Introduction

Dilute soda lakes represent the most alkaline naturally occurring aquatic ecosystems with pH values of up to 12 [13]. These ecosystems are characterized by the presence of large amounts of NaHCO3 and Na2CO3 and in many cases by low concentrations of Mg2+ and Ca2+ because of the insolubility of those cations as carbonate minerals under alkaline conditions. Conditions suitable for the formation of soda lakes are found in arid and semiarid climate zones [11, 40], where intense evaporative concentration rates exceed inflow rates such that salts accumulate [13]. Soda lakes are regarded as being among the worlds most productive environments, as the access to dissolved inorganic carbon in form of HCO −3 and CO 2−3 for primary producers is unlimited [6]. In most cases, alkaliphilic cyanobacteria are the dominating primary producers. Aerobic heterotrophic bacteria have been reported to reach extremely high numbers of 107 to 108 cells mL−1 [12, 16], and bacterial secondary production can be assumed to exceed values measured thus far for natural aquatic ecosystems. Seasonal variation in bacterioplankton biomass and production of alkaline aquatic ecosystems are poorly examined, as well as their controlling environmental factors. Plate count numbers of organotrophic bacteria were shown to remain more or less constant over the year in two deep African soda lakes, despite marked seasonal changes in salinity due to periods of heavy rain [14]. But to our knowledge, there is no information available on the seasonal changes of total bacterioplankton in shallow soda lakes.

The soda pools investigated here are located in Eastern Austria, within the area of the national park Neusiedler See-Seewinkel (47°45′–49′, 16°47′–53′), above the largest mineral water deposit in Europe [24]. The mineral solutes ascending with the groundwater flux [24] formed these shallow soda pools (max. depth: 50 cm) with pH values ranging from 8.5 to 11. Na+ is the dominating cation, and HCO −3 , CO 2−3 , Cl−, and SO 2−4 represent the major anions. The salinity of the pools can vary strongly over the seasons, but also within years, ranging from 0.2% to >4.0% (w/v) before complete evaporation, as they are often exposed to periods of severe aridity, typical for the Pannonian climate in eastern Austria. Loss of salts can then occur by wind deflation. Chemophysical parameters such as salinity, alkalinity, and temperature may have a strong impact on the seasonal changes of bacterial activity in these pools. High solar radiation and regional winds resuspending a substantial portion of the sediment are considered additional important external factors.

The aim of the study was to elucidate the special environmental features characterizing these environments and enabling the development of such abundant and productive bacterial populations in the five pools over a seasonal cycle. Therefore, a variety of ecological parameters were measured along with heterotrophic bacterial numbers and secondary production, and the main factors controlling the bacterial community in the pools were extracted by multivariate comparisons.

Materials and Methods

Study Site and Sampling

Samples were collected from five soda pools in biweekly intervals during summer and once every 6 weeks during the winter season from May 2000 to May 2001 (Fig. 1). According to the trophic classification system proposed by Forsberg and Ryding [9], the investigated soda pools are extremely hypertrophic ecosystems (Table 1). The Oberer Stinkersee (OS) is the pool with the highest total salt concentrations and a wind-independent permanent turbidity, whereas the Unterer Stinkersee (US) has a varying wind dependent turbidity and a varying concentration of humic substances leading to dark brown water color in periods of massive decay of plant material. The Illmitzer Zicksee (ZL) is characterized by a low turbidity caused by the fine sediment covered by an intense biofilm preventing resuspension of the sediment particles. The Lange Lacke (LL) and the Wörthenlacke (WL) exhibit a varying wind dependent turbidity. Despite their proximity they feature large differences in their chlorophyll a content and their amount of total suspended solids (Table 1).

Table 1 Basic characterization of the five investigated soda pools
Full size table
Figure 1
figure 1

Location of the five investigated saltwater pools within the national park Neusiedler See—Seewinkel. Sampling sites are marked by arrows. Villages are black.

Full size image

Chemophysical Parameters

Conductivity (WTW, LF 330), temperature, pH (Seibold Wien, GHM) and oxygen (Seibold Wien, SHO/2) were measured in situ three times along 20-m transects toward the center of the pools. Along the transects three 1-L water samples were collected and integrated for further analysis. Wind velocity was arbitrarily estimated according to the Beaufort wind scale (http://www.zetnet.co.uk/sigs/weather/Met_Codes/beaufort.htm ).

For the determination of total suspended solids (TSS), a defined volume of sample water was filtered through premuffled glass-fiber filters (GF/C; Whatman; England) and dried to constant weight. To obtain the inorganic and organic fraction, the samples were further combusted in a muffle furnace. Total phosphorus was determined photometrically after dissolution of the unfiltered sample with potassium peroxydisulfate, using the molybdenium-blue method according to Strickland and Parsons [35]. After 1:10 dilution with distilled water, the ion concentrations in the samples were measured with ion chromatography. All columns and chemicals were supplied by Dionex (Sunnyvale, CA). For anions, AS-4A columns with AG-4A precolumns and for cations CS-12A columns with CG-12A precolumns were used. As displacing liquid for anions a sodium carbonate buffer (6 mM Na2CO3 + 12 mM NaHCO3) and for cations an 18 mM methanesulfonic acid solution was used. The ions were detected with a conductivity detector after removing the anions with the appropriate suppressors (AMMS-II) and for cations (CSRS-Ultra 4mm). Alkalinity (ALK; mmol L−1) was calculated by the following equation:

On two occasions the carbonate and bicarbonate concentrations were measured with HCl titration to calculate the dissociation constant (pK) of the [CO 2−3 ] ⇔ [HCO −3 ] equilibrium (2):

The mean of the calculated pK values from each site (n = 10) was inserted in the following equations to calculate the carbonate and bicarbonate concentrations of the other sampling dates:

Salinity (SAL; mg L−1) was determined by the addition of the measured and calculated salt concentrations:

Dissolved Organic Carbon and Chlorophyll a

For the determination of dissolved organic carbon (DOC), subsamples of the integrated water samples were filtered through precombusted Whatman GF/C filters. DOC was determined using a Shimadzu TOC 5000 carbon analyzer (Shimadzu Corporation, Tokyo, Japan). Chlorophyll a was extracted with 90% ethanol (1 h at 80°C) and was measured spectrophotometrically (Hitachi U-2000) [30].

Bacterial and Cyanobacterial numbers

For the determination of bacterial and cyanobacterial numbers, 15-mL samples were preserved with formalin (4% v/v final concentration) for direct epifluorescene microscopy. Each sample consisted of three 5-mL subsamples collected along the 20-m transects. Cells were stained with DAPI (0.01% v/v final concentration) and filtered through black polycarbonate membrane filters (Millipore, Ireland) of 0.2 µm pore size [29]. Bacterial (UV excitation: 340–380 nm, barrier filter: 430 nm) and cyanobacterial numbers (red light excitation: 515–560 nm, barrier filter: 580 nm) were counted with a Leitz Diaplan microscope. Bacterial cell volumes (V; µm3) of at least 100 cells were determined for representative samples by eyepiece micrometer to calculate the average cellular carbon content (C; fg C cell−1) after Norland [27]. Cyanobacteria in representative samples were taxonomically classified after Komarek and Anagnostidis [23].

Bacterial Secondary Production

To measure bacterioplanktonic secondary production in the five saltwater pools the [3H]leucine incorporation method, initially developed by Kirchman et al. [18] and Simon and Azam [34], was used and the protocol of Kirschner and Velimirov [20] was followed. A wide range of leucine concentrations ranging from 30 to 240 nM was tested twice at each sampling site. The resulting incorporation velocities were iteratively fitted to the hyperbola function of the Michaelis–Menten enzyme kinetics by using nonlinear regression analysis (Delta Graph 4.0, Delta Point Inc., USA). The plots were used for the calculation of the theoretical maximal uptake velocity (V max) and of the half-saturation constant (K m). The mean of all calculated K m values (n = 10) was used to calculate V max for all other sampling dates from the [3H] leucine uptake velocity (V) at a substrate concentration (S) of 60 nM [3H]leucine according to Equation (6) [8].

[3H]Leucine incorporation rates (leuinc) were converted to carbon production (BSP) according to Simon and Azam [34] by using the following equations:

where BPP = bacterial protein production; 2 = isotope dilution as recommended by the authors; 100/7.3 = 100/mol% of leucine in protein; 131.2 = molecular weight of leucine.

where 0.86 and 1.18 represent the factors requested for converting bacterial protein production into BSP, taking into account that 86% of the weight of the aminoacids is the carbon moiety and that 18% of the amino acids are not detected by [34]. The specific growth rate (µ; [h−1]) and the doubling time (g; [h]) were determined by the following equations (9, 10):

where BN0 = bacterial numbers at time 0, corresponding to the bacterial numbers determined for this sampling point via the epifluorescence technique, and BNbsp-bacterial secondary production expressed in cells L−1 measured within the 0.5 h incubation interval. The doubling times were then calculated after:

Microautoradiography

Because of the high concentration of cyanobacteria, microautoradiography was used to determine whether cyanobacteria were capable of leucine uptake under conditions similar to the determination of BSP (180 nM 3H-leucine, 30 min, in situ temperature). On two occasions during summer, 1-mL samples from each sampling site were incubated with 30 µL [3H]leucine and stopped with 60 µL TCA (5% v/v final concentration). After staining with DAPI (0.01% v/v final concentration) the samples were diluted 10×–50× to obtain optimal cell densities for microscopic examination. The protocol by Fuhrman and Azam [10] was followed, which allows simultaneous visualization of developed silver grains and DAPI-stained bacteria, avoiding interference of the silver grains with viewing the bacteria. Bacteria and cyanobacteria were viewed by epifluorescence microscopy, and clusters of silver grains (≥5 grains) were viewed by phase contrast microscopy.

Statistical Analysis

Data were analyzed according to Zar [39]. A nonparametric Tukey test was applied for two independent samples and a Kruskal–Wallis test for more than two independent samples to compare the five saltwater pools. Principal component analysis (PCA, varimax rotated with Kaiser normalization) was performed to evaluate the parameters responsible for the seasonal fluctuations of the investigated environmental factors in the saltwater pools. For multiple linear stepwise regression analyses and PCA analyses, data not meeting the requirements of homoscedaticity and normal distribution (Shapiro–Wilks test) were log10-transformed after adding 1 to the variable. Further nonparametric Spearman rank correlations were performed for the data of each pool. For calculation of annual averages, monthly means were used. All statistical analyses were performed using the software SPSS 10.0 for Windows.

Results

Physical and Chemical Parameters

The water temperature during the investigation period varied from 1.2°C to 34.4°C, following the typical pattern observed in lakes located in the temperate climate zone. The conductivity in the five habitats increased with the summer evaporation to maximal values of 32 mS cm−1 (ZL) and decreased with the autumn and winter rainfalls and the subsequent groundwater raising (Fig. 2A). The conductivity was highly correlated with salinity (r s = 0.89, p < 0.001) and alkalinity (r s = 0.63, p < 0.001) and was a good indicator for the evaporative concentration. Salinity varied from 2.4 g L−1 to 31.3 g L−1 (Table 1) and was dominated by chloride, sulfate, carbonate, and bicarbonate anions and sodium cations. The proportion of the carbonate and bicarbonate anions comprised up to 60% of total salinity (data not shown) and was responsible for the observed high pH values up to 10.89 in the pools. The high alkalinity (Fig. 2B) was in turn responsible for the relatively low calcium and magnesium concentrations with mean values of 47 mg L−1 and 148 mg L−1, respectively. The peak of conductivity and alkalinity in spring 2000 at US was observed when the part of the US which was used as sampling site for the first four sampling dates desiccated and thereafter an adjacent sampling site was chosen.

Figure 2
figure 2

Annual cycle (8 May 2000 to 28 May 2001) of conductivity (A), alkalinity (B), and DOC (C) in the five investigated soda pools: Oberer Stinkersee (OS), Unterer Stinkersee (US), Illmitzer Zicklacke (ZL), Lange Lacke (LL), and Wörthen Lacke (WL). Conductivity values represent the mean of three measurements; standard errors are smaller than the symbols used. Missing September values for ZL are due to complete evaporation of the pool.

Full size image

DOC and Chlorophyll a

DOC values and conductivity were highly significantly correlated (r s = 0.80, p < 0.001). The highest DOC value coincided with the highest conductivity value in ZL before complete evaporation (Fig. 2C). Chlorophyll a values ranged from 0 to 430 ng L−1 with the lowest concentration in the ZL (annual average: 8.2 µg L−1) and the highest concentration in LL (annual average: 197 µg L−1; Table 1).

Cyanobacterial and Bacterial Numbers

Cyanobacterial numbers ranged from 0 to 307 × 109 cells L−1 with highest values observed at OS and LL in early August 2000 and early May 2001, respectively. US exhibited two peaks with values of more than 170 × 109 cells L−1 in July and September (Fig. 3A). ZL and WL featured the lowest median cyanobacterial numbers (Table 2). During the winter months cyanobacterial numbers were low (<10 × 109 cells L−1) in all pools except for LL in February, where numbers of 54 × 109 cells L−1 were observed. Cyanobacteria were dominated by unicellular Synechococcus species and filamentous forms belonging to the genus Cyanospira. Colony-forming species were frequently observed after long periods of stable weather conditions (personal observation). However, the phytoplankton consisted not only of cyanobacteria, whose numbers explained 36% of the chlorophyll a variation, but also of diatoms and euglenoids.

Table 2 Microbial characterization of the five investigated soda pools
Full size table
Figure 3
figure 3

Annual cycle (8 May 2000 to 28 May 2001) of cyanobacterial numbers (A), bacterial numbers (B), and bacterial secondary production (C) in the five investigated soda pools: Oberer Stinkersee (OS), Unterer Stinkersee (US), Illmitzer Zicklacke (ZL), Lange Lacke (LL), and Wörthen Lacke (WL). For (A) and (B), values represent the mean of an integrated sample ± one standard error of10–20 microscopic fields. For (C) values represent the mean of triplicate samples ± one standard deviation. Missing September values for ZL are due to complete evaporation of the pool.

Full size image

In all pools two peaks of bacterial numbers occurred in early and late summer with values up to 270 × 109 cells L−1 (Fig. 3B). Low numbers (<40 × 109 cells L−1) were observed after a strong decrease in July and during the whole winter season. A subjection of the dataset to a Kruskal–Wallis test revealed that the five soda pools were statistically different in bacterial (χ2 = 27, p < 0.001) and cyanobacterial numbers (χ2 = 41, p < 0.001). Bacterial and cyanobacterial numbers were consistently greater in the soda pools with higher concentrations of total suspended solids (Tables 1 and 2) as indicated by Tukey tests and by highly significant positive correlations of TSS with bacterial numbers (r s = 0.76, p < 0.001) and cyanobacterial numbers (r s = 0.78, p < 0.001), which were also significantly intercorrelated (r s = 0.85, p < 0.001).

Bacterial Carbon Production (BSP)

From all performed saturation experiments (n = 10) a mean half-saturation constant (K m) of 31.8 nM leucine (SE: 2.5 nM) was calculated and further used to calculate the maximal leucine uptake (V max) from the leucine incorporation rates (V) at 60 nM (S) for all other sampling dates. In all experiments production rates reached saturation at a concentration of 180 nM (data not shown). Rates of bacterial production ranged over three orders of magnitude during the investigation period, from 1 to 738 µg L−1 h−1 (Table 2). The annual maximal BSP rates were measured at OS, US, and LL in the middle of July, coinciding with a strong decrease in bacterial numbers and temperature. However, highly significant positive correlations of BSP with bacterial numbers (r s = 0.59, p < 0.001) and temperature (r s = 0.58, p < 0.001) were found, as well as with total suspended solids (r s = 0.47, p < 0.001) and DOC (r s = 0.45, p < 0.001). At all sites high rates >30 µg C L−1 h−1 of bacterial secondary production were measured from April to October, when water temperature was mostly above 20°C. With one exception (January, LL) low rates <30 µg C L−1 h−1 were recorded from November to March. Over the year, the rates of bacterial production were not statistically different between the five study sites, as indicated by Kruskal–Wallis test (χ2 = 4.5, p > 0.1).

The specific growth rates of the bacterioplankton in the five pools varied strongly over the seasons from 0.004 to 1.65 h−1 (Table 2) corresponding to doubling times of 8 days and 25 min, respectively. The microautoradiography experiments showed that the majority of silver grains in the leucine autoradiograms were associated with bacteria (data not shown). In all incubations with [3H]leucine only a small fraction (<0.5%) of silver grains was associated with cyanobacteria (Fig. 4), suggesting that even a high cyanobacterial biomass hardly biased the measurements of bacterial secondary production in the five saltwater pools. “Labeled” bacteria were determined once for each habitat, ranging from 40% (OS) to 90% (WL) of all DAPI-stained bacteria (data not shown). These results represent minimum estimates of the percentage of active cells in the five studied pools, because leucine was probably not utilized by all bacteria.

Figure 4
figure 4

Example of a microautoradiogram of autofluorescent picocyanobacteria from OS after the addition of 180 nM 3H-leucine (1200× magnification). The overwhelming majority of picocyanobacterial cells are not labeled. Arrows indicate sporadic autofluorescent cells with 3H-leucine label incorporation. The other silver grains in the leucine autoradiogram were associated with bacteria (at UV excitation 340–380 nm, barrier filter 430 nm).

Full size image

Principal Component Analysis

In order to elucidate the main environmental factors describing the systems of the five soda pools, a PCA of the whole data set was performed. Four principal components (PC) with an eigenvalue >1 were extracted (Table 3, Fig. 5). PC 1 explained 45.2% of the observed variance and consisted of inorganic and organic suspended solids, total phosphorus, chlorophyll a, and bacterial and cyanobacterial numbers. PC 2 explained 17.3% of the observed variance and contained salinity, conductivity, and DOC. PC 2 contained also the variable wind with a negative sign, which can be interpreted that in periods with high evaporative concentration the weather situation is mostly calm and viceversa. PC 3 (temperature and BSP) and PC 4 (alkalinity and pH) were of less importance, explaining only 10.0 and 7.6% of the variance, respectively. PC 1 can be described as a turbidity factor as it includes all particulate variables. However, it should be mentioned that the variables chlorophyll a and cyanobacterial numbers represent the manifested result of primary productivity, which is tightly coupled to the annual light cycle and which can be seen as an integrated component of PC 1. PC 2, on the other hand, can be described as a concentration factor as it contains all dissolved variables.

Table 3 Principal component analysis showing the principal components (PC) with an eigenvalue >1 and their explained variance
Full size table
Figure 5
figure 5

Two-dimensional plot of the principal component analysis (PCA) (rotation: varimax with Kaiser normalization) performed for the whole data set of all investigated soda pools. cond, conductivity; doc, dissolved organic carbon; sal, salinity; alk, alkalinity; bsp, bacterial secondary production; ptot, total phosphorus; sso, organic suspended solids; ssi, inorganic suspended solids; bn, bacterial numbers; cyb, cyanobacterial numbers; temp, temperature; chla, chlorophyll a.

Full size image

The PCAs of the single pools showed slightly diverging pictures than the PCA performed with the whole data set (Table 4). PC 1 and PC 2 could be described as the turbidity factor and concentration factor, together explaining more than 56% and 73% in OS and US, respectively. At ZL, PC 1 included both the turbidity and the concentration factor, explaining 55% of the total variance alone, and PC 2 consisted of total phosphorus and alkalinity. A strong biotic component (PC 2) explaining 20% of the total variance was found at LL, whereas PC 1 (33% explained variance) is interpreted as a combination of suspended solids and phytoplankton. The turbidity and the concentration factor were of inverted importance in WL. PC 1, explaining 52% of the observed variance, additionally included BSP, chlorophyll a, and total phosphorus.

Table 4 Principal component analysis of the five investigated soda pools showing the principal components (PC) with an eigenvalue >1 and their explained variance
Full size table

Discussion

Soda Pools as Sites of Extreme Bacterial Abundance and Growth

The soda pools of the national park Neusiedler See—Seewinkel appear to be sites of extremely high microbial activity and biomass. In past studies on hypertrophic aquatic environments highest bacterial numbers up to 356 × 109 L−1 were reported by Kilham [16] for African alkaline lakes. Grant et al. [12] reported that aerobic heterotrophic bacteria can reach numbers of 100 × 109 L−1 in dilute soda lakes, and Zinabu and Taylor [42] found maximal bacterial numbers of 117 × 109 L−1 in a variety of different Ethiopian soda lakes. These values are thus of the same magnitude as our measurements (Table 5). Ecosystems similar to the Austrian soda pools exist in Eastern Europe, Russia, and East Africa, but to the best of our knowledge no information is available on heterotrophic bacterial production and growth rates. The highest BSP rates found in the literature, up to 129 µg C L−1 h−1, were reported by Boon [5] for the Australian billabongs—values which were far exceeded by our measurements. White et al. [42] reported maximal µ values of 0.36 h−1 for freshwater ecosystems and 1.28 h−1 for saltwater habitats, which also included data from an artificial aquaculture pond. In Table 5 ecosystems are listed which exhibit similarities to the Austrian soda pools in salinity, trophy, depth, turbidity, and/or alkalinity. With the exception of Big Soda Lake, all of the mentioned environments are shallow and highly productive. The study of Big Soda Lake was chosen because it provides the only data on bacterial production in aquatic alkaline environments. The presented environments in Table 5 exhibit significantly lower bacterial production and growth rates than in this study, except for the Indus River Delta: Bano et al. [3] reported an unprecedented bacterial growth rate of 1.0 h−1 for a natural assemblage, coinciding with a dense cyanobacterial bloom. For the East African soda lakes it is conceivable that growth rates of magnitude similar to those observed for the Austrian soda pools may be encountered, but no data are available. We suppose that the combination of shallowness, the existence of growing algal blooms, and high temperature is probably the driving factor enabling growth rates of µ ≥ 1.

Table 5 Bacterial specific growth rates, abundance, chlorophyll a, dissolved organic carbon, pH, and salinity in selected estuarine areas, river deltas, and soda lakes
Full size table

The extremely high specific growth rates of up to 1.65 h−1, corresponding to doubling times of 25 min, suggest that bacterioplankton in the pools has the potential to grow at close to maximal rates. It was shown for laboratory cultures of Escherichia coli and other Enterobacteriaceae that the minimal time necessary for cell division under optimal nutrient and temperature conditions is in the range of 15–20 min [22]. We are aware that uptake rates of radiotracers do not directly measure growth. However, we believe that this approach provides a good estimate of bacterial growth rates, when conversion factors are chosen carefully and methodological limitations are taken into account. A recommended isotopic dilution of 2 [34] was applied for converting radioactive leucine uptake into carbon production. It is generally assumed that even at substrate concentrations leading to saturation of the incorporated label, intracellular isotope dilution does still occur. However, it is conceivable that in these extremely hypertrophic ecosystems bacteria do not synthesize leucine in their cells when external sources are available at high concentrations. Therefore, when no isotopic dilution may be taken into account, the maximal µ would then amount to 1.0 h−1 (LL) and 0.65 h−1 (ZL), respectively, corresponding to g = 42 min and g = 64 min. This is still only approximately two to four times slower than the minimal doubling time of Escherichia coli. Other sources of error are the conversion factors used in the calculations for converting cell volumes into carbon content and leucine incorporation into carbon production, i.e., mol% of leucine in protein and the cellular protein carbon moiety. However, when bacterial abundance is determined by the direct-count method, the calculated growth rates are most probably underestimated because of the presence of nonactive cells [17]. As indicated by our microautoradiography results, the percentage of active bacteria in the pools varied from 40% to 90%, which would increase the calculated growth rates by a factor up to 2.5, corresponding to µ > 1.65 h−1. Microautoradiograms, on the other hand, indicated clearly that cyanobacteria could be ruled out as contributing significantly to the leucine uptake, which would have led to an overestimation of heterotrophic bacterial growth rates. Only a very small percentage of the silver grains were found to be associated with cyanobacterial cells (<0.5%) in all samples (data not shown). This finding is in contrast to the conclusion by Kamjunke and Jähnichen [15] that measurement of bacterial production in highly eutrophic environments via the leucine method may not be suitable due to significant uptake of leucine by cyanobacteria. Also, Nilsson and Sundback [25] showed that cyanobacteria are able to take up amino acids. However, these authors investigated amino acid uptake by Microcystis sp. and Phormidium sp. two genera which have not been detected so far in saline soda lakes. Summing up, we believe that the observed maximal µ values ≈1 in the soda pools are realistic estimates and that even higher growth rates may be encountered.

The “Fluid Sediment Concept”

An important factor influencing the bacterial communities in the pools is represented by the high turbidity dependent on the character of the sediment and on the high Na+ values in combination with high pH values. The fine sandy sediments of OS and US are more easily resuspended by the action of wind than the silty (LL, WL) and clay (ZL) sediments (Table 1). In addition, the clay sediments in ZL are stabilized by a thick biofilm during the major part of the year. Inorganic suspended solids in the pools consist mainly of colloidal clay minerals; iron, manganese, and aluminum oxides; and amorphous silicates and carbonates, which possess a high density of superficial negatively charged O− groups [24]. These groups together with the Na+ ions build an electrostatic double layer responsible for the stability of turbidity [37]. Organic particles adsorb to the surfaces of the inorganic suspended solids and in turn remain suspended in the water column [7]. These observations lead us to characterize the turbid water columns of the pools as “fluid sediments.” Especially in OS, the pelagic heterotrophic bacterial numbers and production rates were only 10× and 3× lower than in the sediment (Kirschner et al., in prep.). Normally, these parameters differ by three orders of magnitude in shallow aquatic environments [20]. However, the observed doubling times were in the lowest range observed for water-column bacteria (see above) and thus significantly lower than normally observed in sediments [21]. Several speculations exist for the reasons of the long doubling times of benthic bacteria. Sander and Kalff [32] speculated that lower bacterial loss rates lead to higher cell numbers in sediments, and thus to longer doubling times. Törnblom and Boström [36] proposed that benthic bacterial populations consist of a high percentage of dormant or empty cells. The results of this study clearly suggest that after resuspension, bacteria attached to sediment particles have better access to utilizable organic matter and oxygen, leading to the conclusion that benthic bacteria are most probably limited by the availability of nutrients and oxygen.

Regulation of Bacterial Populations over the Seasons

The annual fluctuations of the bacterioplankton abundance in the soda pools were strongly influenced by the semiarid climate. High temperatures during summer and low precipitation rates led to an evaporative concentration of salinity and DOC. The highest DOC and conductivity values were recorded in ZL in August, shortly before complete evaporation, concomitantly with maximal bacterial numbers and production rates in this pool. However, PCA results indicated that turbidity seemed to be more important for the annual variation of bacterial communities and the description of the soda pools than the concentration of dissolved solutes. Nevertheless, most variables contained in both PC 1 and PC 2 were significantly positively intercorrelated, showing that evaporation positively affected the turbidity via a volume decrease of the water body. By this, the water body becomes shallower and wind effects gain a stronger importance for sediment resuspension. In addition, evaporative concentration increases the Na+ content and supports the electrostatic adsorption of suspended solids. Temperature was obviously of minor importance in describing the variance of the soda pools, but was clustered with BSP rates, and also a significantly positive correlation to BSP rates was observed (r s = 0.58, p < 0.001). Against expectation, wind velocity was not clustered with the variables contained in PC 1, and no significant correlation with total suspended solids was found. This leads us to the conclusion that even light winds can lead to significant sediment resuspension because of the shallowness of the soda pools. The PCA results of the single pools underline the differences between the five investigated soda pools despite of their geographical proximity. Although the turbidity and the concentration factor are essential controlling forces for all pools, they have a different importance for the description of each ecosystem. It is obvious that the proposed model of “fluid sediment” fits better to OS and US because of the combination of relatively coarse sediments with high Na+ concentrations and pH values, which is corroborated by the PCA, where the turbidity factor was of most significance for the variance of both systems. Such preconditions were only partly met in ZL because the clay sediment and the intense biofilm stabilizing the sediment prevent wind-induced sediment resuspension. In LL and WL biotic factors gain more importance in describing the variance of these systems, because the fine particle size of the sediment and the lower salinity values allow a stable formation of a “fluid sediment” only in periods with strong wind impact.

Over all investigated pools, bacterial numbers, cyanobacterial numbers, and bacterial production rates were strongly intercorrelated, as well as the concentration of total suspended solids. These strong correlations emphasize once more the key role of wind-induced sediment resuspension and salinity-enabled formation of a “fluid sediment” in regulating the bacterial community in the soda pools. In addition, bacterial production was also strongly influenced by temperature, as normally observed in aquatic ecosystems [1, 19, 32, 38]. Although DOC was correlated with BSP, we do not assume that organic or inorganic nutrients were limiting for the bacterial community, because of the observed high specific growth rates. Additionally, Shiah and Ducklow [33] pointed out that nutrient supply is not a limiting factor for bacterial secondary production when a strong correlation between temperature and BSP is observed.

A strong top-down control can also be supposed to exist in the saltwater pools regulating the bacterial and cyanobacterial numbers. Cladoceran and copepod densities of more than 2 × 103 L−1 and also ciliate abundances of >1 × 106 L−1 were frequently observed (personal observation), representing efficient potential regulators of the bacterial populations. In addition, viral lysis is a possible important regulating factor. Especially in eutrophic marine ecosystems, mortality due to viral lysis was assumed to account for more than 100% of the bacterial and cyanobacterial production [26]. Because of their extremely high abundances, it is also conceivable that bacteria and cyanobacteria are regulating their maximal population density via quorum sensing [2, 4].

Summing up, the investigated soda pools in the national park Neusiedler See—Seewinkel in Eastern Austria harbor extremely abundant populations of heterotrophic prokaryotes, and their productivity values are among the highest reported in the literature. We believe that the combination of dense phytoplankton blooms, high temperature, high turbidity, and evaporative nutrient concentration is the prerequisite enabling the development of such extremely productive microbial populations in a natural aquatic environment.