Precipitation of calcium carbonate in a shallow polymictic coastal lake: assessing the role of primary production, organic matter degradation and sediment mixing

Eleven sampling sites from different parts of the lake (Fig. 1) were selected for the lake water analyses. This number of sites was necessary because single points could be unrepresentative for such a complex lake. Cieśliński (2013) showed that coastal lakes along the Polish coast are characterized by appreciable short-term hydrochemical variability due to marine inflows. It was believed that sampling of the lake water in different sections of the lake would allow to minimize the effect of these short-term and local fluctuations.

The water was sampled on 13 December 2007, 10 April 2008, 10 July 2008, and 17 September 2008. The samples for C isotope composition in dissolved inorganic carbon (DIC) (δ¹³C_DIC) were poisoned with HgC_l2.

The pH and temperature were determined in surface (5-10 cm below the surface), bottom (5-10 cm above the bottom), and pore waters (from the 5 cm surface layer of bottom deposits) with a WTW Multi 350i sensor. The measurements in the surface and bottom waters were made in situ, while the pore waters were analyzed immediately after the retrieval.

Pore waters, together with surface sediments, were collected using a gravity corer (Tylmann 2007) and extracted by centrifugation (3000 rpm) in the laboratory within 1-2 days after collection.

The bicarbonates (HCO₃^-) were titrated with 0.1 M HCl using methyl orange as an indicator.

Concentrations of Ca²⁺ were determined with the method of EDTA titration with regard to murexide in pH = 10. The accuracy of the analyses was ±10%.

δ¹³C_DIC in lake water was analyzed in 2 ml aliquots. CO₂ was released from the samples by adding H₃PO₄. The isotopic measurements were carried out using a Delta ir-MS attached to the MultiFlow system (Micromass Ltd., UK). Results were reported relative to the PDB reference material, using external standards.

The sediment samples for the stable carbon isotope composition of CaCO₃ were collected from the 5 cm surface layer of sediments obtained with a gravity corer in July 2008 (Tylmann 2007).

The sediment samples were lyophilized and homogenized using an agate mill. The carbonate content was determined by thermal combustion (Heiri et al. 2001) and recalculated to CaCO₃, using the formula CaCO₃[%] = 2.27 LOI₉₂₅, where LOI₉₂₅ represents the content of CO₂ released during decomposition of calcite at 925°C (Heiri et al. 2001).

The mineralogical form of carbonate in seven samples was determined in powdered untreated sediment by X-ray diffraction using a VRD6 diffractometer. The diffractograms were recorded by the reflection method using CuK_α radiation.

For the purpose of stable C isotope measurements in sedimentary calcite, the organic matter was removed via oxygen plasma ashing (Lebau et al. 2014) using an Emitech K1050X Plasma Asher. OM-free carbonate powders were then reacted with 100% phosphoric acid at 70°C using Gasbench II connected to a ThermoFinnigan Five Plus mass spectrometer. All values are reported in per mil relative to V-PDB by assigning δ¹³C values +1.95‰ to international standard NBS19 and -46.6‰ to international standard LSVEC, respectively. Reproducibility and accuracy were monitored by replicate analysis of laboratory standards calibrated to NBS19 and LSVEC and was ±0.04-0.07‰.

Carbonate alkalinity (CA) was calculated as follows:

where: (HCO₃^-) and (CO₃^2-) denote activities of HCO₃^- and CO₃^2-, respectively.

The saturation index for calcite (SI_calc) was calculated as:

where:

(Ca²⁺) – activity of Ca²⁺ in lake waters

K_c – equilibrium constant for the reaction of calcite precipitation.

Partial pressure of CO₂ (pCO₂) was computed according to (Kelts & Hsu 1978)

where

(H₂CO₃) – activity of carbonic acid in lake waters

K_CO2 – Henry’s law constant for CO₂

Activities of H₂CO₃ and CO₃^2- were calculated on the basis of temperature-calibrated pH and the concentration of HCO₃^-, using equations adopted from (Kelts & Hsu 1978). Temperature-dependent equilibrium constants in the carbonate system K₁:

K₂:

and K_c (calcite) were calculated according to (Kelts & Hsu 1978).

Ionic strength (I), used for ionic activity calculations, was obtained from the concentrations of Na⁺, K⁺, Mg²⁺, Ca²⁺, HCO₃^-, Cl^-, SO₄^2- and NO₃^- published elsewhere (Woszczyk et al. 2011a).

K_CO2 was calculated using data provided by Eby (2004).

SI_calc in Lake Sarbsko waters varied greatly in time and space (Fig. 2). Surface and bottom waters were relatively similar in terms of SI_calc and showed increasing saturation from December 2007 to July 2008 and a decrease between July 2008 and September 2008. In December 2007, the SI_calc values in the lake water column were mostly negative. In April 2008 and July 2008, the SI_calc increased to 0.3-0.8, respectively. In September 2008, it dropped again and at a few sites even negative values were obtained.

Figure 2

With respect to the SI_calc, pore waters behaved differently from the lake water column. The highest values were obtained in December 2007 and the minimum occurred in July 2008. In September 2008, the SI_calc increased sharply.

The pH of surface and bottom waters increased from December 2007 to the maximum in July 2008, and it dropped in September 2008. In pore waters, the pH was fairly stable from December 2007 to April 2008 and decreased in July 2008. In September 2008, the pH significantly increased.

Ca²⁺ in surface and bottom waters demonstrated a decreasing trend during the whole study period. In pore waters, the maximum concentrations of Ca²⁺ were obtained in December 2008. After that there was a sharp decline in April 2008 and July 2008, followed by a slight enrichment in Ca²⁺ in September 2008.

The highest concentrations of Mg²⁺ in surface and bottom water occurred in December 2007. Then Mg²⁺ was depleted until September 2008. The pore waters were enriched with Mg²⁺ with respect to overlying waters. The highest values of Mg²⁺ in pore waters were observed in December 2007 and the lowest occurred in July 2008.

Mg²⁺ behaves conservatively in relation to salinity (expresses as Cl^- concentrations), whereas Ca²⁺ appears to be a non-conservative constituent of the lake waters (Fig. 2). Thus, the concentrations of Ca2+ vary in time and space due to biogeochemical processes rather than simple physical mixing.

In the surface and bottom waters, the CA showed a slight decrease from December 2007 to April 2008. In July, there was a reasonable increase in CA in bottom waters while in the surface waters, the average CA remained similar to April 2008. The lowest values of CA occurred in September 2008. In the pore water, the CA showed an overall decreasing trend from December 2007 to July 2008 and increased in September 2008.

δ¹³C_DIC in surface, bottom and pore waters changed in a similar manner and showed increasing values from December 2007 to April 2008. In December 2007, the δ¹³C_DIC in pore waters was mostly positive, however, showed a considerable variability throughout the lake. In July 2008, there was a sharp decline in δ¹³C_DIC to highly negative values and in September 2008, the δ¹³C_DIC increased in the lake water.

The content of CaCO₃ in the surface deposits was between <5 and 28%. Spatial distribution of CaCO₃ throughout the lake is shown in Fig. 3. The only carbonate phase is calcite. On the basis of the low Mg/Ca ratio in water (between 0.06 and 0.37), the calcite can be classified as low-Mg calcite (Müller et al. 1972).

Figure 3

The calcite from the surface sediments of Lake Sarbsko had relatively uniform δ¹³C_calc values of -1.55 to -1.98‰.

Hodell et al. (1998) established that δ¹³C_calc preserves δ¹³C_DIC from the period of the maximum rate of carbonate accumulation. This is because there is rather low fractionation between DIC and calcite, and the latter tends to be slightly enriched with 13C with respect to DIC (Leng & Marshall 2004). In Polish lakes, the δ¹³C_calc values in the surface deposits are 1.3-4.1‰ more positive than those of δ¹³C_DIC (Wachniew & Różański 1997; Piotrowska & Hałas 2009). Similar values were obtained in other geographical regions (Hollander & McKenzie 1991; Hodell et al. 1998; Herczeg et al. 2003; Myrbo & Shapley 2006; Barešić et al. 2011). Consequently, the time of CaCO₃ precipitation can be estimated by matching δ¹³C_calc values to the equilibrium δ¹³C_DIC. However, the interpretation of δ¹³C_calc values in the sediments of coastal lakes is not that simple because there is no certainty that the sedimentary calcite is in isotopic equilibrium with ambient water. The sediments of coastal lakes are vigorously mixed and consequently calcite crystals, precipitated in different conditions and characterized by different stable C isotope signatures, can be redeposited in the surface layer of the sediments. In addition, the sedimentary calcite might be prone to selective dissolution which decreases the δ¹³C_calc values in the residual calcite (Herczeg et al. 2003). Thus the measured δ¹³C_calc values represent the product of mixing of these calcites. Moreover, non-equilibrium conditions of precipitation in the lake can be related to the intense primary production (Barešić et al. 2011).

Despite the above limitations, it seems that on the basis of δ¹³C_calc and δ¹³C_DIC, it is possible to determine the approximate time of calcite precipitation in Lake Sarbsko. This is because sedimentary calcite is enriched with 13C with respect to DIC, even if the precipitation occurs under non-equilibrium conditions (Barešić et al. 2011). Consequently, slightly negative values of δ¹³C_calc (-1.55 to -1.98‰) would indicate that the large part of CaCO₃ was precipitated from water characterized by more negative mean δ¹³C_DIC, which occurred during the summer 2008. This conclusion seems to be true even if the calcite is affected by partial dissolution in the sediments. Herczeg et al. (2003) estimated that during dissolution, the δ¹³C_calc signatures in Blue Lake (Australia) became 0.4‰ lower. When this value is added to δ¹³C_calc measured in Lake Sarbsko sediments, we obtain -1.15 to -1.58‰ which is still too low for δ¹³C_DIC in December 2007, April 2008 and September 2008.

Some evidence for the close relationship between δ¹³C in sedimentary calcite in the lake and the lake water chemistry is provided by the fact the δ¹³C_calc-based reconstruction of salinity well agrees with other geochemical and paleoecological proxies (Bechtel et al. 2007; Woszczyk et al. 2010).

The concentrations of Ca²⁺ and CA as well as the values of SI_calc in pore waters in December 2007 were distinctly higher than in surface and bottom waters. Moreover, concentrations of both Ca²⁺ and CA were at their maxima throughout the study period. The high concentrations of these chemical species could have developed owing to microbial decomposition of organic matter (e.g. SO₄^2- reduction, methanogenesis) and/or dissolution of CaCO₃ in the sediments. The contribution of microbial reactions in delivering DIC to the pore waters is not confirmed unambiguously by δ¹³C_DIC. In the case of sulfate reduction, δ¹³C_DIC is expected to be negative (Myrbo & Shapley 2006) and methanogenesis leaves DIC with very positive δ13CDIC (up to ca. +20‰ in pore waters and +13‰ in lake water column) (Gu & Schelske 1996; Gu et al. 2004). On the other hand, the weak correlation of HCO₃^- and Ca²⁺ (R2=0.20) and the average molar ratio of HCO₃^-/Ca²⁺ of only ca. 1.6 argue against enhanced dissolution of CaCO₃, because this process releases 2 moles of HCO₃^- per 1 mol of Ca²⁺ (Jahnke & Jahnke 2004). It is thus possible that the chemical composition of pore waters in December 2007 was affected by intrusion of brackish water from the Baltic Sea and/or from the Łeba river. The ingression was shown on the basis of the distribution of chlorides in the lake (Woszczyk et al. 2010), however, the salinity change was appreciably low. The inflow of saltwater could also be responsible for the increased concentrations of Mg²⁺, lower HCO₃^-/ Ca²⁺ ratio, and higher SI_calc (Berner 1971).

Between December 2007 and April 2008, the concentrations of Ca²⁺ and CA decreased, while they remained relatively unchanged in the water column. In addition, the SI_calc significantly decreased. Such a change in water chemistry could be due to the removal of Ca²⁺ from the sediments via precipitation.

Despite the relatively high values of SI_calc (0.20-0.70), indicating supersaturation of calcite in the pore waters between December 2007 and April 2008, precipitation of CaCO₃ is rather unlikely to occur. The precipitation of calcite within this period is unsupported by δ¹³C_calc values in the lake sediments (Table 1), which seem too low to be in equilibrium with the pore water DIC in December 2007 and April 2008. In addition, the precipitation of CaCO₃ is usually accompanied by a decrease in pH (Soetaert et al. 2007), while in December 2007 and April 2008 this parameter was nearly the same.

It is believed that precipitation of calcite was hindered by the appreciably high concentrations of dissolved organic carbon (DOC), Mg²⁺ (0.3-0.9 mmol l-1) and/or PO₄^3- (0.4-9.1 μmol l^-1; (Woszczyk et al. 2011a)) in the lake waters which inhibited nucleation and crystal growth of calcite (Ramisch et al. 1999; Barešić et al. 2011). The inhibitory effect of organic carbon on CaCO₃ formation can be observed in the Szczecin Lagoon, where the high contribution of the Oder-derived organic carbon results in low CaCO₃ sedimentation throughout the major part of the lagoon (Osadczuk 1999). Müller et al. (2006) report that the presence of 0.3 mmol Mg²⁺ l^-1 is sufficient to increase the solubility of CaCO₃. Therefore, when higher concentrations of Mg²⁺ build up, higher saturation levels would be required to initiate crystallization of CaCO₃. Therefore, it seems reasonable that the changes in the concentrations of Ca²⁺ between December and April were caused by diffusion/advection to overlying waters and not by the precipitation of calcite.

During summer, the pore waters became strongly undersaturated with respect to calcite, which created a considerable contrast to surface and bottom waters demonstrating the maximum CaCO₃ saturation. The state of undersaturation of pore waters with respect to CaCO₃ is usually related to decomposition of organic matter (Wu et al. 1997; Müller et al. 2006), which inevitably leads to a decrease in pH and consequently to a decline in SI_calc. Moreover, the enhanced oxidation of sedimentary organic matter has been invoked to explain highly negative shifts in δ¹³C_DIC due to recycling of ¹²C-enriched DIC from organic materials (Herczeg & Fairbanks 1987; Wachniew & Różański 1997; Leng & Marshall 2004).

Despite the pore waters being highly undersaturated with respect to calcite in July 2008, there was no evidence of extensive dissolution of CaCO₃ resulting in an increase in water pH, in increased concentrations of Ca²⁺ and CA (Müller et al. 2006), or in a positive shift in δ¹³C_DIC (Herczeg et al. 2003; Myrbo & Shapley 2006). Such changes occurred in September 2008. Moreover, in September 2008, the concentrations of Ca²⁺ and CA revealed a strong positive correlation (R2=0.99) and the average molar HCO₃^-/ Ca²⁺ ratio was 2.1. This value is very close to the stoichiometric relationship between the products of calcite dissolution according to the reaction CaCO₃ + CO₂ + H₂O → Ca²⁺ + 2HCO₃^-.

As mentioned above, the sediments of Lake Sarbsko contain appreciable amounts of CaCO₃ (Table 1). A portion of CaCO₃ was dissolved in pore waters between July and September 2008, but another part resisted dissolution. It was not possible, in this study, to estimate the fraction of calcite subjected to dissolution; however, it may vary in a very broad range. For example, in Lake Sempach (Switzerland), about 4.5% calcite is dissolved annually (Müller et al. 2006). In Lake Soppensee (Switzerland) and Lake Gościąż (Poland), ca. 80% of precipitated calcite is buried, the rest being re-dissolved (Wachniew & Różański 1997; Gruber et al. 2000). On the other hand, in Lake Lugano (Switzerland/ Italy) and Blue Lake (Australia), only 24% and 12% of epilimnetic calcite, respectively, is preserved (Ramisch et al. 1999; Herczeg et al. 2003). Among many factors conducive to the dissolution of calcite, the intensity of aerobic mineralization of organic matter (Müller et al. 2003), the small crystal size (Teranes et al. 1999) and the presence of chemical impurities within the crystals (Müller et al. 2006) seem to be the most important.

Ca²⁺, CA, SI_calc and δ¹³C_DIC in surface and bottom waters of Lake Sarbsko show similar temporal patterns, indicating that the lake waters are constantly well mixed. The SI_calc values indicate the increasing saturation with respect to calcite from December 2007 to July 2008 and the declining saturation from July onwards. The state of supersaturation and crystallization of CaCO₃, occurring at SI_calc >0.2 (Baumgartner et al. 2006), are favored between April and July 2008.

The rapidly decreasing concentrations of Ca₂₊ and CA together with the decreasing SI_calc between July and September 2008 indicate the ongoing removal of CaCO₃ from the lake water column due to precipitation of calcite. The summer precipitation is very common and can be attributed to photosynthetic activity of phytoplankton (Teranes et al. 1999; Müller et al. 2006). This process seems to produce the major quantity of sedimentary CaCO₃ in Lake Sarbsko, which is supported by δ¹³C_calc ranging from -1.55‰ to -1.98‰ (Table 1). Such values of δ13Ccalc seem to be close to the equilibrium with summertime δ¹³C_DIC of -7.07 to -3.54‰. Keeping in mind the limitations of δ¹³C_calc signatures in Lake Sarbsko, described in section 5.1, the time of calcite precipitation in the lake can be determined only approximately. It seems reasonable to conclude that formation of calcite in the lake occurred in the water column between late spring and mid-summer. With this respect, Lake Sarbsko is similar to the majority of Polish freshwater lakes (Wachniew & Różański 1997; Tylmann et al. 2012). Additional evidence for the precipitation of CaCO₃ from the lake water column during summer is provided by the satellite image of the lake taken in July 2005, showing the widespread whiting (Fig. 4).

Figure 4

Between April and July 2008, δ¹³C_DIC in Lake Sarbsko waters dropped by ca. 10‰ (Fig. 2). Such a rapid change might be linked to the removal of ¹³C by precipitating calcite and also to the enhanced delivery of ¹²C. There are two processes occurring within the lake in which ¹²C can be added to the lake waters: degradation of organic matter and chemically enhanced invasion of atmospheric CO₂ in the lake (Wachniew & Różański 1997). Owing to the coincidence between the decline in δ¹³C_DIC and the decrease in the concentration of dissolved O₂ in bottom waters (Woszczyk et al. 2011a), it seems that the decomposition of organic matter is a more likely cause of the observed changes. The depletion in O₂ indicates the consumption of oxygen in the lake water column and/or diffusion of O₂ to the sediments where OM is respired. During oxic respiration, there is no fractionation of carbon (Shoemaker & Schrag 2010), so this process can account for a highly negative shift in δ¹³C_DIC (Leng & Marshall 2004).

On the other hand, chemically enhanced invasion of atmospheric CO₂ is observed even in net heterotrophic lakes during periods of intense primary production, when pH is very high (>9) and pCO₂ is much below the atmospheric equilibrium of 360 μatm (3.65 × 10-4 bar) (Herczeg & Fairbanks 1987). In addition, chemical enhancement requires calm meteorological conditions with wind speeds of up to 3-4 m s^-1 (Wanninkhof & Knox 1996). In Lake Sarbsko in July 2008, pCO₂ was between 1.8 × 10^-4 and 3.5 × 10^-4 bar at most of the studied sites (3 sites had higher values), especially in the central and eastern part of the lake. Thus, the invasion of atmospheric CO₂ was thermodynamically favored. However, the chemical enhancement of the invasion does not appear to significantly affect the δ¹³C_DIC under pH values of 8.27-8.62 and in the conditions of high agitation of the water by wind waves. For pH values between 8 and 9 coupled with moderate to high wind stress, the fractionation during CO₂ transfer across the air-lake water interface is estimated to be between -0.1 and -3.8‰ (Herczeg & Fairbanks 1987). On the other hand, if the CO₂ transfer is preferentially due to molecular diffusion, the fractionation is between +6.8 and +9.2‰, which can lead to a slight increase in δ¹³C_DIC (Herczeg & Fairbanks 1987).

The coincidence of the precipitation of CaCO₃ with the oxidation of OM as well as the invasion of atmospheric CO₂ is not easy to explain at this stage of the research, especially that both processes are known to lower pH (Soetaert et al. 2007) and thus create unfavorable conditions for calcite to precipitate. There are many studies showing that degradation of OM in waters is accompanied by dissolution of CaCO₃ (Wachniew & Różański 1997; Wu et al. 1997; Müller et al. 2006; Myrbo & Shapley 2006; Tylmann et al. 2012). One can only speculate that the rate of respiration in Lake Sarbsko was too low to stop the precipitation.

The precipitation of calcite in temperate lakes is known to occur until autumn (Teranes et al. 1999; Tylmann et al. 2012) if photosynthesizing phytoplankton is active. In September 2008, however, the SI_calc values in most studied sites in Lake Sarbsko were relatively low (<0.1), arguing against precipitation.

The cessation of calcite precipitation in late summer might be linked to the relatively low pH and low CA of lake waters, resulting from the intense decalcification in the preceding period. It is well established that the precipitation of CaCO₃ reduces pH and alkalinity (Soetaert et al. 2007) and, consequently, decreases the state of saturation with respect to calcite (Hodell et al. 1998).

Alternatively, the precipitation of CaCO₃ from lake waters could be affected by the intense wind-driven mixing of lake waters and sediments which occurred in Lake Sarbsko between July and September 2008 (Woszczyk et al. 2011a). The deep turbulent mixing had a profound effect on the carbonate saturation in the lake via changes in pCO₂ and pH of surface, bottom, and pore waters. In mid-summer there was a steep pCO₂ gradient between the pore and overlying waters, maintaining the large differences in pH. The surface and bottom waters showed pCO₂ of 1.8 × 10^-4 – 5.9 × 10^-4 bar, while pore waters had 12.1 × 10^-4 – 275.4 × 10^-4 bar. Within the two months, owing to intense mixing, pCO₂ in surface and bottom water increased to 4.0 × 10^-4 – 12.5 × 10^-4 bar and became similar to that in pore waters (5.3 – 13.4 × 10^-4 bar, with one outlier of 101.4 × 10^-4 bar). The decrease in pCO₂ in pore water induced the increase in pH and consequently in SI_calc. On the other hand, the increase in pCO₂ in the surface and bottom waters reduced pH and SI_calc values in the former. The average value of pCO₂ for the whole lake dropped from 50.4 × 10^-4 bar in July to 10.7 × 10^-4 bar in September.

There is at least a threefold explanation of such a dramatic decrease in pCO₂. First, some parts of CO₂ could have been used for dissolution of calcite. Second, Balmer & Downing (2011) postulate that vertical mixing fuels the primary production in shallow lakes by transferring the nutrients from the sediments to overlying waters. Elevated concentrations of chlorophyll in Lake Sarbsko in August 2008 (Woszczyk et al. 2011a) indicate that the enhanced bioproductivity in the lake might have contributed to a drop in pCO₂. Third, it seems that the major change in pCO₂, however, was due to evasion of CO₂ to the atmosphere. It is well established that vertical mixing of water stimulates the emission of CO₂ from lakes whenever the CO₂ concentrations in lakes are higher than atmospheric pCO₂ (Eugster et al. 2003; Jonsson et al. 2008; Balmer & Downing 2011).

Despite the fact that it is not possible to quantify the individual contributions of the above three mechanisms to the pCO₂ decrease in the lake, the combination of CaCO₃ dissolution, the increased primary production, and the evasion of CO₂ from the lake provides the most plausible explanation of the 5-6‰ increase in δ¹³C_DIC between July and September 2008 (Wachniew & Różański 1997; Herczeg et al. 2003; Leng & Marshall 2004; Myrbo & Shapley 2006).

Both the content and spatial distribution of CaCO₃ in the sediments of Lake Sarbsko show the overall similarity to the Vistula Lagoon and Curonian Lagoon (Pustelnikovas 1998; Uścinowicz & Zachowicz 1996), i.e. high concentrations of CaCO₃ are found within the fine grained (muddy) facies and co-precipitate with the phytoplankton-derived TOC (Woszczyk et al. 2014). At the same time, there is no relationship between the content of CaCO₃ within the surface deposits and the composition of water.

It seems, therefore, that hydrodynamic processes play the major role in the distribution of CaCO₃ in the lake. The processes involved in the composition and distribution of the surface sediments in Lake Sarbsko have been discussed by Woszczyk et al. (2014).