Phosphorus balance of Lake Tiefwarensee during and after restoration by hypolimnetic treatment with aluminum and calcium salts

Abstract

Between 2001 and 2005, the recovery of Lake Tiefwarensee from eutrophic to mesotrophic state was successfully accelerated by the stepwise hypolimnetic addition of 137 g aluminum and 154 g calcium per square meter of profundal sediment. In response to the treatment, an 8-cm sediment cover was formed, which almost completely suppressed the phosphorus (P) release from the sediments, and is still present. The spatial variability of the sediments was analyzed at eight sampling points at different lake depths. With increasing lake depth, soluble reactive phosphorus decreased in the pore water, whereas the total phosphorus (TP) increased in treated sediment. Total P in the upper sediment layer (0–10 cm) increased by about 3 tons during the treatment period, consistent with the simultaneous decrease in the water from 0.223 mg/L in 1998 to 0.013 mg/L in 2005 (annual mean values for the whole water body). After initial settling, the drastic TP decrease in the water column can be attributed to an increase in the sediment P-binding capacity, which is related to a decrease of the mobile P pool (NH₄Cl-TP) and a strong increase in the Al:P ratio in sediment. In the 3 years after completion of the treatment, the lake TP concentration was well described by the Vollenweider model, indicating that a sustainable state of nutrient equilibrium was achieved.

Key words:

• Al
• Ca
• lake restoration
• P balance
• P retention

Despite external load reductions, many lakes have either failed to recover from eutrophication or their recovery was delayed for many years (Marsden 1989, Jeppesen et al. 2005). Lake Tiefwarensee recovery was delayed because of the long hydraulic residence time and the ability of its sediments to release stored nutrients, such as phosphorus (P), over long periods. Various in-lake restoration measures have been developed and implemented to overcome such problems (e.g., Cooke et al. 2005). Some of them aim to increase the P-binding capacity of the sediments by aeration, the addition of nitrate (Foy 1986, Søndergaard et al. 2000) or the addition of P-binding substances such as Al^{3 +}, Fe^{3 +} or Ca^{2 +} (Reitzel et al. 2005, Cooke et al. 1993, Prepas et al. 2001). Although deep and stratified lakes in particular suffer from delayed recovery, only a few such restoration case studies have been described in detail. Moreover, knowledge about the effect of in-lake measures on the sediment's P-binding capacity is still incomplete (Reitzel et al. 2005, 2007, de Vicente et al. 2008, Dugopolski et al. 2008).

In Tiefwarensee, nutrient loading peaked in the 1980s due to intensive farming in the watershed and the inflow of industrial and municipal sewage. During spring circulation in 1983, total phosphorus (TP) concentration in the lake water reached 0.79 mg/L (Weiβ 1985), causing constraints for the different users. Since then, major efforts have been made to reduce external nutrient loading from both waste water treatment and extensive agricultural land use in the watershed, resulting in a large decrease of TP concentration in Tiefwarensee. Nonetheless, the TP concentrations during spring circulation of 1998–2001 remained higher than expected at 0.17 mg/L, indicating the persistence of a hyper eutrophic state. The reason for this persistence was probably the continuing high level of internal P recycling from the sediments.

Between 2001 and 2005, a new restoration measure of combined hypolimnetic Al and Ca(OH)₂ treatments was performed to reduce the P release from deep sediments (Koschel et al. 2006). The restoration project was accompanied by extensive limnological investigations before, during and after the chemical treatments. We describe and discuss:

	the chemical restoration process at Tiefwarensee;
	its effects on P retention in the sediment and P balance of the whole lake; and
	the further development of the lake's trophic state using lake models based on the Vollenweider model (Vollenweider 1976) and a one-box model (Schauser et al. 2003).

Materials and methods

Study site

Tiefwarensee, a dimictic hardwater lake, is situated in the Mecklenburg Lake District of Germany, near the city of Waren, about 120 km north of Berlin. Tiefwarensee has three main tributaries: the City Moat (Stadtgraben, Z1), New Moat (Neuer Graben, Z2), and inflow from Lake Melzer See (Z3; Fig. 1). Based on the size and structure of the catchment area and its morphometry (Table 1), Tiefwarensee's potential trophic state is mesotrophy (LAWA 1999), which meets the requirements of the Water Framework Directive (WFD, EC Parliament and Council 2000).

Table 1 Morphometric characteristics of Tiefwarensee.

Parameter	Value
Surface area	1.41 km^2
Mean depth	9.6 m
Maximum depth	23 m
Shoreline length	8.58 km
Catchment area	20.48 km^2
Subsurface catchment area	15.4 km^2
Groundwater formation rate	2.817 L/s · km^2

Figure 1 Bathymetric map of Tiefwarensee (geographical position: 53°32′N; 12°41′E), including tributaries and location of the deep water aerator (TIBEAN) and monitoring stations. With permission from the Environmental Ministry of the State of Mecklenburg-Vorpommern, Department of Integrated Environmental Protection and Sustainable Development – Lakes Project 2001.

Restoration technique

To control P release from the sediment and to increase P retention in the sediment, we used a new restoration technique that combines additions of hypolimnetic sodium aluminate and calcium hydroxide with deep water mixing and aeration. The chemicals were added into the hypolimnion at a depth of 10 m (for equipment details see Koschel et al. 2006). Between 2001 and 2005, 8.14 * 10⁵ kg of NaAl(OH)₄ solution (HTFIX alkalisch 10%, BRENNTAG GmbH, Germany, 151.3 g Al^{3 +}/L) and 1.74 * 10⁵ kg of Ca(OH)₂ (Rüdersdorfer Zement GmbH, Germany, mixed on site to 30% in lake water) were added to the lake in several cycles per year (Table 2). Each cycle consisted of 2 weeks of chemical additions, 2 weeks without treatment but with operation of the deep water aerator to enhance mixing, and 2 weeks without either treatment or aeration to encourage sedimentation. Adapted fisheries management was conducted as additional restoration support. Planktivorous cyprinid fish were removed and piscivorous fish were stocked, but this was of minor importance for the restorations success and is not the subject of the work presented here (Mehner et al. 2008).

Table 2 Treatment cycles and amounts of chemicals added to Tiefwarensee. Relative amounts per square meter sediment surface find in parentheses.

Year	Month	Aluminum [+]^(1) (g Al^3+/m^2(2))	Calcium Ca(OH)2 [t] (g Ca^2+/m^2) ^1
2001	July	11.7 (19)
	Aug/Sep		45 (40)
2002	May/June	12.6 (21)
	July		44 (40)
	Aug/Sep	11.9 (19)
2003	June	12.2 (20)
	July/Aug		42 (38)
	Sep/Oct	12.2 (20)
2004	July	11.7 (19)
	Aug/Sep		43 (38)
2005	June	11.5 (19)
Total amount		83.8 (137)	174 (154)

Sampling procedures and methods

Water samples from the epilimnion (mixed sample: 0–5 or 0–10 m, depending on the epilimnion thickness) and hypolimnion (20 m) from sampling point P1 (maximum water depth), as well as surface water samples from tributaries Z1 and Z3 (Fig. 1), were collected monthly for analysis of soluble reactive phosphorus (SRP) and TP beginning 1998, except in 2000, when only four measurements were taken. During summer stratification in 2001 to 2005 sampling was increased to biweekly intervals (from monthly).

Hydrological investigations of the catchment (inflow velocity of the tributaries, groundwater, atmospheric deposition, water sampling and P analysis on Z2) were performed by the Neubrandenburg State Department of Environment and Nature (StAUN) and by HGN Hydrogeologie GmbH Co.

The SRP in lake water was analyzed in subsamples filtered in the laboratory immediately after sampling through 0.45-μ m membrane filters photometrically (695 nm) with SnCl₂ method (Foss Tecator 1997) on a flow injection analyzer (FIA Star 5010, Foss Tecator, Sweden). The TP was likewise analyzed after digestion of the unfiltered sample with 0.18 M K₂S₂O₈ (30 min, 134°C in a steam autoclave).

During the summer stratification period, sediment from P1 was analyzed at least three times annually in all years since 1998 except 2000. To assess the horizontal variability of sediment phosphorus in 2001, 2002 and 2007, samples were taken from sampling points P1–P8 (Fig. 1). On each sampling date, four sediment cores (6-cm dia) per sampling point were collected using a Uwitec® gravity corer. For the analysis of the SRP concentration above the sediment, some of the water overlying the sediment cores was subsampled immediately to avoid concentration changes during the transport of the cores. The sediment cores were sliced (0–1, 1–2, 2–4, 4–6, 6–8 and 8–10 cm) in the laboratory, and cores from the same location were pooled by horizons.

Pore water from each horizon was obtained by centrifugation (5 min, 13,000 g, 4°C) and filtration (0.45-μ m pore size). The SRP concentrations in pore water were analyzed photometrically with a Nanocolor photometer (Macherey & Nagel, Germany) following the molybdenum blue method of Murphy and Riley (1962). The amount of TP in the particular fraction of the sediment was obtained after digestion of ignited (550°C, 2 h) sediment (0.18 M K₂S₂O₈, 30 min, 134°C in steam autoclave) in the same way.

Dry weight was measured by drying the sediment at 105°C for 24 h. Organic matter and CaCO₃ were determined as loss on ignition at 550°C and 900°C, respectively. Release of P from the sediment was calculated from the SRP concentration gradients between the sediment surface and the overlying water according to the modified Fick's first law of diffusion (Sinke et al. 1990, Gonsiorczyk et al. 1997):

in which J = diffusive flux (mg/m^2.d); Φ = porosity (calculated from the composition of dry sediment); D_i = diffusion coefficient (m²/d; interpolated from data of Li and Gregory (1974) to D _i = (3.267 + 0.1617 * T(°C)) * 10⁻⁵ for the respective temperatures T above the sediments); and dc/dx = linear gradient between the concentrations in the 0–1 cm layer of the sediments and in the overlaying water.

Sediment P binding forms were characterized using the fractionation scheme of Psenner et al. (1984) with the modifications by Hupfer et al. (1995). Six P species were identified: NH₄Cl-TP (pore water P, loosely adsorbed to surfaces); BD-TP (redox-sensitive P, mainly bound to Fe and Mn compounds); NaOH-SRP (inorganic P bound to metal oxides mainly of Al and Fe); NaOH-NRP (organically bound P); HCl-TP (Ca- and Mg-bound P); and residual P (P_res) consisting of organic and other refractory P.

The content of freshly precipitated Al in the sediment was determined photometrically with pyrocatecholviolet in the NaOH extract of the P fractionation after acidification and dilution (Wauer et al. 2004).

Calculation of the P balance

The external phosphorus load (P_in) was calculated from the inflow-weighted TP values in the tributaries, the atmospheric deposition (30 kg P/km² · year; LAWA 1999) and from the groundwater inflow. Due to lack of measured inflow velocities in 2008 the hydraulic residence time and P_in were calculated as mean values 2006/2007, because the average annual rainfall in this region in 2008 was between that of 2006 and 2007. Phosphorus output (P_out) was calculated using the mean TP concentrations in the surface water and by assuming that the water-inflow equals the outflow. The lake P content in the water column (P_lake) was calculated from TP concentrations in the epilimnetic mixed sample and the hypolimnetic sample from 20 m, assuming an epilimnion depth range of 0–8 m (corresponding to a volume of 8.3 × 10⁶ m³ and a hypolimnetic volume of 4.6 × 10⁶ m³). The hypolimnetic P accumulation (P_accu) during summer stagnation was calculated from SRP concentration changes in the hypolimnion from the beginning to end of the summer stagnation period based on the SRP values in 20 m.

In 2001, 2002 and 2007 the spatial variability of the SRP-concentrations in porewater and TP content in the sediment was investigated at different horizontal sampling points (Fig. 1). Considering the growing cover on the sediments, depth-weighted mean values were compared as 2001: 0–1 cm sediment; 2002: 0–2 cm sediment; and 2007: 0–10 cm sediment. Assuming that the treated sediments were located at lake depths below 10 m (i.e., 0.61 km²) and that there was a homogeneous sediment composition in this region before treatment, the area-weighted mean SRP (SRP_mean) and TP (TP_mean), represented in mg/L and mg/kg dry weight, respectively, can be empirically calculated by linear regression from the P1 data as follows:

The mean values SRP_mean and TP_mean were used to calculate the P release (P_rel) and the P content in sediment (P_sed) during and after the restoration period using the P1 sediment data.

The P release (P_rel) was calculated using the summer means for P diffusion with the simplifying assumption that the annual P release occurred only from sediments below 10 m lake depth (0.61 km²) during the stratification period lasting 160 d/yr. The P diffusion out of the sediment, as basis for P_rel, was calculated from the SRP gradient between sediment and water, according to equation 1. The 0–1 cm sediment layer SRP was adapted according to equation 2, whereas the concentration in the overlying water was calculated using the original P1 data because the SRP in the overlying water was indistinguishable at sampling points P1–P8. The P_sed was calculated for the sediment surface (0–10 cm) below a water depth of 10 m (0.61 km²).

Statistical tests were made with the program SPSS 14. The trends in time were compared by linear regression followed by ANOVA testing the significance of the slope.

Results

External phosphorus load and P content in water

The external phosphorus load (P_in) was about 300 kg P/yr between 1998 and 2002 (Table 3). Compared to about 9000 kg P in 1983 (Weiβ 1985), this was an immense load reduction. The city of Waren has continued its efforts to minimize Tiefwarensee's external load in the last years. In 2002, for example, a constructed barrage in the City Moat (Stadtgraben; Fig. 1) formed a wetland area resulting in prolonged hydraulic residence time and a further decrease in P_in since 2003 (Table 3). The annual external P load calculations are variable because short-term events (such as rainfall and storms) are stochastic and are not represented in the monthly monitoring data.

Table 3 Phosphorus budget parameters of Tiefwarensee from 1998 to 2008: external load (P_in), P outflow (P_out), P content (P_lake), P retention (P_ret), P accumulation in the hypolimnion (P_accu), P release from the sediment (P_rel), and P content of the upper sediment (0–10 cm) (P_sed).

Year	Hydraulic residence time (years)	P in (kg/year)	P out (kg/year)	P lake (kg)	P ret ^1 (kg/year)	P accu (kg/summer stagnation)	P rel (kg/year)	P sed (kg)
1998	4.73	325	404	2877		2268	493
1999	4.77	291	329	2909	−70	2125	750	9231
2000	4.88	277	365	2934	−113
2001	4.73	312	275	1793	1179	−152	366	6897
2002	4.73	304	140	640	1317	−271	223	9444
2003	4.86	254	66	258	570	5	82	7979
2004	5.39	198	34	179	243	0	27	9541
2005	5.52	178	32	171	154	5	22	10411
2006	5.52	148	37	228	54	83	0.2	12898
2007	5.44	191	47	274	99	166	458	12800
2008	5.48 ^2	170 ^2	47	300	96	225	77

Before restoration, the P content of the lake (P_lake) at the end of summer stagnation was dominated by the hypolimnetic accumulation of P during summer stagnation (Fig. 2). When Al addition started in July 2001, P accumulation in the hypolimnion had already begun and was immediately interrupted by the treatment, whereas the epilimnion still was unaffected in 2001. The epilimnion was only influenced starting with the mixing period between 2001 and 2002, as manifested by the decreased P concentration at the beginning of 2002. In the following years of the treatment, the P concentrations decreased further, and no hypolimnetic accumulation of phosphorus was observed during the summer stagnation period. In 2008, the annual epilimnetic mean TP was 0.020 ± 0.004 mg/L (n = 12), corresponding to 18% of the levels in 1999 (0.112 ± 0.048 mg TP/L, n = 10). Consequently, P_lake decreased from 2934 kg in 2000 to 1793 kg in 2001, with further decreases to below 200 kg until 2005. Since 2006 it has slightly increasing and reached 300 kg in 2008 (Table 3). Accordingly, P_lake decreased to nearly 10% of the levels measured in 1998–2000.

Figure 2 TP concentration in Tiefwarensee from 1998 to 2007.

Temporal changes in sediment

Before restoration, the SRP concentrations in pore water increased drastically with sediment depth, and there was a steep gradient between the sediment and the overlying water (Fig. 3). During and after sediment treatment, the SRP profiles showed much lower concentrations, and the gradient between sediment and water almost completely disappeared by 2002. The SRP pore water concentration in the upper sediment layer (0–8 cm) after sediment treatment was about 0.02 mg/L. The added chemicals obviously formed a P-absorbing cover, growing to a thickness of about 8 cm in 2005 that remained constant in the following years. Hence, computed P diffusion from the sediment decreased drastically (Table 4), causing a decrease in the annual P release from the sediment (Table 3).

Table 4 Estimated mean P diffusion rates (mg P/m²·d) from the sediment of Tiefwarensee during the summer stagnation period at sampling point P1 as computed from porewater SRP concentrations and equation 1.

Year	1998	1999	2001 ^1	2001 ^2	2002	2003	2004	2005	2006	2007	2008
P diffusion (mg/m^2·d) ± SD	5.05 ± 4.44	7.68 ± 1.15	25.27 ± 15.1	0.21 ± 0.13	0.21 ± 0.20	0.05 ± 0.07	0.01 ± 0.04	0.01 ± 0.05	0.28 ± 0.14	1.50 ± 2.46	0.32 ± 0.12
n	3	2	2	3	4	3	3	3	2	3	2

Figure 3 Profiles of SRP in pore water, TP and Al (extracted by 1M NaOH) in the sediment of Tiefwarensee before (1989–2001), during (2002, 2005) and after (2006, 2007) sediment treatment with Al and Ca.

The upper sediment layers showed a tendency to smaller relative dry weight after Al and Ca treatment. Therefore, the sediment data used were related to the wet matter volume for comparisons before and after addition of the chemicals. In the TP profile, a maximum of 600 μ g P/cm³ was observed (Fig. 3). That maximum moved downward to a depth of 6–8 cm during treatment and has remained there nearly unchanged since the end of treatment in 2005.

Mean Al content in the top 8 cm of the sediment increased from < 0.05 mg/cm³ before the treatment to 2.77 mg/cm³ in 2007 forming a maximum between 2 and 8 cm sediment depth (Fig. 3). The maximum corresponds to the P-absorbing cover, apparent as decreased SRP porewater concentrations and increased TP content in sediment. But the shape of the Al maximum in the sediment profile differs from the TP maximum (Fig. 3).

The P_sed increased by about 3 tons from 1999 to 2007, corresponding to the simultaneous decrease in P_lake (Table 3). The TP content in the upper 10 cm of sediment at P1 increased significantly (R² = 0.219; F = 4.759, p = 0.043, n = 19) from 1998 (185 μ g/cm³) to 2007 (336 μ g/cm³; Fig. 3). The TP increase in the sediment of Tiefwarensee during the restoration period occurred mainly in the NaOH-SRP fraction (depth weighted mean of 0–10 cm sediment from 22 to 193 μ g/cm³, R² = 0.500; F = 7.988, p = 0.022, n = 10), whereas the NH₄Cl-TP decreased (from 17 to 4 μ g/cm³, R² = 0.712; F = 19.796, p = 0.002, n = 10) in the same time (Fig. 4). All other P-binding forms did not change significantly in this time. In particular, the temporarily bound P pool amounted to about half of the TP concentration in the upper sediment horizons before chemical treatment, and nearly the same distribution of P-binding forms was found in the sediment horizon between 4 and 6 cm in 2002 and between 8 and 10 cm in 2007 after treatment (Fig. 4). That finding is consistent with the thickness of the newly formed, TP-rich sediment cover and low SRP concentrations in pore water.

Figure 4 Distribution of the P-binding forms in the sediments of Tiefwarensee before (28 Sept. 1998), during (1 Aug. 2002) and after sediment treatment with Al and Ca (17 Oct. 2005, 14 May 2007).

Discussion

The aim of the restoration measures was to accelerate the recovery of Tiefwarensee to a mesotrophic state (i.e., to an annual mean TP concentration < 0.03 mg/L in the lake water). This goal was achieved, as evidenced by the annual mean TP concentration of 0.023 mg/L in 2008 (Fig. 2). Following the summer mean Secchi disk depths increased from 2.75 m before the treatment (1998–1999) to 4.30 m (2006–2008). The chl-a concentration in the epilimnion (summer mean) decreased from 9.7 mg/m³ (1998) to 5.0 mg/m³ in 2008 (Mehner et al. 2008).

According to the P budget (Table 3), the largest changes in P_lake as well as in P_sed occurred from before (1998–2000) to after restoration (2006–2007). The difference of about 3 tons for each parameter between 1999 and 2006–2007 equates to the P pool transferred from the water into the sediment during the restoration period.

The P release from sediment (P_rel) decreased during the restoration period, and this result agrees with the development of hypolimnetic P accumulation (P_accu; Table 3). Before restoration, extreme hypolimnetic P enrichment was observed during the summer stagnation period (Fig. 2). This enrichment decreased immediately after the chemical treatments in 2001. The annual treatment cycles started in July 2001 and May 2002, respectively (Table 2). The negative P_accu values in 2001 and 2002 suggest that during the first years of treatment the hypolimnetic P concentration increased throughout the stagnation period due to P release from the sediments, but that the chemicals precipitated the released phosphorus. In the following years, the sediment cover sealed very effectively throughout the year. Since 2006, P_accu and P_rel have moderately increased again.

The TP concentrations during the spring circulation period (TP_spring) of Tiefwarensee were compared with those predicted by the Vollenweider equation (Vollenweider 1976) based on the external P load (Fig. 5). The Vollenweider model applies when there is a steady state between P load and P concentration in the lake. Because Tiefwarensee was in transition (eutrophication until 1983, followed by recovery) the Vollenweider equation initially overestimated TP_spring in 1983, when the calculated external load peaked, at 1.1 mg/L (vs. 0.79 mg/L measured TP_spring; Fig. 5). Internal lake processes, such as sedimentation, partially reduced phosphorus from the euphotic zone between 1983 and 1991, but we cannot attribute the reduction to a particular mechanism for lack of data. Despite the external P load reduction from the mid-1980s, the measured TP_spring values still remained much higher than the predicted values in 1991. This lag, most often associated with internal P loading, is a well-known phenomenon (Sas 1989, Cooke et al. 2005) and in some cases, the decreases in P input may not be large enough to shift the lake out of the eutrophic state (Carpenter et al. 1999).

Figure 5 Comparison of measured TP concentrations during the spring circulation period of Tiefwarensee with the values predicted using the Vollenweider equation TP_spring = TP_in/1 + √τ_w, where TP_in is the flow-weighted TP influent concentration (mg/m³) and τ_w is the mean hydraulic residence time (years; Vollenweider 1976), and predicted by equation 4 (Brett and Benjamin 2008). Note the axis break between 1990 and 1998.

During the years of sediment treatment (2001–2005), TP_spring initially reached the value predicted by the Vollenweider model in 2003 and fell then slightly below the predicted values because of the complete suppression of P release and the continuing P precipitation in the hypolimnion. In the last years, TP_spring in Tiefwarensee has been reasonably well described by the Vollenweider model (Fig. 5), suggesting that a new equilibrium has been reached.

A recent analysis of models to predict lake P content based on P load, morphometric and hydraulic data (Brett and Benjamin 2008) found that the best prediction results (84% of the investigated lakes) are achieved using the empirical equation derived from the Vollenweider model:

in which TP_spring = TP concentration during spring circulation (mg/L); and TP_in = mean TP concentration in the inflow (mg/L);τ _w = hydraulic residence time (year).

In the years after the start of external load reduction (1991–2000), equation 4 has also underestimated the measured values, but fits them better than the Vollenweider equation (Fig. 5). The coefficient 1.12 in equation 4 characterizes the net loss due to sedimentation of P-containing particles in the lake (i.e., one or more first-order processes occurring within the lake); however, like the Vollenweider equation, it works only under steady-state conditions. In the treatment and post-treatment years from 2002 on, equation 4 overestimated the TP concentrations (Fig. 5), suggesting that the internal lake processes in Tiefwarensee were not predicted well by the model.

Contrary to Sas' (1989) expectations for deep lakes, we found negative P_ret values (i.e., a net P release from the sediment on an annual basis for the deep Tiefwarensee before the sediment treatment in 1999 and 2000; Table 3). The P_ret soared because of (1) P precipitation and sedimentation from the lake water and (2) a decrease in P release from the sediment. From 2003 to 2006, P_ret decreased further. The low P retention in 2006 results from the greater changes in P_lake after the completion of treatment in 2005. Since 2007, P_ret is stabilizing at close to 100 kg.

The sediments contain many times more P than the water body, and changes in the sediment P content may cause large-scale changes in the water. The P-binding forms in sediments characterize the potential for P release from the sediments. They may be divided into two fractions: permanently bound P and mobile P, which is dissolved during early sediment diagenesis and released from the sediments by diffusion (Hupfer and Lewandowski 2005). The permanently bound P pool, including P_res, HCl-TP and NaOH-SRP, generally remains relatively constant as sediment depth increases. The potentially mobile P pool is defined as temporarily bound P, including NH₄Cl-TP, BD-TP and NaOH-NRP; it decreases with sediment depth due to mineralization (Reitzel et al. 2005). The higher the potentially mobile P pool, the higher the P-release potential of the sediment. Of all analyzed sediment characteristics, significant changes due to the restoration measure were found only for the increase of TP and NaOH-SRP, and decrease of NH₄Cl-TP in Tiefwarensee. These changes indicate a decrease in the sediment's P-release potential or an increase in the sediment's P-binding capacity.

The sediment P-binding capacity was increased by adding Al as a P-binding partner. Relating the added amount of 84 tons Al to the 3 tons P removed from water during the treatment, the molar Al:P ratio is 32. In the upper 10 cm sediment surface we found the molar ratio Al:P = 20 ± 10 at P1 in 2006 and 2007, while Al:P = 1 ± 0,4 was determined in 2001 before the treatment. The measured Al:P ratio of 20 in the sediment cover is below the ratio 32 (added Al:P removed from water), which means that the added Al not only adsorbed P from the water but also from the sediment. Assuming the formation of AlPO₄, the Al:P ratio may achieve 1 (Cooke et al. 2005), indicating still more capacity to bind P to Al in the sediment of Tiefwarensee. Lewandowski et al. (2003) found a molar ratio for Al:P of 2.1 in the sediment of Lake Süsser See about 10 years after treatment with Al salts, and the P-adsorption capacity of the Al containing sediment layer was not exhausted in this lake.

The P retention depends on sediment parameters of P binding-partner in connection with redox and pH conditions, such as iron and sulfate (Caraco et al. 1989, Jensen et al. 1992, Roden and Edmonds 1997, Søndergaard et al. 2003), oxygen (Pettersson 1998, Moosmann et al. 2006), nitrate (Andersen 1982, Søndergaard et al. 2000), and the remineralization of organically bound phosphorus (James et al. 2002, Hupfer and Lewandowski 2005, Reitzel et al. 2007). Nevertheless, P retention is usually calculated as net estimate from input, output and changes in the P content of the lake water (see overview in Hupfer and Lewandowski 2008). The inclusion of P release or P-release potential of the sediment (calculated from the P-binding forms; Nürnberg 1988, Dillon and Evans 1993, Schauser et al. 2003) may help determine P retention during transition periods. The quantification of the mobile P pool allows us to estimate the lakes recovery time as well as the required dose of P-binding partners.

The one-box model introduced by Gächter and Imboden (1985) and modified by Schauser et al. (2003) was used to predict the further development of Tiefwarensee. This model uses the net P sedimentation coefficient σ = P_ret/P _lake and the morphometry-conditioned stratification factor β = P_out/P_lake* τ_w (β = 0.866, mean value 2003–2008 in Tiefwarensee) to calculate the TP content of a lake in the stationary state as

The coefficient σ, which depends on the P-binding capacity of the sediment, is expected to increase after restoration. The increased P-binding capacity was shown by decreased mobile P (NH₄Cl-TP) and increased Al:P ratio in the sediment. In Tiefwarensee, the prerestoration σ value of −0.03 (mean value for 1999 and 2000) increased to 2.21 in 2003 and 0.32 in 2008. The influence of σ is evident when comparing the predicted TP_lake values calculated from TP_in (Table 3) with and without treatment with the variation of σ (Fig. 6). We assumed TP_in to remain constant from 2008 on and σ to increase to 0.15 due to the external load reduction alone. Starting from 0.223 mg/L in 1998, the lake's TP concentration is predicted to decrease to 0.056 mg/L due to the external load reduction alone (σ = 0.15). The coefficient σ increased stronger in response to the chemical treatments. This increase is the key to achieving stabilization at a lower TP concentration. According to the predictions (equation 5), TP_lake will stabilize at 0.047 mg/L despite the chemical treatments if σ remains at 0.15 and will reach 0.030 mg/L only if σ increases to 0.33 and continues at that level (Fig. 6).

Figure 6 Concentrations of TP_lake measured from 1998 to 2008 and prediction of the development of TP_lake in Tiefwarensee according to the modified one-box model by Schauser et al. (2003) with and without chemical treatment assuming that P_in remains at the 2008 value (Table 2) and that σ is 0.15 or 0.33 from 2009 onward.

Conclusions

The rapid trophic shift of Tiefwarensee in response to restoration measures and the limnological data collected during the restoration period enabled us to explain the lake's response mainly based on the changes in the sediments. Sediment parameters are essential for determination of a lake's P-mass balance and for modeling. Lake P-retention analyses should include morphometry, inflow and outflow data as well as sediment data in consider both sedimentation and P-release processes.

The restoration of Tiefwarensee is an example that proves the need for internal measures in addition to external P load reduction to overcome eutrophication. The increased sediment P-binding capacity due to the hypolimnetic treatment was the key to reach and to stabilize a lower trophic state in Tiefwarensee.

Acknowledgments

We thank S. Appel, K. Jöhnk, R. Petzold, and K. Pohlmann for helpful discussions and U. Beyer, J. Dalchow, R. Degebrodt, C. Kasprzak, U. Mallok, and M. Sachtleben for technical assistance. G. Nürnberg and two anonymous reviewers are acknowledged for their helpful comments on a former version of the manuscript, and S. Wandrey is acknowledged for the linguistic improvements. The project received financial support from the Mecklenburg-Vorpommern State Department of Agriculture, Environment and Consumer Protection.

Notes

¹Added as NaAl(OH)₄ solution.

²Sediment surface below 10 m depth (0.61 km²).

¹P_ret = P_in – P_out− dP_lake.

²mean value 2006/2007.

¹Before treatment.

²After start of the treatment.