PII: S0043-1354(00)00413-9

Wat. Res. Vol. 35, No. 6, pp. 13991406, 2001 # 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter

THE POTENTIAL FOR PHOSPHORUS RELEASE ACROSS THE SEDIMENTWATER INTERFACE IN AN EUTROPHIC RESERVOIR DOSED WITH FERRIC SULPHATE
R. G. PERKINS* and G. J. C. UNDERWOOD
Department of Biological Sciences, John Tabor Labaratories, University of Essex, Wivenhoe Park, Colchester, Essex CO4 3SQ, UK (First received 22 March 2000; accepted in revised form 25 August 2000) Abstract}Alton Water, Suolk, UK is a pumped storage reservoir that has a history of cyanobacterial blooms. Dosing of the input water with ferric sulphate to control external phosphorus loading has occurred since 1983. A detailed study of the sediment chemistry of the site was carried out between May 1995 and July 1997. Sequential phosphorus fraction analysis indicated a decrease along the length of the reservoir in sediment labile phosphorus content from 0.62 to 0.08 mg P g1 dw and iron-bound phosphorus content from 3.22 to 0.46 mg P g1 dw. These gradients positively correlated with water column chlorophyll a concentrations reported in a parallel study. Labile and iron-bound sediment phosphorus contents were in a dynamic equilibrium due to diusional release, contributing to internal loading to the water column. Equilibrium phosphorus concentrations (EPC) determined from phosphorus adsorption capacity (PAC) experiments were lower inside the bunded region (0.010.03 mg P-PO4 l1) where iron content was greatest compared to outside the bund (0.150.20 mg P-PO4 l1) suggesting greater potential for diusional release of phosphorus at the latter site. PAC experiments indicated that anaerobic and pHmediated loadings were of less importance than diusional release, although the latter may have contributed to internal loading in the main reservoir. Sulphate concentrations may act to increase the potential for anaerobic internal loading near to the pumped input in microstratied sediment. Sediment iron content decreased from 250 13.1 to 51 4.0 mg Fe g1 dw across the line of a constructed bund at the north-west end near to the pumped input, which indicated successful control of dispersal of the ne ferric oc. The management implications with regard to phosphorus loadings indicated by these results are discussed. # 2001 Elsevier Science Ltd. All rights reserved Key words}internal loading, reservoir, phosphorus adsorption capacity

NOMENCLATURE

Chl a dw EPC HEPES PAC wt

chlorophyll a dry weight equilibrium phosphorus concentration N-2-hydroxyethylpiperazine-N0 -2-ethanesulphonic acid Phosphorus adsorption capacity weight
INTRODUCTION

Nutrient loading to freshwater systems can be from external or internal sources. External loading to reservoirs is largely dominated by abstracted river inputs rather than run o from the catchment and precipitation. This is particularly true for pumped storage reservoirs in East Anglia, UK where annual precipitation is low and local topography results in small catchment areas for each reservoir. The limiting
*Author to whom all correspondence should be addressed. Tel.: +44-1206-873-345; fax: +44-1206-873-416; e-mail: rgperk@essex.ac.uk

nutrient in freshwater reservoirs is principally phosphorus (OECD, 1982). Reduction in the external phosphorus loading for control of algal biomass can be achieved by the use of ferric dosing. This technique involves the addition of ferric sulphate or alternatives to the pumped input, to precipitate dissolved particulate and orthophosphate in the incoming water. Wahnbach reservoir, Germany has a particularly eective ferric dosing system, coupled with ltration to remove the ferric/phosphorus oc, resulting in pumped inputs to the reservoir with phosphate concentrations below 5 mg l1 (Bernhardt and Classen, 1982). Where external loading has contributed to the internal phosphorus store in the reservoir, the reduction in external supply through management can increase the importance of internal phosphorus loading from the sediment (e.g. Van de Molen and Boers, 1994). The processes inuencing phosphorus release from the sediment have been extensively reviewed (e.g. Bostrom et al., 1988). Physico-chemi cal processes aecting such release include

1399

1400

R. G. Perkins and G. J. C. Underwood

temperature, pH, redox potential, reservoir hydrology and environmental conditions. Typically an increase in temperature acts to reduce adsorption of phosphorus by mineral complexes in the sediment (Berkheiser et al., 1980; Redshaw et al., 1990). An increase in pH can reduce the availability of binding sites on ferric complexes due to competition between hydroxyl ions and the bound phosphate ions (Andersen, 1975; Drake and Heaney, 1987; De Montigny and Prairie, 1993). Anaerobic release of phosphorus due to the reduction and solution of ferric to ferrous iron also results in phosphate release as described by the Einsele and Mortimer model (Bostrom, 1984). Sulphate can increase the eects of anaerobic conditions due to a double reaction of ferric iron with sulphate and sulphide to form ferrous iron and iron sulphide (Bostrom et al., 1988). Concomitant increase in pH with sulphate reduction of the ferric iron and polyphosphate breakdown result in phosphorus release to the overlying water (Caraco et al., 1993). These processes are complicated by the inuence of biological processes such as mineralisation, resulting in a complex system governing the potential release of phosphorus across the sedimentwater interface. Trophic gradients along reservoirs, based on variation in chlorophyll a (Chl a) and total phosphorus concentrations, are well documented (Jones et al., 1990; Thornton et al., 1982; Lind et al., 1993; Soyupak and Gokcay, 1994; Vrba et al., 1995). Such gradients usually occur where water column depth increases away from the main external input, typical of many man-made reservoirs (Lind et al., 1993). At Alton Water reservoir, Suolk, UK, gradients along the length of the reservoir were observed in phosphorus concentrations with a correlated gradient in algal biomass, measured as Chl a concentration (Perkins, 1999; Perkins and Underwood, 2000). After determination of the nature of the limiting nutrient, a number of questions relating to eective management of a site can be asked. (1) What are the sources of the limiting nutrient? (2) Are the control measures used to reduce the external phosphorus loading eective? (3) What is the potential for internal loading and (4) what factors are likely to have the greatest inuence on such loadings? To answer these questions a detailed study of the site is required, including a study of the sediment, in particular analysis of phosphorus and iron content. Experimental manipulation of the sediment can then be employed to determine under which conditions the potential for phosphorus release to the overlying water is greatest. In this study, routine sampling of the sediment from Alton Water reservoir was undertaken for the period May 1995 to July 1996. This was followed by experimental manipulation of sediment samples following the methods for phosphorus adsorption capacity described by Hwang et al. (1976), Slater and Boag (1978) and Redshaw et al. (1990). The project was initiated due to historical problems of cyano-

bacterial blooms in the reservoir, culminating in a bloom of Coelosphaerium sp. in 1992 which caused considerable water supply problems. The pumped input at this site is the main external source of phosphorus and is dosed with ferric sulphate to control the phosphate loading. Simple mass balance calculations following Sas (1989) estimated in excess of 90% removal of the external phosphorus loading (comprising 91% orthophosphate) due to ferric dosing, with the resulting ferric/phosphorus oc sedimenting out within a bunded region (Fig. 1) at the top end of the reservoir (Perkins, 1999). The study also determined the eectiveness of the bund in preventing dispersal of the oc into the main reservoir.
SITE DESCRIPTION

Alton Water reservoir (TM 150 361) is a pumped storage reservoir lying 6 miles south of Ipswich, Suolk. It is lled by pumped abstraction from the River Gipping over winter and used for potable water over summer. Inputs are dosed with ferric sulphate and are pumped into the top end of the reservoir while water is removed through the drawo tower in the main basin (Fig. 1). Ferric oc dispersal is controlled by a bund 400 m down the reservoir. The top end is shallow with a mean depth inside the bund of 1 m, increasing gradually along the reservoir with a maximum depth at full capacity of 18 m in the main basin, which is articially mixed using six pneumatic helices to prevent stratication. The mean depth is 9 m.
SAMPLING AND METHODS

Routine sampling Three replicate cores (length 50 cm, internal diameter 7.5 cm) were taken on a monthly basis from May 1995 to

Fig. 1. Alton Water reservoir, Suolk, UK, showing sample site positions relative to the pumped input and draw o tower. Sample sites with increasing distance down the length of the reservoir were inside the bund (site 1), outside the bund (site 2), open water site (site 3) and in the main basin (site 4).

Phosphorus release from reservoir sediment July 1996 from each of four sample sites (Fig. 1) using a Jenkins corer. Core length was limited to 40 cm to maintain a depth of 10 cm overlying site water. Cores were returned to the laboratory where this overlying water was siphoned o and the sediment cores were sectioned into 3 cm depth intervals and stored in sealed acid-washed plastic containers at 48C. Hosomi and Sudo (1992) suggested that the top 3 cm of sediment regulates phosphorus dynamics across the sedimentwater interface. Narrower sections were considered unsuitable as the ne, unconsolidated nature of the sediment resulted in some mixing of the core. Data are presented for the surface sections only as no pattern with core section depth was observed. Each sample site was analysed in situ for water temperature, pH and redox potential using a Hannah portable redox meter and a WPA CD7000 portable eld pH and temperature meter. Freeze-dried sediment samples were analysed for Chl a concentration using cold methanol extraction and spectrophotometric measurement at 665 nm with correction for turbidity at 750 nm. Samples were acidied to correct for phaeophorbide type pigments (Lorenzen, 1967). Percentage water content (wt/wt) was analysed by loss on drying at 808C for 24 hours. Dried samples were then incinerated at 5508C in a mue furnace for 1 hour to determine the percentage ash free dry weight (wt/dw). Freeze-dried samples were also analysed for iron and calcium content using an atomic adsorption spectrophotometer (Varian AA-1275). Samples were digested along with a standard estuarine sediment (NBS standard reference material 1646) and using standard additions to test for accuracy. Initial digests used nitric acid at 1508C (Allen et al., 1974) which removed 50% of the iron in the standard sample. This was improved by digesting in aqua regia, 3 parts concentrated hydrochloric to 1 part concentrated nitric acid, which digested 70% of the standard sample iron content and did not interfere with the accuracy as measured by standard additions. Wet sediment samples equivalent to 0.5 g dw were analysed for labile phosphorus, iron/aluminium bound phosphorus and calcium bound phosphorus using sequential extractions in ammonium chloride (1 M, pH 7), sodium hydroxide (0.1 M) and hydrochloric acid (0.5 M) using the methods of Hieltjes and Lijklema (1980). After neutralisation and centrifugation (4000 rpm for 15 min) samples were measured for orthophosphate using the ascorbic acid method (APHA, 1995). Phosphorus adsorption capacity (PAC) Sediment samples were collected as above from sites 1 and 3 in March and July 1996 for PAC experiments and determination of equilibrium phosphorus concentrations (EPC) (Hwang et al., 1976; Redshaw et al., 1990). Site 1 was considered a high ferric content sediment based on routine sampling, lying 100 m inside the bund. Site 3 was considered a low ferric site, positioned 500 m outside the bund. Three experimental designs were carried out; (1) analysis of phosphorus adsorption from dierent initial phosphate concentrations, from which adsorption isotherms were constructed, (2) the same design at three dierent pH, and (3) adsorption under anaerobic conditions. A wet sub-sample of sediment equivalent to 0.5 g dw was placed in an acid washed container with 100 ml of

1401

phosphate solution (anhydrous KH2PO4). Concentrations of 0, 0.5, 1, 5 and 10 mg P-PO4 l1 were used in March, with the 10 mg P-PO4 l1 omitted in July. Incubations at 208C in an orbital shaker at 100 rpm lasted 24 h with regular measurement of the overlying phosphate concentration. pH was monitored and decreased slightly from pH 7.1 to 6.8. The eect of pH was analysed using a 5 mg P-PO4 l1 phosphate solution adjusted to pH 7, 8.5 and 10 buered using HEPES and sodium hydroxide. As before, pH was monitored and decreased slightly by 0.1 pH units in each case. July sediments were incubated anaerobically with 0,1 and 5 mg P-PO4 l1 de-gassed (using oxygen-free nitrogen) phosphate solutions for 3 h with continuous de-gassing. Phosphate concentration was measured as before at time zero and after 1 and 3 h. Oxygen concentration was monitored by Winkler titration (APHA, 1995) and was maintained at 12% of saturation or below. A second set of anaerobic incubations were carried out in July including addition of 5 mM sulphate solution over a 7 day period at 208C with regular mixing. A wet sediment sample of 20 g was placed in an acid-washed container with 80 ml of site water of known initial phosphorus concentration, de-gassed with oxygen free nitrogen for 3 h. Samples had either sulphate solution added to result in a nal concentration of 0.5 mM or the same volume of de-gassed site water. Again oxygen concentration was monitored by Winkler titration (APHA, 1995) and maintained below 12% saturation. Sub-samples were removed at time zero and after 7 days for PAC analysis and determination of EPC as described above for aerobic incubations. A second subsample was centrifuged for analysis of the water fraction for phosphate concentration, to determine if release from the sediment had occurred. Statistical analysis Data were tested for heteroscedasticity by Bartletts test for homogeneity of variances. Where necessary data were transformed by log transformation (x=log (x+1)) (Zar, 1996). One-factor and two-factor analysis of variance (ANOVA) were used to test for signicant variations within the data. One-factor analysis was then followed by post-hoc Tukey test to determine the cause of variation. Two populations were compared for signicant dierence in means of samples taken using the Students t-test. Statistical software used was SYSTAT 7.

RESULTS

Sediment characteristics No signicant temporal variation was observed in the eld survey sediment data based on the monthly sampling strategy used. No seasonal patterns to data were observed (except water temperature) and mean values for each site are presented only. Chl a concentration and ash-free dry weights (Table 1) decreased signicantly down the length of

Table 1. Sediment characteristics from four sites in Alton Water reservoir sampled May 95 to July 96, in triplicate sediment core sections (mean s.e., n=30). The bund is located between sites 1 and 2 Sample site 1 2 3 4 Distance from input (m) 300 500 1000 5000 Mean depth (m) 1 2 4 14 Chl a concentration (mg Chl a l1) 68 9.5 51 7.3 70 10.0 36 7.4 % ash-free dry wt 17 1.1 13 1.4 12 0.1 9 0.5 % Water content 87 1.2 72 0.8 81 5.4 72 0.7

1402

R. G. Perkins and G. J. C. Underwood

Table 2. Summary of in situ physical data measured in water samples overlying sediment sampled May 95 to July 96 (mean s.e., n=30) Sample site 1 2 3 4 Redox potential mV, mean s.e. 142 140 172 178 8.3 11.3 13.9 18.2 pH, mean s.e 8 8 8 8 0.1 0.6 0.6 0.1 pH maxima, Aug 95 9.2 9.4 11.3 8.6 Temperature, Aug 95, 8C 23 27 26 22

Table 3. Mean sediment iron, calcium and phosphorus fraction contents for surface core sections sampled May 95 to July 96 (mean s.e., mg g1 dw) Site 1 2 3 4 Fe 250 13.1 51 4.0 42 2.7 39 1.9 Ca 108 17.1 143 27.6 173 19.5 109 8.2 Labile-P 0.62 0.068 0.10 0.008 0.11 0.007 0.08 0.004 Iron bound P 3.22 0.353 0.70 0.073 0.46 0.055 0.46 0.099 Calcium bound P 3.33 0.313 0.77 0.082 0.53 0.060 0.46 0.047

Fig. 2. Gradients along the length of the reservoir for mean water column summer Chl a (line) concentration in 1995 and sediment labile (closed columns) and iron-bound P contents (open columns).

P concentration at each sample site and concentrations of phosphorus in each fraction correlated signicantly (r=0.576, p50.01). Sample sites showed high Fe : P ratios of 64, 63, 89 and 83 : 1 for sites 14, respectively. A parallel study of water chemistry and nutrient limitation between April 1995 and September 1997 reported a gradient down the length of the reservoir of mean summer phytoplankton Chl a concentration (Perkins and Underwood, 2000). The same spatial gradient was observed for labile-P and iron-bound P in the sediment (Fig. 2). Mean summer water column Chl a concentration correlated signicantly with both labile and iron-bound P (r=0.94 and 0.92 respectively, p50.01).

the reservoir (F1(3, 860)=8.556 and 51.240, respectively, p50.001, post-hoc Tukey test). Water content was signicantly higher for the ne ferric oc sediment inside the bund (site 1) compared to sites outside the bund (F1(3, 860) = 113.652, p 5 0.001, post-hoc Tukey test). Mean redox potential of the overlying site water (Table 2) was below 200 mV for all sample sites, with no signicant spatial variation. pH peaked in August 1995, though no signicant variation was observed between sites. Temperature followed a seasonal trend with maxima in August 95 and a minimum temperature of 48C at all sites in February 96. Sediment iron content (Table 3) decreased signicantly down the length of the reservoir (F1(3, 671)=280.277, p50.001, post-hoc Tukey test). Calcium content did not show any signicant spatial variation. All phosphorus fractions (Table 3) decreased signicantly down the length of the reservoir with higher mean concentrations inside the bund compared to outside the bund (F1(3, 761)=29.016, (F1(3, 761)=61.187, F1(3, 761)=57.579 for labile, ironbound and calcium-bound P, respectively, p50.001, post-hoc Tukey test). Labile-P concentration was equivalent to approximately 20% of the iron-bound

Phosphorus adsorption capacity PAC experiments were carried out using sediment from inside the bund (site 1) and site 3. Sediment from site 1 collected in March 97, was seen to adsorb almost all the overlying phosphorus (>96%) within 2 h of incubation (Table 4). All phosphate added was adsorbed at equilibrium indicating the binding sites for phosphate were not saturated. Sediment from site 3 adsorbed signicantly less phosphorus compared to site 1 (t22=2.116, p50.05, Students t-test) with equilibrium reached after 24 h. Phosphorus adsorption capacity isotherms were constructed from these data (Slater and Boag, 1972; Hwang et al., 1976; Redshaw et al., 1990) to determine the equilibrium phosphorus concentrations (EPC) for the two sediments. Data for site 3 are presented in Fig. 3 with the tted logarithmic trendline to determine the intercept when no uptake or loss of phosphorus from the sediment has occurred. The EPC for site 1 sediment was 0.01 mg P-PO4 l1, lower than the mean annual orthophosphate concentration in the overlying water at this site of 0.08 0.011 mg P-PO4 l1 (Perkins and Underwood, 2000). Site 3 had an EPC of

Phosphorus release from reservoir sediment

1403

Table 4. Phosphorus adsorbed at equilibrium as a percentage of the initial phosphorus added to sediment slurries (means of 3 replicates s.e.) by sediment from sites 1 and 3 collected March 97 Initial concentration mg P-PO4 l1 0.5 1.0 5.0 10.0 Adsorption by site 1 sediment, % 96 0.5 98 0.5 100 0.1 100 0.1 Adsorption by site 3 sediment, % 67 4.6 78 2.9 83 2.7 70 6.1

Fig. 3. Phosphorus adsorption isotherm constructed for site 3 sediment slurries collected March 97.

0.15 mg P-PO4 l1. This is lower than the mean annual orthophosphate concentration in the overlying water at this site of 0.20 0.002 mg P-PO4 l1. There was no signicant eect of pH on phosphorus adsorption by the site 1 sediment with over 98% of phosphate adsorbed after 2 h incubation (Fig. 4a). However site 3 sediment showed a signicant reduction in phosphate adsorbed at pH 10 (F1(1, 7)=7.912, p50.05, post-hoc Tukey test) with a reduction from 83 2.7 to 41 6.0% (Fig. 4b). Sediment samples collected in July did not dier signicantly for the amount of phosphate adsorbed or in EPC values calculated (0.03 and 0.20 mg PPO4 l1 for sites 1 and 3, respectively). As for routine sampling, this suggests an absence of a seasonal pattern to data. There was also no signicant eect of pH on phosphorus adsorption by sediment from either site, even at pH 10. Anaerobic incubation had no eect on phosphate adsorption by the sediment from either site. No signicant variation between phosphate adsorbed in anaerobic compared to aerobic conditions after 3 h incubation was observed. Despite this the EPC obtained for site 1 in anaerobic incubations increased to 0.09 mg P-PO4 l1. Site 3 sediment showed no change in EPC with a value of 0.16 mg P-PO4 l1 obtained. In treatments with no phosphate addition to the overlying water the net ux of phosphate out of site 1 sediment did not vary signicantly between aerobic (14 0.5 mg P-PO4 g1 dw h1) and anaerobic incubation (16 3.6 mg P-PO4 g1 dw h1). However site 3 sediment showed a signicant increase in phosphorus ux out of the sediment in anaerobic treatment above that for aerobic treatment (t6=8.892, p50.001, Students t-test) increasing from 8 0.4 to 14 0.8 mg P-PO4 g1 dw h1. This contra-

Fig. 4. Phosphorus adsorbed by site 1 sediment (a) and site 3 sediment (b) collected March 1997 at increasing pH, with 5 mg P-PO4 l1 initial overlying phosphate concentration.

dicts the absence of any change in EPC observed for this site. The release of phosphorus from sediments due to anaerobic reduction of iron therefore appears unlikely at site 1, the site with highest sediment iron content. Site 3 sediment showed the potential for some anaerobic release despite having lower sediment iron and iron-bound P contents (Table 3). Phosphorus release from anaerobic slurries with sulphate addition Site 1 slurries with no phosphate addition showed no signicant dierence in phosphorus release when incubated anaerobically for 7 days for treatments with and without sulphate addition with net uxes out of the sediment of 0.05 and 0.03 mg P-PO4 g1 dw h1, respectively. Site 3 sediment showed no net phosphate ux out of the sediment without sulphate addition, but with sulphate addition a net release of

1404

R. G. Perkins and G. J. C. Underwood

0.003 mg P-PO4 g1 dw h1 occurred. However this release was an order of magnitude smaller than that for site 1 sediment, representing a very low rate of release in extreme conditions (constant mixing, reducing conditions for 7 days and 5 mM sulphate concentration). Anaerobic release of phosphorus enhanced due to ferric iron reduction in reactions to form sulphide is therefore unlikely for sediments from these two sites. Equilibrium phosphorus concentrations were determined for sediment slurries from both sites at time zero and after 7 days incubation. EPC of site 1 sediment signicantly increased from 0.08 to 0.15 mg P-PO4 l1 with no sulphate addition and from 0.13 to 0.50 mg P-PO4 l1 with sulphate addition (F2(4, 5)=12.375, p50.05). EPC was signicantly higher with sulphate addition throughout the period of incubation (F2(1, 5)=14.068, p50.05). Sediment from site 3 showed no signicant variation in EPC over time or between treatments with or without sulphate addition, with an EPC of 0.29 0.040 mg PPO4 l1. Sulphate may therefore act to increase the potential for diusional release of phosphorus from site 1 sediment in long-term anaerobic conditions, but did not enhance an already higher potential for release from site 3 sediment.

DISCUSSION

Gradients in sediment contents of labile and ironbound phosphorus along the length of Alton Water agreed well with those previously observed for the water column (Perkins and Underwood, 2000). Sediment Chl a and ash-free dry weights signicantly decreased down the reservoir, probably due to the decrease in the algal biomass in the water column and hence a lower load of sedimenting algal cells. In addition the pumped input may have increased the organic content of the north-west end of the reservoir. Sediment water content at this end was greatest due to the ne nature of the ferric oc. Inspection of the sediment inside the bund showed a very ne oc to a depth of up to 2 m in places. This could act as a large store of phosphorus were it to spread into the rest of the reservoir and become available for release. Gradients in sediment phosphorus fractions also showed decreases down the length of the reservoir, primarily due to the sharp decrease in labile and ironbound fractions across the position of the bund. Signicant correlations between Chl a in the water column and labile and iron-bound phosphorus were observed indicating a possible link between algal biomass and phosphorus availability in the form of internal loading. With distance down the reservoir, internal loading would be of increasing importance as the external pumped input would be utilised at the north west end, with any remaining phosphorus diluted greatly by the increasing water volume down

the reservoir. Sedimental iron content decreased from 250 13.1 to 51 4.0 mg P g1 dw between sites 1 and 2. This shows the bund to be an ecient method in the control of ferric oc dispersal down the reservoir. The potential availability of phosphorus for internal loading appears to decrease with distance down the reservoir, due to the lower sediment phosphorus content. However the lower labile phosphorus content would be matched by the larger area of sediment in the main basin acting as a store of phosphorus. The labile fraction may be in a dynamic equilibrium in the sediment as it accounts for 20% of the iron-bound fraction at all sites. This is despite the large ratios of Fe : P observed, which argue against saturation of binding sites on the ferric mineral complexes in the sediment. Such ratios well exceed the value of 7 : 1 at which saturation may occur (Lennox, 1984). Diusional release may be occurring along a concentration gradient from the sediment via the pore water to the water column. Whether this phosphorus is then re-adsorbed in the surface layers of the sediment or whether it passes into the water column is unclear. Aluminium sulphate dosing experiments in two eutrophic lakes indicated a strong correlation between labile phosphorus and ironbound phosphorus (by bicarbonate dithionite extraction). It was concluded that the iron-bound pool gradually contributed to the labile pool on an equilibrium basis according to solubility products, hence acting as a phosphorus source under aerobic as well as anaerobic conditions (Rydin and Welch, 1998). Despite redox potentials below 200 mV in the overlying water, at which ferric iron can become reduced facilitating phosphate release due to a reduction in binding sites (Bostrom et al., 1988), anaerobic incubations of the sediment did not indicate a high potential for phosphorus release. There was no decrease in site 1 PAC data for sediment from inside the bund (site 1) and the EPC only increased from around 0.01 to 0.03 mg P-PO4 l1 to 0.09 mg P-PO4 l1. However, this increase in EPC was sucient to exceed the orthophosphate concentration typically observed at this site in a parallel study of the water column (Perkins and Underwood, 2000) suggesting some potential for diusional release in anaerobic conditions such as microstratication in the shallow water inside the bund. Outside the bund (site 3) no signicant variation in PAC data or EPC were observed, however phosphorus release to phosphate-free overlying water did increase in anaerobic compared to aerobic incubations. In contrast Redshaw et al. (1990) reported net uxes out of sediment from Ardleigh reservoir, Essex, UK, in anaerobic conditions with initial phosphate concentrations of 2.5 mg P l1 indicating anaerobic release could overcome fairly strong diusion gradients into the sediment. Overall the data from this study did not support anaerobic release as a

Phosphorus release from reservoir sediment

1405

potential form of internal loading. However, even a small release of phosphorus in the main body of the reservoir could be signicant due to the much larger surface area of sediment compared to inside the bund. Thus, small increases in release of phosphorus may be enough to support some algal biomass. The addition of sulphate in anaerobic incubations resulted in conicting data. Fluxes of phosphorus out of the sediment were negligible in all incubations. However, both sites observed an increase in EPC over 7 days, with the largest increase for site 3. Caraco et al. (1993) showed that high sulphate concentrations led to both a decrease in iron concentration and an increase in phosphorus release acting to reduce the Fe : P ratios in the water column of North American lakes. The presence of sulphate may be important in the bunded region and it is suggested that sulphate concentrations should be monitored regularly, not just in the pumped input, but along the reservoir at the north-west end. It is surprising that the EPC increased most at site 3, the site with lower ferric iron and iron-bound phosphorus contents compared to site 1 inside the bund. Site 3 also showed an increase in phosphorus release to phosphate-free water in anaerobic incubations, suggesting that the potential for anaerobic release may increase outside the bund. The use of pneumatic helices to prevent stratication in the deep main basin is therefore particularly important. This form of management may also act to prevent dominance by vacuolate and colonial cyanobacteria due to the extent of mixing below the euphotic depth. The incubation of sediment with increased pH only resulted in a decrease in phosphorus adsorption at site 3 in March 1996 at the highest initial phosphate concentration. On no occasion did an increase in pH result in phosphorus release. This suggests that ligand exchange between hydroxyl ions and phosphate ions on ferric iron binding sites, was not important in the control of phosphorus release to the water column. This is despite the method employed breaking up the boundary layer of intact sediment which could buer any increase in pH. Drake and Heaney (1987) reported release of labile and ironbound phosphorus from Esthwaite water sediments at pH 10. Lijklema (1980) argued that above pH 8.5 probably no positive binding sites remain on iron hydroxy complexes, greatly reducing the phosphorus adsorption capacity of the sediment. This was not observed in this study, with no eect of pH observed at pH 8.5 and only a decrease in adsorption at the highest initial phosphate concentration on one occasion for site 3 sediment in March 1996. The north-west end of Alton Water still suers from high algal biomass in summer and is classied as eutrophic based on OECD (1982) criteria (Perkins and Underwood, 2000). The source of phosphorus at this site is likely to be from the 10% of phosphorus escaping ferric dosing coupled with internal loading

in the form of diusional release. The latter is most likely due to the ease with which re-suspension of the ne ferric oc can occur. Such re-suspension results in a larger diusion gradient to the low phosphate concentration in the water column, facilitating diusional release of labile and possibly then ironbound phosphorus. Turbulent re-suspension was observed even in light winds, coupled with the disturbance from the pumped input this could cause short periods of high phosphate concentration in the water column initiating the summer algal blooms noted at the north west end of the reservoir (Perkins and Underwood, 2000). Dredging of the sediment in the bunded region should be carried out to increase the water depth to reduce resuspension of the ne phosphorus rich sediment. This would reduce diusional release to the overlying water column and also prevent dispersal of the sediment over the bund and into the main body of the reservoir. The main basin at the opposite end of the reservoir is unlikely to receive a signicant external loading due to the dilution of any phosphorus not utilised for algal assimilation at the north-west end. Internal loading at this site is therefore of greater importance. Labile phosphorus may be released by diusion to the water column. Stratication is prevented at this end of the reservoir by 6 pneumatic helices to mix the water column. Stratication leading to anaerobic phosphorus release is therefore unlikely, unless a boundary layer of water remains adjacent to the sediment surface. Since the algal bloom in 1992, an extensive growth of Elodea has been noted in the reservoir, particularly in the main basin (Perkins, 1999). This may contribute to the control of internal loading by stabilising the sediment and utilising the labile phosphorus pool.

CONCLUSIONS
*

Gradients in sediment labile and iron-bound phosphorus content positively correlated with that observed for algal biomass in the overlying water column. Labile and iron-bound phosphorus were in dynamic equilibrium with a high potential for diusional release acting as internal phosphorus loading in the reservoir. Ferric dosing has resulted in Fe : P ratios greatly exceeding the point below which phosphate binding sites would be saturated. Eective control of ferric oc dispersal has been achieved by the bund at the north-west end of the reservoir, with iron content decreasing from 250 13.1 to 51 4.0 mg P g1 dw across the position of the bund. At the north-west end anaerobic and pH mediated internal loading appear less important than diusional release. The combined eect of

1406

R. G. Perkins and G. J. C. Underwood Hwang C. P., Lackie T. H. and Huang P. M. (1976) Adsorption of inorganic phosphorus by lake sediments. J. Water Pollute Control Fed. 48(12), 27542760. Jones J. R., LaPerriere J. D. and Perkins B. D. (1990) Limnology of Walker Lake and comparisons with other lakes in the Brooks Range, Alaska (USA). Verh. Int. Verein. Limnol. 24, 302308. Lennox L. J. (1984) Lough Ennell: laboratory studies on sediment phosphorus release under varying mixing, aerobic and anaerobic conditions. Freshwater Biol. 14, 183187. Lijklema L. (1980) Interaction of orthophosphate with iron(III) and aluminum hydroxides. Am. Chem. Soc. 14, 537541. Lind O. T., Terrell T. T. and Kimmel B. L. (1993). Problems in reservoir trophic-state classication and implications for reservoir management. In Comparative Reservoir Limnology and Water Quality Management, eds. M. Straskraba, J. G. Tundisi and A. Duncan, pp. 5767. Kluwer Academic Publishers, Netherlands. Lorenzen G. J. (1967) Determination of chlorophyll and phaeopigments: Spectrophotometric equations. Limnol. Oceanogr. 12, 343346. OECD (1982) Eutrophication of waters. Monitoring, assessment and control. OECD, Paris. Perkins R. G. (1999) Phosphorus limitation in Alton Water reservoir. Ph.D. Thesis, University of Essex. Perkins R. G. and Underwood G. J. C. (2000) Gradients of chlorophyll a and water chemistry along an eutrophic reservoir with determination of the limiting nutrient by in situ nutrient addition. Water Res. 34, 713724. Redshaw C. J., Mason C. F., Hayes C. R. and Roberts R. D. (1990) Factors inuencing phosphate exchange across the sediment-water interface of eutrophic reservoirs. Hydrobiologia. 192, 233245. Rydin E. and Welch E. B. (1998) Aluminum dose required to inactivate phosphate in lake sediments. Water Res. 32(10), 29692976. Sas, H. (1989). Lake Restoration by Reduction of Nutrient Loading. Academia Verlag, Richarz GmbH, St. Augustin. Slater S. J. E. and Boag A. J. (1978) The phosphorus status of the sediments of three eutrophic lakes in Victoria. Aust. J. Mar. Freshwater Res. 29, 26274. Soyupak S. and Gokcay C. (1994) Assessment of eutrophi cation for Keban Dam Reservoir. Fresenius Environ Bull. 3, 181186. Thornton F. W., Kennedy R. H., Magoun A. D. and Saul G. E. (1982) Reservoir water quality sampling design. Water Resour. Bull. 18(3), 471480. Van der Molen D. and Boers P. (1994) Inuence of internal loading on phosphorus concentration in shallow lakes before and after reduction of the external loading. Hydrobiologia. 275/276, 379389. Vrba J., Vyhnalek V., Helzlar J. and Nedoma J. (1995) Comparison of phosphorus deciency indices during a spring bloom in a eutrophic reservoir. Freshwater Biology 33, 7381. Zar J. H. (1996) Biostatistical Analysis. Prentice-Hall, Upper Saddle River, NJ, USA.

sulphate and anaerobic conditions could be important in microstratied sediment. In the main basin diusional release may be supported by anaerobic internal loading. Internal loading is probably the most important source of phosphorus for algal biomass at this site. Management policy should include dredging of the ne ferric oc in the bunded region.

Acknowledgements}We thank Dr. D. W. G. Daldorph of Anglian Water Services. RGP was funded by Anglian Water Plc.

REFERENCES

Allen S. E., Grimshaw H. M., Parkinson J. A. and Quarmby C. (1974) Chemical Analysis of Ecological Materials. Blackwell Scientic Publications, Oxford. Andersen J. M. (1975) Inuence of pH on release of phosphorus from lake sediments. Arch. Hydrobiol. 76(4), 411419. APHA (1995) Standard Methods for the Examination of Water and Wastewater, 19th ed, eds A. D. Eaton, L. S. Clesceri and A. E. Greenberg, American Public Health Association, Washington, DC. Berkheiser, V. E., Street, J. J., Rao, P. S. C. and Yuan, T. L. (1980) Partitioning of inorganic orthophosphate in soilwater systems. Agricultural Experiment Station, University of Florida, Gainesville, Article No. 2396. Bernhardt H. and Clasen J. (1982) Limnological eects of the elimination of phosphorus from the Wahnbach Reservoir. Water Sci. Technol. 14, 397406. Bostrom B. (1984) Potential mobility of phosphorus in dierent types of lake sediment. Int. Rev Gesamten. Hydrobiol. 69, 457475. Bostrom B., Andersen J. M., Fleischer S. and Jansson M. (1988) Exchange of phosphorus across the sedimentwater interface. Hydrobiologia. 170, 229244. Caraco N. F., Cole J. J. and Likens G. E. (1993) Sulphate control of phosphorus availability in lakes. Hydrobiologia. 253, 275280. De Montigny C. and Prairie Y. (1993) The relative importance of biological and chemical processes in the release of phosphorus from a highly organic sediment. Hydrobiologia. 253, 141150. Drake J. C. and Heaney S. I. (1987) Occurrence of phosphorus and its potential remobilization in the littoral sediments of a productive English lake. Freshwater Biol. 17, 513523. Hieltjes A. H. M. and Lijklema L. (1980) A simplied phosphorus extraction scheme. J. Environ Qual. 9(3), 405407. Hosomi M. and Sudo R. (1992) Development of the phosphorus dynamic model in sediment}water system and assessment of eutrophication control programs. Water Sci. Technol. 26(78), 18811900.