An investigation of the origin and mobility of phosphorus in freshwater sediments from Bort-Les-Orgues Reservoir, France V. Ruban,* S. Brigault, D. Demare, and A.-M. Philippe Laboratoire Central des Ponts et Chausse�es, B.P. 19, 44340 Bouguenais, France Received 22nd March 1999, Accepted 25th May 1999 In order to assess the origin and the potential mobility of phosphorus (P) in the sediment of the Bort-Les-Orgues Reservoir, France, two sequential extraction schemes, i.e., the SMT (modified Williams method)and the Golterman schemes, were compared. Finally, the potential mobility of P in this sediment was estimated from results of sequential extraction.The SMT method appeared to be more satisfactory than the Golterman method, which is in accordance with results from a study currently carried out in the framework of the European programme Standards, Measurements and Testing.Iron-bound P and organic P were the dominant forms of P in the sediment; these forms are likely to be released at the sediment/water interface in case of anoxia and could diVuse into the water column, thus increasing the risk of eutrophication in this sensitive reservoir.The P stock (330±66 t) is not negligible and should be taken into account in any restoration project of the reservoir. The SMT procedure seems promising and will provide, in the near future, a valuable tool for water managers in the field of lake restoration. potential mobility of P species, they also allow the assessment Introduction of the origin of the sediment phosphorus. Water management of lakes, reservoirs and watercourses has In this paper, two sequential extraction schemes, modified become an increasing problem due to the development of Williams (SMT) and Golterman, aimed at determining the eutrophication.This excessive supply of phosphorus (mainly origin of phosphorus in the sediment of the Bort-Les-Orgues from anthropogenic origin) to water systems, induces a degra- Reservoir (France) are compared.Finally, the potential dation of their quality through the proliferation of algae, thus mobility of phosphorus in this sediment is assessed. hindering various water uses. Much eVort and money have been spent on combating this form of pollution, but eutrophication remains one of the most important water quality prob- Description of the study site lems in Europe. This clearly appears through the Urban The hydroelectric reservoir of Bort-Les-Orgues is located in Wastewater European Directive of May 21, 1991, in which the upper part of the Dordogne River, not far from its source; eutrophic waters are defined as ‘sensitive areas’ requiring it is the largest reservoir of the French Massif Central.The stricter standards for phosphorus and nitrogen discharge to drainage basin is 1745 km2 (1010 km2 for the Dordogne River, receiving waters. 735 km2 for the Rhue River). The reservoir is 20 km long with Phosphorus is regarded as the key factor in freshwater a mean depth of 43.7 m and a maximum depth of 110 m. For eutrophication.1–3 It has two origins, i.e., external and internal. the normal elevation of exploitation (542.5 m above sea level ), Phosphorus from external origin comes from diVuse sources it has an area of 10.91 km2 and a volume of 477 ×106 m3.It (natural, agricultural ) or point sources (industrial and domshould be noted that these data apply for 1996, the year during estic eZuents). This nutrient is also released from the sediment, which this study was carried out.The main set of characteristics which acts as an internal source.4 Sediment can contribute regarding hydrology, land use and population are given in phosphate to the overlying waters at levels comparable to the Ruban and Demare.12 external source.5 However, not all the forms of P are likely to be released and thereby increase eutrophication. The ability of a sediment Sampling and preservation to store or release P has repercussion on the trophic state of a lake, hence on its productivity.The study of the sediment Three sediment cores were taken in the reservoir in 1997, behaviour is, therefore, a key factor which helps understanding down to a depth of 25 cm, the CEMAGREF sampler was the P cycle in rivers and lakes. It is also of prime importance used.A surface sample (0–5 cm) was also collected by means in the framework of water management and restoration of of a grab sampler. The cores were sliced as follows : 0–2 cm, water bodies. 2–5 cm, 5–10 cm and then every 5 cm to the bottom. The The various forms in which P occurs in sediments has samples were put into glass bottles and stored at a temperature received considerable attention during the past decades and of 4 °C until analysis. The sediment was passed through a many extraction schemes have been developed. Good extrac- 200 mm sieve in order to eliminate coarse plant fragments and tion methods are operationally defined on the basis of the then freeze dried.Though not perfect, this technique induces reactivity of a particular phase in a given extractant.There little change in the sediment and allows reproducible analyses are two types of methods: the ‘strong’ ones using HCl– even after several months, which is not always the case with NaOH6–9 and the ‘soft’ ones in which chelators such as nitrilo fresh sediment. triacetic acid or EDTA are used.10,11 pH and redox potential (Eh) were measured in situ by means of Ponselle probes (Ponselle PVRT).Extraction methods not only provide information on the J. Environ. Monit., 1999, 1, 403–407 403with de-ionised water. A quality control procedure was applied Sample analysis throughout the diVerent steps of sample preparation and The following determinations were carried out on the fraction analysis. The quality control procedure mentioned was that in of the sediment with a particle size of less then 200 mm: (i) loss use at the Water Division of Laboratoire Central des Ponts et on ignition (calcination at 550 °C for 2 h); (ii) water content Chausse�es.The materials and equipment (e.g., glassware, (drying at 105 °C until stable weight); (iii) particle size distri- balances, pH meters, pipettes, spectrophotometers) were regubution ( laser diVraction, Malvern Mastersizer MS1005) larly checked and the traceability of the samples was ensured Kjeldahl nitrogen13 (distillation with concentrated H2SO4); thanks to data reported in specific books.Furthermore, blanks (iv) iron, calcium (FAAS,Varian Spectra A300); (v) organic and standards were regularly run with each series of analyses carbon14 (oxidation with potassium dichromate); (vi) total in order to check the validity of the data.phosphorus (persulfate digestion+colorimetry according to Murphy and Riley15); and (vii) sequential extraction of P. Characteristics and origin of the sediment Two operationally defined extraction schemes were used in this study, i.e., a modified version of the Williams method Fig. 1 shows the main characteristics of the sediment in the (SMT), which has been proposed in the frame of the Bort-Les-Orgues Reservoir. The sediment was mainly silt (87% ‘Standards, Measurements and Testing’ programme of the on average) and was homogeneous with depth.Furthermore, European Commission, and the Golterman method (G). The there was little variation from upstream to downstream. initial version of the Williams method and the Golterman Organic carbon slightly increased from upstream (5.4%) to scheme are widely recognised for the determination of P downstream (6.2%); these values are in the range of concencompounds in lake sediments,10,11,16–18 and the SMT protocol trations for sediments in reservoirs from crystalline watersheds seems promising19 because it is relatively easy to implement, (4.9–6.4%) as reported by Rofes.20 can be used for all kinds of sediments (either calcareous, Calcium concentration was low (7.2 mg g-1), whereas iron siliceous or organic), and has a good reproducibility.concentration was high (58 mg g-1). Again, these elements The SMT method is composed of five steps and involves were homogeneous with depth (Table 2).As the iron concenthe of aqueous solutions of NaOH and HCl19 (Table 1). tration was much higher than the calcium concentration, iron The following forms are determined : iron-bound P, calcium- would probably be a key factor in phosphorus exchange at bound P, inorganic P, organic P and total P. the sediment/water interface. Note that the abundance of iron The Golterman method, (G) utilises chelators (Ca–EDTA, is characteristic of crystalline regions such as the French Na2EDTA) and is carried out at a pH similar to the sediment Massif Central.pH. It also comprises five steps as shown in Table 1: labile P, The C/N ratio is a good indicator of organic carbon iron-bound P, calcium-bound P, acid soluble organic P and sources.17 Sediments of terrestrial origin have a C/N ratio of reductant organic P.around 14–20 (by weight), while phytoplankton, the major Both methods have shortcomings, e.g., the partial resorption source of autochthonous organic matter in lakes, produces of P extracted by NaOH onto CaCO3 in the case of SMT, ratios of from 5 to 9. In the Bort-Les-Orgues sediment, C/N and possible solubilisation of a very small part of organic averaged 13 upstream and 11 downstream.These values are matter by NaOH in the first step. In the case of the G method, slightly higher than the average for the Pareloup Reservoir chelators interfere with P determination and the method is (9.8) located about 100 km south from Bort-Les-Orgues. This very long and lacks precision. ratio results from the mixing of two pools of organic matter The detection limits were 0.01 mg g-1 for the forms of P (OM), autochthonous and allochthonous, on which is superand total phosphorus (TP), 1 mg g-1 for Fe and 0.1 mg g-1 imposed biogeochemical modification of the particulate matter.for Ca. Allochthonous organic matter prevailed but it can be seen Five replicate determinations were made for each form of that autochthonous material increased downstream.P (for both methods). The coeYcient of variation ranged from 0.5 to 5%. For the SMT scheme, the precision of the method Sequential extraction of P is ±5% for HCl–P, inorganic P and concentrated HCl extractable P, ±10% for NaOH–P and organic phosphorus (OP). Comparison of the SMT and Golterman methods Regarding the G method, the precision is around ±20%, ±50% for labile P because of the very small quantities Total phosphorus (TP) concentrations in the surface sediment (SMT method) are presented in Fig. 2. Although TP concen- extracted in this step. The Murphy and Riley colorimetric method15 was used to trations were rather high (2–2.5±0.1 mg g-1 on average), these values are comparable to those reported by Bostro�m measure P in the extracts. The quantification of phosphate was based upon calibration curves of standard solutions.These et al.21 for lake sediments and they are in the upper range of concentrations for sediments from crystalline reservoirs as curves were determined several times during the period of analysis. reported by Rofes.20 There was no significant tendency with depth as shown in Table 2.The small variations observed All the reagents used were analytical-reagent grade; the glassware and plasticware were soaked in 0.3% HCl and rinsed probably resulted from changes in phosphorus inputs over Table 1 Reagents used and corresponding P fractions in the two sequential extraction procedures tested in this study Step 1 Step 2 Step 3 Step 4 Step 5 SMT NaOH 1 mol L-1, HCl 1 mol L-1 HCl 3.5 mol L-1 HCL 1 mol L-1, HCl 1 mol L-1 iron-bound P, +3.5 mol L-1, +calcination, inorganic P +calcination, bioavailable Ca-bound P, total P organic P, partly non-available available Golterman H2O, labile P, Ca–EDTA+dithionite Na2–EDTA H2SO4 NaOH 2 mol L-1, bioavailable 0.05 mol L-1, 0.1 mol L-1, 0.25 mol L-1, reductant organic P, iron-bound P, Ca-bound P, acid soluble non-available bioavailable non-available organic P bioavailable 404 J.Environ. Monit., 1999, 1, 403–407Fig. 1 Characterization of the sediment in the Bort-Les-Orgues Reservoir. time (dry year/wet year), but these changes remained small, since P concentration in the Rhue River is higher than that of other tributaries of the lake. The influence of the Rhue River within the precision of the method.Besides, TP concentrations tended to increase from upstream (1.99±0.1 mg g-1) to down- was observed up to station 3, 2 km from the dam.23 In order to know whether the diVerence between upstream stream (2.55±0.13 mg g-1), as did organic C concentrations. This has been observed in other reservoirs, e.g., in St Etienne and downstream TP was significant, a statistical test was carried out.Because the number of observations was small, Cantale`s Reservoir.22 In the case of Bort-Les-Orgues, the increase in P could result from inputs from the Rhue River the non-parametric test of White was used. For a=0.05, the J. Environ. Monit., 1999, 1, 403–407 405Table 2 Concentration of the forms of P, Fe, and Ca in a sediment core collected at station 4 (deeper part of the lake) Depth/ Ca/ Fe/ TP/ HCl–P/ NaOH–P/ OP/ cm mg g-1 mg g-1 mg g-1 mg g-1 mg g-1 mg g-1 0 7.3 5.8 2.69 0.38 1.55 0.67 2–5 7.2 59 2.65 0.42 1.60 0.62 5–10 7.1 58 2.60 0.40 1.62 0.65 10–15 7.2 58 2.70 0.40 1.50 0.69 15–20 7.2 57 2.72 0.42 1.65 0.63 20–25 7.3 59 2.67 0.42 1.60 0.63 Mean 7.2 58 2.67 0.41 1.59 0.65 SD 0.08 0.75 0.05 0.02 0.05 0.03 Fig. 2 Total phosphorus concentrations in the Bort-Les-Orgues Reservoir. TP was measured by the SMT and Golterman extraction schemes, and by persulfate digestion. Five replicates were made, the mean value and standard deviation are reported on the figure. hypothesis of equality between the means was rejected, therefore, it can be admitted that upstream P values are diVerent from downstream P values.Fig. 3 shows the distribution of phosphorus forms at the four stations in the Bort-Les-Orgues sediments, the SMT and Golterman methods were used. Concentrations were homogeneous within a core (Table 2), consequently, average values are presented here. The test of White was again applied to test the diVerence between the two methods. As shown in Table 1, step 1 (SMT) and 2 (G) are equivalent, so are step 2 (SMT) and 3 (G), and step 4 (SMT) and 4+5 (G).Consequently, for each station, HCl–P (SMT)/Ca–P (G), NaOH–P (SMT)/Fe–P (G), OP (SMT)/acid soluble OP+reductant OP (G) were compared. For a=0.05 the comparison of means carried out according to this test showed a significant diVerence. This confirms that the SMT and Golterman methods are not equivalent.The validity of the SMT scheme and the accuracy with which it has been carried out may be tested by comparing the sum of the single fractions with the total (concentrated HCl–P) measured separately.24 Such a comparison shows that TP-S(NaOH–P+HCl–P+OP)=0.05, the mean diVerence is thus negligible. Furthermore, TP values measured by persulfate Fig. 3 Sequential extraction of phosphorus (SMT and Golterman digestion were close to those obtained following the SMT schemes) at the 4 stations in the Bort-Les-Orgues Reservoir.Five protocol. replicates were made, the mean value and standard deviation are reported on the figure. As TP was not measured separately in the G method, it is not possible to check the validity of this scheme as was done for the SMT scheme.However, it has to be noted that the are in agreement with a series of interlaboratory comparisons currently carried out in the framework of the European project sum of the diVerent fractions yielded by the G method is significantly lower than TP measured by persulfate digestion SEPHOS (sequential extraction of phosphate in freshwater sediments) supported by the Standards, Measurements and and SMT method (Fig. 2). It seems reasonable to conclude then, that the SMT method gives better results in the case of Testing Programme of the European Commission.19 The discrepancy between the two methods enhances the necessity of the Bort-Les-Orgues sediments. This is not surprising as the G method was mainly designed for calcareous sediment and a harmoned sequential extraction scheme, which would enable water managers and researchers to compare their the Bort-Les-Orgues sediments are crystalline.Also, it is important to note that the better results of the SMT scheme results. 406 J. Environ. Monit., 1999, 1, 403–407(18 mg m-2 d-1) was measured through a laboratory experiment12 before the reservoir was emptied in 1995. The partial emptying of the reservoir induced a reduction of P release due to the oxidation of iron compounds to which P was adsorbed.Also, it is important to keep in mind that a large releasable P stock exists in the reservoir, which could delay the recovery of the lake even when measures to reduce P inputs are taken. Quantification of the P stock, together with the release rate will help in predicting this restoration delay. Acknowledgements The authors thank the Adour-Garonne Water Agency as well as the Councils of the Cantal, Corre`ze and Puy-de-Do�me Fig. 4 Relative distribution of phosphorus according to the SMT Departments for their financial support; they are also grateful protocol, in the Bort-Les-Orgues Reservoir. to E.P.I.DOR (Etablissement Public Interde�partemental de la Dordogne).The help of the CEMAGREF in core sampling The relative percentages of phosphorus forms according to was very much appreciated. the SMT method are presented in Fig. 4 for the Bort-Les- Orgues Reservoir. NaOH–P was 60% of TP, OP 24%, and References HCl–P 16%. The potentially releasable P (NaOH–P+OP) in the sediment is around 80%, which is in agreement with 1 R. A. Vollenweider, Technical Report OECD, DAS/CSI 68-27, previous results12.OECD, Paris, 1968. Therefore, in the Bort-Les-Orgues Reservoir, phosphorus is 2 OECD, Synthesis report (Eutrophisation des eaux. Me�thodes d’e�valuation et de lutte), OECD, Paris, 1982. mainly of anthropogenic origin (NaOH–P+OP partly), 3 P. C. M. Boers and D. T. Van der Molen, Eur. Water Pollut. whereas Ca–P, which is of detrital origin, accounts for less Control, 1993, 19.than 20%. These results also show that most sediment P in 4 B. Bostro�m and K. Pettersson, Hydrobiologia, 1982, 170, 229. the lake is allochthonous. Brigault and Ruban25 have shown 5 G. Premazzi and A. Provini, Hydrobologia, 1985, 120, 23. that among the sources of allochthonous P, about 30% comes 6 J. D. H. Williams, J.-M.Jaquet and R. L. Thomas, J. Fish. Res. from domestic or industrial eZuents (mostly NaOH–P), 60% Bd. Can., 1976, 33, 413. 7 J. D. H. Williams, T. Mayer and J. O. Nriagu, Soil Sci. Soc. Am. from agriculture (NaOH–P+OP). J., 1980, 44, 462. 8 A. H. M. Hieltjes and L. Lijklema, J. Environ. Qual., 1980, 3, 405. Estimation of the phosphorus stock in the sediment 9 R. Psenner, R. Pucsko and M.Sager, Arch. Hydrobiol., Suppl., 1985, 70, 111. An estimation of the sediment stock in the reservoir of Bort- 10 H. L. Golterman and A. Booman, Verh. Int. Ver. Limnol., 1988, Les-Orgues was made. 23, 904. Taking into account the volume (477×106 m3) and the 11 H. L. Golterman, Hydrobiologia, 1996, 5, 1–9. mean depth (43.7 m) of the reservoir (the surface of the 12 V. Ruban and D.Demare, Hydrobiologia, 1998, 373/374, 349. sediment was estimated to 10.91 km2) and assuming that P is 13 AFNOR, NF T90 110, in Qualite� de l’eau, recueil de normes franc�aises, AFNOR, Paris, 1996. active in the upper 5 cm of the sediment,26 it is possible to 14 AFNOR, NF X31 109, in Qualite� des sols, recueil de normes calculate the available P stock, since the concentration is franc�aises, AFNOR, Paris, 1996.known. The sediment stock (5 cm) was 545 500 m3, with a 15 J. Murphy and J. P. Riley, Anal. Chim. Acta, 1962, 27, 31. mean water content of 75%, the volume of dry sediment was 16 J. Armengol and A. Vidal, Ergeb. Limnol., 1988, 30, 61. 130 625±10% m3. The density of the sediment was 1.1, hence 17 P. G. Manning and K. R. Lum, Can. Mineral., 1983, 21, 121.the sediment mass was 150 012 t. The concentrations of 18 R. L. Thomas, S. Santiago, V. Gandais, L. Zhang and J.-P. Vernet, Water Pollut. Res. J. Can., 1991, 26, 433. NaOH–P and OP were 1.62 and 0.58 mg g-1, respectively, 19 V. Ruban, J. F. Lopez-Sanchez, P. Pardo, G. Rauret, H. Muntau and their mass in the sediment was then: and P. Quevauviller, J. Environ. Monit., 1999, 1, 51. 20 G. Rofe`s, in Gestion inte�gre�e des milieux aquatiques, Presses de mNaOH--P=1.62 10-3×150 012=243±48 t l’ENPC, Paris, 1994, pp. 89–99. mOP=0.58 10-3×150 012=87±18 t 21 B. Bostro�m, M. Jansson and C. Forsberg, Ergebn. Limnol., 1982, 18, 5. The stock of P likely to be released averaged 330±66 t, which 22 V. Gandais-Ruban, D. Demare and A. Fournier, Bull. Labo. Pont is far from negligible if compared to the annual TP input of Chausse�es., 1993, 186, 79. 121 t.25 However, this value is only indicative as it supposes 23 S. Brigault, E. Sacchi, R. Gonfiantini and G. M. Zuppi, C. R. Acad. Sci., Ser. IIa: Sci. Terre Plane`tes, 1998, 327, 397. that the release occurs on the whole reservoir, which is 24 A. Barbanti and G. P. Sighinolfi, Environ. Technol. Lett., 1988, probably not the case. 9, 117. Nurnberg27 has shown that significant correlations between 25 S. Brigault and V. Ruban, Water, Air, Soil Pollut., 1999, to be anoxic release rates and sediment P fractions can be used to published. predict phosphate release rates in lakes with known sediment 26 H. S. Jensen, P. Kristensen, E. Jeppesen and A. Skytthe, P concentrations. Applying this relationship to the sediment Hydrobiologia, 1992, 235/236, 731. 27 G. Nurnberg, Can. J. Fish. Aquat. Sci., 1988, 45, 453. of Bort-Les-Orgues leads to a release rate of about 5 mgm-2 d-1, which is typical of meso-eutrophic lakes in this area. It has to be noted that a much higher flux Paper 9/02269D J. Environ. Monit.,