Accepted Manuscript Long-term efficiency of lake restoration by chemical phosphorus precipitation: Scenario analysis with a phosphorus balance model Michael Hupfer, Kasper Reitzel, Andreas Kleeberg, Jörg Lewandowski PII:

S0043-1354(15)30101-9

DOI:

10.1016/j.watres.2015.06.052

Reference:

WR 11394

To appear in:

Water Research

Received Date: 3 May 2015 Revised Date:

27 June 2015

Accepted Date: 30 June 2015

Please cite this article as: Hupfer, M., Reitzel, K., Kleeberg, A., Lewandowski, J., Long-term efficiency of lake restoration by chemical phosphorus precipitation: Scenario analysis with a phosphorus balance model, Water Research (2015), doi: 10.1016/j.watres.2015.06.052. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Graphical abstract

ACCEPTED MANUSCRIPT 1

Long-term efficiency of lake restoration by chemical phosphorus precipitation: Scenario

2

analysis with a phosphorus balance model

3

Michael Hupfer1, Kasper Reitzel2, Andreas Kleeberg1*, Jörg Lewandowski1

4

1

5

D-12587 Berlin, Germany

6

2

7

Odense M, Denmark

8

Corresponding author: Michael Hupfer, Tel. ++49/+30/64181605, Fax ++49/+30/64181682,

9

[email protected]

RI PT

SC

University of Southern Denmark Odense, Institute of Biology, Campusvej 55, DK-5230

M AN U

10

Leibniz-Institute of Freshwater Ecology and Inland Fisheries Berlin, Müggelseedamm 301,

*Now at: State Laboratory Berlin-Brandenburg, Kleinmachnow, Germany

12

TE D

11

Abstract

14

An artificial increase of phosphorus (P) retention in lakes with a long residence time and/or a

15

large mobile sediment P pool by adding P binding chemicals can drastically shorten the time

16

these lakes require to reach water quality targets. Suitable tools to optimize timing and extent

17

of external and internal measures are lacking. The one-box model, a mass balance tool for

18

predicting the P trend in the water under different management options was applied to highly

19

eutrophic Lake Arendsee (a = 5.14 km2, zmax = 49 m), Germany. Mass developments of blue

20

green algae and increasing hypolimnetic oxygen deficiencies are urgent reasons for restoring

21

Lake Arendsee. Detailed studies of P cycling and scenario analyses with the one-box model

22

led to the following conclusions: i) immediate improvement of the trophic state is only

AC C

EP

13

1

ACCEPTED MANUSCRIPT possible by in-lake P inactivation because of the long water residence time (56 years); ii) a

24

gradual external P load reduction, even if the effect is delayed, will assure the sustainability of

25

the scheduled Al application beyond one decade; iii) a twofold precipitation reduces the risk

26

of failure compared to a singular application with an overdose related to the relevant internal

27

P pools.

28

Key words

29

Sediment, phosphorus retention, one-box model, eutrophication, phosphorus binding

30

chemicals, lake restoration

M AN U

SC

RI PT

23

31

1. Introduction

33

Phosphorus (P) as the limiting nutrient for primary production is often the driver of the

34

ecological deterioration of temperate freshwater systems (Smith and Schindler, 2009; Withers

35

et al., 2014). The introduction of surface water quality targets, and the legislative pressure for

36

their implementation, e.g. as part of the EU Water Framework Directive (WFD), has enforced

37

a debate for geo-engineering using P-binding chemicals (PBC) in lake ecosystems (Spears et

38

al. 2013a, 2014; Mackay et al., 2014). In the EU territory, about 36% of all reported WFD

39

lakes (on a surface area basis) fail to meet the target of a ‘good ecological status’ (Spears et

40

al., 2013a).

41

Many lake managers and scientists argue that external load reduction is the only sustainable

42

way to improve water quality, and the precondition for supplemental in-lake measures

43

(Mehner et al., 2002; Schauser & Chorus, 2007; Jensen et al., 2015). However, often the P

44

load cannot be lowered to levels necessary to effectively control the trophic state within an

45

acceptable time frame or budget. Additionally, the reduction of external P sources has not

AC C

EP

TE D

32

2

ACCEPTED MANUSCRIPT brought the expected improvement of water quality in many lakes because of processes in the

47

catchment or in the lake itself that delay the response (Jeppesen et al., 2005; Schippers et al.,

48

2006). The latter authors reported a study of a coupled catchment-shallow lake model

49

considering the duration of buffer-related time delays. Results show that the most important

50

buffer was the percolation of the soil layer, which may cause a delay of 150-1700 years

51

depending on agricultural P surplus levels. The surface soil layer in contact with runoff water

52

accounted for a delay of 5-50 years. However, the buffering capacity of the lake water was

53

negligible whereas buffering in the lake sediment postponed the final lake equilibrium for

54

several decades (Schippers et al., 2006). Therefore, PBCs are receiving more attention

55

because the enrichment of mobile P in lake sediments has often been identified as the most

56

important reason for the delay of water quality improvement after external P load reduction

57

(Mehner et al., 2008; Søndergaard et al., 2013; Zamparas and Zacharias, 2014; Jensen et al.,

58

2015).

59

In general, the ‘good ecological status’ can be achieved within an appropriate time frame by

60

implementing internal measures into lake management concepts as follows: first, by rapidly

61

decreasing available P in water and sediment (Mackay et al., 2014); second, by creating a

62

positive feedback mechanism (Benndorf, 2008; Schallenberg and Sorrell, 2009) leading to

63

self-stabilization of the lake ecosystem at the desired quality, e.g. formation of a macrophyte-

64

dominated clear water state; and third, by preventing negative symptoms of a too high trophic

65

state (Uhlmann et al., 2011). Benndorf (2008) argued that internal restoration techniques

66

could partly compensate excessive external loads, and simultaneously decrease the cost of

67

reaching quality targets since internal measures could be cheaper and faster than external

68

measures for the equivalent P mass reduction. The point sources and the internal P are easier

69

to control than the non point sources.. It is not always obvious whether control measures

70

should focus on reducing external or internal P loads, or whether both should be attempted,

AC C

EP

TE D

M AN U

SC

RI PT

46

3

ACCEPTED MANUSCRIPT and the PBC doses required to achieve objectives need to be resolved on a case-by-case basis

72

(Pilgrim et al., 2007).

73

The addition of PBCs promotes geochemical conditions which increase the net sink function

74

of sediments by enhancing P sedimentation rates and/or decreasing P release rates. New

75

substances and mixtures with improved characteristics have been designed and tested (Hickey

76

and Gibbs, 2009; Spears et al., 2013b; Lurling and Oosterhout, 2013) but all of these

77

substances are defined by a finite P binding capacity under given environmental conditions.

78

Under consideration of the acid buffer capacity and alkalinity of the water, respectively, the

79

necessary dose is often determined by the P pool in the water. , But the retrospective

80

evaluation of chemical inactivation measures has only shown a weak relation between the

81

dose per unit water volume or area and the long-term effects (Welch and Cooke, 1999;

82

Smeltzer et al., 1999; Huser et al., 2011). An increasing number of recent studies include the

83

mobile P pool in the sediment (Rydin and Welch, 1999; Reitzel et al., 2005; de Vicente et al.,

84

2008). Therefore, special attention is given to determining the available P pool pragmatically

85

by assigning P fractions to temporary or permanent P pools (Rydin, 2000; Reitzel et al.,

86

2005). Long-term improvements were observed when ‘a sufficient’ amount of PBC was

87

added (e.g., Jensen et al., 2015). However, what is sufficient? The dose of PBC depends on

88

the additional P binding capacity required over time, and thus from the development of the P

89

balance including future P import. Empirical nutrient models like the Vollenweider Model

90

(Vollenweider 1976) are well proven tools for the management of eutrophic lakes. The

91

statistical analysis of a large data set allows the prediction of P concentration and trophic state

92

under a new steady state after external P load reduction or estimations of critical threshold

93

values for the external P loading. However, this model does not explicitly include P retention

94

in the sediment. The development during the transitional phase after restoration, adaptation

95

time after external P load reduction and the impact of internal measures cannot be estimated

96

with this kind of models. Contrary to this, mechanistic models based on the nutrient mass

AC C

EP

TE D

M AN U

SC

RI PT

71

4

ACCEPTED MANUSCRIPT 97

balances can be used to evaluate the impact of external and internal measures on the P

98

development (Schauser and Chorus, 2007; Wauer et al., 2009; Grüneberg et al., 2011).

99

We use the highly eutrophic Lake Arendsee, North Germany, with a long water residence time, as a case study to predict the future P concentration following a planned P inactivation

101

measure. Experience with the application of PBCs in a lake as large as L. Arendsee, which is

102

three times larger than Lake Delavan, USA (46.4×106 m3), the largest lake treated by PBC to

103

date, are not reported in the scientific literature (Huser et al., this issue).

104

The main objective of the present study is to provide a simple reliable tool for lake managers

105

to optimize the interplay between external and internal measures for decreasing P availability

106

in the water. The one-box model (Gächter and Imboden, 1985; Sas, 1989) was applied in

107

order to: i) predict the speed and sustainability of external P load reduction versus internal P

108

inactivation, ii) determine the optimal timing and dose of a PBC treatment, and iii) evaluate

109

the effects of single dosage (equivalent to the internal P pools) compared to overdosage in the

110

form of singular or repeated application under different P load reduction scenarios. We use

111

scenario analyses to determine the necessary P fixation capacities over time as the basis for

112

selecting an appropriate PBC.

113

2. Methods

114

2.1 Study site

115

Lake Arendsee (area 5.14 km2, max. depth 49 m, mean depth 29 m) is situated in Northern

116

Germany (52°53′21′′ N, 11°28′27′′ E) (Hupfer and Lewandowski, 2005). This dimictic hard

117

water lake was originally fed solely by groundwater. Nowadays, four ditches draining

118

adjacent agricultural fields additionally discharge into the lake and an artificial runoff channel

119

transports water out of the lake (Meinikmann et al., 2015). At least since the middle of the last

120

century, the lake has been strongly eutrophied (Scharf, 1998). The lake volume weighted total

121

P (TP) concentration averaged 184 ± 7 µg L-1 (2005-2014, n=10);

AC C

EP

TE D

M AN U

SC

RI PT

100

5

the mean epilimnetic TP

ACCEPTED MANUSCRIPT concentration during growing season was 96 ± 16 µg L-1The water quality is impaired by

123

occasional low transparency with Secchi depth falling below 1 m and mass developments of

124

phytoplankton dominated by cyanobacteria such as Planktothrix rubescens, (DC. Ex

125

Gomont); or diazotrophic Anabaena flos-aquae, Bory de St.-Vincent and Aphanizomenon

126

flos-aquae, (L.) during summer. The assessment based on the phytoplankton community has

127

indicated a ´bad ecological status’ so that the demand of the EU WFD for a good ecological

128

status cannot be achieved at present. Additionally, dissolved oxygen (O2) in the hypolimnion

129

at the end of summer stratification has continuously decreased over the last four decades

130

(Shatwell et al. 2013). The volume-weighted O2 concentration between 20 and 48 m

131

decreased from 4.76 ± 0.80 mg O2 L-1 (1976-1980) to 1.83 ± 0.85 mg O2 L-1 (2010-2014).

132

Simultaneously, the upper border of the layer with concentration less than 2 mg O2 L-1 shifted

133

upwards from 42.4 ± 2.9 m to 33.3 ± 2.6 m depth.

134

The above-ground catchment area (29.5 km2) is dominated by agriculture (52.1%) and

135

forestry (30.6%). The town Arendsee is situated directly on the south west shore (Fig.1). The

136

sum of the different, separately determined external P sources was 1,560 kg yr-1 (0.303 g m-2

137

yr-1; Meinikmann et al., 2015). More than 50% of this total P load is imported by groundwater

138

enriched in P while passing below the town Arendsee. The recent anthropogenic P input via

139

groundwater is one order of magnitude higher than the estimated input based on natural

140

background P concentrations. According to Meinikmann et al. (2015), the P load in

141

groundwater is highest followed by atmospheric deposition (19 %), water fowl (maximum 13

142

%), and drainage from agriculture (12 %). Previous in-lake restoration measures in Lake

143

Arendsee were not successful. Hypolimnetic withdrawal (1976-1990) and the capping of

144

profundal sediments by mechanical resuspension of calcareous mud from the littoral (autumn

145

1995) have not shown any significant decrease of P (Hupfer et al., 2000). The current

146

restoration intention aims to decrease the mean TP concentration at least to 60-80 µg L-1

AC C

EP

TE D

M AN U

SC

RI PT

122

6

ACCEPTED MANUSCRIPT 147

(based on German guidelines for the implementation of WFD) so that P-limiting conditions

148

for phytoplankton growth will prevail during the vegetation period.

149

2.2 Lake water phosphorus investigations

151

Lake Arendsee has been monitored since 1976. The P balance was calculated using TP

152

concentrations in water samples taken at 0, 5, 10, 15, 20, 30, 40, 45 and 48 m depth (see

153

Hupfer and Lewandowski, 2005).

154

2.3 Sediment phosphorus investigations

155

Gradient method: The mobile P pool was determined from undisturbed sediment cores

156

repeatedly taken with a modified Kajak sampler (UWITEC, Mondsee) at the deepest site of

157

the lake between 2000 and 2014. TP profiles were determined at 1 cm vertical resolution from

158

these cores, down to15 cm or at least to calcareous mudlayer. The mobile P pool in the

159

sediment was calculated as the difference between TP in each uppermost layer and the TP in

160

the background layer below the depth of endpoint of early diagenesis (EP), where no further

161

TP decrease is found (Hupfer and Lewandowski, 2005; Carey and Rydin, 2011; Grüneberg et

162

al., 2014).These differences were multiplied with the dry mass in the respective layer and

163

summed to yield the mobile P mass per unit area.

SC

M AN U

TE D

EP

AC C

164

RI PT

150

165

Fractionation method: Additionally, undisturbed sediment cores were taken twice from the

166

deepest point of the lake and 800 m east of this position where the lake is 42 m deep

167

(September 2007 and March 2008). The uppermost 5 cm of sediments were sliced into 0.5

168

and 1 cm layers. The sediment was fractionated according to Psenner et al. (1986) modified

169

by Hupfer et al. (1995). The mobile P Pool was calculated as the sum of P forms potentially

170

contributing to P release, i.e. loosely adsorbed P (NH4Cl-P), redox-sensitive P (BD-P), and 7

ACCEPTED MANUSCRIPT organic-bound P (NaOH, non reactive P = nrP) (Rydin et al., 2000; Reitzel et al., 2005). The

172

same cores were used to determine the mobile P based on the TP gradient in the sediment as

173

described above.

174

Bulk method: The P retention rate was determined by sediment cores taken at 4 to 7

175

randomly selected sites ten times between February 2000 and September 2014. The deposited

176

material above the calcareous mud from 1995 (lake restoration measure, see above) was

177

separated as a single layer before dry mass and TP content were determined.

178

Chemical analysis: Sedimentary P forms were characterized by the sequential extraction

179

scheme according to Psenner et al. (1984) modified by Hupfer et al. (1995). Total P in dried

180

sediments was determined as soluble reactive P (SRP) after digestion with H2SO4 and H2O2

181

for 10 h (Zwirnmann et al., 1999). SRP was photometrically determined by the molybdenum

182

blue method (Murphy and Riley, 1962) using a segmented flow analyzer (Skalar Sanplus,

183

Skalar Analytical B.V., De Breda).

184

2.5 Calculation of phosphorus mass balance parameters (Plake, Pin and Pexp)

185

The P mass of the whole water body (Plake), the upper layer (0-15 m, maximum extension of

186

epilimnion), and the lower layer (15-48 m, hypolimnion) was calculated using the vertical

187

profile of TP concentrations and the volume of the corresponding water layers (Hupfer and

188

Lewandowski, 2005). The external P load (Pin) equals the sum of net P sedimentation (Psed), P

189

export from the lake by surface water and groundwater outflow (Pexp), and changes in the P

190

inventory of the lake water (∆Plake) (Meinikmann et al., 2015). All components in equation 1

191

have the unit tonnes P per year (t yr-1).

AC C

EP

TE D

M AN U

SC

RI PT

171

192 193

Pin = Psed + Pexp + ∆Plake

(1)

194 8

ACCEPTED MANUSCRIPT ∆Plake was derived from the mean linear trend of the amount of P in the whole lake from 1995

196

to 2014. Psed was calculated from dated sediment cores (using calcareous mud deposited in

197

1995) taken at different water depths and referenced to the lake area deeper than 30 m (3.0

198

km2). Pexp is based on mean TP concentrations in the upper layer (0-15 m) from 1995 until

199

2014. The respective outflow water volume was based on the lake’s water residence time with

200

the simplified assumption that precipitation onto and evaporation from the water surface are

201

equal.

202

2.6 One-box model

203

The one-box model introduced by Gächter and Imboden (1985) was used to predict the

204

development of P concentration of Lake Arendsee under different management scenarios.

205

Assuming steady state conditions, the P balance (eq. 1) can be expressed as follows:

206

Pin = σ Plake + β Plake/τ

207

where the following coefficients were used: τ is the theoretical water residence time (yr), σ is

208

the net P sedimentation (Psed) divided by Plake (yr-1), β is the stratification factor, i.e. quotient

209

of the annual mean outflow and annual mean P concentration of the lake (dimensionless).

210

The mean P concentration at steady state (cPstat) depends on the mean concentration of the

211

external P load (cPin), the water residence time (τ), the stratification factor (ß) and the net

212

sedimentation coefficient (σ):

213

cPstat= cPin/(ß + τ σ)

214

The mean P concentration in years under non-steady conditions (e.g. transition states from

215

one to another cPstat) was calculated using the difference between the current P concentration

216

(cP0) and cPstat as follows:

(2)

AC C

EP

TE D

M AN U

SC

RI PT

195

(3)

9

ACCEPTED MANUSCRIPT 217

cP(t) = (cP0–cPstat) * e{–(β/τ + σ) * t} + cPstat

218

The influence of PBC addition on the P concentration was simulated by (1) setting the net

219

sedimentation coefficient (σ) to a maximum value in the year when the PBC was applied to

220

reach a target concentration of 20 µg P L-1, (2) increasing the P binding capacity of the

221

sediment in the years after PBC addition by temporarily increasing σ. The simulation starts in

222

the year after PCB addition with σ = 0.3 representing P gross sedimentation (P loss in the

223

epilimnion without P release) and then linearly decreasing it again to the original σ at least

224

over a period of 15 years. Scenarios with external P load (Pin) reduction consider it only

225

realistic to decrease the unusually high P input in groundwater.

226

3. Results and Discussion

227

3.1 Phosphorus balance

228

Determination of P pools. The mean annual Plake increased slightly over the last four

229

decades. At present, Plake is stable at about 27.3 tons P, representing a TP concentration of 186

230

µg L-1 (Fig. 2). The small standard deviations around the annual mean show that the seasonal

231

variation of TP in the water body is relatively low. However, the internal P dynamics show a

232

strong vertical redistribution of P due to substantial loss of P from the epilimnion and an

233

equivalent accumulation of P in the hypolimnion during summer stratification; the respective

234

mean P release was 1.41 ± 0.28 g m-2 (n = 5, 2010-2014) representing a P release rate of 8.69

235

± 1.43 mg m-2 d-1. Sedimentary TP sharply decreased from the uppermost to deeper layers.

236

During diagenesis, the TP content converges to about 1.1 mg g dw-1 within the first 5 to 8 cm

237

(equivalent to about 2.4 kg m-2 to 4.4 kg m-2 dry mass, Fig. 3). The repeated collection of

238

sediment cores enables the determination of the P content and changes in the former surface

239

layer as it is buried by freshly settled sediment to be observed in real time. The mobile P pool

240

determined by the P gradient method varied between 0.84 g m-2 and 1.47 g m-2 (n = 5, 1995,

AC C

EP

TE D

M AN U

SC

RI PT

(4)

10

ACCEPTED MANUSCRIPT 2000-2014, Fig. 3, Table 1). This approach allows P mobility to be determined under natural

242

conditions rather than exclusively by its chemical solubility using P extraction schemes (e.g.,

243

Psenner et al., 1984). Alternatively, the mobile P pool was calculated by the vertical

244

distribution of P fractions down to 5 cm depth from the same cores used for the gradient

245

method. This direct comparison shows that the mobile P fractions exceed the mobile P

246

determined by the gradient method by a factor of 4 (Table 1). Table 1 contains further cases

247

where a direct comparison of both methods was possible. This result shows that the

248

potentially mobile P pool, determined by chemical extraction, probably drastically

249

overestimates the real amount of mobile P, especially when deeper sediment horizons are

250

considered, where diagenesis is almost complete. This is also supported by Reitzel et al.

251

(2007), who found the nrP pool to constitute a mixture of labile and recalcitrant organic P

252

compounds. Rydin (2000) also found that NaOH-nrP can be resistant to degradation and

253

should not be included in the "mobile-P" pool. Additionally, many studies have shown that

254

the redox sensitive P fraction (BD-P) is to a large extent immobile even in deeper sediment

255

horizons under strongly anoxic conditions (Grüneberg and Kleeberg, 2005). On the other

256

hand, the gradient method is only applicable in sediments with a diagenetically induced

257

decrease of TP down core. For comparison, Table 1 shows the mobile P at two lake sites at

258

two different occasions, i.e. in late summer and during the subsequent winter. The seasonal

259

variability of mobile P pool in L. Arendsee was relatively low and distinct differences

260

between the two sites could not be detected. Therefore the main sampling point is

261

representative of a large area of the profundal zone. Compared to other lakes listed in Table 1,

262

the mobile P of L. Arendsee is very low. Especially in the sediments of shallow and small

263

lakes, mobile P is higher than in deeper stratified lakes. In L. Arendsee, the small mobile P

264

pools seem to be in contrast to the high P release rates during summer because the mobile P in

265

the sediment alone is not sufficient to explain the observed hypolimnetic P accumulation

266

during one summer. Thus, L. Arendsee is an example showing how high hypolimnetic P

AC C

EP

TE D

M AN U

SC

RI PT

241

11

ACCEPTED MANUSCRIPT accumulation is driven by a continuous flux of settling P, rather than a large inventory of

268

mobile P in the sediment (Hupfer & Lewandowski 2005). In summary, the amount and the

269

seasonal variation of mobile P in the sediment of L. Arendsee is low and can actually be

270

neglected with respect to dose calculations (see Table 1).

271

Determination of P fluxes. Plake increased on average by 0.26 t yr-1 between 1995 and 2014

272

(Fig. 2). Within the same period the calculated Pexp was on average 0.37 t yr-1. The P net

273

sedimentation based on sediment core investigations has not changed significantly during

274

recent years (2000-2014). The P amount retained (above the calcareous mud from 1995)

275

increased linearly with time (Fig. 4). The longer the time elapsed since 1995, the lower the

276

influence of the mobile P pool on the reliability of calculating P retention using sediment

277

cores. The linear relationship in Fig. 4 could be used to calculate the mobile P pool, which is

278

the y-axis intercept. The value calculated this way (1.04 g m-2) agrees well with the value

279

determined by the gradient method (0.95 ± 0.32 g m-2, n = 5, see also Table 1). The slope of

280

the line represents the P retention rate (0.34 g m-2 yr-1) and is in concordance with former P

281

retention rates determined using dated sediment cores (Hupfer and Lewandowski, 2005). A

282

value of Psed of 1.0 t yr-1, which was calculated using the lake area below 30 m depth, is used

283

for further considerations. Considering reference depths of 15 m and 40 m as limits for this

284

calculation, Psed could theoretically vary between 1.24 t yr-1 and 0.7 t yr-1, respectively. Based

285

on eq. (1) ∆Plake, Pexp, and Psed were summed to yield Pin = 1.63 t yr-1. This load is only slightly

286

higher than the external P load (Pin) of 1.56 t yr-1 determined as sum of all individual P inputs

287

(Meinikmann et al., 2015). In general, differences could be explained by inevitable

288

uncertainties of both approaches and by the different periods of time considered. The single P

289

sources were monitored only for a short period of at most three years. On the other hand, the

290

mass balance approach demands a longer time span to get reliable results. Uncertainties in the

AC C

EP

TE D

M AN U

SC

RI PT

267

12

ACCEPTED MANUSCRIPT mass balance approach according to eq. (1) include the water residence time and the

292

representative area chosen for the P retention rates determined by sediment cores.

293

Our analysis of water and sediment data has shown that certain input data for the one-box

294

model can in part be provided by alternative ways (Table 2). Direct measurements of external

295

P sources are often not available so that sediment core investigations on dated cores can

296

substitute the time-consuming observation of all input paths. In contrast, the Vollenweider

297

model as an example of an empirical model, would drastically overestimate the external P

298

load for L. Arendsee (Table 1). The high Plake compared to Pin, and the mobile P pool in the

299

sediment implicates that 1) the lake internal P pool is the potential starting point for

300

management measures, and 2) the sediment cannot have a delayed effect. The directly

301

measured external P load and the mass balance-based net sedimentation (Table 2) were used

302

in following management scenario analyses (section 3.2).

303

3.2 Scenario Analyses

304

Fundamental management options for restoring L. Arendsee were compared in Figure 5a.

305

Scenario A0 shows cPlake following a 50% reduction of Pin over a period of five years. Due to

306

its long water residence time of 56 years, there would be a significant delay in the lake’s

307

response to changes of Pin. It will take many years to reach the target P concentration range

308

and the lake will not fulfil the EU WFD water quality standards in the near future. In contrast,

309

a PBC dose sufficient to inactivate the P inventory in the water and sediment (single

310

application) is expected to cause a rapid response. According to scenario B0, a single dose

311

application of PBC without Pin reduction causes an abrupt decrease of cPlake. The

312

sustainability of the application is limited; the target P concentration is exceeded again within

313

less than 10 years. A single application of PBC is only sustainable in combination with a

314

reduction of Pin (scenario AB). In this case, the reduction of Pin could be started before, during

315

or even after the in-lake measure. Figure 5b shows the impact of the time between beginning

AC C

EP

TE D

M AN U

SC

RI PT

291

13

ACCEPTED MANUSCRIPT the Pin reduction and applying a single dose of PBC. Starting Pin reduction earlier than the

317

PBC application has little effect (AB-a) on the course of Plake compared to when internal and

318

external measures begin simultaneously (AB). A delayed start of Pin reduction (AB-b) is only

319

tolerable if this occurs within five years after chemical P precipitation. A longer delay of

320

starting Pin reduction will to lead target P concentrations being exceeded within 10 years.

321

Scenarios AB1 and B1 show the effect of overdosage with and without simultaneous Pin

322

reduction (Fig. 5 c). The chosen overdose is 1.5 times the dose required to eliminate the

323

mobile P pool in the water and sediment (single dose). The overdosage aims to establish an

324

extra binding capacity available for the expected external loading in years to come.

325

Interestingly, the overdose in B1 (without Pin reduction) resulted in lower cPlake than in AB

326

(single doses with Pin reduction) during the first 15 years after application of the PBC.

327

Comparing both scenarios shows that overdosing can compensate the absence of Pin

328

reduction. The abrupt increase of Plake in scenario B1 ten years after the PBC application is

329

explained by the exhaustion of additional P binding capacity in the sediment. Scenario AB1

330

shows the most sustainable result with 50% Pin reduction and an overdosage of PBC. Scenario

331

Β2 is based on a twofold application of PBC within a period of five years (Fig. 5 d). Without

332

P load reduction, the twofold addition of PBC does not change the longevity but decreases

333

cPlake during the period immediately after application compared to the same dose applied in a

334

single treatment (B1). The second treatment would bind P from the catchment during the

335

time between the two treatments.

336

Based on the P course scenarios the respective required PBC over time was determined for

337

three different cases (Fig. 6). Using the same overdose, it takes longer to exhaust the P

338

binding capacity in AB1 (with P load reduction) than in B1 (without P load reduction). Such

339

calculations could be helpful to select an appropriate PBC fulfilling these requirements under

340

the given circumstances. The model used is a suitable tool for selecting appropriate

341

management options. The largest model uncertainty lies in realistically predicting the net P

AC C

EP

TE D

M AN U

SC

RI PT

316

14

ACCEPTED MANUSCRIPT sedimentation coefficient (σ). In this context it is to consider that the development of natural σ

343

(without consideration of added PBC) might change if cPlake decreases drastically. Contrary to

344

the simplified model assumption Psed and Plake probably cannot be linearly extrapolated to

345

different trophic states due to changes in the vertical P transport (e.g., lower intensity of

346

calcite precipitation) and altered sedimentary retention processes (e.g., improved redox

347

conditions). Contrary to the case in many other lakes, the addition of PBC in L. Arendsee

348

does not lock the mobile P pool that accumulates in the sediment but improves the retention

349

capacity for the settling P. Therefore, the expected changes in the phytoplankton composition

350

after restoration could influence P transport such that a higher proportion of P is already

351

released during sedimentation. In this case an overdose is inefficient because the contact of

352

PBC and settling P is limited by the short P circuit in the water body. Although the additional

353

retention capacity due to application of PBCs can be well quantified by the dose, the

354

effectiveness and temporal availability could be influenced by (1) capping with newly settled

355

sediment materials (Lewandowski et al., 2003), (2) aging and crystallization of PBC or other

356

biogeochemical inhibitions (Berkowitz et al. 2006; deVicente et al., 2008) and (3) focusing of

357

the added PBC (Huser 2012). The model by Lewandowski et al. (2003) showed that the

358

proportion of released P that is fixed in the PBC layer decreases exponentially with increasing

359

sediment accretion above. According to this diffusion-based calculation, only 20% of the P

360

released at the sediment surface can be bound in the P sink layer if covered by 4 cm of new

361

sediment. Due to the uncertainties linked with a single overdose application, a partitioned, i.e.

362

twofold PBC application of the same dose would reduce the risk of failure. The scenario

363

analyses help to determine the necessary allocation of additional binding capacity to reach the

364

P targets over a certain period. This kind of P balance calculation is a prerequisite for deciding

365

whether inactivation by PBC is an appropriate option (Schauser et al., 2003; Wauer et al.,

366

2009). The model is not limited to lakes with an annual P retention in the sediment but also

367

applies in lakes where the sediments act as an annual P source, leading to net sedimentation

AC C

EP

TE D

M AN U

SC

RI PT

342

15

ACCEPTED MANUSCRIPT rates < 0. The one-box model is recommended for all kinds of lake restoration measures

369

designed to decrease cPlake because the relevant processes can easily be described by the input

370

parameters. The model can be extended by implementing processes rates and the temporal

371

dynamics of input parameters. Nevertheless, the model cannot replace the necessary system

372

studies but is well suited for comparing the efficiency of different options. After quantifying

373

the required increase in P binding capacity, a suitable PBC and dosage can be selected based

374

on the biogeochemical conditions, the lake’s buffer capacity (alkalinity), and a cost benefit

375

analysis.

SC

RI PT

368

376

4. Conclusions

378

The one-box model is a simple and robust tool for predicting the effectiveness of different

379

external and internal management options designed to decrease the P concentration in

380

eutrophied lakes. The most important step before using the model to optimize the dosage and

381

timing as well as longevity of chemical P inactivation in lakes designated for restoration is to

382

measure or estimate the pools and fluxes of P. The required input data for the one-box model

383

can be provided by different ways including sediment analyses. Data from the present case

384

study and from other lakes have shown that the potentially mobile P pool in the sediment, as

385

determined by chemical fractionation, often overestimates the real P pool necessary to be

386

considered for P inactivation. Scenario analyses with data of a stratified lake have shown that

387

(1) an immediate improvement of the trophic state of lakes with a long water residence time is

388

only achievable through in-lake P inactivation, (2) a strong decrease of the external P load is

389

not always a precondition for internal P inactivation; a delayed start of external measures can

390

still assure the sustainability of in-lake measures, and (3) an overdose of a single or a repeated

391

application is not only effective in lakes with a large sedimentary P pool, but can also partly

392

compensate insufficient P load reduction. The precision of the one-box model is limited

393

because it is difficult to predict the P retention over longer periods of time, particularly under

AC C

EP

TE D

M AN U

377

16

ACCEPTED MANUSCRIPT 394

changed trophic conditions or altered lake chemistry after chemical addition. The one-box

395

model can be extended by implementing specific process rates relevant for the P balance.

396

Acknowlegements

398

We thank Christiane Herzog (IGB) for her conscientious laboratory work. We are grateful to

399

Sylvia Jordan, Thomas Rossoll, Matthias Rothe and Catherin Neumann for field sampling.

400

Thanks to Tom Shatwell for constructive comments and language improvements. Thomas

401

Gonsiorczyk (IGB) is acknowledged for providing sediment data form Lake Stechlin. Part of

402

this study was funded by the State Agency for Flood Protection and Water Management

403

Saxony-Anhalt (LHW) and by the German Research Foundation (DFG, HU 740/5-1). Kasper

404

Reitzel was supported by the Villum Kann Rasmussen Centre of Excellence: Centre for Lake

405

Restoration (CLEAR). We thank the Department of Lake Research of the Helmholtz Centre

406

for Environmental Research (UFZ) and the LHW for providing monitoring data. The

407

manuscript was improved due to inspiring criticism and helpful comments by two anonymous

408

reviewers.

SC

M AN U

TE D

EP AC C

409

410

RI PT

397

411

References

412

Benndorf, J., 2008. Ecotechnology and emission control: Alternative or mutually promoting

413

strategies in water resources management? International Review of Hydrobiology 93 (4–

414

5), 466–478.

415 416

Berkowitz, J., Anderson, M.A., Amrhein, C. 2006. Influence of aging on phosphorus sorption to alum floc in lake water. Water Research 40(5), 911-916. 17

ACCEPTED MANUSCRIPT 417 418

Carey, C.C., Rydin, E., 2011. Lake trophic status can be determined by the depth distribution of sediment phosphorus. Limnology and Oceanography 56 (6), 2051-2063. de Vicente, I., Huang, P., Andersen, F.O., Jensen, H.S., 2008. Phosphate adsorption by fresh

420

and aged aluminum hydroxide. Consequences for lake restoration. Environmental Science

421

& Technology 42 (17), 6650–6655.

RI PT

419

Egemose, S., de Vicente, I., Reitzel, K., Flindt, M.R., Andersen, F.O., Lauridsen, T.L.,

423

Sondergaard, M., Jeppesen, E., Jensen, H.S., 2011. Changed cycling of P, N, Si, and DOC

424

in Danish Lake Nordborg after aluminum treatment. Canadian Journal of Fisheries and

425

Aquatic Sciences 68(5), 842-856.

427

Gächter, R., Imboden, D.M., 1985. Lake restoration. In: Stumm, W. (ed.) Chemical processes

M AN U

426

SC

422

in lakes. New York: Wiley, p. 363–88

Grüneberg, B., Kleeberg, A., 2005. Benthic phosphorus forms and transformations during

429

neutralization of acid mining lakes. In: Serrano L, Golterman H (eds) Phosphates in

430

sediments. Backhuys Publishers, Leiden, pp 127–137

TE D

428

Grüneberg, B., Rücker, J., Nixdorf, B., Behrendt, H., 2011: Dilemma of non-steady state in

432

Lakes - Development and predictability of in-lake P concentration in dimictic lake

433

Scharmützelsee (Germany) after abrupt load reduction.- International Review of

434

Hydrobiology 96 (5): 599-621

EP

431

Grüneberg, B., Dadi, T. Lindim, C., Fischer, H., 2014. Effects of nitrogen and phosphorus

436

load reduction on benthic phosphorus release in a riverine lake. Biogeochemistry 123 (1-

437

2), 185-202.

AC C

435

438

Hickey, C.W., Gibbs, M.M., 2009. Lake sediment phosphorus release management: Decision

439

support and risk assessment framework. New Zealand Journal of Marine and Freshwater

440

Research 43 (3), 819–856.

18

ACCEPTED MANUSCRIPT 441

Hupfer, M., Lewandowski, J., 2005. Retention and early diagenetic transformation of

442

phosphorus in Lake Arendsee (Germany) - consequences for management strategies.

443

Archiv für Hydrobiologie 164 (2), 143–167. Hupfer, M., Gächter, R., Giovanoli, R., 1995. Transformation of phosphorus species in

445

settling seston and during early sediment diagenesis. Aquatic Sciences 57 (4), 305–324.

446

Hupfer, M., Pöthig, R., Brüggemann, R., Geller, W., 2000. Mechanical resuspension of

447

autochthonous calcite (Seekreide) failed to control internal phosphorus cycle in an

448

eutrophic lake. Water Research 34 (3), 859–867.

SC

RI PT

444

Huser, B., Brezonik, P., Newman, R., 2011. Effects of alum treatment on water quality and

450

sediment in the Minneapolis Chain of Lakes, Minnesota, USA. Lake and Reservoir

451

Management 27 (3), 220–228.

452 453

M AN U

449

Huser, B., 2012. Variability in phosphorus binding by aluminum in alum treated lakes explained by lake morphology and aluminum dose. Water Research 46, 4697-4704. Huser, B., Harper, H., Reitzel, K., Jensen, H. S., Egemose, S., Rydin, E. Hupfer, M., Pilgrim,

455

K.. Factors controlling the longevity and effectiveness of aluminum salt addition to restore

456

lakes via reduction of sediment phosphorus release. Water Research (this issue).

TE D

454

Jensen, H.S., Reitzel, K., Egemose, S., 2015. Evaluation of aluminum treatment efficiency on

458

water quality and internal phosphorus cycling in six Danish lakes. Hydrobiologia 751 (1),

459

189–199.

AC C

EP

457

460

Jeppesen, E., Søndergaard, M., Jensen, J.P., Havens, K.E., Anneville, O., Carvalho, L.,

461

Coveney, M.F., Deneke, R., Dokulil, M.T., Foy, B., Gerdeaux, D., Hampton, S.E., Hilt, S.,

462

Kangur, K., Köhler, J., Lammens, E., Lauridsen, T.L., Manca, M., Miracle, M.R., Moss,

463

B., Noges, P., Persson, G., Phillips, G., Portielje, R., Schelske, C.L., Straile, D., Tatrai, I.,

464

Willen, E., Winder, M., 2005. Lake responses to reduced nutrient loading - An analysis of

465

contemporary long-term data from 35 case studies. Freshwater Biology 50 (19), 1747–

466

1771. 19

ACCEPTED MANUSCRIPT 467

Lewandowski, J., Schauser, I., Hupfer M., 2002. The importance of sediment studies in the

468

selection of restoration measures. Hydrologie und Wasserbewirtschaftung 46: 2-13 (in

469

German with English abstract). Lewandowski, J., Schauser, I., Hupfer, M., 2003. Long-term effects of phosphorus

471

precipitations with alum in hypereutrophic Lake Süsser See (Germany). Water Research

472

37 (13), 3194–3204.

474

Lürling, M., van Oosterhout, F., 2013. Controlling eutrophication by combined bloom precipitation and sediment phosphorus inactivation. Water Research 47: 6527-6537.

SC

473

RI PT

470

Mehner, T., Benndorf, J., Kasprzak, P., Koschel, R., 2002: Biomanipulation of lake

476

ecosystems: successful applications and expanding complexity in the underlying science.

477

Freshwater Biology 47: 2453-2466.

M AN U

475

Mehner, T., Diekmann, M., Gonsiorczyk, T., Kasprzak, P., Koschel, R., Krienitz, L., Rumpf,

479

M., Schulz, M., Wauer, G., 2008. Rapid recovery from eutrophication of a stratified lake

480

by disruption of internal nutrient load. Ecosystems 11 (7), 1142−1156.

483 484 485 486

a potential source of lake eutrophication. Journal of Hydrology 524, 214–226. Murphy, J., Riley, J.P., 1962. A modified single solution method for determination of

EP

482

Meinikmann, K., Hupfer, M., Lewandowski, J., 2015. Phosphorus in groundwater discharge –

phosphate in natural waters. Analytica Chimica Acta 27 (1), 31–36. Pilgrim, K.M., Huser, B.J., Brezonik, P.L., 2007. A method for comparative evaluation of

AC C

481

TE D

478

whole-lake and inflow alum treatment. Water Research 41 (6), 1215–1224.

487

Psenner, R., Pucsko, R., Sager, M., 1984. Die Fraktionierung organischer und anorganischer

488

Phosphorverbindungen von Sedimenten - Versuch einer Definition ökologisch wichtiger

489

Fraktionen. Archiv für Hydrobiologie/Supplement 70, 111–155.

490

Reitzel, K., Hansen, J., Andersen, F.Ø., Hansen, K.S., Jensen, H.S., 2005. Lake restoration by

491

dosing aluminum relative to mobile phosphorus in the sediment. Environmental Science &

492

Technology 39 (11), 4134–4140. 20

ACCEPTED MANUSCRIPT 493

Reitzel, K., Ahlgren, J., DeBrabandere, H., Waldeback, M., Gogoll, A., Tranvik, L., Rydin, E.

494

(2007) Degradation rates of organic phosphorus in lake sediment. Biogeochemistry 82 (1),

495

15-28. Rydin, E., Welch, E.B., 1999. Dosing alum to Wisconsin lake sediments based on in vitro

497

formation of aluminum bound phosphate. Lake and Reservoir Management 15 (4), 324–

498

331.

501 502

(7), 2037-2042.

SC

500

Rydin, E., 2000. Potentially mobile phosphorus in lake Erken sediment. Water Research 34

Sas, H., 1989. Lake restoration by reduction of nutrient loading. Expectations, Experiences, Extrapolations. Academia-Verlag Richarz, Sankt Augustin, 497 pp.

M AN U

499

RI PT

496

503

Schallenberg, M., Sorrell, B., 2009. Regime shifts between clear and turbid water in New

504

Zealand lakes: Environmental correlates and implications for management and restoration

505

New Zealand Journal of Marine and Freshwater Research 43 (3), 701–712.

508 509

TE D

507

Scharf, B.W., 1998. Eutrophication history of Lake Arendsee (Germany). Palaeogeography, Palaeoclimatology, Palaeoecology 140 (1–4), 85–96. Schauser, I., Chorus, I., 2007. Assessment of internal and external lake restoration measures for two Berlin lakes. Lake and Reservoir Management 23:366-376.

EP

506

Schauser, I., Lewandowski, J., Hupfer, M., 2003. Decision support for the selection of an

511

appropriate in-lake measure to influence the phosphorus retention in sediments. Water

512

Research 37 (4), 801–812.

AC C

510

513

Schippers, P., van de Weerd, H., de Klein, J., de Jong, B., Scheffer, M., 2006. Impacts of

514

agricultural phosphorus use in catchments on shallow lake water quality: about buffers,

515

time delays and equilibria. Science of the Total Environment 369 (1–3), 280–294.

516

Shatwell, T., Jordan, S., Ackermann, G., Dokulil, M., Rücker, J., Scharf, W., Wagner, A.,

517

Kasprzak, P., 2013. Long-term monitoring oft the impact of climate change and

21

ACCEPTED MANUSCRIPT 518

eutrophication on lakes and reservoirs. Korrespondenz Wasserwirtschaft 6 (12): 729-736

519

(in German with English abstract).

520 521

Smith, V.H., Schindler, D.W., 2009. Eutrophication science: Where do we go from here? Trends in Ecology and Evolution 24 (4), 201–207. Smeltzer, E., Kirn, R.A., Fiske, S., 1999. Long-term water quality and biological effects of

523

alum treatment of Lake Morey, Vermont. Lake and Reservoir Management 15 (3), 173–

524

184.

527 528

SC

526

Søndergaard, M., Bjerring, R., Jeppesen, E., 2013. Persistent internal phosphorus loading during summer in shallow eutrophic lakes. Hydrobiologia 710 (1), 95–107. Spears, B.M., Dudley, B., Reitzel, K., Rydin, E., 2013a. Geo-engineering in lakes: a call for

M AN U

525

RI PT

522

consensus. Environmental Science & Technology 47, 3953–3954. Spears B.M., Lürling, M., Yasseri, S., Castro-Castellon, A.T., Gibbs, M., Meis, S.,

530

McDonald, C., McIntosh, J., Sleep, D., Van Oosterhout, F., 2013b. Lake responses

531

following lanthanum-modified bentonite clay (Phoslock®) application: an analysis of

532

water column lanthanum data from 16 case study lakes. Water Research 47 (15), 5930–

533

5942.

TE D

529

Spears, B.M., Maberly, S.C., Pan, G., Mackay, E., Corker, N., Douglas, G., Egemose, S.,

535

Hamilton, D., Hatton-Ellis, T., Huser, B., Li, W., Meis, S., Moss, B., Lürling, M., Phillips,

536

G., Yasseri, S., Reitzel, K., 2014. Geo-engineering in lakes: a crisis of confidence?

537

Environmental Science & Technology 48 (17), 9977–9979.

539

AC C

538

EP

534

Uhlmann, D., Paul, L., Hupfer, M., Fischer, R., 2011. Lakes and Reservoirs. Treatise on Water Science, Vol. 2: The Science of Hydrology 01/2011, Elsevier, 157–213.

540

Vollenweider, R.A., 1976. Advances in defining critical loading levels for phosphorus in lake

541

eutrophication. Mem. Ist. Ital. Idrobiol. 33, 53–83.Wauer, G., Gonsiorczyk T., Hupfer,

542

M., Koschel, R., 2009. Phosphorus balance of Lake Tiefwarensee during and after

22

ACCEPTED MANUSCRIPT 543

restoration by hypolimnetic treatment with aluminum and calcium salts, Lake and

544

Reservoir Management, 25 (4), 377-388

547 548 549 550

alumn. Lake and Reservoir Management 15 (1), 5–27. Withers, P.J.A., Neal, C., Jarvie, H.P., Donnacha, G., Doody, D.G., 2014. Agriculture and

RI PT

546

Welch, E.B., Cooke, G.D., 1999. Effectiveness and longevity of phosphorus inactivation with

Eutrophication: Where Do We Go from Here? Sustainability 6, 5853–5875.

Zamparas, M., Zacharias, I., 2014. Restoration of eutrophic freshwater by managing internal nutrient loads. Science of the Total Environment 496, 551-562

SC

545

Zwirnmann, E., Krüger, A., Gelbrecht, J., 1999. Analytik im Zentralen Chemielabor.

552

Jahresbericht des IGB (Leibniz-Institut für Gewässerökologie und Binnenfischerei) 9, 3–

553

24.

AC C

EP

TE D

M AN U

551

23

ACCEPTED MANUSCRIPT

Table 1. Mobile P pool in the sediments of L. Arendsee in comparison to sediments from other lakes. The mobile P Pool was calculated by the gradient method (see Hupfer et al., 2005) and by the fractionation method (e.g., Reitzel et al., 2005). The fractionation method considered defined

RI PT

depths of 0-5 cm and 5-10 cm. EP: depth of the endpoint of early diagenesis (see Fig. 3), mix- mixis type, p- polymictic, di- dimictic, mmonomictic, tr- trophic state, hy- hypertroph, eu- eutroph, me- mesotroph, o- oligotroph1 - present study; original data were drawn and recalculated from 2 - Grüneberg et al. (2015), 3 - Hupfer et al. (1995), 4 - Lewandowski et al. (2002), 5 – Egemose et al. (2011), 6 - Reitzel et al. (2005), 7-

L. Arendsee (D)

km2 5.14

m 49

di

Lower Havel (D)* L. Sempach (CH) L. Auensee (D) L. Scharmützel (D L. Nordborg (DK) L. Sønderby

11.8 14,5 0.12 12.1 0.55 0.08

10.5 87 7.8 29.5 8.5 5.7

p m di di di di

L. Stechlin

4.23

69.5

di

*riverine lake

tr Sampling date depth m eu 2008-09 49 40 2009-02 49 40 2000-08 49 eu 2011-10 7 eu 1992-06 87 hy 1999-08 7.8 me 2003-05 29.5 eu 2006-07 8.5 hy 2001-02 5.0 2.0 o/me 2015-06 69.5

Mobile P pool TP gradient Fractionation EP 0-5 cm 0-10 cm -2 -2 mg m cm mg m 867 5 3687 915 5 3236 747 5 3480 861 5 3499 837 3 2836 16100 10 7458 19424 2400 3 4546 9338 2781 6 3966 321 5 2428 5646 1042 8 3277 6537 2736 14 4347 8351 13607 14 9786 15860 5356 8 5115 5497

M AN U

mix

TE D

zmax

EP

area

AC C

Lake

SC

Gonsiorczyk (unpubl.).

24

Ref.

1

2 3 4 1 5 6 7

ACCEPTED MANUSCRIPT

Table 2. Overview about the parameter relevant for P management measures and the calculations with the one-box model in L. Arendsee [1]

External P load

Method Pin

Ref.

Data for L. Arendsee

[1]

1.56 t yr-1*

[2]

1.63 t yr-1

Pin = cPlake ∗ zm/τ ∗ √ ∗  ∗ 

[3]

4.33 t yr-1

[1]

56 years*

[2]

184 µgP L-1*

Sum up of single external P sources

M AN U

Balance Pin = Psed + Pexp + ∆Plake

SC

Parameter

RI PT

Meinikmann et al. (2015), [2] this study, [3] Vollenweider (1976), [4] Hupfer & Lewandowski (2005) *used for scenario analyses in section 3.2

Water residence time

τ

Water balance

Lake P concentration

cPLake

Mean annual value of volume weighted cPlake

27.0 t

Mobile P pool

Sed Pmobil

Sedimentary TP gradient

[4], [2]

2.85 t

Net sedimentation

σ

P retention in sediments

[2]

0.037

EP

TE D

(2005-2014)

[2]

0.044*

[2]

0.77*

coefficient

Mass balance (steady state)

Stratification factor

ß

AC C

Psed= Pin -Pexp

cP0-15m/cPLake (2005-2014)

25

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Fig. 1 Map of the subsurface and the above ground catchment area of L. Arendsee and the usage structure. The flow direction of groundwater (grey arrows) is from southeast to northwest passing the town Arendsee before reaching the lake (see Meinikmann et al., 2015). The included bathymetric map of L. Arendsee shows the main sampling station at the deepest site of the lake.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

TE D

Fig. 2 Long-term trend of the annual means of total phosphorus (TP) in water of L. Arendsee as volume weighted concentration (left axis) and as whole mass (right axis) (522 sampling dates).

AC C

EP

Error bars represent the standard deviation.

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 3 Total P content in the sediments layers versus cumulative dry mass per area (CDMA) over a twenty year period before and after capping with calcareous mud (dark grey layer) at the

EP

main sampling point. The mobile P was calculated by the difference between the end point TP of early P diagenesis and the TP in the layers above this point (shadow area) (see also Tab. 1);

AC C

extended from Hupfer and Lewandowski (2005).

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 4 Total phosphorus retained in the profundal sediments (Deposited P) since end of the year

EP

1995 (as dated by the calcareous capping layer, compare Fig. 3). Randomly sampled sediment cores at water depths deeper than 40 m, between 2000 and 2014. Each data point is the average of 4 to 7 cores sampled at the same date. Error bars represent the standard deviation. Deposited P

AC C

increased linearly over time with 0.33 g m-2 yr-1 (n= 10, r2= 0.969). The intersection with the second axis indicates the mobile P pool of 1.04 g m-2.

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig. 5 Prediction of cPlake under different management options with the one-box model. a)

EP

Simulation of basic options: Scenario A0: 50% P load reduction within 5 years; B0: P inactivation by chemical P precipitation, single dose; AB: P inactivation by PBC and 50% P

AC C

load reduction within 5 years). b) Simulation of time shifts between external and internal measures: AB-a: external P load reduction is finished before internal P inactivation; AB-b: external measures begin 5 years after internal P inactivation. c) Effect of PBC overdoses (1.5 times of single doses) without (B1) and with (AB1) external P load reduction. d) Effect of twofold PBC application without external P load reduction (B2) (same doses as B1 and AB1). Green area: range of target cPlake for L. Arendsee.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

TE D

Fig. 6 Temporal development of required P binding capacity after adding of phosphorus binding chemicals (PBC) expressed as P equivalents (tons) for the scenarios with single application without (B1) and with simultaneous P load reduction (AB1) in comparison with two-fold application without P load reduction (B2). Necessary binding capacities: a-mobile P pool in the

AC C

EP

sediment, b-P in the lake water, c-increasing of P retention in the sediment

ACCEPTED MANUSCRIPT

Highlights Enhanced P retention by restoration measures shortens the adaptation time of lakes

•

The one-box model is a useful tool to study scenarios of different measures

•

If water residence time is long only in-lake P inactivation is immediately effective

•

Gradual external P load reductions assure the sustainability of P precipitation

•

The risk of twofold precipitation is lower than single applications with an overdose

AC C

EP

TE D

M AN U

SC

RI PT

•