Nine years of phosphorus management with lanthanum modified bentonite (Phoslock) in a eutrophic, shallow swimming lake in Germany

ABSTRACT

Epe TS, Finsterle K, Yasseri S. 2017. Nine years of phosphorus management with lanthanum modified bentonite (Phoslock) in a eutrophic, shallow swimming lake in Germany. Lake Reserve Manage. 33:119–129.

Eutrophication is threatening shallow lakes, their functioning and ecosystem services. In recent years, lanthanum modified bentonite (LMB) has been increasingly applied to eutrophic lakes to enforce recovery processes by increasing sediment phosphorus (P)-sorption capacity and reducing P in the water column. In this long-term study, we examined the water quality characteristics of a polymictic, eutrophic swimming lake in central Germany, Lake Bärensee, an artificial, excavated lake suffering from frequent cyanobacterial blooms caused by nutrient enrichment, mostly from swimmers, runoff, and P release from the sediments. Bärensee was first treated with LMB in 2007. This restoration method is based on the strong ionic bond formed between lanthanum (La) and phosphate (PO₄), which results in the formation of rhabdophane (LaPO₄ nH₂O) in the sediment. Smaller reapplications of LMB were conducted in 2010 and 2013 when P concentrations exceeded a defined level as a result of frequent nutrient inputs, primarily by swimmers. The intensity of mineralization processes decreased, preventing large algal and cyanobacterial blooms. Post-treatment chlorophyll a, total nitrogen, and ammonium concentrations indicate that the productivity of the lake has been limited by recurrent small treatments with LMB. Because of this P management, the lake could be used continuously for swimming and recreation.

KEYWORDS:

• Lake management
• lanthanum modified bentonite
• phosphorus inactivation
• swimming lake
• trophic control

Water ecosystems provide a range of different goods and services, collectively defined as ecosystem services, including provisioning, regulatory, cultural, and supportive functions (Vanneuville and Werner 2012, Hering et al. 2015). These functions are threatened by the process of eutrophication, which is considered the world's most widespread water quality problem (Lürling et al. 2016). The growth of excessive and sometimes toxic phytoplankton biomass in lakes negatively affects the provision of many ecosystem services (Jeppesen et al. 2009, Carvalho et al. 2013).

The deterioration of water quality as a result of eutrophication has been an important issue in the implementation of the European Water Framework Directive (WFD) 2000/60/EC (EU 2000, Reyjol et al. 2014). The WFD requires that waterbodies in the EU achieve a good ecological status within a given time scale and sets out clear deadlines for meeting environmental objectives (end of first, second, and final management cycle in 2015, 2021, 2027, respectively; Hering et al. 2010). Good ecological status and accordingly good ecological potential for heavily modified waterbodies (Borja and Elliott 2007) is defined as a status deviating only slightly from undisturbed reference conditions. Despite the increased efforts implemented during the past decade, eutrophication remains a central problem (EC 2015). The persisting high external loads and internal lake mechanisms prevent or delay recovery of many lakes in the EU (Søndergaard et al. 2007).

Shallow lakes are especially vulnerable to eutrophication because they are less responsive to reductions in external nutrient loads (Cooke et al. 2005). These lakes are prone to environmental changes and thus can easily switch from a clear water state with macrophyte dominance to a phytoplankton dominated turbid state (Scheffer and van Nes 2007). Because of increased benthic–pelagic coupling, the internal nutrient load, and in particular the internal P load, plays a decisive role in determining whether the lake is in a clear or turbid state (Scheffer et al. 1993, Beklioğlu et al. 2011).

Small shallow lakes in urban and suburban areas play an important role as recreational and cultural facilities and ensure significant economic benefits (Braat and de Groot 2012). Nevertheless, intensive recreational usage and pressures from densely populated catchments can severely impact these waterbodies and enhance the eutrophication process. For instance, each swimmer contributes up to 94 mg P per day to the lake (Dokulil 2014). The reduction of external nutrient loading is important but often not sufficient on its own to preserve ecosystem services and guarantee that a waterbody can be used for recreation (Cooke et al. 2005). Thus, additional internal measures such as biomanipulation (Søndergaard et al. 2007), physical manipulation (Gächter and Müller 2003), and sediment treatment (Welch and Cooke 2005, Yasseri and Epe 2015) are commonly employed to affect trophic cascades and to target the internal P load.

In-lake management actions that manipulate geochemical processes using materials to control eutrophication and obtain a desired chemical and/or ecological response have recently been described as geoengineering (Lürling et al. 2016). Most recently, the use of Phoslock, a lanthanum modified bentonite (LMB), has become a commonly employed geoengineering tool to manage eutrophic freshwater systems (Copetti et al. 2015). LMB binds soluble reactive phosphorus (SRP) in the water column and the sediment under a wide range of environmental conditions (Haghseresht et al. 2009, Dithmer et al. 2015, 2016b).

This study documents the post-application conditions of Bärensee and focuses on water quality parameters that mark the trophic status of the lake. We tested the hypothesis that the treatment changed both chlorophyll a (Chl-a) and the total phosphorus (TP) concentrations and consequently affected the nitrogen (N) balance in this intensely used recreational lake. Secondly, we assessed the targeted use of LMB as a management tool to control eutrophication and maintain ecosystem functions of shallow lakes suffering from exceeding anthropogenic pressure.

Study site

Bärensee (50°15′22.9″N, 8°95′39.2″E) is an artificial polymictic lake (main morphometric and physicochemical characteristics in Table 1) situated east of Frankfurt am Main, in Hessen, Germany. The lake is surrounded on 3 sides by Hessen's largest campsite (4000 permanent campers) and by a mixed forest on its eastern side. The lake has no inflow or outflow and a small fetch. Eutrophication issues date back to the 1990s, leading to hypertrophic state in 2004. During this period, cyanobacterial blooms caused regular swimming bans and degraded the recreational value of the area. The process of restoration commenced with different measures aimed at minimizing external inputs of nutrients. The sewer system of the campsite was upgraded, overhanging trees were cut back, and the condition of the onsite sanitary facilities was improved. Furthermore, cyprinids were no longer introduced in the lake, and the food web was manipulated by strengthening predatory fish.

Table 1. Main morphometric and physicochemical (before application of LMB) characteristics of Lake Baerensee.

Characteristics	Unit	Value
Watershed area	ha	10
Surface area	ha	6
Mean depth	m	2.63
Maximum depth	m	3.8
Volume	10³ m³	156
Trophic state	—	hypertrophic
Conductivity	µS/cm	260–370
pH	Std. units	8–9
Carbonate	mg/L	0.8–4
Bicarbonate	mg/L	123–155
TOC	mg/L	14
DOC	mg/L	12
TP	mg/L	0.08
TN	mg/L	2.1

In recent years, the lake has been intensively used during summer for swimming, fishing, and recreation. Between 2007 and 2015, an average of 20,000 swimmers have visited the lake each swimming season (May–Sep), not including the use of the lake by the 4000 permanent residents of the campsite. For our study, we assumed the daily average of swimming events during the 5 months of the swimming season to be 220.

Materials and methods

Sampling and analyses

Water samples were taken in strict compliance with the German guidelines (LAWA 2003) at the deepest point in the lake from 1 and 3 m depths using a peristaltic pump and hose. Sampling frequency was reduced for budgetary reasons; however, the lake was still sampled on at least 5 occasions in spring and 3 times in summer and in winter in all years (Table 2). Total nitrogen (TN) and total inorganic N (TIN) concentrations were analyzed according to the national industry standard (NIS) 13395:1996–12 (EN-ISO 1996) and by flow analysis and spectrometric detection. TP and SRP concentrations were determined according to NIS 6878:2004–09 (EN-ISO 2004); dissolved silica (SiO₂) by spectrometric detection according to NIS 38405–21:1990–10 (DIN 1990); DOC concentrations according to NIS 1484:1997–08 (EN 1997); Chl-a concentrations according to NIS 38412–16:1985–12 (DIN 1985), based on the ethanol extraction spectrometric method; and metal concentrations by optical emission spectrometry (ICP-OES) according to NIS 11885:2009 (EN-ISO 2009).

Table 2. Annual mean values of samples from 1 and 3 m depth. Variables in italics were considered for principal component analysis (PCA). Number of samples (n), annual mean values (mean), minimum (min), and maximum (max) values are shown. The n of total inorganic nitrogen (TIN), ammonium nitrogen (NH₄-N), dissolved organic carbon (DOC), total nitrogen (TN), soluble reactive phosphorus (SRP), calcium (Ca), and dissolved silica (SiO₂) is identical to the n shown for total phosphorus (TP). Values were set as limit of detection when they fell below. Capital letters (A, B, C, and D) indicate time periods used in tests for significant differences and classification in PCA. LMB reapplications are represented with a slash and were conducted in May 2010 and in March 2013.

		2007	2008	2009	2010	2011	2012	2013	2014	2015
		A	B	B	B/C	C	C	C/D	D	D
	n	9	7	4	6	6	5	5	5	5
Chl-a (µg/L)	mean	35.9	19.6	19.2	9.3	15	14.3	8.6	6.8	14
	min	6.2	10.1	9.8	1	1	1.8	1	1.2	8.2
	max	76.3	27.3	35	19.4	31.1	30.2	15.4	11.3	25.8
	n	9	7	5	6	6	5	5	5	5
TP (mg/L)	mean	0.061	0.033	0.042	0.028	0.034	0.037	0.033	0.036	0.052
	min	0.024	0.025	0.028	0.023	0.026	0.029	0.029	0.031	0.029
	max	0.089	0.065	0.057	0.039	0.043	0.047	0.036	0.049	0.078
TIN (mg/L)	mean	0.65	0.42	0.24	0.13	0.08	0.10	0.15	0.18	0.17
	min	0.2	0.14	0.04	0.04	0.04	0.05	0.04	0.1	0.06
	max	1.53	0.79	0.45	0.35	0.15	0.13	0.27	0.3	0.26
NH4-N (mg/L)	mean	0.55	0.32	0.14	0.1	0.05	0.06	0.11	0.07	0.08
	min	0.11	0.14	0.04	0.04	0.04	0.04	0.04	0.04	0.04
	max	1.4	0.62	0.27	0.35	0.09	0.10	0.25	0.19	0.15
DOC (mg/L)	mean	10.7	8.74	8.52	8.25	8.40	7.69	7.71	9.22	8.9
	min	9.1	8.15	7.7	7.6	7.25	7.2	7	8.25	8.25
	max	12	9.45	10.5	9.2	9.5	8.35	8.45	9.95	10.4
TN (mg/L)	mean	1.81	1.22	1.06	1.03	0.76	0.89	0.89	0.87	0.94
	min	1.25	0.94	0.75	0.4	0.59	0.85	0.55	0.55	0.6
	max	2.85	1.5	1.5	1.95	1	0.95	1.25	1.05	1.45
SRP (mg/L)	mean	0.013	0.005	0.005	0.005	0.005	0.005	0.005	0.005	0.016
	min	0.005	0.005	0.005	0.005	0.005	0.005	0.005	0.005	0.005
	max	0.037	0.005	0.005	0.005	0.005	0.005	0.005	0.005	0.022
Ca (mg/L)	mean	43.5	56.1	53.6	54.8	55.6	51.7	51.5	53.2	54.4
	min	37.3	52.5	49.0	51.5	50.5	48.5	49.0	47.5	48.5
	max	50.5	58.5	58.0	56.5	62.0	55.0	55.0	59.0	60.5
SiO2 (mg/L)	mean	8.4	3.0	3.9	2.8	5.0	2.5	4.0	4.1	6.0
	min	3.1	0.4	1.7	0.4	0.4	0.9	1.4	0.4	0.4
	max	12.0	5.4	6.0	8.3	11.0	5.5	6.3	7.3	9.9

Sediment samples were collected in 2007 using an Ekman-Birge grab sampler (Hydro-Bios; Kiel). Samples of the upper 10 cm of sediment were extracted from the grab sample using a modified 50 mL syringe. Sediment samples were cooled at 4 C and transported to the laboratory where they were prepared for further chemical analysis. Subsamples were used to determine the dry weight according to NIS 11456/EN 14346 (ISO 2007). Elements were analyzed using an ICP-OES (ICAP 6000, Thermo Scientific) according to NIS 11885:2009 (EN-ISO 2009). Sequential phosphorus (P) extraction of the sediment was conducted according to Hupfer et al. (1995) and Psenner et al. (1988) to determine the release-sensitive sediment P pool. This mobile P pool (P_mob) consists of the sum of labile, immediately available P-, reductant soluble organic P, reductant soluble Fe/Mn-bound P, and the organic-bound P fraction (Søndergaard et al. 2003). The method is described in its entirety in Meis et al. (2012).

Statistical analyses

Statistical analyses were conducted with OriginPro 9.1 (Origin Lab Corporation, Northampton, MA). Data were checked for normality and heteroscedasticity. Concentrations of Chl-a, TP, and TIN were distributed normally (Kolmogorov–Smirnov test, P > 0.05) but violated the criteria of homoscedasticity (Levene test, P < 0.001). Therefore, a non-parametric Kruskal–Wallis ANOVA (KWA) was used to test for significant variation in concentrations between previously defined time periods. Time periods were created as follows. Samples in period A (n = 9) were taken in 2007, the year of the initial LMB application; B samples (n = 14) cover the period from 2008 to the first reapplication in June 2010; and C samples (n = 16) cover the period between the first and the second reapplication of LMB in March 2013. Samples in period D (n = 14) were taken after the second reapplication of LMB until the end of 2015. When significant variation was detected, the non-parametric Mann–Whitney U (MWU) test was used to identify the time periods between which the significant differences in concentrations occurred.

The strongest trends were plotted and the information condensed by conducting a principal component analysis (PCA) using correlation. Pearson correlation matrix revealed strong intercorrelation between TN and ammonium (NH₄-N; correlation coefficient 0.78; P < 0.01), so only NH₄-N was included in the analysis. The variables TP, Chl-a, NH₄-N, DOC, calcium (Ca), and SiO₂ were 1n transformed and automatically standardized to zero mean and unit variance (Quinn and Keough 2002). Principal components (PCs) with eigenvalues ≥1 (Kaiser–Guttmann criterion) were considered for interpretation (McCune et al. 2002). Pearson's correlations between PC scores and variables were performed to classify variables with either the first component (PC 1) or second component (PC 2).

Results

Treatment conditions

Our pre-treatment analysis revealed a TP content of 700 mg/kg dry weight (DW) in the upper 10 cm of the sediments. Sequential P extraction showed that 140 mg/kg DW, or 20%, of the sediment P was potentially mobile (P_mob). In June 2007 an initial total dose of 11,500 kg of LMB was applied to the lake to minimize TP concentrations in the waterbody and to bind the majority of P_mob in the sediment. This initial dose equates to an LMB load of 1900 kg/ha. In comparison to lakes listed by Spears et al. (2016), this initial dose is considered low.

Because external inputs of nutrients were expected to be high due to intensive lake use by swimmers, small adjusted maintenance applications were required. This strategy was developed in consultation with stakeholders and was aimed at avoiding swimming bans and maintaining the improved trophic state of the lake. Two smaller reapplications were therefore undertaken in May 2010 (3000 kg LMB) and March 2013 (3000 kg LMB). During the whole period of this P management, the average costs of the treatment were €5000/yr (US$5481/yr; exchange rate 20 Oct 2016). Considering only the campsite and the open-air bath, an annual revenue of >€300,000 (US$328,887; exchange rate 20 Oct 2016) was generated.

Responses in TP, Chl-a, and TIN concentrations following the LMB applications

KWA indicated significant variation (P < 0.01) in TP concentrations (H₃ = 13.36), TIN concentrations (H₃ = 26.2), and Chl-a concentrations (H₃ = 17.33) between the different time periods. For TP, post hoc tests revealed 2 homogenous groups: (1) sample period A, the year of the initial application in 2007, and (2) sample periods B, C, and D (Fig. 1a). Mean TP concentrations decreased from 0.061 mg/L in period A to 0.036 mg/L in period B. By reapplications of LMB, mean TP concentrations were controlled in period C (0.032 mg/L) and in period D (0.041 mg/L).

Figure 1. Range of (a) total phosphorus concentrations, (b) chlorophyll a concentrations, and (c) total inorganic nitrogen concentrations among temporal subgroups (A–D). Samples in period A (n = 9) were taken in 2007, the year of the initial LMB application. B (n = 14) comprises samples from 2008 to the first reapplication in June 2010. C (n = 16) covers the period between the first and the second reapplication of LMB in March 2013. Samples in period D (n = 14) were taken after the second reapplication of LMB up to end of 2015. Lines within the boxes mark the median. Lines below and above the median indicate the 25^th and 75^th percentiles, respectively; whiskers below and above the boxes indicate the 10^th and the 90^th percentiles, respectively; and dots mark minimum and maximum concentrations of the range. Groups that do not share the same small letter (a, b, c; above the boxes) are significantly different (Mann–Whitney U-test; P < 0.05).

Post hoc tests on Chl-a revealed a similar pattern of 2 homogenous groups: (1) sample period A and the period until the first reapplication B, and (2) sample periods C and D (Fig. 1b). Although not significant, mean Chl-a concentrations decreased from 35.9 µg/L in period A to 19.2 µg/L in period B. A significant decrease followed from B to C (11.7 µg/L) and period D (10.1 µg/L). Local federal authorities investigated the lake regularly and did not observe massive blooms of cyanobacteria. Consequently, swimming bans have not occurred since 2007.

Regarding TIN concentrations, post hoc tests revealed 3 homogenous groups: (1) period A, (2) period B, and (3) period C and D. Mean TIN concentrations in Bärensee significantly decreased from 0.65 mg/L in period A to 0.31 mg/L in period B and 0.11 mg/L in period C, respectively. After the second reapplication (period D = 0.16 mg/L) TIN concentrations ranged within the same level as period C (Fig. 1c). The mean annual values indicated decreases in PO₄-P, TN, and SiO₂ and a slight increase in Ca; however, these changes were not significant (Table 2).

Assessing temporal development in main lake water characteristics

PCA was conducted to capture the variation in ln transformed data covering the whole sample period. PC 1 and PC 2 explained a large proportion of variance and were considered for interpretation and visualization in the distance biplot (Fig. 2). Results indicated a positive correlation in TP, NH₄-N, Chl-a, DOC, and SiO₂ concentrations with PC 1. Ca concentrations were negatively correlated to PC 1, which represents the general changes in biomass and mineralization. Chl-a and NH₄-N concentrations were positively correlated, whereas TP concentrations were negatively correlated to PC 2 (Table 3), which represents seasonal variation in the dataset (e.g., lower TP concentrations in winter). The biplot also indicated correlation between the variables. TP was correlated with DOC and SiO₂, and NH₄-N was correlated with Chl-a. Temporal development was coded by classification of each single sampling with its time period.

Figure 2. Results of principal component (PC) analysis for selected lake water characteristics depicted as a biplot along first (PC 1) and second (PC 2) principal component. Samples are classified with time periods (A–D); hatched areas cluster samples from period A and B visually. NH₄-N = ammonium nitrogen concentration; Chl-a = chlorophyll a concentration; DOC = dissolved organic carbon concentration; SiO₂ = dissolved silica concentration; TP = total phosphorus concentration; Ca = calcium concentration; EV = eigenvalue; CV = cumulative variance explained. Variables were ln transformed before analysis.

Table 3. Classification of ln transformed variables with the axes by Pearson's correlation coefficients of principal components (PC) scores and variable scores. Asterisks denote significant correlations (α < 0.01).

	Chl-a	TP	NH4-N	DOC	Ca	SiO2
PC 1	0.60^*	0.76^*	0.55^*	0.80^*	−0.81^*	0.75^*
PC 2	0.53^*	−0.41^*	0.67^*	−0.07	0.32	−0.13

Discussion

Early responses of lake water quality following the initial LMB application

LMB applications are undertaken to bind SRP in the water column and to cap bioavailable forms of P in the sediments. In summer, lake water SRP concentrations are lower than in winter months in most lakes. In summer, sediments release P, which is often picked up by phytoplankton (Spears et al. 2007, Nürnberg 2009). For administrative reasons, the initial application of LMB at Bärensee was undertaken in summer 2007 rather than during the winter or late autumn, as was originally recommended. Consequently, a high proportion of P was incorporated in the biomass as TP at the time of application. Nevertheless, the initial dose eliminated the remaining SRP concentrations to the level of detection. This reduction was expected and was also observed by Márquez-Pacheco et al. (2013), Bishop et al. (2014), Spears et al. (2016), and Yamada-Ferraz et al. (2015). The mean TP concentration decreased during the first year (Fig. 3) and was reduced by almost 50% in the second year. This reduction in TP occurred progressively as the biomass died and released P as SRP. The proportion of SRP not immediately reused by algae in the waterbody was bound by LMB at the sediment surface. Similarly, Nürnberg and LaZerte (2016) observed a delay in TP reduction following LMB treatments during periods when most P has been taken up into the biomass.

Figure 3. Development of mean annual chlorophyll a (Chl-a) concentrations and mean annual total P concentrations depicted in a scatter plot. Pre (n = 3) is only considered as a separated group in this plot and relates to available data from 2007 before application. Trophic states were set with reference to Vollenweider and Kerekes (1982).

Shallow^{Table 3} eutrophic lakes are largely influenced by sediment–water interactions exhibiting positive P retention during winter, negative P retention during part of the summer, and a trend from P limitation in spring to N or light limitation later in the year (Søndergaard et al. 2003, Kolzau et al. 2014). Because the availability of SRP as a nutrient source for phytoplankton was restricted after the initial application of LMB, an influence on Chl-a concentrations was expected. The decrease of the phytoplankton biomass following LMB treatment can commence at different intervals. Compared to the study of Nürnberg and LaZerte (2016), who described shallow hypertrophic Swan Lake in Canada, the onset of the treatment success at Bärensee started earlier. Gunn et al. (2014) studied the effects of an LMB application to shallow Loch Flemington in Scotland in 2009 and found similar reductions of TP (from 0.06 to 0.031 mg/L) and Chl-a concentrations (from 51 to 25 µg/L) in the first year after application. By contrast, however, an immediate reduction of Chl-a and TP concentrations was observed by Lürling and van Oosterhout (2013) at Lake Rauwbraken, Netherlands, where a combination of flocculent polyaluminium chloride (PAC) and LMB was used to remove a bloom of cyanobacteria (Aphanizomenon flos-aquae).^{Figure 3}

Decreases in TN are associated with decreases in the phytoplankton biomass and can be considered a positive effect of lake restoration activities (Søndergaard et al. 2007, Novak and Chambers 2014). In our dataset, Chl-a and TN concentrations were positively correlated (Pearson's correlation coefficient 0.55; P < 0.01), and mean TN concentrations decreased during the first post-treatment years. Among the TIN components, which are direct food sources for phytoplankton, NH₄-N significantly decreased with time. High NH₄-N concentrations often indicate nutrient enrichment and increased internal loading (Nürnberg 2007, Nikolai and Dzialowski 2014, Nürnberg and LaZerte 2016). Our results indicated that both these factors were reduced as a result of the initial treatment at Bärensee.

The effects of LMB applications on DOC concentrations have recently been discussed by Spears et al. (2016). DOC also slightly decreased in our dataset in the post-application period (Fig. 2) and corresponds to findings of Nürnberg (1996) and Nürnberg and Shaw (1998) because DOC is correlated to TP concentrations. Laboratory experiments of Lürling et al. (2014) indicated that LMB may interact with components of DOC, and the interactions between SRP, lanthanum (La), and DOC may influence post-application responses of lakes. Dithmer et al. (2016a) detected a delay in P binding by LMB in low alkalinity water with high DOC concentrations in their incubation trials. They described that P uptake continued during the whole incubation period, almost overcoming the negative effect of high DOC. The water of Bärensee can be considered to be strongly buffered with a high alkalinity, high pH, and low concentrations of humic acids. Regular algal blooms, which release algal exudates as well as decompose organic matter in the sediments, also contribute to elevated DOC concentrations (Bertilsson and Jones 2003, Wyatt et al. 2014). We observed no immediate interference of DOC or delay in binding SRP caused by DOC in post-application periods, indicating that the autochthonous-produced dissolved organic matter had no or limited La chelating properties. We consider the decrease of DOC after the first application to be an indirect consequence of the reduced productivity, analogous to the indirect decrease of Chl-a and TIN concentrations.

The slight reduction of SiO₂, the dissolved nutrient for diatoms, may be another indirect effect of the P inactivation. In many lakes, diatoms bloom in spring and consequently lower the SiO₂ concentrations in the waterbody. The proportion of diatoms decreases in summer, and their frustules sink to the sediment surface where SiO₂ can be released in the process of internal loading (Wetzel 2001, Douglas et al. 2016). The correlation between SiO₂ and TP and its reduction indicates that (1) diatoms may be more competitive in P-reduced situations, and (2) mineralization processes are attenuated by LMB.

The main driver of these indirect responses was the reduction of TP as a result of the first LMB treatment. The reduction of available P in the water column and sediments resulted in reduced phytoplankton production and diminished mineralization and decomposition. This study confirms the importance of the sediment P-pool in shallow lakes and its influence on nutrient development (Cooke et al. 2005, Søndergaard et al. 2007, Meis et al. 2013).

Effects of phosphorus management with smaller reapplications

The strategy of undertaking maintenance reapplications was formulated at the outset of the Bärensee restoration program because nutrient inputs from swimmers were expected to be a major nutrient source. Values quantifying nutrient inputs by swimmers are rare. We based our estimate on values given by Dokulil (2014). Compared with these studies and the data provided by local authorities, our estimate is probably an underestimate of the number of swimmers using the Bärensee and the nutrient inputs from this source. Nevertheless, an additional input of 3.15 kg TP per year (which equates to the addition of 0.02 mg/L TP per year) by swimmers will clearly have an effect on the trophic state of this small waterbody. Time period C (May 2010–Feb 2013) and time period D (Mar 2013–Dec 2015) comprise 6 swimming seasons, which would equate to a total nutrient input by swimmers of 18.9 kg TP. Furthermore, swimmers stir up sediments, an action that will have a greater impact on shallow lakes than on deeper waterbodies.

As a result, we focused on TP concentrations in post-application periods and set an annual mean of 0.035 mg/L TP as the unofficial target. This value marks the transition between a mesotrophic and eutrophic state according to limits set by the Organisation for Economic Co-operation and Development (OECD; Vollenweider and Kerekes 1982). Although concentrations of Chl-a, TN, and TIN were either still decreasing or remained at low levels at the beginning of period C, an application of 3000 kg LMB was undertaken as a precautionary measure to interrupt a trend toward increased TP levels evident in 2009. Regional mean winter temperatures in 2008/2009 and mean spring and autumn temperatures in 2009 were significantly higher than the moving 30-year average (DWD 2016). These higher temperatures may have led to enhanced mineralization and probably attracted more swimmers.

During period C, TP concentrations remained near the target level, and elevated TP concentrations in autumn 2012 led to the second reapplication of 3000 kg LMB at the beginning of swimming season in 2013. The positive effect of this application was evident in the following year, although the extraordinarily warm summers in 2014 and 2015 (DWD 2016) may have contributed to the increased TP concentrations we measured. Chl-a concentrations seem to have a delayed response to higher TP concentrations (Schindler et al. 2008, Novak and Chambers 2014), however, indicating that TP was not mainly incorporated in algae, and higher concentrations may be a result of suspended matter, excretion or other organic compounds in the waterbody (Reitzel et al. 2012). Additional parameters were recently added to the regular monitoring to determine the reasons for these developments. In addition, a further reapplication of 4000 kg LMB was undertaken in February 2016, which eliminated SRP and reduced TP concentrations to 0.031 mg/L by April 2016 (Institut Dr. Nowak, May 2016, unpubl. data).

Implications for lake managers

The main aim of the application of LMB to Bärensee was to address the unrestricted use of the lake as a swimming lake. This meant that phytoplankton growth should be restricted and the dominance of cyanobacteria avoided (EU 2006); both objectives have been achieved since the beginning of the treatment. Nevertheless, fishing is still permitted on the lake and, although the feeding of fish is prohibited, the use of bait lures, a common method to attract cyprinids, may still be occurring. The cyprinids, and especially the common carp (Cyprinus carpio), have a negative impact on the P-budget by increasing the sediment mixing depth (Huser et al. 2016). Multiple pressures are still exerted on lake water quality, but the external load of Bärensee has been reduced as much as is practical, and there is no intention to restrict numbers of swimmers per day or season.

Phosphorus management at Bärensee can be linked to the conceptual model for the use of P-capping agents in lake remediation projects by Meis et al. (2013). This model hypothesizes that calculating the exact effective dose due to inherent complexities of P-cycling mechanisms is impossible, and proposes the application of multiple smaller doses to reach defined water quality targets and to improve cost effectiveness. The work in Bärensee indicates that reducing the trophic level with an initial dose of LMB and the subsequent target-oriented control of TP by reapplications is a feasible way to ensure long-term water quality in an intensively used swimming lake.

Our findings demonstrate that the reduction of P alone can efficiently reduce effects of eutrophication and are in line with the conclusions stipulated by Schindler et al. (2008) and Schindler (2012). These authors evinced that a reduction of P sources controls algal blooms and other eutrophication effects, independent of other nutrient loads and recovery of culturally eutrophied lakes, is often delayed due to phosphorus-saturated sediments. Therefore, targeting the internal P-load as described in our study is a valuable part of maintaining lake function. A simple and relatively inexpensive monitoring program has been used as a basis to implement fast and responsible measures aimed at maintaining trophic levels within a defined target range. Setting the targets is critical and should be done in close consultation with local authorities and the stakeholders to ensure a pragmatic and realistic communication of the remediation process.

Conclusions

•	The initial application of LMB eliminated SRP concentrations immediately and reduced TP concentrations by 50% in the first post-treatment year, although the measure was undertaken in summer. Chl-a concentrations dropped in the first year after application.
•	Reductions of TIN concentrations indicate reduced internal loading and diminished mineralization processes as an indirect response to the reduced biomass.
•	Constant nutrient inputs through intensive use required the development of a P management strategy from the outset of the restoration program. Smaller maintenance applications were undertaken twice to bind P when mean annual TP concentrations exceeded a defined target of 0.035 mg/L.
•	Chl-a concentrations dropped in the reapplication period, and TIN concentrations remained at reduced levels. Swimming bans and massive cyanobacterial blooms have been avoided since the restoration measures commenced.
•	The reduction of trophic level with an initial dose of LMB and control of new inputs of P by smaller reapplications is a feasible way to ensure long-term water quality in an intensively used swimming lake. Monitoring programs and close consultation with the stakeholders can be used as the basis to implement fast and responsible measures to maintain trophic levels within a defined target range.