Variation of phosphorus loss from a small Catchment in south Devon, UK

Agriculture, Ecosystems and Environment 79 (2000) 143–157

Variation of phosphorus loss from a smallCatchment in south Devon, UK

Richard McDowell∗,1, Stephen TrudgillDepartment of Geography, University of Cambridge, Downing Place, Cambridge CB2 3EN, UK

Received 16 June 1999; received in revised form 23 November 1999; accepted 23 November 1999

Abstract

The application of fertilizers and manure in excess of plant requirements has resulted in an accumulation in soil phosphorus(P), and increased potential for P loss. To develop P-based catchment management plans, we need to be able to estimatethe impact of soil P on water draining a catchment. A 12-month investigation (August 1997–July 1998) determined thetemporal change of soil P forms and soluble reactive P (SRP) in stream discharge in a small catchment of mixed landuse(cereal crops (Triticum aestivumandHordeum sativum), root crops (Solanum tuberosumandBrassicasp.), grassland (Phleumpratense) and woodland (largelyCastanea sativa)), in south Devon, UK. This included monthly sampling of soils for sodiumbicarbonate extractable P (Olsen P), calcium chloride extractable P (CaCl2-P) on wet and air-dry soil, organic carbon andpH. Also available were weekly data for stream discharge and SRP concentration during 1987–1989 and 1994–1998, whichenabled an 8-year mean to be calculated for each month. All forms of soil P exhibited seasonal variation, with a late summermaximum and late winter minimum. Olsen P and CaCl2-P were related by a quantity–intensity relationship. Above a certainvalue of Olsen P, termed the change point, CaCl2-P increased more per unit Olsen P than below this point. The change pointremained virtually constant throughout the year, never deviating more than 5 mg kg−1 from a mean value of 31 mg kg−1 OlsenP. Changes in stream SRP concentrations for the monthly means for 8 years data were correlated only with CaCl2-P from drysoil. Plots of cumulative SRP export against cumulative discharge over the 8-year data set suggested that SRP loss was limitedby the supply of SRP from the soil matrix. Olsen P for root and cropping soils was twice that needed for maximum yields.Thus, to reduce SRP loss, P fertilizer applications should be stopped to allow Olsen P to decrease below the change point.The use of CaCl2-P and the change point has the potential to form the basis of simple environmental management planningat the catchment scale. © 2000 Elsevier Science B.V. All rights reserved.

Keywords:Phosphorus loss; Catchment; Change point; Soil P; SRP

∗ Corresponding author. Tel.:+44-1223-333399;fax: +44-1223-333392.E-mail address:[email protected] (R. McDowell)

1 Present address: USDA-ARS, Posture Systems and WatershedManagement Laboratory, Building 3702, Curtin Road, UniversityPark, PA 16802-3702, USA.

1. Introduction

Phosphorus (P) is a primary factor in the nutrition ofplants and the eutrophication of surface waters. Soilswith a high soil P concentration will generally giverise to high P concentrations in runoff (surface andsub-surface). Pote et al. (1996) showed that P concen-tration in surface runoff was closely correlated (r≈1)to water extractable P in the topsoil. Smith et al. (1995)

0167-8809/00/$ – see front matter © 2000 Elsevier Science B.V. All rights reserved.PII: S0167-8809(99)00154-1

144 R. McDowell, S. Trudgill / Agriculture, Ecosystems and Environment 79 (2000) 143–157

related the increase in soluble P in drainflow from agrassland catchment to a rise in soil Olsen P concentra-tions. Heckrath et al. (1995) showed that the concen-tration of dissolved (<0.45mm) reactive P, dissolvedorganic P and total P were linearly related to Olsen Pabove a certain point (at about 60 mg kg−1 Olsen P).This point, termed the ‘change point’, is also evidentwhen comparing molybdate reactive P (MRP), essen-tially inorganic P, in 0.01M CaCl2 extracts with OlsenP in laboratory extractions (Brookes et al., 1997). Thiscomparison has been suggested, in the absence of datafor P concentrations in drainage water, as an indicatorto predict soil P concentrations above which P lossesto drainage water become environmentally significant.

Seasonal changes in inorganic P (Pi) concentrationsextractable from field soils under various vegetationtypes is well known. A 3-fold increase in extractablePi during summer was found in an unfertilized plotin Scotland by Smith (1959). Concentrations of Piextracted using anion exchange resin were greatestduring the summer in arable and grassland ley plotsat Rothamsted and Woburn, UK (Garbouchev, 1966).Similar variation occurred in arable and grasslandplots sequentially extracted with resin, NaHCO3 andNaOH from sandy soils in Denmark (Magid andNielsen, 1992). Saunders and Metson (1971) andGoodwin et al. (1998) noted that seasonal variationof Pi extracted by anion exchange resin and Olsen’sreagent (NaHCO3) from improved grassland soilsexhibited a similar trend to Pi in 0.01M CaCl2 soilextracts and soil solution. A seasonal variation in wa-ter extractable Pi concentrations occurred in an arablesoil (Kuo and Jellum, 1987). Shand et al. (1994)showed the maximum concentration of Pi in soil so-lution of three P-deficient Cambisols in NE Scotlandoccurred in summer (August). Such seasonal varia-tion could affect the timing of soil sampling to assessfertilizer requirements, the amount of P available forplant uptake and the loss of P in runoff (surface orsub-surface).

The Slapton Wood catchment in south Devon, UKis a small (0.94 km2) second order tributary of theRiver Gara which drains into Slapton Ley, the largestbody of freshwater in south-west England. Since late1969, continual monitoring of stream discharge andweekly water sampling at this site, along with threeother catchments has occurred, aimed at quantifyinginputs of water, sediments and solutes into the lake. Of

the total annual runoff, only 1.15% is surface-runoff(Troake and Walling, 1973). During winter, a lag timeof several days occurs between rainfall and the maxi-mum baseflow in stream discharge. This is indicativeof a catchment where sub-surface runoff is favorable(Burt et al., 1983; Burt, 1988). It is therefore likelythat P concentrations in runoff may reflect soil P con-centrations if little water bypasses the soil in macro-pore flow. However, it is recognised that a significantamount of total P loss and stream P concentration maybe as particulate P (PP) or desorbed from PP in sur-face runoff during intense rainfall.

Seasonal dynamics in catchments cause changes instream P concentrations. Many researchers have es-tablished a linear relationship between the log of flowrates and the log of stream dissolved or soluble reac-tive P (SRP) concentrations (e.g., Lennox et al., 1997).Large SRP concentrations occur during periods of lowflow in summer (Leinweber, 1998). However, mostSRP load is lost during periods of high flow in winter.Xue et al. (1998) used the theory of first-order kinet-ics in a plot of cumulative P export versus cumulativeflow to describe P loss in tile drainage. They suggestedthat the exhaustion of P from the readily leachablesoil P pool caused a decrease in cumulative P exportcompared to cumulative flow. Hodgkinson and With-ers (1998) used the same type of plot to show thatmost P was lost during winter. During a single runoffevent, P concentrations will be affected by soil wa-ter residence times and dilution, but the potential forphosphorus loss in runoff and for plant nutrition willalways be affected by the amount of soil P, and thatwhich can be rapidly released into soil solution.

To develop P-based catchment management plans,we need to be able to estimate the impact of soil Pon water draining a catchment. This study aims tocompare and investigate the temporal change in soilP concentrations, an indicator for P loss (the changepoint) and P concentrations in runoff in the short term(1997–1998) and on average over 8 years (1987–1989and 1994–1998) by measuring: (1) the concentrationof key plant available and soil solution P fractions, (2)calculating the change point between 0.01M CaCl2and Olsen P concentrations and (3) SRP stream con-centration and load. Accompanying measurementsthought to influence the movement of P from the soilwere made of, pH, organic matter and the kinetics ofP release.

R. McDowell, S. Trudgill / Agriculture, Ecosystems and Environment 79 (2000) 143–157 145

2. Materials and methods

2.1. Site

The Slapton Wood catchment in south Devon,England, is the lowest tributary of the River Gara(Fig. 1). Impermeable Dartmouth slates underlie thesmall drainage basin (0.94 km2). The slates haveweathered to give a regolith of 1–2 m and freelydraining acid brown earth soils (Eutric Cambisol), ofpH 4–5, with 300–400 g kg−1 clay, 300–400 g kg−1

silt and the remainder being sand. The catchment isdominated above a 90 m contour by a plateau withslopes below 5◦, largely under arable farming. Below

Fig. 1. The Slapton Catchments.

90 m the valley side slopes are much steeper up to25◦ under grass or wood.

Climate data (rainfall and temperature) representa-tive of the catchment is measured at a nearby FieldCentre 800 m away. The catchment has a mild, moist,maritime climate with mean temperatures of 5.7◦C inFebruary and 15.8◦C in July and an annual mean tem-perature of 10.5◦C. On average, temperatures fall be-low freezing only 22 times a year. The long-term meanrainfall is quoted by Ratsey (1975) as 1051 mm per an-num. Mean annual rainfall for the period 1961–1985was 1035 mm. On average, 1 mm of rainfall falls 135days of the year and 10 mm of rainfall falls 35 daysof the year (Ratsey, 1975). Although the catchmentreceives enough moisture for agriculture, mean an-nual evaporation exceeds 400 mm resulting in droughtconditions during some warm months (Troake andWalling, 1973).

2.2. Stream analysis

Stream discharge was measured using a 120◦ thinplate V-notch weir and an Ott R16 stage recorder. Spotsamples of stream water, taken at weekly intervals arefiltered through a glass fibre filter paper (GF/C) andstored at 4◦C until analyzed for P. This is defined asSRP, essentially orthophosphate.

A continuous discharge and SRP data set for theperiod 1987–1989 and 1994–1998, without missingvalues of longer than 2 weeks is used in this study.Outliers in SRP and discharge (defined as >3 standarddeviations from the geometric mean) were removedfrom the data set. This accounted for less than 0.5%of all observations. Monthly SRP load averaged overthe 8 years was calculated using an interpolation pro-cedure (Walling and Webb, 1982):[∑n

i=1(CiQi)∑ni=1Qi

]QrK (1)

whereCi is the instantaneous individual SRP concen-tration (mg l−1), Qi is the instantaneous discharge attime of sampling (l s−1), Qr is the mean discharge forthe time of record (l s−1) andK is the conversion fac-tor to take account of period of record.

This method has been employed as an estimate ofnitrate load for the Slapton Wood catchment (Burtet al., 1988). It is not the best method identified by


Walling and Webb (1982) of calculating loads. How-ever, given the poor nature of rating curves betweenSRP and flow which could be used to estimate loadby extrapolation (r2 generally<0.10, data not given)and the frequency of sampling, this method was seenas most appropriate. However, it must be recognizedthat without more intensive sampling, some amountof error is to be expected.

2.3. Soil sampling and analyses

Six soil samples of the 0–7.5 cm depth of 10 fields(three grassland (Phleum pratense), cereal cropping(Triticum aestivumandHordeum sativum), root crop-ping fields (Solanum tuberosumand Brassica sp.)and one woodland field (largelyCastanea sativa))were taken at monthly intervals from August 1997 toJuly 1998, excluding January 1998. Each sample wassieved to 2 mm. SRP was determined in filtered 0.01MCaCl2 extracts of wet and air-dried soils (CaCl2-P)and 0.5M NaHCO3 (adjusted to pH 8.5 with NaOH[Olsen P]) extracts of air-dried soils. Soils had beenshaken for 30 min using soil:solution ratio of 1:20for Olsen P (Olsen et al., 1954) and 1:5 for CaCl2-P(Schofield, 1955). Soils were extracted field moistwith 0.01M CaCl2 in order to minimize disturbanceof soil solution chemistry. Results for 0.01M CaCl2extracts of wet soils are presented on an air-dry basisfollowing a gravimetric moisture determination. Mi-crobial biomass P was determined on moist soils ofthe cereal, root and grassland landuses from the Junesampling (not enough woodland soil was available),using the method of Brookes et al. (1982). Organiccarbon (C) was determined on oven dry soil (Grewalet al., 1991) and soil pH in water using a 1:2.5 soil:solution ratio. The kinetics of SRP release in 0.01MCaCl2 extracts of air-dry soils at either low and highOlsen P for each landuse from the June sampling wasdetermined on filtered extracts after 2, 10, 30, 120,300 and 1440 min and fitted to an expanded Elovichequation (Polyzopoulos et al., 1986):

Q = [ln(ab) + ln(t + c)]

b(2)

whereQ is the amount (mg kg−1) of released P at timet, t is the desorption time (min) anda, b and c areconstants.

The equation was fitted using nonlinear regression(SPSS v6.0) and the fit assessed through a linear plotof observed versus predicted values giving anr2 value.The r2 value is given, since nonlinear regression doesnot yield truer2 values.

A preliminary analysis of variance using Minitab(v8.0) showed that four groups of data could separateaccording to their landuse. The management of eachfield has remained relatively unchanged during the 8years of stream data. Data from each landuse werepooled (three fields each for soils under cereal crops,root crops and grassland), resulting in an acceptablestandard error of the mean commonly less than 5%from 18 samples covering an area of 1.2 ha. Samplesfrom the woodland landuse commonly had a standarderror less than 9%.

3. Results

3.1. Soil characteristics

Table 1 gives the mean, standard error of the meanand range of pH and organic C (g kg−1) for each lan-duse during the study period and microbial biomass P(mg kg−1 Olsen P) for the cereal, root and grasslandsoils from the June sampling. No systematic or sig-nificant variation in pH of organic C between monthswas noted. Microbial biomass P was greatest in thegrassland soils and greater in the root soils than thecereal soils. During the study period, pH ranged from3.47 in the woodland soil to 7.10 in soil under cere-als. Organic C ranged from 0.26 g kg−1 in a grasslandsoil to 1.87 g kg−1 in a woodland soil. On average,soils under root and cereal cropping had a significantlyhigher pH (p<0.01) and lower organic C than soilsunder grassland or woodland. One field that had beenused for cropping 3 years before being sown to grasscaused low values of organic C in grassland.

Curve fits and data for the kinetics of P desorptionis shown in Fig. 2. The kinetics of P release was de-scribed by the expanded Elovich equation. While it isbelieved that no one equation can describe the desorp-tion kinetics of all soils, the expanded Elovich equa-tion has been found to fit desorption rates well over awide range of soil types (Raven and Hossner, 1994).The r2 values for any of the fits were never less than


Table 1Mean (± standard error of the mean), with range given below, for topsoil (0–7.5 cm) pH in water (1:2.5 ratio), organic C and microbialbiomass P for each landuse

Landuse pH Organic C (g kg−1) Microbial biomass P(mg kg−1 Olsen P)

Cereals 6.23 (0.08) 0.41 (0.20) 26.9 (2.9)5.22–7.10 0.28–0.52 18.7–39.5

Root 6.01 (0.06) 0.38 (0.17) 32.7 (2.4)5.12–6.97 0.29–0.51 28.5–44.5

Grassland 5.68 (0.04) 0.52 (0.20) 42.5 (5.4)4.83–6.03 0.26–0.85 38.3–55.1

Woodland 4.63 (0.12) 1.08 (0.58) n.d.a

3.47–5.17 0.74–1.87

a n.d.=not determined.

0.951. The quantity of desorbed P and the rate of Pdesorbed was assessed through the quantities and ratesof P released after 30 and 1440 min (Table 2). Thesedata show that more P is released and at a faster ratewith increasing Olsen P.

Fig. 2. Concentration of CaCl2-P with time for low and high Olsen P grassland, cereal and root soils (symbols) fitted to the expandedElovich equation (line).

3.2. Soil moisture and temperature

Soil moisture, temperature and rainfall for eachmonth during the study period are shown in Fig. 3.A total of 1175 mm of rain fell on the catchment,


Table 2The initial and final quantity and rate of desorbed P for soils of each landuse at one or two Olsen P concentrations

Landuse r2 value Olsen P Desorbed soil P Soil P desorption rates(mg kg−1) (mg kg−1 h−1) (mg kg−1 h−1)

Initiala Finalb Initial Final

Grassland 0.972∗∗∗ 20 0.376 0.940 0.752 0.0390.969∗∗∗ 39 0.865 1.729 1.316 0.054

Cereal 0.989∗∗∗ 23 0.418 0.964 0.836 0.0400.951∗∗∗ 47 1.234 1.512 2.468 0.063

Root 0.990∗∗∗ 26 0.451 0.977 0.902 0.0400.982∗∗∗ 55 1.429 1.767 2.857 0.074

Woodland 0.997∗∗∗ 8 0.376 0.602 0.752 0.025

a Initial=after 30 min.b Final=after 1440 min.∗∗∗ Indicates significant at thep<0.01 level.

compared to a long-term mean of 1051 mm (Ratsey,1975). Despite the greater than average annual rain-fall, only 9.4 mm fell during February compared toa long-term average of 86 mm. This caused low soil

Fig. 3. Monthly (± standard error of the mean (SEM) I) soil moisture for each landuse, rainfall and soil temperature at 25 cm.

moisture content during February in the cereal androot landuses. No cultivation and the presence of apermanent root mat or canopy in the grassland andwoodland soils imparts a resilience to low rainfall by


decreasing evaporative soil moisture loss. In contrastto February, greater than average rainfall (120 mmfor November and 122 mm for January) fell duringNovember and January affecting soil moisture espe-cially in the grassland landuse. The seasonal patternin soil temperature at 25 cm was normal, except for alower than average temperature for April. Soil tem-perature was greatest in August and lowest in January.

3.3. Soil P forms

The monthly concentrations and variation in OlsenP and wet and dry CaCl2-P for each landuse, isshown in Fig. 4. Mean values of Olsen P withineach landuse generally decreased in order fromroot>cereal>grassland>woodland for each month.Within each landuse there is a clear seasonal patternin Olsen P concentration, which has been noted pre-viously, and related to the control and release of P bythe microbial biomass (Seeling and Zasoski, 1993).The maximum concentration of Olsen P occurred ineither July or August, while the lowest concentrationoccurred in winter.

Phosphorus extracted from wet and dry soil by0.01M CaCl2 was an order of magnitude lower thanOlsen P. Mean values of wet and dry CaCl2-P withineach landuse generally decreased in order fromroot>cereal>grassland>woodland for each month.Both wet and dry CaCl2-P concentrations exhibiteda seasonal variation, with a summer maximum anda winter minimum. The exception was the wood-land soil that had no obvious pattern. The variationand difficulty in detecting the low concentrations ex-tracted may have caused this. There was a rise in wetCaCl2-P in the root, cereal and grassland soils, whichcorresponded to the application of superphosphatefertilizer in March. However, the seasonal variationand summer maxima in dry CaCl2-P and Olsen Pconcentrations for root, cereal and grassland soils isprobably not caused by a spring fertilizer applicationsince neither rose during March or April.

3.4. Change points

From Fig. 4 it is evident that an approximate 6-foldincrease in Olsen P between woodland and root soilsfor August is paralleled by a 6-fold increase in dry

CaCl2-P. However, an approximate 5-fold increasein Olsen P between woodland and grassland soils inFebruary only gives an approximate 3-fold increase indry CaCl2-P. This is caused by the non-linear quantity(Olsen P)–intensity (CaCl2-P) (Q/I) relationship thatexists between the two measurements (McDowell andCondron, 1999). This relationship can be described bysorption isotherms (e.g. Freundlich), but a split-linemodel has also been used (Brookes et al., 1997)which gives a change point above which environmen-tally significant levels of P may be lost. This changepoint is attractive as a management tool for farmersand policy makers alike. Fitting a split-line modelto monthly dry CaCl2-P data shows that the changepoint has a mean value of 31 mg kg−1 Olsen P anda range of 26–36 mg kg−1 Olsen P (Table 3). Thereis no apparent seasonal variation in the value of thechange point. However, the number of soil samplesthat lie above the change point varies according toseason. For example, Fig. 5 shows the split line modelfitted to data from February and June. The summermaximum associated with Olsen P and dry CaCl2-Presults in more soil samples above the change point,and consequently at risk of accelerated P loss (butrequiring moving water to be lost), than soils abovethe change point in winter. However, high flow ratesin winter means that more P is lost than in summermonths (Figs. 4, 6 and 7). Although following thesame pattern as dry CaCl2-P, much more variationwas associated with fitting the split-line model to cal-culate change points for wet CaCl2-P against Olsen P(data not shown).

3.5. Seasonal variation in stream P concentrationand load

Monthly SRP concentration in stream discharge forthe 1997–1998 water year (beginning in August) andfor the mean of 8 years (1987–1989 and 1994–1998),along with the P load for the mean of 8 years is seenin Fig. 4. Cumulative discharge and P load for theSlapton Wood stream for the mean of 8 years and forthe 1994–1995 and 1997–1998 water year (beginning1 August) is seen in Fig. 6. These data show that P loadparallels discharge and that the major loss of P occursduring winter when discharge is at its maximum. Inthe 1994–1995 water year, high discharge rates during


Fig. 4. Monthly concentration (± standard error of the mean (SEM) I) and variation in Olsen P and wet and dry CaCl2-P for each landuse,along with the monthly SRP concentration and load for the mean of 8 years stream discharge and monthly SRP concentrations for the1997–1998 water year.

January and February generated 40% of the annualload. In contrast, low discharge rates during the samemonths in 1998 generated 19% of the annual P load(Fig. 6). The mean annual loss of P over 8 years isequivalent to about 7.5% of the P fertilizer applied tothe catchment.

Temporal variation in discharge P concentration isseen in Fig. 4 along with soil P concentration. Alongwith the large variation in mean values, there appearsto be a late summer maximum and late winter min-imum in discharge P concentration. The correlationbetween discharge SRP concentration and P in soil


Fig. 5. Split line model fitted to CaCl2-P vs Olsen P for the February and June soil sampling. Numbers indicate the change point for eachmonth and the dotted lines the total range of change points over the 12-month sampling period.

Table 3Monthly calculated change points for theQ/I relationship of OlsenP vs dry CaCl2-P

Month Change point value(mg kg−1 Olsen P)

August 1997 33September 1997 31October 1997 26November 1997 26December 1997 36January 1998 n.d.a

February 1998 28March 1998 33April 1998 32May 1998 30June 1998 34July 1998 33

a n.d.=not determined.

extracts is given in Table 4. These data show that al-though the cause and effects relationships are not clear,and that P in soil extracts do not necessary representthe catchment as a whole, they suggest that changesin mean stream SRP concentration over an 8 year pe-riod were related to changes in CaCl2-P concentra-tion. Correlations of P in soil extracts with stream SRPconcentrations during the August 1997–July 1998 pe-riod were low and may have been caused by a combi-nation of unusually low rainfall during February andhigh variation in stream SRP concentration associatedwith the low number of samples (n=47) compared tothe 8 year period (n=300).

Variation in P export for the mean of 8 years andfor the water year of 1994–1995 and 1997–1998 isshown in Fig. 7 as a plot of cumulative SRP export


Fig. 6. Cumulative SRP export and flow for the mean of 8 years, the 1994–95 and 1997–98 water years.

against cumulative discharge. Variation from a linearrelationship occurs twice around late autumn and latewinter. Cumulative SRP export plateaus where P sup-ply is becoming exhausted. This effect is not as dis-tinct during the 1997–1998 water year compared tothe 1994–1995 water year possibly due to unusuallylow rainfall and discharge in February 1998.

4. Discussion

4.1. Seasonal variation of P forms

Fluctuations in the concentration of soil P arecaused by several factors. Increases arise from theaddition of inorganic fertilizers or organic manures

and the decomposition of plant material and organicmatter by microbial activity. Decreases are caused bythe uptake of P by crops, by leaching and the immobi-lization of P in microbes and soil constituents. Thesein-turn are affected by climatic conditions such astemperature and rainfall and soil characteristics suchas pH and variation in soil mineral constitution.

The application of superphosphate during Marchwas reflected in a rise in wet CaCl2-P, followed by adrop in April. Blakemore (1966) showed that the appli-cation of fertilizer to a silty-clay loam soil at Rotham-sted (Harpenden, UK) reacted quickly and only raisedCaCl2 extractable P above pre-application concentra-tions for 1 or 2 months. Increasing the moisture con-tent of field moist soils (wetting-up), and extractingthe soils within a few days of collection is expected to


Fig. 7. Cumulative SRP export against cumulative discharge for the mean of 8 years data (A), the 1994–95 and 1997–98 water years (B).

minimize any extreme changes in soil solution chem-istry (Qian and Wolt, 1990). The least disturbance tosoil solution chemistry would be expected in winterand spring months when soil moisture content is great-est (Fig. 3).

Air-drying soil is known to affect the release ofP. The death of microorganisms has been reported tocause an increase in NaHCO3-extractable P (Brookes

Table 4Correlation coefficients for either mean monthly stream P discharge of 8 years of the 1997–1998 year and Olsen P and wet and dryCaCl2-P for each landuse

Landuse Olsen P (mg kg−1) Dry CaCl2-P (mg l−1) Wet CaCl2-P (mg l−1)

8 year mean monthly stream P dischargeCereal 0.36 0.66∗ 0.64∗Root 0.49 0.68∗ 0.63∗Grassland 0.15 0.60∗ 0.59Woodland 0.11 0.64∗ 0.32

1997–1998 mean monthly stream P dischargeCereal 0.30 0.09 0.30Root 0.09 0.12 0.27Grassland 0.04 0.05 0.31Woodland 0.05 0.20 0.12

∗ Indicates significant at thep<0.05 level.

et al., 1982). Sparling et al. (1987) showed that a con-siderable proportion of Olsen P might be adsorbed Preleased from the microbial biomass during air-drying.Re-wetting air-dried soil also stimulates microbial ac-tivity (Hunt et al., 1989) and results in the mineral-ization of organic N (Larsen and Widdowson, 1968).However, P behaves differently, and released P maybe taken up by microorganisms and released later


(Birch, 1964). Grassland soils contained more organicC and extractable microbial biomass P than either thecereal or root soils (Table 1). However, in generalCaCl2-P from dry soils was less than CaCl2-P fromwet soils. Dehydration of soil has been shown to in-crease P adsorption (Haynes and Swift, 1985; Bram-ley et al., 1992; Baskaran et al., 1994). Olsen andCourt (1983) suggested that re-wetting exposes newsurfaces containing native P and unreacted adsorptionsites. Marked increases in CaCl2-P are therefore notexpected, unless unreacted sites near saturation.

In the field, soil moisture content changes slowly,especially in those landuses with a crop canopy or rootmat. This would allow the microbial biomass time toadapt. Indeed, Tate et al. (1991) has shown that noseasonal variation occurred in the microbial biomassof two pasture soils of different P fertility. Magid andNielsen (1992) proposed that physiochemical changesassociated with soil moisture might mask any bio-logical cycling. Several authors have noted a winterminimum and summer maximum in P concentrationsoccur in coarse textured soils where a large part ofthe soil volume dries out during the summer (Smith,1959; Weaver et al., 1988; Magid and Nielsen, 1992).A winter maximum and summer minimum in soil Pconcentrations have been noted in fine textured soils,where P concentrations may be controlled by the re-duction and release of P from ferric hydroxides duringwet months (Jensen et al., 1998). The soils at SlaptonWood are weekly structured, loose and friable, but alsocontain well-marked biopores and are free draining.Both preferential flow and soil moisture deficits areknown to occur at Slapton Wood (Coles and Trudg-ill, 1985). These physical effects coupled with slowplant growth or even death during periods of low soilmoisture in warmer months will cause P to becomeconcentrated in the soil solution.

4.2. Seasonal variation in P runoff

Soluble reactive P in stream runoff is a directfunction of the concentration of available soil P forleaching and water flow (surface and sub-surface). Ifneither is present then SRP loss will not occur. Fig.7 shows that loss of SRP into the stream is directlyaffected by discharge and Table 4 shows that the con-centration of SRP in the stream from an 8 year period

is correlated to the concentration of CaCl2-P extractedfrom dry soil and to a lesser extent CaCl2-P extractedfrom cereal and root wet soils. Neither, Olsen P orCaCl2-P extracted from wet grassland or woodlandsoils were significantly correlated to the concentra-tion of SRP in stream discharge during 1997–1998 orthe 8 year mean, nor was CaCl2-P extracted from wetor dry soil correlated to the concentration of SRP instream discharge from 1997–1998 (Table 3). When asoil is enriched with P fertilizer or leached of P, thereare changes in both the concentration of P in solution(intensity) and the supporting labile P pool (quantity).However, the amount of change in either differs bythe slope of the sorption isotherm orQ/I relationshipbetween the two, which is an expression of the soilsP buffer capacity (Bache and Williams, 1971). OlsenP represents a quantity measurement of P associatedwith Al and Fe hydroxides and Ca phases (Schoenauand Karamanos, 1993), whereas CaCl2-P representsthe concentration of P immediately available for plantuptake or lost by leaching.

Since soil extracted with CaCl2-P measures the con-centration of P in the soil solution, we may then expectthis to be sensitive to short term changes in soil condi-tions and therefore better correlated to the 1997–1998year than the mean of the 8 years SRP stream data.This was not the case. Rainfall during the 1997–1998period was unusual, characterized by very low rain-fall during February and higher than average rainfallin November and January. This was reflected in dis-charge (Fig. 6). However, concentrations of CaCl2-Pwere similar to SRP in stream discharge for the meanof 8 years. This implies that readily leachable soil Phas maintained a memory of soil physical conditionsanalogous to the mean of 8 years. Discharge from theSlapton Wood catchment is characterized by a largeamount of base flow, supplied with P by the soil ma-trix. Physical conditions in the soil matrix control-ling P concentration such as soil moisture will changeslowly especially in those soils with a root mat or cropcanopy (Fig. 3).

During the summer months more soils were abovethe change point and consequently at a much higherCaCl2-P concentration than during winter. At this timedifferences between CaCl2 extractions of wet and drysoil is lowest. The kinetics of P release (Table 2, Fig. 2)shows that soils under root cropping were quickest torelease P. This landuse also has the greatest correlation


to SRP concentrations in stream discharge. However,while this suggests that soils under root cropping aremost closely linked to SRP in stream discharge, with-out knowing the proportion of water flowing throughroot soils and the influence of grassland soils nearerthe stream (Fig. 1), we cannot make any prediction forthe amount of SRP supplied by them.

Plots of cumulative SRP export against cumu-lative flow show two contrasting situations for the1994–1995 water year and the 1997–1998 water yearcompared to the mean of 8 years (Fig. 7). The shapeof the plot for the 1994–1995 water year is best de-scribed by linking two or three first order kineticcurves, where the rapid loss of SRP is stemmed bySRP supply and the curve begins to plateau. The poolof readily leachable P then begins to be replenished bydesorption, mineralization and fertilization ready forthe next period of rapid loss. The major period of lossduring the 1994–1995 year and the mean of 8 yearsoccur during late autumn, when soil P concentrationsand rainfall are high. Compared to 1994–1995 andthe mean of 8 years data, the plot for the 1997–1998water year is nearly linear, especially during winterwhen most SRP is lost. This suggests that SRP exportfrom the catchment in 1997–1998 is not related to theamount of SRP that was in the catchment.

4.3. Management implications

It is possible to present correlations betweenchanges in landuse, climate and management on theamount and concentration of SRP in streams (Smithet al., 1995). However, if we are to create predictivemodels or policy tools then the causal links, whichdepend upon the hydro-chemical processes withinthe soil, must be understood. For practical reasonsthere is also a need to maximize land area describedby the least number of soil samples, but within anacceptable standard of error. This is critical if theresults are to be used in a farm scale environmentalmanagement tool such as a ‘P index’ (Gburek et al.,1996), or used at larger scales within a GIS system.The results presented here show that for cereal androot landuses there is a common change point, abovewhich P intensity increases greatly compared to plantavailable Olsen P, than if below. There is some ev-idence to show that a change point also exists for

grassland soils, which is affected by pH (McDowelland Condron, 1999). Any predictive tool that is basedon plant-available (Olsen) P is presented with a prob-lem, namely ‘what happens once the change pointis exceeded?’ Data presented here shows that whilethe increase or decrease in P intensity (CaCl2-P) andquantity (Olsen P) varies according to season, thereis a constant change point.

Further work is required to assess the variability ofP intensity with topography and the sensitivity of Plosses. Clearly we need to understand where withina catchment a soil sample of high P intensity is ofconcern. The data from this site suggests that on av-erage (over 8 years), P transport within the catchmentis limited by SRP supply rather than by water flow(Fig. 6.). Soils, which are ‘high risk’ in terms of sup-plying P, are those above the change point with rapidP release kinetics. In general, the root and some ofthe cereal soils were well above the Olsen P recom-mended for maximum yield for potatoes (Solanumtuberosum) (25 mg kg−1 Olsen P), sugar beet (Betavulgaris) (20 mg kg−1 Olsen P) and winter wheat(Triticum aestivum) (20 mg kg−1 Olsen P) (Johnstonet al., 1986). Consequently, the first step to mitigat-ing P losses within this catchment would be to allowOlsen P to decrease to half their present values bystopping P fertilizer application.

5. Conclusions

This study has found that all measurements of soilP (Olsen P and wet and dry CaCl2-P) exhibited a sea-sonal variation, with a maximum concentration in latesummer and a minimum concentration in late winter.Olsen P and CaCl2-P form a quantity–intensity rela-tionship that exhibited a change point in Olsen P, abovewhich CaCl2-P increases more per unit Olsen P thanif below. The change point of 31 mg kg−1 Olsen P didnot change with season, but the number of CaCl2-Pdata points above the change point increased in thesummer. Changes in stream SRP concentrations forthe monthly mean of 8 years were correlated withCaCl2-P in dry soil. A plot of cumulative SRP exportagainst cumulative discharge suggested that SRP ex-port is limited by the supply of SRP from the soil ma-trix. The Olsen P for root and cropping soils is twicethat needed for maximum yields and approximately


33% greater than the change point. Thus, to reduceSRP loss, P fertilizer applications should be stoppedto allow Olsen P to decrease below the change point.The use of CaCl2-P and the change point has the po-tential to form the basis of simple environmental man-agement planning at the catchment scale.

Acknowledgements

We thank AGMARDT (New Zealand) and St. JohnsCollege, Cambridge (UK) for funding and J.J. Cocksand sons for land access.

References

Bache, B.W., Williams, E.G., 1971. A phosphate sorption indexfor soils. J. Soil Sci. 22, 289–301.

Baskaran, S., Bolan, N.S., Rahman, A., Tillman, R.W., Macgregor,A.N., 1994. Effect of drying of soil on the adsorption andleaching of phosphate and 2,4-Dichlorophenoxyacetic acid.Aust. J. Soil Res. 32, 491–502.

Birch, H.F., 1964. The effect of 2:4-Dinitro-phenol on phosphorustransformations during humus decomposition. Plant Soil 21,391–394.

Blakemore, M., 1966. Seasonal changes in the amounts ofphosphorus and potassium dissolved from soils by dilutecalcium chloride solutions. J. Agric. Sci. 66, 139–146.

Bramley, R.G.V., Barrow, N.J., Shaw, T.C., 1992. The reactionbetween phosphate and dry soil I. The effect of time,temperature and dryness. J. Soil Sci. 43, 749–758.

Brookes, P.C., Heckrath, G., De Smet, J., Hofman, G.,Vanderdeelen, J., 1997. Losses of phosphorus in drainage water.In: Tunney, H., Carton, O.T., Brookes, P.C., Johnston, A.E.(Eds.), Phosphorus Loss from Soil to Water. CAB InternationalPress, Cambridge, England, pp. 253–272.

Brookes, P.C., Powlson, D.S., Jenkinson, D.S., 1982. Measurementof microbial biomass phosphorus in soil. Soil Biol. Biochem.14, 319–329.

Burt, T.P., Butcher, D.P., Coles, N., Thomas, A.D., 1983. Thenatural history of Slapton Ley nature reserve XV. Hydrologicalprocesses in the Slapton Wood catchment. Field Stud. 5, 731–752.

Burt, T.P., Arkell, B.P., Trudgill, S.T., Walling, D.E., 1988. Streamnitrate levels in a small catchment in south west England overa period of 15 years (1970–1985). Hydrol. Process 2, 267–284.

Burt, T.P., 1988. Seasonality of subsurface flow and nitrateleaching. Catena Sup. 12, 59–65.

Coles, N., Trudgill, S.T., 1985. The movement of nitrate fertiliserfrom the soil surface to drainage waters by preferential flow inweakly structured soils, Slapton, South Devon. Agric. Ecosyst.Environ. 13, 241–259.

Garbouchev, I.P., 1966. Changes occurring during a year in solublephosphorus and potassium in soil under crops in rotationexperiments at Rothamsted, Woburn and Saxmundham. J. Agric.Sci. 66, 399–412.

Gburek, W.J., Sharpley, A.N., Pionke, H.B., 1996. Identificationof critical source areas for phosphorus export from agriculturalcatchments. In: Anderson, M.G., Brooks, S.M. (Eds.), Advancesin Hillslope Processes Vol. 1. Wiley, Chichester, UK, pp.263–282.

Goodwin, M.J., Parkinson, R.J., Williams, E.N.D., Tallowin,J.R.B., 1998. Soil phosphorus extractability and uptake inCirsio-Molinietum fen-meadow and an adjacentHolcus lanatuspasture on the culm measures, north Devon, UK. Agric. Ecosyst.Environ. 70, 169–179.

Grewal, K.S., Buchan, G.D., Sherlock, R.R., 1991. A comparisonof three methods of organic carbon determination in some NewZealand soils. J. Soil Sci. 42, 251–257.

Haynes, R.J., Swift, R.S., 1985. Effects of air-drying on theadsorption and desorption of phosphate and levels of extractablephosphate in a group of New Zealand acid soils. Geoderma.35, 145–157.

Heckrath, G., Brookes, P.C., Poulton, P.R., Goulding, K.W.T., 1995.Phosphorus leaching from soils containing different phosphorusconcentrations in the Broadbalk experiment. J. Environ. Qual.24, 904–910.

Hodgkinson, R.A., Withers, P.J., 1998. Tile drains as a pathway forphosphorus loss in small catchments. Practical and Innovativemeasure for the control of agricultural phosphorus losses towater. OECD Workshop, Greenmount College of Agricultureand Horticulture, Northern Ireland, pp. 150–151.

Hunt, H.W., Elliot, E.T., Walter, D.E., 1989. Inferring trophictransfers from pulse-dynamics in detrital food webs. Plant Soil115, 247–259.

Jensen, M.B., Hansen, H.C.B., Nielsen, N.E., Magid, J., 1998.Phosphate mobilization and immobilization in two soilsincubated under simulated reducing conditions. Acta Agric.Scand., Sect. B, Soil Plant Sci. 48, 11–17.

Johnston, A.E., Lane, P.W., Mattingly, G.E.G., Poulton, P.R., 1986.Effects of soil and fertilizer P on yields of potatoes, sugarbeet, barley and winter wheat on a sandy clay loam soil atSaxmundham Suffolk. J. Agric. Sci. Camb. 106, 155–167.

Kuo, S., Jellum, E.J., 1987. Influence of soil characteristicsand environmental conditions on seasonal variations ofwater-soluble phosphate in soils. Soil Sci. 143, 257–263.

Larsen, S., Widdowson, A.E., 1968. Chemical composition of soilsolution. J. Sci. Food Agric. 19, 693–695.

Leinweber, P., 1998. Phosphorus in soil and surface water from theHarle catchment: results of a two-year monitoring programme.Practical and Innovative measure for the control of agriculturalphosphorus losses to water. OECD Workshop, GreenmountCollege of Agriculture and Horticulture, Northern Ireland, pp.80–81.

Lennox, S.D., Foy, R.H., Smith, R.V., Jordan, C., 1997. Estimatingthe contribution from agriculture to the phosphorus load insurface water In: Tunney, H., Carton, O.T., Brookes, P.C.,Johnston, A.E. (Eds.), Phosphorus loss from soil to water. CABInternational Press, Oxon, U.K. pp. 55–76.


Magid, J., Nielsen, N.E., 1992. Seaonal variation in organic andinorganic phosphorus fractions of temperature-climate sandysoils. Plant Soil 144, 155–165.

McDowell, R.W., Condron, L.M., 1999. Developing and predictorfor phosphorus loss from soil. In: Currie, L.D. (Ed.), BestSoil Management Practices for Production (Fertilizer and LimeResearch Centre 12th Annual Workshop). Massey University,Palmerston North, New Zealand, pp. 153–164.

Olsen, R., Court, M.N., 1983. Effect of wetting and drying of soilson phosphate adsorption and resin extraction of soil phosphate.J. Soil Sci. 33, 709–717.

Olsen, S.R., Cole, C.V., Watanabe, F.S., Dean, L.A., 1954.Estimation of available phosphorus in soils by extraction withsodium bicarbonate. USDA Circ. 939, US Gov. Print Office,Washington, USA.

Polyzopoulos, N.A., Keramidas, V.Z., Pavlatou, A., 1986. On thelimitations of the simplified Elovich equation in describing thekinetics of phosphate sorption and release from soils. J. SoilSci. 37, 81–87.

Pote, D.H., Daniel, T.C., Sharpley, A.N., Moore Jr, P.A., Edwards,D.R., Nichols, D.J., 1996. Relating extactable soil phosphorusto phosphorus losses in runoff. J. Environ. Qual. 60, 855–859.

Qian, P., Wolt, J.D., 1990. Effects of drying and time of incubationon the composition of displaced soil solution. Soil Sci. 149,367–374.

Ratsey, S., 1975. The climate at Slapton Ley. Field Stud. 4, 191–206.

Saunders, W.M.H., Metson, A.J., 1971. Seasonal variation ofphosphorus in soil and pasture. NZ J. Agric. Res. 14, 307–328.

Schoenau, J.J., Karamanos, R.E., 1993. Sodium bicarbonate-extractable P, K, and N. In: Carter, M.R. (Ed.), Soil Samplingand Methods of Analysis, Lewis Publishers, Boca Raton, pp.51–58.

Schofield, R.K., 1955. Can a precise meaning be given to‘avialable’ soil phosphorus. Soils Fertil. 18, 373–375.

Seeling, B., Zasoski, R.J., 1993. Microbial effects in maintainingorganic and inorganic solution phosphorus concentrations in agrassland topsoil. Plant Soil 148, 277–284.

Shand, C.A., Macklon, A.E.S., Edwards, A.C., Smith, S., 1994.Inorganic and organic P in soil solutions from three uplandsoils. Plant Soil 159, 255–264.

Smith, A.M., 1959. Soil analysis and fertilizer recommendation.Proc. Fertil. Soc. 57, 1–40.

Smith, R.V., Lennox, S.D., Jordan, C.J., Foy, R.H., McHale, E.,1995. Increase in soluble phosphorus transported in drainflowfrom a grassland catchment in response to soil phosphorusaccumulation. Soil Use Manage. 11, 204–209.

Sparling, G.P., Milne, J.D.G., Vincent, K.W., 1987. Effectof soil moisture regime on the microbial contribution toOlsen phosphorus values. NZ J. Agric. Res. 30, 79–84.

Tate, K.R., Speir, T.W., Ross, D.J., Parfit, R.L., Whale, K.N.,Cowling, J.C., 1991. Temporal variations in some plant andsoil P pools in two pasture soils of widely different P fertilitystatus. Plant Soil 132, 219–232.

Troake, R.P., Walling, D.E., 1973. The natural history of SlaptonLey nature reserve VII. The hydrology of the Slapton Woodstream. Field Stud. 3, 719–740.

Walling, D.E., Webb, B.W., 1982. The design of samplingprogrammes for studying catchment nutrient dynamics. Proc.Hydrology of Research Basins, Bern, Germany. Sonderh.Landeshydrologie, 747–758.

Weaver, D.M., Ritchie, G.S.P., Anderson, G.C., Deeley, D.M.,1988. Phosphorus leaching in sandy soils I. Short-term effectsof fertilizer applications and environmental conditions. Aust. J.Soil Res. 26, 177–190.

Xue, Y., David, M.B., Gentry, L.E., Kovacic, D.A., 1998.Kinetics and modelling of dissolved phosphorus export from atile-drained agricultural watershed. J. Environ. Qual. 27, 917–922.

Documents

Variation of phosphorus loss from a small Catchment in south Devon, UK