Water Quality Assessment of a Meromictic Lake Based on Physicochemical Parameters and Strontium Isotopes ( 87 Sr / 86 Sr) Analysis: A Case Study of Lubi ´nskie Lake (Western Poland)

: In 2017, hydrochemical surveys of meromictic Lubi´nskie Lake (W Poland) and its water inﬂows were carried out. The lake experienced complete mixing in 2008 due to a series of orkan winds, and since 2015, intensifying worsening of water quality in the lake has been observed. Our aim was to determine the degree of transformation of Lubi´nskie Lake based on water chemistry and to identify the source of pollution of the lake using strontium isotopes ( 87 Sr / 86 Sr) as a new chemical tracking tool. The physicochemical analysis conﬁrmed the meromictic character of the lake. The comparison with previous studies (2003 and 2008) showed signiﬁcant year-to-year di ﬀ erentiation, indicating intensifying eutrophication of the lake’s water, both in the epilimnion and the hypolimnion. Nine spring niches, directly supplying the lake, provide water with very high phosphorus and nitrogen concentrations (up to 10 kg of nitrogen and 0.9 kg of phosphorus daily). The strontium isotopes ( 87 Sr / 86 Sr) analysis indicated that the lake’s water was supplied mostly by the springs, and recharge from deep aquifers is of secondary importance. Moreover, strontium isotope data and the relationship between Sr and Cl content support the ﬁnding that the high load of nutrients is of anthropogenic origin and reaches the lake through springs.


Introduction
Climate changes triggered by human activity have caused profound alterations in lake ecosystems, including imbalances in the water budget, eutrophication, and increased sediment resuspension, which lead to the worsening of water quality and the loss of habitats [1][2][3]. These negative changes have affected both large water bodies, such as the Great Lakes or Lake Chad [4,5], and small lakes in remote areas [6,7]. Anthropogenically driven eutrophication is considered as one of the most critical threats for water quality in lakes, resulting in plankton blooms, decreased water clarity, oxygen deficiencies, and finally, loss of habitats [8]. It is mostly induced by an increased supply of nutrients coming from manure and mineral fertilizers applied on crop fields. Nutrients reach the lake through ground water weathering into water and soil and becomes available for further uptake by plants and animals [38]. Since the strontium isotope composition of surface water results from the mixing of strontium derived from various natural (bedrock, soil, precipitation, sea spray) and anthropogenic (industrial emissions, sewage, mineral fertilizers) sources with different isotopic ratios, the 87 Sr/ 86 Sr ratio might be applied to trace strontium sources [39][40][41]. Although strontium isotopes are commonly applied in studies of lakes, works investigating meromictic lakes are scarce [42].
In 2004, Pełechaty et al. [43] described a new meromictic lake in mid-Western Poland: Lubińskie Lake. According to Pełechata et al. [44], this lake experienced complete mixing in 2008 due to a series of storm winds. Since then, intensifying worsening of water quality (i.e., increase in nutrient concentration, phytoplankton blooms, decrease of water visibility) in the lake has been observed. Additionally, in 2015, increased fertilization of fields, mainly with liquid manure, occurred within the catchment area.
Our aim was to determine the degree of transformation of Lubińskie Lake based on water chemistry and to identify the source of pollution of the lake using strontium isotopes ( 87 Sr/ 86 Sr) as a new chemical tracking tool. We hypothesized that complete mixing of the water column after storm winds triggered the eutrophication process, which was then reinforced by the inflow of nutrients from the catchment.

Study Object
Lubińskie Lake is a small lake with the surface area of 22.7 ha located in western Poland (52 • 18 32" N, 14 • 54 25" E), in Lubuskie Lakeland (Figure 1). Its catchment area only covers 5.5 km 2 . It is one of 21 lakes located in the post-glacial tunnel valley running by Gronów, Rzepin, and Torzym towns. The valley spreads NW-SE but in the southern part, its direction changes to W-E, so the basin of the water body has an arched shape. In the southern part of the lake, a bay has formed, separated from the main pool with a 60 m wide isthmus [45]. The maximum depth of the lake is 21.6 m [46,47], whereas the mean depth is 8.8 m [47]. This results in steep slopes of the lake bottom. Detailed parameters and morphometric indices of the lake are presented in Table 1. The west and north-east slopes are particularly steep. The slopes of the bay, however, are gentle, and its depth does not exceed 5 m. The north and west shores of the lake are directly adjacent to the edge of the valley and have the form of steep scarps, over a dozen meters high. The south and north sides are milder.  Table 1. Parameters and morphometric indices of Lubińskie Lake (according to Jańczak [47]).

Parameters and Indices Unit Value
Surface area ha 22.7  Table 1. Parameters and morphometric indices of Lubińskie Lake (according to Jańczak [47]). The bottom of the lake is mostly made of fine-or medium-grained sands, and at some points also of gravels. In the bay and in the north part of the lake, the bottom is mostly covered with gyttjas [45]. The lake basin cuts through alternating deposits of fluvial sands and gravels [48] whose thickness fluctuates between several and over a dozen meters, which facilitate the infiltration of water from the catchment. In the lower part, the basin probably also cuts through the deposits of post-glacial clay (Figure 2), whose thickness in the open pit located approximately 1 km north of the lake is up to 35 m. Above and below the layer of clay there are approximately 10 m thick deposits of fine-and medium-grained sands, whose waters directly feed the lake in the form of groundwater upwellings [48]. As evidenced on an orthophotomap made in 2010 [49], woodlands and farmlands prevail in the catchment, covering the area of 2.7 and 2.6 km 2 , respectively ( Figure 3). The remaining area is covered by waters and the urban territory of Lubin town. The lake is surrounded by forests from all sides, but on the west and south side the forest buffer is very narrow, only between about a dozen and several dozens of meters from the lake shore. Woodlands predominantly occur in the eastern part of the catchment, while arable lands are located in the western part. As evidenced on an orthophotomap made in 2010 [49], woodlands and farmlands prevail in the catchment, covering the area of 2.7 and 2.6 km 2 , respectively ( Figure 3). The remaining area is covered by waters and the urban territory of Lubin town. The lake is surrounded by forests from all sides, but on the west and south side the forest buffer is very narrow, only between about a dozen and several dozens of meters from the lake shore. Woodlands predominantly occur in the eastern part of the catchment, while arable lands are located in the western part.

Materials and Methods
The lake was investigated in April and August 2017, both for comparison with the previous results in 2003 [43] and 2008 (unpublished data) and with regard to the potential meromictic nature of the lake. In both months, the electrolytic conductivity (EC) and pH were measured in the deepest part of the lake, using an Elmetron CX-701 meter (Elmetron Sp.j.). The Secchi disk was used for visibility. Oxygen concentration and temperature were measured in the vertical profile using an Elmetron CX-401 meter. In order to confirm the mixing type, additional measurements for oxygen and temperature were taken at the end of November 2017.
For further chemical analysis, the samples were collected in 1-l plastic bottles (one from the surface and one from the above-bottom layer-using an electric pump) and kept in a portable refrigerator. Then, the samples were kept in a refrigerator (at 4 °C) until the remaining chemical analyses were performed at the Department of Water and Soil Analysis, Faculty of Chemistry, Adam Mickiewicz University in Poznań.
Concentrations of total nitrogen (TN) were determined in the filtered samples by hightemperature combustion (HTC) using a Shimadzu TOC-L Total Organic Carbon analyzer (Shimadzu, Japan). For TN determination, the analyzer was equipped with a TNM-L TN unit (Shimadzu, Japan). The concentrations of NO2 − , NO3 − , and NH4 + ions were determined by the ion chromatography method with a conductivity detector (Shimadzu, Japan), using Dionex AS22 with AG22 columns. Total water hardness was determined by the versenate method. In order to determine Ca 2+ and Mg 2+ ion concentrations, a Shimadzu ion chromatograph with conductivity detection was applied, using Dionex IonPac SCS 1 Silica Cation Separator Column.

Materials and Methods
The lake was investigated in April and August 2017, both for comparison with the previous results in 2003 [43] and 2008 (unpublished data) and with regard to the potential meromictic nature of the lake. In both months, the electrolytic conductivity (EC) and pH were measured in the deepest part of the lake, using an Elmetron CX-701 meter (Elmetron Sp.j.). The Secchi disk was used for visibility. Oxygen concentration and temperature were measured in the vertical profile using an Elmetron CX-401 meter. In order to confirm the mixing type, additional measurements for oxygen and temperature were taken at the end of November 2017.
For further chemical analysis, the samples were collected in 1-l plastic bottles (one from the surface and one from the above-bottom layer-using an electric pump) and kept in a portable refrigerator. Then, the samples were kept in a refrigerator (at 4 • C) until the remaining chemical analyses were performed at the Department of Water and Soil Analysis, Faculty of Chemistry, Adam Mickiewicz University in Poznań.
Concentrations of total nitrogen (TN) were determined in the filtered samples by high-temperature combustion (HTC) using a Shimadzu TOC-L Total Organic Carbon analyzer (Shimadzu, Japan). For TN determination, the analyzer was equipped with a TNM-L TN unit (Shimadzu, Japan). The concentrations of NO 2 − , NO 3 − , and NH 4 + ions were determined by the ion chromatography method with a conductivity detector (Shimadzu, Japan), using Dionex AS22 with AG22 columns. Total water hardness was determined by the versenate method. In order to determine Ca 2+ and Mg 2+ ion concentrations, a Shimadzu ion chromatograph with conductivity detection was applied, using Dionex IonPac SCS 1 Silica Cation Separator Column. The inductively coupled plasma optical emission spectrometer (ICP-OES) ICPE-9820 (Shimadzu, Japan) was used for the qualitative and quantitative detection of total phosphorus (TP). Prior to the analysis, the Sigma-Aldrich (USA) containing 1000 mg L −1 P in 10% nitric acid (comprising hydrofluoric acid traces) was used for the calibration of ICP-OES. In order to preserve the standard/sample conditions, the matrix match method was used.
Soluble reactive phosphorus (SRP) in the water was determined by the molybdate method with ascorbic acid as a reducer. Prior to the analyses, the water samples were filtered through a rinsed 0.45 µm pore size filter (Macherey-Nagel, Germany).
Based on Secchi disc visibility and the concentration of TP in water, Carlson's trophic state index (TSI) was calculated [50]. The TN and TP concentrations were used for the N:P ratio.
The same methodology as above was applied in the case of the analysis of physicochemical parameters of water in springs feeding the lake, also carried out in April and August 2017. Methods described by Pełechaty et al. [43] were applied with regard to the compared years 2003 and 2008.
The flow and discharge of springs flowing out of the lake were measured three times: On 15 March, 15 April, and 15 August, using an electromagnetic flow meter Valeport model 801 with measurement accuracy of ± 0.5% of reading plus zero stability < 0.005 m s −1 .
Separate samples were taken for 87 Sr/ 86 Sr analysis of the dissolved strontium load. Additionally, rain water was collected at Słubice, about 30 km to the west, and groundwater was collected at Boczów (about 2.5 km to the north-east) in order to evaluate its contribution to the Sr budget of the lake. A total of 16 samples were taken. Grab samples were taken from the springs. Lake water was sampled with an electric pump from the deepest part of the lake and from the south-eastern bay ( Figure 1). In the former, it was gathered at points from the surface up to the depth of 20 m in 4 m intervals, whereas in the latter, samples were collected from the surface and the depth of 4 m. The rain was caught on the roof of Collegium Polonicum of Adam Mickiewicz University to a precleaned rain collector. The obtained sample encompass rainwater from several rain events during two weeks. The groundwater was collected from water intakes. All the samples were taken unfiltered to precleaned polyethylene bottles (250 mL) after flushing the bottles with the sampled water. The samples were filtered through Rotilabo cellulose syringe filters (0.45 µm), acidified with Merck sub-boiled ultrapure HNO 3 (1 mL of acid per 100 mL of water), and stored at 4 • C immediately after returning to the laboratory.
The analytical part of this work, including chemical separation of Sr and the measurements of Sr isotope ratios, were carried out in the Isotope Laboratory of the Adam Mickiewicz University in Poznań. About 15 mL of water was evaporated on a hot plate in PFA vials. Subsequently, a 340 µL portion of 2N HNO 3 was added to the sample and left overnight to equilibrate. The Sr was separated from matrix elements on PFA columns filled with Sr.Spec resin using the miniaturized chromatographic technique described by Pin et al. [51]. Some modifications in the column size and concentration of reagents were introduced by Dopieralska [52]. Strontium was loaded with a TaCl 5 activator on a single W filament and analysed in the dynamic collection mode on a Finnigan MAT 261 mass spectrometer. During the course of this study, the NBS 987 Sr standard yielded 87 Sr/ 86 Sr = 0.710210 ± 10 (2σ mean on thirteen analyses). The measured ratios were normalized to the nominal value of 0.710240 for the standard NBS 987. Total procedure blanks were below 83 pg.
Statistical analyses were performed using STATISTICA 12 (StatSoft, Kraków, Poland) software. The normality of distributions of the analyzed variables and the homoscedasticity of the samples were tested with the Shapiro-Wilk and Levene tests, respectively. The significance of differences was tested by ANOVA with the Kruskal-Wallis test. p < 0.05 was accepted as being statistically sound. The principal components analysis (PCA) was performed for the statistical analysis of the physicochemical parameters data set. Prior to this analysis, the data (except pH values) was subject to logarithmic transformation (log (1 + x)) to minimize the discrepancies between the empirical distribution of the variables and their theoretical normal variation. Out of the studied environmental variables, speciation forms of nutrients that were highly interrelated were reduced to TP and TN and included in the PCA. Additionally, to reduce the number of variables, a simple factor analysis was applied. Covariation was tested using non-parametric Spearman rank correlation. The significant levels were set to α = 0.05.

Lake Physio-Chemical Characteristics
The thermal and oxygen profiles ( Figure 4) performed during the spring overturn and summer stagnation and repeated during the autumnal overturn indicate the meromictic character of Lubińskie Lake. In all seasons, the monimolimnion zone began at 13 m, which was indicated both by a constant temperature (approx. 4.5 • C) and by anoxia.
Water 2019, 11, x FOR PEER REVIEW 9 of 22 distribution of the variables and their theoretical normal variation. Out of the studied environmental variables, speciation forms of nutrients that were highly interrelated were reduced to TP and TN and included in the PCA. Additionally, to reduce the number of variables, a simple factor analysis was applied. Covariation was tested using non-parametric Spearman rank correlation. The significant levels were set to α = 0.05.

Lake Physio-Chemical Characteristics
The thermal and oxygen profiles ( Figure 4) performed during the spring overturn and summer stagnation and repeated during the autumnal overturn indicate the meromictic character of Lubińskie Lake. In all seasons, the monimolimnion zone began at 13 m, which was indicated both by a constant temperature (approx. 4.5 °C) and by anoxia. The analyzed water properties were typical for eutrophy (TSI = 61), with high nutrients concentration (TN = 3.81 mg N L −1 and TP = 0.13 mg P L −1 , on average), high water mineralization (mean EC = 380 µS cm −1 ), and poor light conditions (mean SD visibility = 2 m; Table 2).  The analyzed water properties were typical for eutrophy (TSI = 61), with high nutrients concentration (TN = 3.81 mg N L −1 and TP = 0.13 mg P L −1 , on average), high water mineralization (mean EC = 380 µS cm −1 ), and poor light conditions (mean SD visibility = 2 m; Table 2).
Comparison with the previous studies showed significant (p < 0.05) year-to-year differentiation for three parameters only: Hardness, TN, and NO 3 − . For all of them, a significant increase in values was observed between 2003, 2008, and 2017. A similar pattern of increase was also observed for TP concentration and trophic indicators: TSI and N:P. The results indicate intensifying eutrophy, both in the epilimnion and hypolimnion waters. Interestingly, in the epilimnion, the N:P ratio increases with increasing TSI (phosphorus limitation), while in the hypolimnion, N:P decreases (nitrogen limitation). Although similar differentiation between the surface (epilimnion) and the above-bottom (hypolimnion) layer was observed in all the study years, the results from 2017 are the most divergent, indicating the highest mineralization and trophy.
The factor analysis showed that pH, EC, hardness, Ca 2+ , Mg 2+ , TN, and TP were the variables significantly differentiating the physicochemical composition of water within the studied periods. Based on these variables, the PCA analysis was performed ( Figure 5). The first two main components explained almost 80% of the observed variance. The PCA output proved pH, hardness, Mg 2+ , TN, and TP (correlated with the first axis, r > 0.80) as well as EC (correlated with the second axis, r > 0.80) to be primarily responsible for the observed variance. (hypolimnion) layer was observed in all the study years, the results from 2017 are the most divergent, indicating the highest mineralization and trophy.
The factor analysis showed that pH, EC, hardness, Ca 2+ , Mg 2+ , TN, and TP were the variables significantly differentiating the physicochemical composition of water within the studied periods. Based on these variables, the PCA analysis was performed ( Figure 5). The first two main components explained almost 80% of the observed variance. The PCA output proved pH, hardness, Mg 2+ , TN, and TP (correlated with the first axis, r > 0.80) as well as EC (correlated with the second axis, r > 0.80) to be primarily responsible for the observed variance.

Site-to-Site Differentiation of Water Chemistry
The water analyses showed that springs provide water with a significantly lower temperature (from about 10 °C in winter to 12 °C in summer) and much higher mineralization (about 50 µS cm −1 higher in spring and about 100 µS cm −1 in summer) than in the lake's water ( Figure 6). This was also reflected in the values of total hardness and concentrations of Ca 2+ and Mg 2+ . In the case of pH and oxygen, however, the lowest values were found in spring waters and the highest in the outflow waters.

Site-to-Site Differentiation of Water Chemistry
The water analyses showed that springs provide water with a significantly lower temperature (from about 10 • C in winter to 12 • C in summer) and much higher mineralization (about 50 µS cm −1 higher in spring and about 100 µS cm −1 in summer) than in the lake's water ( Figure 6). This was also reflected in the values of total hardness and concentrations of Ca 2+ and Mg 2+ . In the case of pH and oxygen, however, the lowest values were found in spring waters and the highest in the outflow waters.  The spring waters were characterized by much higher nutrient concentrations. This is particularly noticeable in the case of TP and SRP as well as TN and NO 3 -N concentrations ( Figure 6). It is worth noting that the waters of the bay directly fed by springs (site P1) had higher values than the waters in site P2. Even higher values were found in outflow waters (site O). Total nitrogen concentrations in spring were almost twice as high (up to 8.2 mg L −1 on average) as in lake waters. The majority of this load was NO 3 -N, the highest values of which were found in spring S1, both in the spring and the summer (24.13 and 22.86 mg L −1 , respectively). The exceptions were: NO 2 -N and N org. -N, which was found to be almost non-existent in spring waters, and lower concentrations of NH 4 -N.

Water Inflow to the Lake
Groundwater upwellings occur both in the littoral zone and in the pelagic zone (Figure 1). They are fed from the shallow aquifer at the depth of 2 to 5 m below ground level and from the aquifer occurring at the depth below 10 m and isolated from the surface with post-glacial clay deposits. The gravity springs feeding the lake are permanent. In the littoral zone of the lake, nine spring niches were recorded, from which waters are directly supplied to the lake. The flow discharge rate of the four measured springs was in the range of 1 to 8 dm 3 s −1 ( Table 3). The other flow discharge rates were estimated at below 0.5 dm 3 s −1 . The total flow discharge rate of all the springs is on average 20 dm 3 s −1 . In the southern part of the lake, two zones of sublacustrine springs were also recorded. They are approximately 10-20 m from the shoreline. However, it is difficult to measure their flow discharge. It is estimated to be about 50% of the inflow from overground springs, but this value may be underestimated. The discharge rate of the stream flowing out of the lake in the spring, directly after the disappearance of the ice cover was 40 dm 3 s −1 , and in the summer, 21 dm 3 s −1 . The difference between the amount of water feeding from the springs and the outflow from the lake indicates additional feeding with groundwater from aquifer outcrops on the bottom of the lake.

Strontium Isotope Composition in Water
The profiles of 87 Sr/ 86 Sr and Sr content for Lubińskie Lake are shown in Figure 7, and Sr isotope and elemental characteristics for components of the hydrologic system of the lake are given in Table 4. Since the bay yields a uniform Sr isotope composition and Sr concentration throughout the water column, it will not be discussed in detail. The lake is characterized by a narrow range of 87 Sr/ 86 Sr values, between 0.710741 and 0.710798. Thus, it is isotopically homogeneous throughout the water column (Figure 7a). Similarly, Sr content of lake water is highly uniform in the profile and changes from 0.124 to 0.148 mg L −1 (Figure 7b). It is worth mentioning that the highest concentration of dissolved Sr occurs in the lower part of the water column. As can be expected, the stream draining the lake has the 87 Sr/ 86 Sr ratio close to that of surface lake water, but is enriched in Sr (Figure 7). column (Figure 7a). Similarly, Sr content of lake water is highly uniform in the profile and changes from 0.124 to 0.148 mg L −1 (Figure 7b). It is worth mentioning that the highest concentration of dissolved Sr occurs in the lower part of the water column. As can be expected, the stream draining the lake has the 87 Sr/ 86 Sr ratio close to that of surface lake water, but is enriched in Sr (Figure 7).    The springs show a wide range of 87 Sr/ 86 Sr values from 0.710412 to 0.711851 and relatively diverse Sr dissolved load (0.127-0.175 mg L −1 ). The strontium isotope composition of the groundwater falls within the range for the springs ( 87 Sr/ 86 Sr values of 0.710693-0.710906), yet it has a clearly higher Sr content (0.241-0.266 mg L −1 ). Finally, the rain is the least radiogenic component (with the lowest 87 Sr/ 86 Sr ratio) and yields 87 Sr/ 86 Sr value of 0.708879.

Water Chemistry Analysis
The conducted research showed that Lubińskie Lake is currently in the state of higher trophy, which is indicated by the concentrations of nutrients and low Secchi disk visibility. The comparison with previous years of research shows clearly worsening of the water quality. In 2003, despite high nutrient concentrations, the lake maintained the clean water state (visibility over 4 m) and was classified as a charophyte lake [43,53].
The results of the physicochemical and algological tests showed that the meromictic character of the lake maintained since 2003 was disturbed between January and March 2008 [44]. At the end of February 2008, the waters in the lake mixed completely. At that time, a series of storm winds went across the western part of Poland ("Paula", "Zizi", and "Emma"). The complete restoration of the meromictic conditions lasted until June 2008 [54] (Pukacz et al. 2010). Although the complete mixing of lake waters lasted only a few months (which is confirmed by measurements carried out in the subsequent years), it probably contributed to the release of nutrients and other substances (e.g., methane and hydrogen sulfide) deposited for years in the monimolimnion [55]. This resulted in gradual worsening of the condition of the whole ecosystem. One of the visible effects was the worsening of light conditions and gradual disappearance of submerged vegetation (Pukacz 2008(Pukacz -2018. Although meromictic stratification was restored in the lake, further increase in nutrients concentration was observed.'The study showed that an important source of phosphorus and nitrogen (especially in the form NO 3 -N and NH 4 -N) supplied to the lake is the spring waters. Based on the average concentrations and water supply from the springs, it can be estimated that each day, approximately 10 kg of nitrogen and 0.9 kg of phosphorus can be introduced to the lake. However, these values may be underestimated by approximately 25%, because they do not take into account the load supplied with discharges for which the volume was estimated, and the load supplied through the springs in the bottom of the lake.
The increased supply of those elements must be associated with the change in the use of land in the catchment area of Lubińskie Lake. Since 2015, in the fields located in the SE part of the catchment, corn, and rapeseed have been cultivated, which are not only fertilized with mineral fertilizers but also, in the spring, with organic fertilizers (own observations). Taking into account the directions of inflow of groundwater and the very high concentrations of nutrients in spring waters, it is a very probable cause of the increased inflow of nutrients to the lake.
Interestingly, the values of the studied parameters in the SE bay, where most springs are located, were not significantly higher than in the central basin. This may result from the small surface area of the lake and from feeding it with inflows located in the slope of the basin. Feeding the lake through springs located in the bottom of the lake basin may have an impact, not only on the lower quality of the lake waters, but also on disturbed stratification and crenogenic meromixis [56].

Strontium Isotopes Perspective
Strontium isotope composition coupled with Sr content enables the identification of reservoirs controlling the 87 Sr/ 86 Sr ratio of water and mixing processes involved [27,36,57]. There is a strong correlation between 87 Sr/ 86 Sr and 1/Sr for the springs dataset, which points to mixing between the two Sr reservoirs (Figure 8). The first reservoir is characterized by a relatively radiogenic 87 Sr/ 86 Sr ratio and moderately low 1/Sr (high Sr concentration), whereas the second reservoir has a less radiogenic 87 Sr/ 86 Sr ratio and higher 1/Sr (lower Sr content). Since the springs emerge from shallow Quaternary aquifers, the first reservoir is represented by silicates, mostly glacial deposits. It controls the isotopic composition of the S1 spring. Since the investigated springs are charged by atmospheric waters [58], we identify the second reservoir with rainwater, which strongly influences the Sr isotope budget of the S2-S4 springs. As shown by Raiber et al. [59], rain can be an important source of Sr to shallow groundwater. Strontium isotope composition of rainwater at Słubice ( 87 Sr/ 86 Sr value of 0.708879) is less radiogenic than that of modern seawater ( 87 Sr/ 86 Sr about 0.70917; [60]) being the source of rainwater, likely due to the addition of atmospheric dust. Inland deviation of precipitation 87 Sr/ 86 Sr ratio is a well-documented phenomenon [59,61,62]. Typically, rainwater is not rich in Sr [59,61,63]. Zieliński et al. [25] provide Sr concentration of about 0.001 mg L −1 for rain and snow at Poznań, about 140 km east of Lubińskie Lake. However, the admixture of dust elevates the Sr content [59]. Therefore, Sr dissolved load of tens of µg L −1 for the rain at Słubice can reasonably be assumed. The role of rain as one of the major sources of Sr for the springs is supported by a clear relationship between Cl and Sr in the springs dataset (Figure 9), indicating their common origin. Chloride ions are a useful atmospheric input reference to the hydrosystem, because in the dissolved load of the hydrosystem they originate from seasalt and saltrock dissolution [64,65]. The role of rain as one of the major sources of Sr for the springs is supported by a clear relationship between Cl and Sr in the springs dataset (Figure 9), indicating their common origin. Chloride ions are a useful atmospheric input reference to the hydrosystem, because in the dissolved load of the hydrosystem they originate from seasalt and saltrock dissolution [64,65].  Figure 9. Relationship between Cl and Sr in samples collected from the springs and Lubińskie Lake. Chlorine content in the lake was measured only for surface and near bottom waters (see Table 4).
However, they can also result from various human activities (domestic sewage, fertilizers, and deicing salts) [66]. Since the Lubińskie Lake drainage basin is free of evaporates, they do not participate in the weathering reactions. Consequently, the general trend of relationships between Cl and Sr can be explained by the contribution of Cl from the atmospheric and anthropogenic sources. Since crop fields occur in hinterland of the spring area, the latter is likely represented by mineral fertilizers. This assumption is supported by the high load of nitrates (NO3) carried by the springs (Figure 6). Additionally, sewage leaking from surrounding villages may also provide some load of Cl.
Recently, Zieliński et al. [25] provided evidence for the common lowering of the present-day 87 Sr/ 86 Sr values of Noteć River water in western Poland due to the impact of fertilizers used in agriculture. The estimated scale of the lowering was not higher than 0.0006 to 0.0007 of the observed 87 Sr/ 86 Sr ratios in the downstream portion of the Noteć River. Given that the input of fertilizers-related Sr to the springs waters was recognized, the lowering of the lake water isotopic ratio should be assumed. Figure 8 shows that data points for Lubińskie Lake in general plot in the springs domain on the 87 Sr/ 86 Sr versus 1/Sr diagram; thus, crucial contribution of spring-derived Sr is identified. Surprisingly, spring S1, the second largest yet richest in Sr, has a negligible impact on the Sr isotopic composition of the lake. Instead, springs S2-S4 with lower flow rates and Sr contents seem to Figure 9. Relationship between Cl and Sr in samples collected from the springs and Lubińskie Lake. Chlorine content in the lake was measured only for surface and near bottom waters (see Table 4).
However, they can also result from various human activities (domestic sewage, fertilizers, and deicing salts) [66]. Since the Lubińskie Lake drainage basin is free of evaporates, they do not participate in the weathering reactions. Consequently, the general trend of relationships between Cl and Sr can be explained by the contribution of Cl from the atmospheric and anthropogenic sources. Since crop fields occur in hinterland of the spring area, the latter is likely represented by mineral fertilizers. This assumption is supported by the high load of nitrates (NO 3 ) carried by the springs (Figure 6). Additionally, sewage leaking from surrounding villages may also provide some load of Cl.
Recently, Zieliński et al. [25] provided evidence for the common lowering of the present-day 87 Sr/ 86 Sr values of Noteć River water in western Poland due to the impact of fertilizers used in agriculture. The estimated scale of the lowering was not higher than 0.0006 to 0.0007 of the observed 87 Sr/ 86 Sr ratios in the downstream portion of the Noteć River. Given that the input of fertilizers-related Sr to the springs waters was recognized, the lowering of the lake water isotopic ratio should be assumed. Figure 8 shows that data points for Lubińskie Lake in general plot in the springs domain on the 87 Sr/ 86 Sr versus 1/Sr diagram; thus, crucial contribution of spring-derived Sr is identified. Surprisingly, spring S1, the second largest yet richest in Sr, has a negligible impact on the Sr isotopic composition of the lake. Instead, springs S2-S4 with lower flow rates and Sr contents seem to contribute to it the most. This suggests additional input of shallow groundwater with 87 Sr/ 86 Sr ratios similar to these springs below the water table due to the lack of other springs along the lake's shore. Indeed, we found several seepages at the bay's bottom.
If the isotopic composition of Lubińskie Lake was solely related to the springs-derived Sr, data points for the lake water would lie on paths between the springs. Most of the data points, however, are moved aside the mixing trends, mainly towards lower Sr concentrations but without noticeable changes in 87 Sr/ 86 Sr, and evidence the contribution of less concentrated waters. Apart from rainwater supplied to Lubińskie Lake through the springs, as inferred above, it also reaches the lake directly. Therefore, it is likely responsible for the dilution of the lake's Sr dissolved load.
The input of dust-derived Sr can account for differences in Sr content between Lubińskie Lake and the stream originating from it. Graustein and Armstrong (1983) [39] show that coniferous foliage is more effective in dust trapping than deciduous foliage. As the stream flows in a coniferous forest surrounding the northern fringe of the lake, dust deposited on the foliage and then incorporated into the soil by throughfall seems to charge soil waters and hence the stream by additional Sr.
Although the lake basin cuts deeply into the aquifer (Figure 2), and the recharge of the lake by groundwater arguably occurs, its impact on Sr isotope composition of the lake water seems negligible. Nonetheless, this impression is only apparent, because 87 Sr/ 86 Sr ratios of the groundwater fall within the 87 Sr/ 86 Sr range for the springs, so the contribution of the former will be masked. The input of groundwater-derived Sr to Lubińskie Lake is suggested by the rise of Sr concentration several meters above the bottom, which is consistent with the lateral input of groundwater enriched in Sr and its downward descend. Due to the meromictic character of the lake, it is, however, limited to its deepest part.

Conclusions
The results of the study indicate that Lubińskie Lake has regained its meromictic character after complete mixing in 2008. That sudden event probably led to the release of nutrients from the monimolimnion layer, which resulted in an increase in trophy and deterioration of water quality in subsequent years. An additional, significant factor affecting the lake's water quality is the nutrient inflow from the catchment. Although the lake is insulated with a belt of trees, water polluted with fertilizers infiltrates into the first aquifer and enters the lake through springs. The difference between the lake supply from springs and the outflow from the lake indicates lake charging with groundwater from underwater outcrops of aquifers. That was also confirmed by strontium isotopes ( 87 Sr/ 86 Sr) analysis. It indicates that water is supplied to Lubińskie Lake mostly by the springs, and recharge from deep aquifers is of secondary importance. Moreover, strontium isotope data and the relationship between Sr and Cl content support finding that high load of nutrients is of anthropogenic origin and reach the lake through springs. We show that the 87 Sr/ 86 Sr composition of lake water supplemented by major ion chemistry is a valuable tool in the investigation of meromictic lakes.