal et al. / The Science of the Total Environment 314 –316 (2003) 121–151

1. Introduction

Over the past 30 or more years, there has beena major drive to study nitrogen dynamics withinsoils and stream waters of the UK uplands inrelation to concerns over land use change andatmospheric deposition of nitrogen oxides adverse-ly impacting water resources and stream ecology(Hudson et al., 1997a).The primary environmental concerns of land use

change in relation to nitrogen are associated withfertilizer applications for grassland improvementand the introduction of conifer forests onto moor-land areas. Under grassland improvement, the mainconcern is the rapidity of fertilizer leachingthrough the soils and into the stream. Forestryissues are more indirect, being associated withnutrient generation following land disturbance dur-ing the planting and harvesting phases of theplantation forestry cycle, due to disruption of thebiogeochemical cycle over the past 50 years(Nealet al., 1998a,b). Plantation forestry has been amajor feature of upland agricultural development.The developments started between and post thetwo World wars in response to the strategic requi-rements of homegrown timber and many of theafforested areas are now reaching maturity.The UK uplands are frequently acidic, acid

sensitive and they have been directly impacted bySO emissions and by forestry developmentx

(UKAWRG, 1988). Nitrogen oxide and ammoni-um add to the acid deposition loading. Whilenitrogen oxides and ammonium have not been themain components of acidification to date, it is ofincreasing importance in terms of acidificationrecovery associated with national and internationallegislation for acid emissions reduction from indus-try. Thus, the proportion of acid deposition fromnitrogen sources increases as SO concentrationsx

decline. Indeed, the proportion is increasing asNO emissions continue at current levels due tox

the opposing trends of decreasing industrial emis-sions and increasing vehicle emissions(RGAR,1997). The situation for ammonium inputs, partic-ularly from agricultural emissions, remains uncer-tain (RGAR, 1997). More recently, there has been

the concern over the generation of nitrogen con-taining ‘greenhouse’ gases.In this paper, the dissolved concentration and

flux of inorganic forms of nitrogen(ammoniumand nitrate) is examined within the context of adetailed study of the hydrogeochemistry of uplandacidic catchments in mid-Wales; the Plynlimonstudy(Neal, 1997a,b; Neal et al., 2001). The workintegrates the findings for nitrate and ammoniumwithin the context of the most detailed study ofrainfall, cloud water, stream water and groundwater undertaken in the UK that links closely withstudies on nutrient dynamics and fluxes and soilsolution chemistry(Reynolds et al., 1988, 1989,1992, 1995; Stevens et al., 1990, 1994, 1997).New information is presented on how coniferharvesting affects nitrate and ammonium waterquality, which is of strategic management concernin relation to forestry policy and practice, for akey type locality in the UK uplands.

2. Study area

This paper relates to wide ranging studies at theCentre for Ecology and Hydrology catchmentresearch sites at Plynlimon in mid-Wales. Keyfeatures of the hydrology, biology and water qual-ity of these sites, and the Plynlimon research havebeen described earlier(Neal, 1997a,b).

3. General characteristics of the area

The study area is the headwater catchments ofthe River Severn, with its three main tributaries,the Afon Hafren, the Afon Hore and the NantTanllwyth. The streams drain a hill top plateaudominated by acid moorland in the upper portionof the catchment(Pumlumon Fawr) and the HafrenForest. Catchment areas for these sites vary fromapproximately 50 to 300 ha. The moorland andforest catchments represent a mixture of uplandacid soil types dominated by peaty podzols withsubsidiary peaty gleys. Deep peat deposits are alsoimportant in the moorland plateau area. The bed-rock comprises fractured Lower Palaeozoic mud-stones, shales and grits.

123C. Neal et al. / The Science of the Total Environment 314 –316 (2003) 121–151

The Hafren Forest comprises mainly Sitkaspruce (Picea sitchensis) with some Norwayspruce(Picea abies), larch (Larix spp.) and lod-gepole pine(Pinus contorta) planted in variousphases from the mid-1940s through to the late1960s. A variety of harvesting techniques has beenused, but, in most areas, only the tree stem hasbeen removed from site leaving the felling debris(stumps, branches and needles) behind.Rainfall averages approximately 2518 mm yry1

with evapotranspirational losses of approximately500–700 mm yr and it is typical for the uplandy1

UK. The rainfall is relatively unpolluted exceptfor moderate, long distance transport of acidicoxides. For all the streams, the flow responses tostorm events are very flashy. For example, theAfon Hafren and Afon Hore have flows varyingfrom approximately 0.01 m s after dry antece-3 y1

dent conditions to approximately 4.5 m s fol-3 y1

lowing major rainfall events. Stream chemistry isvariable and predominantly linked to inputs ofwater from two hydrochemically distinct parts ofthe catchment. Under baseflow conditions, streamwaters are approximately pH 7 and they are cal-cium and bicarbonate bearing. These waters areessentially derived from shallow groundwater areaswhere weathering reactions release calcium fromthe soil: the bicarbonate ions in solution comefrom biogenic sources of carbon dioxide in thesoil which are converted to bicarbonate as theweathering reactions consume hydrogen ions. Inthe case of the Afon Hore, calcium and bicarbonateconcentrations are higher than for the Afon Hafrenand Nant Tanllwyth due to increased weathering(the bedrock contains a higher proportion of cal-cium carbonate, as vein minerals, which weatherrelatively quickly). Storm flow waters are mainlyderived from the acidic and soils of the area. Thisprovides runoff that has low pH and low calciumconcentrations but it is enriched in aluminium dueto mobilisation under acidic conditions. Ground-water is dominated by fracture flow and the chem-istry ranges from acidic waters characteristic ofthe soil zone to moderately alkaline and calciumbicarbonate bearing waters characteristic of bed-rock weathering. Groundwater can show a highlydynamic response similar to the streams, but the

pH can be one unit higher and calcium andbicarbonate concentrations up to an order of mag-nitude higher than in the stream.

4. Monitoring points and sampling programme

Nitrate and ammonium data is available forthree types of water quality monitoring site:(1) Long-term monitoring sites for rainfall,

cloud water and the main tributaries of the head-waters of the River Severn. Rainfall was collectedweekly from open gauges at the top and bottomof the HafrenyHore catchment. Cloud water wascollected using a passive lidded ‘harp type’ systemat one site near the top of the catchment on aweekly basis(Wilkinson et al., 1997). All threetributaries of the upper River Severn were moni-tored and there are five monitoring points(twoeach on the Afon Hafren and Afon Hore and oneon the Tanllwyth). Monitoring was weekly usinggrab samples. Only one site has a catchmentconfined to the moorland area upper Hafren(UHa)while the remaining sites have catchments partiallyto completely forested. They are the lower Hafren(LHa), which includes the moorland drainage fromthe upper Hafren, the entirely forested Tanllwyth(Tan), the upper Hore(UHo), and the lower Hore(LHo). The forest area in the lower Hore catch-ment was clear felled during the mid- to late-1980s(mainly between 1985 and 1989). For theother sites, some localised felling has taken placeover many years to thin and in some cases harvestsmall areas of the crop, and about half of theTanllwyth catchment was felled in February 1996.(2) Seven small catchment studies(-15 ha).

These sites examine the interaction between soiltype and forest harvesting on water quality. Apaired catchment approach with control and manip-ulated forested catchments has been used at twolocations. All the sites are within Hafren Forestand represent two of the main soil types ofconcern, peaty podzols(S2Ho, SE1, SE2, SE3)and peaty gleys(Tan1, Tan2). The manipulatedsites at Plynlimon are S2Ho, SE1, SE2 and Tan1.Of these sites, all but one drain into the AfonHafren—the other site, S2Ho, drains into the AfonHore. The site names for the paired catchments


are given in the text and tables with the suffix ‘f’for felling and ‘c’ for control. For the paired sites,both streams and boreholes were monitoredfortnightly.(3) Fourteen exploratory and six monitoring

boreholes. These were drilled to monitor ground-water chemistry and groundwater levels within theHafren catchment. The exploratory boreholes wereestablished in 1984 throughout the catchment toexamine upper plateau(US1, 2, 3), intermediateslopes(IS1, 2), lower slopes(LS1, 2, 3, 4) andthe valley bottom(VB1) to cover a full range ofgeomorphologic environments. All of these explor-atory boreholes were monitored monthly over 1year except for three deep ones examining waterquality gradients at depth(up to 50 m) whichwere monitored less frequently(Neal et al.,1997a,b). Four monitoring boreholes were installedin 1995 near the paired catchments to provide dataon groundwater quality changes at felling andcontrol sites. These borehole sites have the samename as the stream sampling sites except that thesuffix ‘b’ is used to denote that they representborehole locations. These sites were monitoredfortnightly. Two additional borehole sites weremonitored to examine the influence of felling ongroundwater chemistry. One of these sites providedan extended record to the exploratory boreholerecord(site HA4bf—named LS4 within the explor-atory borehole series) and information on theinfluence of felling. The site was monitored weeklyand felling was undertaken at the same time asSE3. The other site is near the Afon Hafren on anintermediate slope between the upper and lowerHafren gauging points. This site, IS3, provided anadditional ‘control’ with information on drainagefrom intermediate slopes and this site was moni-tored fortnightly.A companion paper to this by Neal et al.(2003)

provides a location map with the location of themonitoring points together with a table that pro-vides catchment information and details of theperiod of sampling.

5. Chemical analysis

For the chemical analysis, the samples were firstfiltered using 0.45mM membranes. In the case of

the stream waters, the samples were filtered in thefield while all other samples were filtered soonafter return to the laboratories. The samples werestored in chromic acid washed glass bottles at 48C in the dark prior to nitrate and ammoniumanalysis using standard colorimetric techniques.The nitrate analytical methodology involves thereduction of nitrate in alkaline solution by hydra-zine in the presence of catalytic amounts of copper.The nitrite produced undergoes a diazonizationreaction with sulphanilamide to give a diazoniumion. This diazonium ion is coupled withN-(1-naphtyl)-ethylenediamine to form a reddish purpleazo dye. The methodology determines nitrate plusnitrite within the water. However, nitrite is unstablewithin the waters of concern, and, for the presentpurpose, the results are presented as ‘nitrate’ ratherthan ‘nitrate plus nitrite’. For ammonium, an indo-phenol blue methodology is used: ammoniumreacts in an alkaline phenol solution to producemonochloramine, which leads to the production,via quinone chloramide, of an intensely colouredindophenol blue ion. A pH greater than 9.6 isrequired for the generation and sodium nitroprus-side is added to catalyse the reaction. For themethodologies, calibration was carried out usingstandard stock solutions prepared from high purity(ANALAR grade) potassium nitrate and ammoni-um sulphate, respectively: the calibration rangesare 0–5 mg-NO l (0–80mM-NO l ) and 0–y1 y1

3 3

0.2 mg-NH l (0–11mM-NH l ). Calibrationy1 y14 4

was undertaken at the start of each days analysisand intermediate and check standards were incor-porated every five samples. Analysis was per-formed in duplicate and the samples werereanalysed if they did not replicate within onedivision of a full scale of 100 units(i.e."0.4mM-NO l and"0.06mM-NH l ).y1 y1

3 4

Hydrological information on rainfall, cloudwater, stemflow, throughfall, stream discharge andgroundwater level was collected to assess theinfluence of hydrology on nitrate and ammoniumconcentrations and fluxes.(1) Rainfall and streamflow: Hydrological data

for Plynlimon hydrological network, maintainedby the catchment section of the Centre for Ecologyand Hydrology, Wallingford (Hudson et al.,


1997a,b), provided weekly rainfall and instanta-neous streamflow values corresponding to the timeof sampling. For rainfall, deposition was estimatedfrom an integrated network of gauges locatedthroughout the catchment. Flow data were collect-ed at four flumes within the catchment: the upperand lower Hore, the lower Hafren and the Tan-llwyth. For all the remaining water quality moni-toring sites, no gauging structures were in placeand so flow data had to be extrapolated from thenearest, most suitable site. For example, for theupper Hafren and the SE1f and SE3c streams inthe Afon Hafren catchment, flow data from thelower Hafren is used. Correspondingly, for theS2Ho site in the Hore catchment, the upper Horeflume data is used and Tanllwyth flume data isused for the Tan1f and Tan2c streams. In all cases,the flow data used is expressed in terms of area-weighted values(mm 15 min ) to allow for they1

different size of the catchments and the absolutedifferences in flow. The catchments shows veryresponsive hydrographs characteristic of both smalland large catchment areas. Although there may besmall (hourly scale) differences in the timings ofthe hydrograph responses to rainfall, these differ-ences will not be significant in relation to themuch longer weekly to fortnightly samplingsinvolved.(2) Cloud water, throughfall and stemflow: Vol-

umes of catch were used to provide an indicationof the relative levels of deposition although, ofcourse, there is no direct measure of the actualtransfer of water flux.(3) Groundwater: The depth to the water table

was determined using an electronically triggered‘dipper’ system. The levels quoted have not beennormalized to a standard datum level such asm.a.s.l.

6. Results

Within this paper, a large amount of data ispresented. For example, long-term information isprovided for rainfall, cloud water and five majorstreams, together with more specialized informa-tion for four other streams and nineteen ground-waters, all of which need tying in with issues such

as heterogeneous behaviour, varying degrees ofdeforestation and climate variability. Thus, thepresentation has had to rely on a concise analysisof the data. To do this, the information is sum-marized in the text using tables that provides astatistical summary of the averages and ranges inconcentration together with illustrations that showsalient temporal and hydrologically related fea-tures. In addition, time series are presented in anAppendix A that shows the full range of responsesas a set of three diagrams. Schemes 1a and 1bprovides, respectively, nitrate and ammonium timeseries for the longest available datasets(rainfall,upper Hore, lower Hore and lower Hafren).Schemes 2a and 2b provides a similar time series,but for the stream datasets of intermediate length(upper Hafren and Tanllwyth) together with timeseries for the upper Hore and lower Hore andHafren to show the trends and seasonality acrossall the major stream sites. The third set of figuresin Appendix A shows the variation in nitrate andammonium concentrations over time for the pairedcatchment experiments: Scheme 3ab for the podzolfell and control sites(SE1f and SE3c); Scheme3cd for the gley fell and control sites(Tan1f andTan2c); Scheme 3e provides information for thesouth2Hore stream.Tables 1 and 2 provide a statistical summary of

nitrate and ammonium concentrations for the var-ious monitoring points: mean and flow weightedmean values are presented with information on theranges(minimum and maximum) in concentration.The average, minimum and maximum informationprovide the general statistical overview while theflow weighted values are of direct relevance whenconsidering fluxes. For the tables, the data havehad to be subdivided for stream and groundwatersites where felling has taken place, as will beshown in more detail later in the paper, felling ledin some cases to initial increases in concentrationsand then subsequent loss. The separation usedbased on visual inspection of the time seriespatterns with a split of the data into pre-fellingand felling periods and, where there is a sufficientdata run, a post-felling period. The felling periodtypically lasted for 1–2 years commencing at thetime of felling.


Table 1Mean and ranges in nitrate concentrations(mM l ) aty1

Plynlimon

Mean Flw mean Min Max

AtmosphereycanopyRainfall 22.7 12.6 0.0 206.5Cloud water 325.8 160.4 7.3 3306.5Throughflow 77.1 54.1 0.0 725.8Stemflow 101.5 67.8 0.0 1161.3

Main streamsUpper Haf 15.5 15.7 0.0 73.4Hafren 20.3 22.6 0.0 96.8Upper Hore 14.9 16.1 0.0 58.0Hore pre-fell 20.3 25.4 5.8 50.0Hore fell 40.1 44.3 6.5 75.8Hore post-fell 16.0 16.6 1.6 51.6

Intermediate size streamTan pre-fell 40.4 43.8 15.3 85.5Tan fell 42.7 52.8 15.3 113.7Tan post-fell 22.5 21.6 1.6 109.7

Small streamsSE1f pre-fell 39.9 43.1 13.7 75.8SE1f fell 184.1 179.1 118.5 306.5SE1f post-fell 73.6 98.0 18.5 142.7SE3c 32.5 29.3 12.1 73.39Tan1f pre-fell 10.7 8.5 0.0 80.6Tan1f fell 61.3 52.0 0.0 179.0Tan1f post-fell 20.0 14.0 4.8 37.9Tan2c 1.0 1.2 0.0 85.5S2Ho pre-fell 18.0 16.0 4.0 33.9S2Ho fell 53.7 55.1 5.6 137.1S2Ho post-fell 14.0 11.8 0.0 74.2

BoreholesHa4b pre-fell 10.9 – 0.0 17.7Ha4b fell 123.7 – 4.0 220.2IS3 11.8 – 3.2 33.9SE1b pre-fell 8.0 – 0.0 33.9SE1b fell 62.9 – 0.0 193.5SE3bc 38.3 – 0.0 71.8Tan1b pre-fell 0.9 – 0.0 37.9Tan1b fell 7.8 – 0.0 44.4Tan2bc 31.6 – 1.6 55.7US1 1.0 – 0.0 6.5US2 21.4 – 11.5 24.2US3 19.5 – 15.0 25.6IS1 10.9 – 4.8 13.5IS2 10.5 – 3.2 15.0IS3 11.8 – 3.2 33.9LS1 13.4 – 6.5 20.6LS2 12.4 – 6.5 16.9LS3 16.1 – 11.5 20.5LS4 10.3 – 0.0 16.1VB1 12.7 – 0.0 35.0

For nitrate, the salient features are:(1) Mean and flow weighted mean values in

rainfall are 23 and 13mM l , respectively, whiley1

the range is 0–207mM l .y1

(2) In contrast, cloud water has high nitrateconcentrations in terms of mean(326 mM l ),y1

flow weighted mean(160mM l ) and range(7–y1

3307mM l ).y1

(3) Throughfall and stemflow show intermediatevalues and ranges to rainfall and cloud water.(4) The highest nitrate concentrations occur for

low volumes of catch(see greater discussion onthis point later in the paper). This is why the meanconcentration is higher than the flow weightedmean concentration.(5) The large streams, excluding the data for

sites where felling has occurred, have a meanyflow weighted mean nitrate concentration of 17.9y20.0 mM l and the range in mean for they1

individual sites is 15.5–20.3y15.7–25.4mM l .y1

The range extends to 40.1y44.3 mM l if they1

felling data are included.(6) For the intermediate to small streams, the

corresponding meanyflow weighted mean, forcatchments not subjected to felling, is higher(30.9y28.4 mM l ) as is the range in mean fory1

all the streams in this category(1.0–40.4y1.2–43.1mM l ). The main Tanllwyth stream and they1

SE1f and SE3c provide the highest mean and flowweighted mean nitrate concentration. The lowestvalues occur also for the Tanllwyth but for thesmall drainage areas.(7) For the small to intermediate size streams,

the range in the meanyflow weighted mean con-centrations was 184y179 mM l , if the fellingy1

information is included.(8) The data show that felling can result in a

major increase in the mean and range in nitrateconcentration within the streams. Such increasesmay be up to an order of magnitude. However,post-felling, concentrations can then return to pre-felling values and even lower. For example, thesouth2Hore stream has pre-fell, fell and post-fellmeanyflow weighted mean nitrate concentrationsof 18y16, 54y55 and 14y12 mM l , respectively.y1

For SE1f and Tan1f, a similar feature is observed,but post-felling mean nitrate concentration doesnot decline to pre-fell levels. In the case of SE1f


Table 2Mean and ranges in ammonium concentrations(mM l ) at Plynlimon: %NHs100=NH y(NH qNO )—also on amM l basisy1 y1

4 4 4 3

Mean Flw mean Min Max %NH4

Atmospheric inputsRainfall 27.17 16.11 0.00 283.3 54.5Cloud water 286.11 157.61 0.89 3000.0 46.8Throughflow 72.08 57.61 0.00 522.2 48.3Stemflow 80.15 62.75 0.22 755.6 44.1

Main streamsUpper Haf 0.65 0.75 0.00 17.3 4.0Hafren 0.86 1.09 0.00 75.6 4.1Upper Hore 0.72 0.58 0.00 5.6 4.6Hore pre-fell 1.66 2.77 0.00 14.3 7.6Hore fell 0.88 1.44 0.00 11.2 2.1Hore post-fell 0.56 0.73 0.00 4.9 3.4

Intermediate size streamTan pre-fell 0.65 0.75 0.00 8.9 1.6Tan fell 1.27 1.52 0.00 62.2 2.9Tan post-fell 0.90 1.32 0.00 14.2 3.8

Small streamsSE1f pre-fell 0.65 0.57 0.00 2.2 1.6SE1f fell 0.94 1.68 0.00 12.4 0.5SE1f post-fell 1.05 0.81 0.00 7.2 1.4SE3c 0.70 0.84 0.00 3.9 2.1Tan1f pre-fell 1.02 0.82 0.00 5.4 50.5Tan1f fell 7.36 9.41 0.22 47.2 40.8Tan1f post-fell 1.50 1.74 0.56 7.4 2.4Tan2c 1.51 1.92 0.22 9.0 7.0S2Ho pre-fell 0.98 0.82 0.00 3.4 5.2S2Ho fell 1.06 1.29 0.00 14.4 1.9S2Ho post-fell 0.91 0.85 0.00 95.6 6.1

BoreholesHa4b pre-fell 2.91 – 0.00 12.2 21.1Ha4b fell 2.08 – 0.00 21.0 1.7IS3 2.24 – 0.00 42.9 16.0SE1b pre-fell 1.72 – 0.00 5.2 17.7SE1b fell 1.89 – 0.00 5.8 2.9SE3bc 2.52 – 0.00 11.8 6.2Tan1b pre-fell 4.42 – 0.71 12.4 83.1Tan1b fell 3.24 – 0.57 34.3 29.3Tan2bc 1.74 – 0.00 8.7 5.2US1 0.84 – 0.00 2.2 45.7US2 2.06 – 0.00 12.0 8.8US3 1.57 – 0.00 5.1 7.5IS1 3.52 – 0.00 8.9 24.4IS2 2.10 – 0.00 6.1 16.7IS3 1.74 – 0.00 33.3 12.9LS1 1.42 – 0.00 3.6 9.6LS2 3.49 – 0.78 8.2 22.0LS3 1.67 – 0.00 3.3 9.4LS4 2.47 – 0.00 6.3 19.3VB1 4.41 – 0.33 10.8 25.8


and Tan1f, the lack of an extensive post-fell recordprobably means that nitrate reductions post-fell hasnot bottomed out as discussed more fully later inthe paper.(9) Felling is not always followed by a signifi-

cant increase in mean nitrate concentration. Thisfeature is shown for the main channel of theTanllwyth, where there is hardly any rise in meanconcentration from the period pre-fell to fell.However, there seems to be a reduction in nitrateconcentration post-fell.(10) The groundwater monitoring sites has

slightly lower mean concentrations(14.2mM l ),y1

excluding the felled data, than the streams althoughthe range in mean across the groundwater moni-toring sites is similar to that for the streams(0.9–38.3 mM l ). If the felling information isy1

included, the mean extends up to 123.7mM l .y1

The felling activity has led to increases in nitrateconcentrations. These increases last for at least 3years. N.B. comparison cannot be made betweenstream and groundwaters on a flow weighted basisas only groundwater level data was available torepresent the hydrology and this is not equate withgroundwater flow.Correspondingly, for ammonium, the salient fea-

tures are:(1) Ammonium concentrations in rainfall

(meanyflow weighted mean 27.2y16.1 mM l ,y1

range 0–283mM l ) are much lower than iny1

cloud water(meanyflow weighted mean 286y158mM l , range of 1–3000mM l ). Throughfally1 y1

and stemflow show intermediate values and rangesto rainfall and cloud water.(2) The streams have mean, flow weighted mean

and ranges in concentration one to two orders ofmagnitude lower than even the rainfall. As withnitrate, ammonium concentrations in rainfall andcloud water have higher concentrations in smallervolumes of catch(see more detailed discussionslater in the paper) and hence flow weighted meansare only about a half those for the mean. There isa much smaller difference for the mean and flowweighted mean concentrations in the stream. Witha 10-fold difference between the rainfall andstream water means, even for the flow weightedaverages, the catchments behave as a major sinkfor the atmospheric input.

(3) The large streams, excluding the data forsites where felling has occurred, have a meanammonium concentration of 1.0mM l (1.3y1

mM l on a flow weighted basis) and the rangey1

in mean for the individual sites is 0.65–1.66mM l . The range remains the same if the fellingy1

data is included as the additional fell site in thiscategory(the lower Hore) has a mean ammoniumconcentration of 0.88mM l (1.2 mM l on ay1 y1

flow weighted basis).(4) For the small to intermediate sized streams,

the meanyflow weighted mean and ranges inmeanyflow weighted mean for ammonium concen-tration is very similar to the larger streams iffelling has not occurred(1.0y1.1 mM l ), withy1

corresponding ranges of 0.57–1.92mM l ). They1

range in mean and flow weighted mean for theintermediate and smaller streams is considerablyextended if the felling data is included(maximum7.36y9.41mM l ).y1

(5) The groundwater monitoring sites has slight-ly higher ammonium concentrations(mean 2.40mM l ), excluding the felled data, than they1

streams and the corresponding range in meanacross the borehole sites are higher(0.84–4.42mM l ). If the felling information is also includedy1

in the analysis, the mean and range remains aboutthe same.(6) Felling resulted in a 7-fold increase in the

mean and range in ammonium concentration forthe small drainage site monitored within the gleycatchment of the Tanllwyth(Tan1f). For this site,post-felling concentrations approximately returnedto pre-felling levels. However, the subsequentreduction for Tan1f may not have ‘bottomed out’due to the lack of a sufficient post-fell data run:pre-fell, fell and post-fell mean and flow weightedmean concentration are 1.0y0.8, 7.4y9.4 and 1.5y1.7 mM l , respectively. For the Tanllwyth bore-y1

hole, no clear change in mean ammoniumconcentration occurred with felling. Further, thereis no clear change in ammonium concentration forthe streamsyboreholes associated with the podzols(SE1f and south2Hore).(7) The increase in mean and flow weighted

mean ammonium concentration for the small Tan-llwyth drainage area Tan1 is observed within themain stream, but the change much smaller, pre-


Fig. 1. Ammonium and nitrate concentration time series forPlynlimon rainfall and cloud water.

fell, fell and post-fell meanyflow weighted meanammonium concentrations being 0.7y0.8, 1.3y1.5and 0.9y1.3mM l , respectively.y1

Ammonium typically makes up only approxi-mately 20% of the inorganic nitrogen in the streamand groundwaters, compared to 50% in the rainfalland cloud water. However, there is a range in%NH mean in the streams and groundwaters4

across sites that seem related to scales and fellingactivity. Both the mean and the range in mean for%NH increase in a sequence large streams to4

small and intermediate streams and then to ground-waters, irrespective of whether or not the fellingdata is included.The averages and ranges in %NH are:4

(1) Large streams: mean 4.9%, range 4.0–7.6%.(2) Small to intermediate size streams: mean

14.7%, range 1.4–50.5%.(3) Groundwaters: mean 20.6%, range 5.2–

83.0%.(4) The highest proportion of ammonium occur

for the Tan1f site on gley where reduction potentialis at its highest and the values are influenced bythe felling activity: the %NH was approximately4

40–50% during the pre-felling to felling periodand 83–29% for the associated borehole. Corre-spondingly, for the main Tanllwyth stream, thepercentage was much lower at 1.8–3.6%.

7. Flow, temporal and felling related changes inconcentration

The variations in nitrate and ammonium concen-trations in the atmospheric inputs, the streams andthe boreholes are related to hydrology, time andextent of felling activity and it is difficult toprovide a simple scheme for describing the infor-mation. Here, the atmospheric inputs are firstconsidered as they have unique characteristics,then the stream and ground waters are dealt within terms of moorlandyforested sites and felledsites.

7.1. Atmospheric inputs

Nitrate and ammonium concentrations vary con-siderably through the year in both rainfall andcloud water. The highest concentrations occur

during the summer months and there are year-to-year variations(Fig. 1). The highest concentrationsand variations in concentration of both nitrate andammonium occur for low volumes of catch. Thiscorresponds to a washout effect where fine cloudwater and particulate materials are removed fromthe atmosphere during wet conditions(cf. Cryer,1986). i.e. with the atmospheric scavenging of alimited supply of ammonium and nitrate, as dryand mist particles, increased rainfall inputsycloudwater-catch results in a dilution of the signal.While the ammonium and nitrate patterns variedfrom year to year, the changes are very similar foreach other. Thus, it seems that the cloud water anddry deposited sources of ammonium and nitrateare of a similar nature. The highest mean concen-trations of nitrate and ammonium occurred duringthe mid-1990s.The concentrations of nitrate and ammonium in

stemflow and throughfall show intermediatebehaviour to that of rainfall and cloud water. The


Fig. 2. An example of the seasonal nature of nitrate concen-tration and temperature variability for moorland and relativelyundisturbed forested sites: the lower Afon Hafren during theperiod 1990–1992.

highest concentrations occur during the summermonths and this reflects a similar feature to thatin rainfall and cloud water, a dilution of atmos-pheric inputs with increasing volumes of catch.However, biochemical cycling will affect the sea-sonal patterns, but there are a large number offactors such as canopy uptakeyleaching: this aspectcannot be examined here.

7.2. Moorland and forested catchments relativelyundisturbed by forest harvesting

For ammonium, there is no visually discernableseasonal or flow related pattern for the moorlandand relatively undisturbed forested catchments andthe data is not plotted here. Nitrate concentrationsshow clear patterns of seasonal and longer-termchange in the streams and smaller changes withinthe groundwaters. However, ammonium concentra-tions in the streams and groundwaters are lowrelative to the inputs although there are occasionaloutlier points of higher concentration.Given the clear patterns for nitrate and the lack

of pattern and the relatively low concentrations forammonium, only the salient features for nitrate aredescribed below: the reader can gain any extrainformation on ammonium from Table 1 and theAppendix A.For the streams, those draining the moorland

and the forested catchments relatively undisturbedby felling show very similar behaviour. There aretwo main features to the patterns observed. Firstly,there are regular seasonal oscillations that areinversely correlated with temperature(Fig. 2). Thelowest nitrate concentrations occur during the sum-mer months when both temperature and biologicaluptake, within the catchment and within thestream, is at its highest. The variation in nitrateconcentration over time is oscillatory, but there issome marked scatter. In part, the scatter is relatedto hydrological conditions and the nitrate concen-trations become almost constant at low to inter-mediate concentration values at high flows.Secondly, there are year-to-year variations innitrate concentrations with higher concentrationsbeing observed during the early 1980s and the midto late 1990s when concentrations are perhaps50% higher than at other times. During these

abnormally high periods, the scatter in the nitrateconcentration data is higher and the sinusoidalpatterns become less distinct.There is no clear visual change in nitrate con-

centrations in the exploratory boreholes that canbe attributed to season(e.g. temperature) orhydrology (water level). Nonetheless, informationfor the longest and most compete time series fora forested site in podzol shows some relationshipbetween the nitrate concentrations and hydrologi-cal factors in that at times of greater water flow,i.e. when the water table is shallower, nitrateconcentrations are higher by a factor of approxi-mately three(;5 mM l at 5 m depth to 15y1

mM l at 3.5 m depth). As shown in earliery1

publications, the groundwater is dominated byfracture flow. This feature, for example, explainswhy the groundwater levels and some chemicaldeterminants show large fluctuations over time(Neal et al., 1997a,b). The changes in nitrateconcentration observed may reflect the removal ofnitrate from within the groundwater zone as resi-dence time increases andyor more nitrate rich soilwaters being transmitted through the groundwatersystem when the catchment is wet.


8. Nitrate and ammonium in streams for fellingareas

With felling, the hydrobiogeochemical cyclingis disrupted. For the majority of the catchmentsstudied, the primary change in nutrient chemistrywith felling is associated with nitrate generation(e.g. Bormann and Likens, 1994). In the case ofdrainage from forested catchments with podzolicsoil, felling results in relatively little change inammonium concentrations in runoff. However, forforested catchments draining gley soils under somecircumstances felling can lead to a marked increasein ammonium concentration. Further, even fornitrate there are some differences with felling forthe podzolic and gley soils.

8.1. The effects of felling podzolic areas on nitrateconcentrations in stream water

The effects of felling on stream water nitrateconcentrations for podzolic system are representedin this paper by three studies. Firstly, there is themain Hore with the upper Hore monitoring pointbeing a control for downstream felling to the lowerHore monitoring point. Secondly, there is a smalltributary of the Hore(south2Hore). Thirdly, twominor tributaries of the Hafren(SE1f and SE3c)have been studied as part of a paired catchmentapproach. The results of these studies are describedbelow and illustrated in Fig. 3 as time series plotsand in Fig. 4 as plots of nitrate concentrationagainst the logarithm of flow.

8.1.1. The HoreThe effects of progressive felling on stream

water nitrate levels show that 4 years of fellingactivity over double the concentrations(from 20to 50 or 60mM l ) during the felling years, aftery1

which time concentrations then decline to pre-felling and even lower values. There are six majorpoints to note from the time series and logarithmof flow plots.(1) There is a clear increase in nitrate concen-

tration with felling even though(a) only a half ofthe catchment was felled and(b) the felling took4 years to complete.

(2) The annual cyclical pattern is maintained,with maximum concentrations occurring duringthe winter with minimum values occurring duringthe summer. However, with felling, there is muchgreater scatter to the seasonal patterns observed.(3) The nitrate pulse associated with felling is

complete within approximately 2–4 years after thecessation of felling. This post-felling reduction innitrate concentration is marked by a gradual annualdecline in both the summer minima and the wintermaxima.(4) There are increases in nitrate concentration

after felling from 1996 to 1998. These increasesare the same for the control catchment and forother streams within the area as discussed earlierin the paper in relation to climatic influences.(5) The increases in nitrate concentrations asso-

ciated with felling are most marked at higher flowsalthough there is high scatter to the data. Fig. 4shows that for the lower Hore, which records thefelling response, there is a ‘L’ shaped area in theconcentration vs. log(flow) diagram. In essence,the upper slope represents an upper limit for thefelling years, while the lower slope represents theperiods pre- and post-felling. The data plotted forthe upper Hore in Fig. 4 illustrate a much flatterpattern for the control.(6) From the evidence above, it is clear that

with felling the nitrate concentrations in the soilincrease with the disturbance of the biogeochemi-cal cycle and during rainfall events, at which timenitrate is flushed from the catchment to the stream.With regards to assessing in detail the impacts

of felling for the lower Hore, a comparison withthe upper Hore control(Fig. 3) is unsatisfactoryas the data record for the upper Hore is shorter interms of nitrate concentration and flow measure-ments. To take things further, here a more detailedcomparison is made for the lower Hore and Hafren,which have equivalent data runs. Indeed, initially,the lower Hafren was meant as a control for thelower Hore. However, differences in bedrockweathering precluded a direct comparison for com-ponents such as the base cations, bicarbonate andpH (Reynolds et al., 1986). Such differences arenot of importance for nitrate, which is not involvedto a major degree with the hydrochemical pro-cesses associated with bedrock weathering. The


Fig. 3. Time series plots for nitrate in streams draining podzolic soils to show the effects of felling activity. There are three plotsincluded. At the top, nitrate data for the upper and lower Hore streams are provided: the upper Hore represents a control to setagainst felling in the lower Hore. The middle plot provides a nitrate time series for the south2Hore. The lower plot provides nitratedata for the SE1f and SE3c streams: SE3c being a control reference for felling on SE1f. For SE3c, there is an incomplete record—the later part of the record is missing as sampling had to be curtailed due to cuts within the programme. For all the plots, the periodof felling is denoted by ‘Fell’. Note that the period of record differs for the different sites. Grid lines covering yearly periods andmajor ‘tick’ marks covering 2-year time steps are used to standardize the plots so that the reader can gauge the respective recordlengths and the felling responses.

time series for the lower Hafren and Hore(Fig. 5)show a similar behaviour as between the upperHore and the lower Hore(except for the pre-felling period where upper Hore data are missing).There are three periods of behaviour. Firstly, forthe period pre-felling, nitrate concentrations aresimilar for both the lower Hafren and Hore. Sec-ondly, during the period of felling activity, there is

a divergence in pattern and nitrate concentrationsfor the lower Hore become approximately two tothree times higher than for the lower Hafren. Thisrepresents the main felling response and it lastsfor approximately 5 years. Thirdly, for the periodfollowing the main felling response, nitrate con-centrations are initially similar for the lower Hafrenand Hore, but gradually concentrations for the


Fig. 4. Plots of the relationship between nitrate concentration and the logarithm of flow for the streams draining forestedyharvestedcatchments with podzolic soil. The two graphs at the top of the page show fell and control information for the Hore while the nextrow of graphs on the page show the corresponding information for the SE sites of the Hafren. Below these, a graph for the south2Horesite is plotted. There is no corresponding ‘small stream’ reference control. The nearest control for the south2Hore is probably theupper Hore(the top right-hand figure to the page) and this is re-plotted against the south2Hore graph for comparative purposes. Atthe bottom of the page, the corresponding plots for the lower and upper Hafren are presented.

lower Hafren become higher than for the lowerHore. These three periods are shown clearly inFig. 5 where the difference in concentrationbetween the lower Hafren and Hore is plotted asa time series. The point to greatest divergence canbe considered as corresponding to the nitrate flux

associated with disturbance of the hydrobiochem-ical cycle. The subsequent convergence of the twodatasets corresponds to the period where the hydro-biochemical cycle re-establishes itself and nitrateis taken up into the newly developing biomass.Presumably, over long time spans the two datasets


Fig. 5. A comparison of the nitrate concentration variations for the lower Hafren and Hore. The topmost graph shows a straighttime series while the lower graph shows the variation in the difference in their nitrate concentrations over time.

meet as a mature second rotation forest ecosystemestablishes itself.

8.1.2. South2HoreWith felling in the autumn of 1989, there is an

initial increase in nitrate concentration. Thisincrease takes place within the same year as thefelling, for the early autumn storms. The increaselasts for approximately 1–2 years: nitrate typicallyincreases from approximately 18 to approximately80mM l although there are clear nitrate concen-y1

tration fluctuations through the year. Subsequent

to this increase, there is a decline in nitrateconcentration from 1990 through to 1994, bywhich time the felling response is largely over.However, there are secondary peaks in concentra-tion from 1996 to 1998, a climate related featuredescribed earlier that is not associated with thefelling. Nitrate concentrations post-felling and postthe 1996–1998 secondary peaks are lower thanpre-felling values(a reduction from;18 to;7mM l ). During the major period of felling dis-y1

turbance, when nitrate concentrations peak, thehighest values are observed during the autumn


when the decomposition products of felling arefirst leached from the catchment. There are alsopeaks in nitrate concentration each year during thewinter, throughout the study period. This corre-sponds, as for the other catchments, to the periodwhen biological uptake is minimal. With regardsto linkages between nitrate concentration and flow,Fig. 4 shows a similar ‘L’ shaped feature to thenitrate concentration vs. log(flow) plot as theHore. For this graph, many of the low nitratepoints lie in a scattered pattern parallel to thelog(flow) axis and the pattern is very similar tothat for the upper Hore control. The reason whyso many points plot in this area is due to the longdata span and the relatively few years when nitrateconcentrations are high. For the felling disturbanceperiod, there seems to be an upper bound to thenitrate concentration vs. log(flow) pattern that islinear in nature and the data is scattered duringthe felling disturbance between this upper boundand the lower bound characterised by the pre- andpost-felling response.

8.1.3. SE1fySE3cThe felling of the SE1 catchment, during Sep-

tember and October 1995, resulted in a majorincrease in nitrate concentration the following yearin response to the first autumnal storms. Duringthis time, concentrations increased from approxi-mately 40 to 180mM l . Nitrate concentrationsy1

then remained high through the winter of 1997and concentrations were even higher during thefollowing winter, values reaching up to approxi-mately 300mM l . Subsequent to this maximum,y1

around January 1998, the concentrations havegradually declined. However, even now, thedecline does not seem to be complete: current(2001) values are approximately 60mM l . Withy1

regards to the relationship between nitrate concen-tration and the logarithm of flow, a similar patternis observed as the Hore and south2Hore, with ‘L’shaped feature for the fell site and a flat responsefor the control site. i.e. the felling response ismarked by increases in nitrate concentration as afunction of flow. This response is a scattered onewith a upper bound corresponding to the period of

high felling response and a lower bound corre-sponding to pre- and post-felling response periods.

8.1.4. Horeysouth2Horeysoutheast comparisonsThe three felling studies for podzolic soils show

major increases in nitrate concentration that decayaway after approximately 2 or 3 years post-felling.The results may be summarized as follows.(1) The variations in nitrate concentration over

time show both a yearly cycle and a disturbanceeffect that is in part linked to the hydrologicalresponse.(2) With felling, nitrate concentrations increase

during the following autumnal storms of the sameor subsequent year(the time lag depends upon theseason during which felling takes place—if fellingis late in the year, the response occurs during thefollowing autumn and if the felling is early in theyear, the autumnal response is in the same year).(3) The major response to felling lasted for

approximately 3 years but there seemed to be alonger-term post-felling decline in nitrate concen-tration to values below the pre-fell concentrations.(4) The increase in nitrate concentration asso-

ciated with felling varies from site to site. Thesoutheast fell site shows the greatest responsewhere clearfelling results in a nitrate increase toapproximately 300mM l . This compares with ay1

clearfelling nitrate response of approximately 100mM l for south2Hore. With felling staged overy1

several years, the nitrate response for the lowerHore is smaller in terms of concentration(a dou-bling in concentration to approximately 50 or 60mM l —the Hore).y1

For ammonium, although the effects are noteasy to discern visually owing to the ‘spiky’ natureof the signal examination of the averaged, pre-fell,fell and post-fell revealed contrasting behaviourfor the three podzolic sites. For the lower Hore,mean and flow weighted means declined from pre-to post-fell with intermediate behaviour at the timeof fell. Thus, in this case there was no indicationof a release of ammonium at the time of felling:e.g., the mean ammonium concentration from pre-fell to fell and post-fell is 1.66–0.88–0.56mM l . For the SE1 stream, the mean ammoniumy1

concentrations increase through the felling


Fig. 6. Time series plots for nitrate and ammonium for the Tanllwyth. The presentation includes paired catchment information forsmall drainage areas(Tan1 and Tan2 sites) as well as for the main channel.

sequence: i.e. mean concentrations changed in thesequence 0.65–0.94–1.05mM l although flowy1

weighted averages peak at the time of felling. Forsouth2Hore, concentrations peaked during the fell-ing period, but the changes involved are small:e.g., for mean and flow weighted mean, the pre-felling, felling and post-felling sequence is 0.98–1.06–0.91mM l and 0.82–1.29–0.85mM l .y1 y1

8.2. The effects of felling a gley area at Plynlimonon nitrate and ammonium concentrations in streamwater

The effects of felling for a gley area at Plynli-mon are represented in the present study with themonitoring in the Nant Tanllwyth catchment oftwo small drainage ditches and the main stem. Theresponses are distinct, they are described separatelybelow, and the patterns of change are shown inFig. 6 together with the relationships to flow inFig. 7.

8.2.1. Localised effects of felling on stream nitrateand ammonium concentrations for the TanllwythThere is a marked increase in nitrate and ammo-

nium concentrations for the small drainage areamonitored during the autumn of the same year asfelling for the Tanllwyth catchment in February1996. The changes are as follows.(1) Nitrate: Nitrate concentrations increased

with felling from approximately 10 to 100mM l during the autumn storm periods of they1

felling year. Nitrate concentrations remained highat similar concentrations for the subsequent yearand then they started to fall up to the present time(spring 2001). Post-felling nitrate concentrations,while much lower than the felling maxima werestill about twice the pre-fell levels. With regardsto the linkages between nitrate concentrations andflow, the results for the Tanllwyth(gley) look verysimilar to those for the Hore and Hafren(podzols)described earlier. In essence, a nitrate concentrationvs. the logarithm of flow shows a ‘L’ shapedfeature with the upper bounding gradient corre-


Fig. 7. Graphs showing the relationship between nitrate concentration and the logarithm of flow for streams draining forest-edyharvested catchments with gley soil. The left side of the page comprise information on nitrate and the right and side comprisesinformation on ammonium. The two graphs at the top of the page represent the SE1f fell sites for the paired catchment while themiddle two graphs provide information for the control(SE3c). The lower two graphs relate to the main stem of the Tanllwyth.

sponding with felling effects and a lower slopelinked to background changes(Fig. 7).

Ammonium: Ammonium concentrationsincreased with felling from less than 5mM l toy1

approximately 60mM l during the autumn stormy1

periods of the felling year. Concentrations thenrapidly declined to pre-felling levels over the next12–18 months. There were no clear pre-fell, felland post-fell relationships between ammoniumconcentration and flow(Fig. 7).Thus, it seems that the disruption of the biogeo-

chemical cycle results in both nitrate and ammo-

nium release to small drainage areas and the mainstem of the Tanllwyth. In the case of nitrate, therelease continued for several years, but in the caseof ammonium, the effects were much shorter lived.

8.2.2. The influence of felling on stream nitrateand ammonium concentrations for the main stemof the TanllwythDespite the large changes in nitrate and ammo-

nium concentration for the small drainage areamonitored in the Tanllwyth, these changes are notvisually observed within the main channel even


though felling took place over about a half of thecatchment(Fig. 6). Rather, the main pattern ofchange is a decline in nitrate concentration in themain channel through the fellingypost-felling peri-od mirrors that for the other streams with orwithout felling. However, to examine if any fellingresponse could be observed, nitrate concentrationswere plotted against the logarithm of flow to seeif the characteristic ‘L’ shaped pattern for a fellingresponse could be observed(Fig. 7). The plots doshow an ‘L’ shaped structure, but the range innitrate concentrations is small and the impacts aremuch smaller than for perturbations found at othersites.Because of the features described for the paired

catchments and the main stem of the Tanllwyth, itseems that either the small areas monitored are notrepresentative of the catchment as a whole or thereare within-catchment or within-river processes thatcounteract the nitrate pulses being generated withinthe soil.With regards to ammonium, concentrations

remained relatively low throughout the period ofmonitoring apart from the occasional ‘spike’. Themajority of these spikes occurred subsequent tofelling and this was clearly indicated by the highflow weighed ammonium concentrations being anorder of magnitude higher for the felling periodcompared to pre-fell times. Subsequent to felling,the flow weighted ammonium concentrationsdeclined to pre-fell levels. The ammonium signalwithin the main channel in some ways resembledthat in the Tan1f stream in that there were shortlived ‘spikes’ at the time of fell, but the ‘spikesdid not correspond to particular days.

9. The effects of felling on nitrate and ammo-nium concentrations in groundwater

With felling of both the podzolic and the gleyareas studied, nitrate concentrations in groundwaterchanged as a function of time and groundwaterlevel, but ammonium concentrations remainedunaffected(Fig. 8). There were two types of nitrateresponse. Firstly, at one of the sites with podzolicsoils (HA4 borehole), nitrate concentrationsincreased from approximately 20 to 180mM ly1

with the felling. The increase was gradual and the

concentrations seem to have flattened after approx-imately 2 years. These high levels have beenmaintained for a further 3 years up to the presenttime. There was no significant fluctuation in nitrateconcentration with groundwater level. Secondly, attwo sites, one podzol(SE1f) and the other gley(Tan1), there was an increase in nitrate concentra-tion with felling. However, this increase is onlyobserved at times of shallow depths to the watertable, i.e. wet high flow conditions. This increasepersisted for at least 3 years post-felling. Thus, itseems that at these sites, there is a relativelylimited volume of storage and that nitrate is beingflushed through the fracture flow system from thesoils. This pattern of behaviour is similar to theIS3 borehole control site except that the responseto changing water level was much more pro-nounced due to the much higher soil water nitrateconcentrations involved with felling.

10. Nitrate and ammonium fluxes

The study of nitrogen fluxes for such hydrolog-ically and hydrobiochemically dynamic catchmentsis difficult. The difficulty comes about for fourreasons. Firstly, the input of nitrogen fluxes com-prises rainfall, cloud water, particle and gaseousdeposition and while rainfall inputs may be rea-sonably well determined from the hydrometricnetwork of gauges, the other components are muchmore difficult to gauge. Secondly, because of thedynamic nature of the hydrograph and nitrogenchemograph, a very detailed sampling programmeis required to pick up the nitrogen dynamicsthrough hydrological events: in the absence of this,flux estimation is much less reliable. Thirdly,nitrogen may be lost from the catchment to theatmosphere with the decomposition of organicmatter, particularly as nitrogen gas. Fourthly, otherdissolved components may be important(e.g.organic nitrogen). However, some reference tonitrogen input–output fluxes are appropriate toillustrate the broad features of within catchmentprocessing and although there will be significantuncertainty over the exact fluxes, reference can bemade to data for other sites to see if the values fitinto a pattern for this type of catchment system.


Fig. 8. Nitrate time series for boreholes representative of felling and control sites on podzolic and gley soils. The podzol soil systemsrelate to the HA4 and SE1fySE3c borehole sites while the gley system refers to the Tan1f and Tan2 sites.

Table 3 provides approximate nitrate and ammo-nium flux estimates for the streams prior to fellingand during the felling phase. For this table, long-term hydrological information for the Plynlimoncatchments has been combined with the flowweighted concentration data for nitrate and ammo-nium. There is no flow-gauging site for the upperHafren, SE1, SE3, Tan1 and Tan2 sites. Informa-tion from the moorland catchment of the adjacentWye is used to estimate long-term runoff infor-mation for the upper Hafren as this area is moor-land and evapotranspiration would be expected tobe lower (Hudson et al., 1997b). For the other

catchments, the long-term mean flows for theRiver Severn catchment is used. Although this isnot ideal, the errors involved are small comparedto others describe above. The results are as follows.For nitrate, there are five main observations to

make. Firstly, there is an approximate balancebetween the rainfall nitrate input and the streamoutputs for the upper Hafren and upper Hore.However, cloud water inputs typically account forabout a third of the nitrate deposition and somecases there may be a balance while in other casesthere may be losses or net effluxes from theindividual catchments(Neal et al., 1997c; Wilkin-


Table 3Approximate nitrate and ammonium flux estimates for the major Plynlimon streams

Flowa Nitrate Ammonium Nitrate flux Ammonium(mm yr )y1 (mM l )y1 (mM l )y1 (kg-N ha yr )y1 y1 (kg-N ha yr )y1 y1

Rainfallb 2518 12.6 16.1 4.44 5.68upper Hafren 2048 15.7 0.75 4.50 0.22Upper Hore 1934 16.10 0.58 4.61 0.17Tanllwyth—pre-fell 1934 43.8 0.75 11.86 0.20Tanllwyth—fell 1934 52.8 8.52 14.30 2.20Lower Hafren 1934 22.6 1.09 6.12 0.30Lower Hore—pre-fell 1934 25.4 2.77 6.88 0.75Lower Hore—fell 1934 44.3 1.44 12.00 0.39Lower Hore—post-fell 1934 5.1 0.91 1.38 0.25

Paired catchments and small streamsSouth2Hore—pre 1934 16.0 0.82 4.33 0.22South2Hore—fell 1934 55.1 1.29 14.90 0.35South2Hore—post-fell 1934 8.6 0.74 2.33 0.20SE3c 1934 29.3 0.84 7.94 0.23SE1f—pre-fell 1934 43.1 0.57 11.67 0.15SE1f—fell 1934 179.1 1.68 46.51 0.46Tan2c 1934 1.2 1.97 0.33 0.53Tan1f—pre-fell 1934 8.5 0.82 2.30 0.22Tan1f—fell 1934 52.0 9.41 14.09 3.81

The flow data are approximate and are taken from the long-term record presented by Hudson et al.(1997b).a

The atmospheric deposition will be larger than the rainfall estimates due to cloud water and dry deposition, which can contributeb

up to an extra 50%.

son et al., 1997). Secondly, there seems to beapproximately a 50% higher output from the lowerHafren and Hore(pre-felling) relative to the rain-fall input. For these catchments, there will be anapproximate input–output balance or a small netoutput flux. Thirdly, the estimated output fluxesfor the Tanllwyth (pre-felling), SE1 (pre-felling)and SE3 are two to three times that of the rainfallinput (7.9–11.9, compared to 4.4 kg NO –3

N ha yr ). It seems that these streams providey1 y1

a net nitrate output, which is unexpected as thiscatchment has the greatest potential for nitratereduction and nitrate concentrations can be partic-ularly low for the small drainage areas in theTanllwyth catchment prior to felling(2.3 and 0.33kgNO –N ha yr for Tan1f and Tan2c, respec-y1 y1

3

tively). Fourthly, for the sites undisturbed by fell-ing, the export fluxes for nitrate are in anincreasing order Tan2c, Tan1f, upper Hore, upperHafren, lower Hafren, lower Hore, SE3, SE1 andTanllwyth. This sequence, with the exception forthe Tan1f and Tan2c sites, corresponds approxi-

mately to one of decreasing altitude. Fifthly, duringthe felling disturbance phase nitrate fluxesincreased significantly(Table 3). Comparing preand felling disturbance phases, the following fea-tures were observed for various felling areas.For the podzolic sites:(1) For the lower Hore, the nitrate flux approx-

imately doubles with felling, from 6.9 to 12.0 kg-N ha yr . Since only a half of the catchmenty1 y1

was felled, this approximately represents a quad-rupling of flux if full deforestation is considered(i.e. a change from 6.9 to 17.1 kg-N ha yr).y1 y1

(2) For south2Hore, the nitrate flux increasedby approximately three and a half times, from 4.3to 14.9 kg-N ha yr .y1 y1

(3) For SE1, the nitrate flux increased approxi-mately 4-fold, from 11.7 to 46.5 kg-N ha yr .y1 y1

(4) Overall, felling led to an average increasein nitrate flux during the major period of distur-bance of the podzol areas monitored by a factorof approximately four(on average a change of10–34 kg-N ha yr ).y1 y1


(5) With regards recovery after the felling dis-turbance, only the records for the lower Hore andsouth2Hore sites are sufficiently long to assess thechanges involved. Table 3 shows the net changesbased on a post-felling record comprising the last2 years of measurement(i.e. the periods post-felling and post the secondary climate influencedpeaks in 1996–1998). The results indicate amarked reduction in nitrate flux: from 6.9 kg-N ha yr pre-fell to 1.4 kg-N ha yr post-y1 y1 y1 y1

fell for the lower Hore and from 16.0kg-N ha yr pre-fell to 5.1 kg-N ha yry1 y1 y1 y1

post-fell for south2Hore.Correspondingly, for the gley sites:

● For the Tanllwyth main stem, the nitrate fluxpossibly increased by approximately 20%(achange from 11.9 to 14.3 kg-N ha yr). Asy1 y1

only a half of the catchment was felled, clearfell disturbance probably represents approxi-mately a 40% increase(i.e. a change from 11.9to 16.7 kg-N ha yr ).y1 y1

● For the Tan1f small drainage area, the nitrateflux increases by a factor of approximately sixtimes, a change from 2.3 to 14.1 kg-N ha yr .y1 y1

● Clearly there is a major discrepancy for nitratefluxes in the case of the main stem and smalldrainage areas in the Tanllwyth. These discrep-ancies are considered later in the paper.

For ammonium, there are three major observa-tions. Firstly, there is a strong uptake of the rainfallinput from the undisturbed catchments(83–97%).With felling, the loss is lower for the Tanllwythcatchment although uptake is still highly signifi-cant (61%). Secondly, the disturbance effects offelling for the Tanllwyth are to increase the ammo-nium flux from 0.2 to 2.2 kg NH –N ha yry1 y1

4

for the main catchment. However, it must beremembered that only 50% of the catchment wasfelled and hence the ammonium fluxes for themain catchment must be doubled(a change from0.2 to 4.2 kg NH –N ha yr ) and the estimatey1 y1

4

of percentage ammonium in the rainfall lost to thecatchment is greatly reduced to a value of approx-imately 26%. For the small drainage area for theTanllwyth, the change in clear fell disturbance flux

for ammonium is similar to that for the maincatchment(a change from 0.2 to 3.8 kg NH –4

N ha yr ). Thirdly, clearfelling changes iny1 y1

ammonium flux for the podzolic catchments aremuch smaller. The changes involved are 0.75–0.39, 0.22–0.35 and 0.15–0.46 kg NH –4

N ha yr for the lower Hore, south2Hore andy1 y1

SE1, respectively. There seems to be little changebetween pre- and post-felling levels.

11. Discussion

Dissolved inorganic nitrogen in the moorlandand forested catchments at Plynlimon are deter-mined by a complex set of hydrological andbiological reactions. There are two main inorganicforms of nitrogen in the streams, nitrate andammonium of which nitrate predominates. Thehydrobiogeochemical behaviour differs for thesetwo components. The salient points are discussedseparately below followed by a comment on dis-solved organic carbon.

12. Nitrate

There are strong seasonal variations in nitrogenconcentrations within the streams across all thecatchments. In terms of nitrate fluxes, there maybe a net uptake—to, a net balance or a net efflux—from the catchment relative to the rainfall andcloud water input depending upon the particularcatchment. The highest variations in nitrate con-centrations and fluxes are probably associated withclimate variability and felling activity. In the caseof climate, it seems that certain perhaps drieryears, such as 1996 and 1998, resulted in unusuallyhigh nitrate concentrations(Neal et al., 2001).However, there are lag effects involved and factorscome into play such as biomass production andturnover rates. Correspondingly, in terms of felling,there is strong evidence that felling disturbanceresults in an increase in nitrate concentration andflux for a few years after disturbance. After thistime, nitrate concentrations and fluxes revert backto pre-felling and then to even lower levels asuptake from the newly establishing vegetationincreases. With felling, the strong seasonal varia-


tions continue, but the magnitude of the changesincreases. The extent of the increase links to theextent of felling in a given year, the soil type andthe total area cropped. For the podzolic sitesmonitored, clear fell results in an increase inconcentration that averages 28–99mM l (rangey1

16–179mM l ) and flux that averages 7.6–23.9y1

(range 6.9–46.5 kg-N ha yr). With regards toy1 y1

the gley sites, there is contrasting behaviour at thesmall and large scales. At the small scale, there isa marked increase in nitrate concentration and flux(8–52mM l and 2–14 kg-N ha yr , respec-y1 y1 y1

tively), about the same as one of the podzolicsites, the south2Hore. However, the changes aremuch smaller for the main river system: for clearfell disturbance, ammonium concentrations changefrom 44 to 62mM l and fluxes change fromy1

11.9 to 16.7 kg-N ha yr , respectively. Thus ity1 y1

seems that either, the more local area studied isnot representative of the catchment as a whole orthe gley catchment at the larger scale is immobi-lising much of the nitrate generated with felling.Gleys certainly have the capacity to reduce nitrateconcentrations to ammonium. Indeed, for the smallcontrol site in the gley, concentrations and fluxesare lower than for any other catchment monitoredwhile the corresponding fell site provides thesecond lowest figure. Further, for gley sites ingeneral, there is a much higher near surface runofffactor because of the impermeableywaterloggednature of this soil type. Both factors can explainthe change. There is a clear difference for theTanllwyth between the higher nitrate releases atthe local scale compared with the large scale. Forthis to occur, it seems that there must be somenear- or within- main-channel uptake of nitrogenand this may well be related to phosphorus releasesduring felling promoting a biological response,phosphorus being the limiting nutrient at Plynli-mon. This aspect is discussed in a companionpaper to this by Neal et al.(2003). However, itmust always be bourn in mind that a very limitednumber of sites have been examined and that thereare marked uncertainties in flux measurement. Inorder to tie down what the actual changes are at acatchment to regional scale, a far greater samplingeffort is needed in terms of number of sitesmonitored, the period of monitoring and the fre-

quency of monitoring. However, set against this isthe practicality of making such measurements andthe relationship between precision and accuracy.Although emphasis has been placed on stream

water measurement, analysis of borehole dataclearly shows that transport of nitrate by ground-water is occurring. The extent of this flow cannotbe gauged here, as there is no hydrogeologicalinformation that allows assessment of flux trans-fers. There are a number of features to note in thisregard.(1) Studies of conservative chemical tracers

(chloride in particular) show a large damping inthe catchment indicating significant groundwaterdamping (Neal and Kirchner, 2000; Kirchner etal., 2000, 2001).(2) The groundwater system is dominated by

fracture flow(Neal et al., 1997a,b).(3) Nitrate concentration variations in the

groundwaters shows features that correspond withthese first two observations. In many cases, thenitrate concentrations increase at shallower depths.This is consistent with an increase in the input ofmore nitrate-laden near surface runoff during wet-ter conditions. This feature is most stronglyobserved for two sites subjected to felling thatshow a strong ‘spiky’ response at shallow depths.However, one groundwater site exhibits a markedincrease in nitrate concentration that persists for atleast 2 years without any change as a function ofgroundwater level. This feature is consistent witha large groundwater reservoir.

13. Ammonium

Ammonium inputs are largely retained withinthe Plynlimon catchments even with felling activ-ity, but ammonium concentrations in stream runoffshowed very erratic behaviour. Felling was fol-lowed by increased in ammonium concentrationsin the stream and this was most prevalent for thegley sites. The clearest indication of the increasewas for the small drainage areas in the gley wherethe felling response lasted for a relatively shortperiod with a sharp decay during the year offelling. However, the detailed mechanism ofchange is not clear, as the main stream did notshow a clear dynamic response.


Within the groundwaters, there was a slightlyhigher proportion of ammonium compared to thestreams and this would be expected in the sensethat there will be increased reduction potential dueto iron, manganese and sulphur within the bedrock.However, the dynamic response observed fornitrate as a function of level was not observed forthe groundwaters.

14. Dissolved organic nitrogen

Although in this paper the emphasis is oninorganic forms of nitrogen, it is important to flagthat nitrogen is also present in dissolved organicforms (DON) as well. Information on DON isfragmentary, but there have been some new region-al studies in the UK uplands including at Plynli-mon (Chapman et al., 1998, 1999, 2001). Theresults for Plynlimon (Chapman et al., 1999),based on monthly sampling of streams drainingmoorland and partly forested catchments fromMarch 1997 to March 1998 show the following:

● DON is on average higher in the streams drain-ing moorland (mean 7 and range 0–14mM l ) to those draining catchments with ay1

mix of moorland and forest(mean 4 and range0–13mM l ).y1

● On average, 86% of the total dissolved nitrogen(TDN) is present as nitrate for the streamsdraining the partly forested areas while thefigure for moorland areas is 72% with ammo-nium making up less than 6%.

● DON shows a yearly cycle for the streamsdraining both the moorland and partly forestedcatchments with the contribution of DON toTDN being higher in the summer(59 and 31%,respectively, May–October) than the winter(10and 3%, respectively, November–April).

● DON correlates with dissolved organic carbonfor streams draining the moorland but not theforest.

Thus, while inorganic forms of nitrogen domi-nate the TDN in the Plynlimon streams monitored,DON can still make up a significant part. Howimportant DON is to TDN across the catchmentin relation to area, temporal variations, soil typeand timber harvesting cannot be judged here. The

uncertainty involved is related to the lack ofdetailed direct measurements and an inability tocalculate values based on DOC measurements dueto the lack of correlation between DOC and DONfor the streams of concern.

15. Conclusion

This paper provides a major overview of thenitrate and ammonium dynamics within rainfall,cloud water streams and groundwaters for thePlynlimon catchments. The results show that thereare significant inputs of both nitrate and ammoni-um in rainfall and cloud water and these concen-trations show major changes from year to year.These inputs are largely ‘pollution’ related andconcentrations show strong dilution with increas-ing volume of catch, a ‘washout’ effect. Whilethere is no data for dissolved organic nitrogen,high levels seem unlikely given the lack of poten-tial sources and the low levels of DOC usuallyencountered. However, dissolved organic nitrogenmay be very high in relation to cycling throughthe vegetation owing to the large DOC levelsencountered. Further, the dissolved organic nitro-gen may well be increasing with time as shownby the increasing patterns observed for DOC inthroughfall. Within the catchment, the inputs arestrongly modified and the biological processesproduce a seasonal effect within the stream fornitrate with a variable input–output balancedepending upon particular circumstances. Nitrate,being in anionic form, is not adsorbed onto theexchange sites on the soil as(a) nitrate has a lowaffinity for adsorption onto surfaces and(b) thereare few anion exchange sites in the soil. Hence, itseems that there is an approximate net balancebetween biological uptake and leaching processessuch as vegetation-leakage from the canopy andorganic mater decomposition. For ammonium,there is a net uptake by the catchment and concen-trations show a marked ‘spiky’ stream responsethat cannot be easily explained within a process-based framework. Since there is a strong uptakeof the atmospheric input, ammonium seems to befertilizing the catchment and, presumably, it isbeing adsorbed by the soil onto cation exchangesites. The inorganic forms of nitrogen seem to


dominate in the stream, but there is also dissolvedorganic nitrogen, which may make up a significantpart of the TDN in the streams. Felling leads to adisruption of the biogeochemical cycle and to therelease of both nitrate and ammonium. There areboth similarities and differences in response tofelling of the podzolic and gley areas studied.Felling for the more oxic podzolic soils lead tolarge increases in nitrate concentrations and fluxesbut much smaller increases for ammonium. In thecase of nitrate, there is a strong uptake by thenewly developing biomass following felling andpossibly, after 10 or more years, the net flux gainby felling disturbance may well be balanced bysubsequent flux loss relative to the undisturbedmoorland and forested catchments as a new steadystate evolves. Felling of the more anoxic gley soilsleads to marked increases in both nitrate andammonium concentrations at the small scalealthough at the larger scale of the catchment outletthe nitrate efflux is much reduced possibly due toincreased biological activity associated with phos-phorus release.

16. Future research needs

The results presented in this paper provideimportant clues for the hydrobiogeochemical func-tioning of nitrogen within UK upland acidic andacid sensitive moorland and forested catchments.However, there is a clear need to investigate thedetails of the changes involved and undertake amore in-depth study of the processes and fluxtransfers involved through the hydrological cycle.These needs include:(1) The acquirement of much longer data runs.

This is required for two reasons. Firstly, there is aneed to establish what the long-term changes innitrate and ammonium fluxes are: the present studyshows that even after 9 years post-felling, thecatchments have not reached steady state condi-tions. Secondly, there is the need to determine towhat degree climatic influences affect concentra-tions and fluxes.(2) A more accurate determination of atmos-

pheric inputs and stream output fluxes. The pres-entation made here does not provide a full processbased assessment of input fluxes, as there are

major uncertainties with regards to cloud water,dry and gaseous inputs of inorganic nitrogen spe-cies. For stream output fluxes there needs to be amuch tighter statistically based analysis as theremay be significant error terms involved usingstandard methodologies both in terms of determin-ing yearly flux and trends in flux(Cooper andEvans, 2002; Cooper and Watts, 2002).(3) An in-depth assessment of dissolved organic

nitrogen. This is needed to allow a full assessmentof nitrogen export from the catchments, a base forestablishing an input-storage-output relationshipfor nitrogen in UK upland acidic and acid sensitivesystems.(4) Measurement of missing components related

to factors such as within-river processing of nitro-gen species and altitudinal influences on nitrogenfluxes is required to identify why there are site tosite variations in nitrogen concentrations andfluxes.(5) Assessment of nitrogen and ammonium

dynamics in gley catchments. Clearly, there areimportant differences between the processesobserved within small areas of gley catchmentsand the main stream. Issues of concern include therates of denitrification, the extent of within-riverprocessing, the linkage between nitrogen and phos-phorus dynamics and hydrologically driven trans-fers of nitrogen fluxes.

Appendix A:

Schemes 1a, 1b, 2a, 2b, 3ab, 3cd and 3e.

References

Bormann FH, Likens GE. Pattern and process in a forestedecosystem. New York: Springer, 1994. p. 226

Chapman PJ, Edwards AC, Reynolds B, Cresser MS, Neal C.Hydrology, Water Resources and Ecology in Headwaters.Proceeding of the HeadWater’98 Conference, MeramyMer-ano, Italy), HIS, 248, 1998. p. 443–50.

Chapman PJ, Edwards AC, Reynolds B, Neal C. The nitrogencomposition of streams draining grassland and forestedcatchments: influence of afforestation on the nitrogen cyclein upland ecosystems. Impact of land-use change on nutrientloads from diffuse sources. Proceeding of IUGG 99 Sym-posium, HS3, Birmingham, July 1999. IAHS, 257, 1999. p.17–26.


Scheme 1a. Nitrate time series for rainfall and upper Hore, lower Hore and lower Hafren streams.


Scheme 1b. Ammonium time series for rainfall and upper Hore, lower Hore and lower Hafren streams.


Scheme 2a. Nitrate time series for upper Hore, lower Hore, upper Hafren, lower Hafren and Tanllwyth streams.


Scheme 2b. Ammonium time series for upper Hore, lower Hore, upper Hafren, lower Hafren and Tanllwyth streams.


Scheme 3ab.(a) Nitrate and ammonium concentration time series for streams draining paired podzol fell and control catchments(SE1f and SE3c). (b) Nitrate and ammonium concentration time series for groundwater associated with a paired podzol fell andcontrol catchments(SE1f and SE3c).

Chapman PJ, Edwards AC, Cresser MS. The nitrogen com-position of streams in upland Scotland: some regional andseasonal differences. Sci Total Environ 2001;265:65–83.

Cooper DM, Evans CD. Constrained multivariate trend anal-ysis applied to water quality variables. Environmetrics2002;13(1):43–53.

Cooper DM, Watts CD. A comparison of river load estimationtechniques: an application to dissolved organic carbon.Environmetrics 2002;13:733–750.

Cryer . Atmospheric solute processes. In: Trudgill ST, editor.Solute processes. Chichester: Wiley, 1986. p. 15–85.

Hudson JA, Gilman K, Calder I. Land use and water qualityissues in the uplands with reference to the Plynlimon study.Hydrol Earth Syst Sci 1997a;1(3):389–397.

Hudson JA, Crane SB, Blackie JR. The Plynlimon waterbalance 1969–1995: the impact of forest and moorlandvegetation on evaporation and streamflow in upland catch-ments. Earth Syst Sci 1997b;1(3):409–427.

Kirchner JW, Feng XH, Neal C. Fractal stream chemistry and

its implications for contaminant transport in catchments.Nature 2000;403:524–527.

Kirchner JW, Feng XH, Neal C. Catchment-scale advectionand dispersion as a mechanism for fractal scaling in streamtracer concentrations. J Hydrol 2001;254:81–100.

Neal C, editor. Water quality of the Plynlimon Catchment(UK). Hydrol Earth Syst Sci 1997a;1(3):381–764.

Neal C. A view of water quality from the Plynlimon watershed.Hydrol Earth Syst Sci 1997b;1(3):743–754.

Neal C, Kirchner JW. Sodium and chloride levels in rainfall,mist, streamwater and groundwater at the Plynlimon catch-ments, mid-Wales: inferences on hydrological and chemicalcontrols. Hydrol Earth Syst Sci 2000;4(2):295–310.

Neal C, Robson AJ, Shand P, Edmunds WM, Dixon AJ,Buckley DK, Hill S, Harrow M, Neal M, Reynolds B. Theoccurrence of groundwater in the Lower Palaeozoic rocksof upland Central Wales. Hydrol Earth Syst Sci1997a;1(1):3–18.


Scheme 3cd.(c) Nitrate and ammonium concentration time series for streams draining paired gley fell and control catchments(Tan1f and Tan2c). (d) Nitrate and ammonium concentration time series for groundwater associated with paired gley fell and controlcatchments(Tan1f and Tan2c).

Neal C, Hill T, Alexander S, Reynolds B, Hill S, Dixon AJ,Harrow M, Neal M, Smith CJ. Stream water quality in acidsensitive upland areas, an example of potential water qualityremediation based on groundwater manipulation. HydrolEarth Syst Sci 1997b;1(1):185–196.

Neal C, Wilkinson J, Neal M, Harrow M, Wickham H, HillL, Morfitt C. The hydrochemistry of the River Severn,Plynlimon, mid-Wales. Hydrol Earth Syst Sci1997c;1(3):583–618.

Neal C, Reynolds B, Adamson JK, Stevens PA, Neal M,Harrow M, Hill S. Analysis of the impacts of major anionvariations on surface water acidity particularly with regardto conifer harvesting: case studies from Wales and NorthernEngland. Hydrol Earth Syst Sci 1998a;2(2–3):303–322.

Neal C, Reynolds B, Wilkinson J, Hill T, Neal M, Hill S,Harrow M. The impacts of conifer harvesting on runoffwater quality: a regional study for Wales. Hydrol Earth SystSci 1998b;2(2–3):323–344.

Neal C, Reynolds B, Neal M, Pugh B, Hill L, Wickham H.Long term changes in the water quality of rainfall, cloudwater and stream water for moorland, forested and clear-

felled catchments at Plynlimon, mid-Wales. Hydrol EarthSyst Sci 2001;5(3):459–476.

Neal C, Reynolds B, Neal M, Hughes S, Wickham H, Hill L,Rowland P, Pugh B. Soluble reactive phosphorus in rainfall,cloud water, throughfall, stemflow, soil waters, streamwatersand groundwaters for the Upper River Severn, Plynlimon,mid-Wales. Sci Total Environ 2003;314–316:99–120.

Reynolds B, Neal C, Hornung M, Stevens PA. Baseflowbuffering of stream water acidity in five mid-Wales catch-ments. J Hydrol 1986;87:167–185.

Reynolds B, Neal C, Hornung M, Hughes S, Stevens PA.Impact of afforestation on the soil solution chemistry ofstagnopodzols in mid-Wales. Water Air Soil Poll1988;38:55–70.

Reynolds B, Hornung M, Hughes S. Chemistry of streamsdraining grassland and forest catchments at Plynlimon, mid-Wales. J Hydrol Sci 1989;34:667–686.

Reynolds B, Emmett BA, Woods C. Variations in stream waternitrate concentrations and nitrogen budgets over 10 years ina headwater catchment in mid-Wales. J Hydrol1992;136:155–175.


Scheme 3e. Nitrate and ammonium concentration time seriesfor the south2Hore stream.

Reynolds B, Stevens PA, Hughes S, Parkinson JA, WeatherlyNS. Stream chemistry impacts of conifer harvesting in Welshcatchments. Water Air Soil Poll 1995;79:147–170.

RGAR. Acid deposition in the United Kingdom, 1992–1994.Fourth report of the Review Group on Acid Rain. Blackhorse Road, London, SE99 6TT: DoE Publications DespatchCentre; 1997. p. 176.

Stevens PA, Adamson JK, Reynolds B, Hornung M. Dissolvedinorganic nitrogen concentrations and fluxes in three BritishSitka spruce plantations. Plant Soil 1990;128:103–108.

Stevens PA, Norris DA, Sparks TH, Hodgson AL. The impactsof atmospheric N inputs on throughfall, soil and streamwater interactions for different aged forest and moorlandcatchments in Wales. Water Air Soil Poll 1994;73:297–317.

Stevens PA, Reynolds B, Hughes S, Norris DA, DickersonAL. Relationships between spruce plantation age, solute andsoil chemistry in Hafren Forest. Hydrol Earth Syst Sci1997;1:627–637.

UKAWRG. United Kingdom Acid Waters Review Group,Second Report. Acidity in United Kingdom fresh waters.London: Her Majesty’s Stationary Office; 1988. p. 1–61.

Wilkinson J, Reynolds B, Neal C, Hill S, Neal M, Harrow M.Major, minor and trace element composition of cloud waterand rainwater at Plynlimon, mid-Wales. Hydrol Earth SystSci 1997;1(3):557–570.

nitrogen in rainfall, cloud water, throughfall, stemflow, stream water and groundwater for the...

Documents