Diatom assemblages as indicators of salinity gradients: a case study from a coastal lake

The diatom analysis was conducted according to the method of Battarbee (1986). Samples were treated with 10% HCl to remove calcium carbonate and washed several times in distilled water. Afterward, the samples were boiled in 30% H₂O₂ in order to oxidize the organic matter. Permanent microscope slides were mounted in Naphrax. The identification of diatoms was performed using a Nikon Eclipse E-200 light microscope at a magnification of ×1000. At least 500 frustules of diatoms were counted on each slide and the relative abundance of each taxon in a sample was calculated. The diatoms were identified according to Krammer & Lange-Bertalot (1986, 1988, 1991a,b), Denys (1991/2), van Dam et al. (1994), Snoeijs (1993), Snoeijs & Vilbaste (1994), Snoeijs & Potapova (1995), Lange-Bertalot & Metzeltin (1996), Snoeijs & Kasproviciene (1996), Snoeijs & Balashova (1998), Lange-Bertalot (1999), Lange-Bertalot & Genkal (1999), Krammer (2002), Lange-Bertalot et al. (2003), Hoffmann et al. (2011), Lange-Bertalot (2011), and Zgrundo et al. (2013). Data from AlgaeBase (Guiry & Guiry 2016) were used for the more recent diatom nomenclature.

The classification of diatoms, taking into consideration their ecological requirements (salinity, trophic state, pH), was based on the OMNIDIA database (Lecointe et al. 1993).

In terms of lifeforms, diatom taxa were grouped according to Denys’ classification (1991/2), focusing mainly on euplanktonic and tychoplanktonic species (including tychoplanktonic, of epontic origin and benthic origin). Salinity preferences of diatoms were based on the classification by van Dam et. al (1994; after van der Werff & Huls, 1954-1974; Table 1). In terms of the trophic spectrum, diatoms were grouped on the basis of Naumann’s classification (1932), and pH classification – according to Hustedt (1939; see Denys 1991/2).

The redundancy analysis (RDA) was used to determine similarities between the species distributions and sampling sites with respect to physical and chemical water parameters: salinity, temperature, pH, dissolved oxygen (DO), nitrogen (ammonium [NH₄⁺], and nitrate [NO₃^-]), phosphorus (phosphate [PO₄^3-] and total phosphorous [P_tot]) and chloride (Cl^-). Diatoms with relative abundance above 2% were included in the analysis (49 species). RDA analysis was performed for all sites throughout the sampling period using the CANOCO 4.5 software (ter Braak & Smilauer 2003). Furthermore, Monte Carlo permutation tests were run to reduce the number of variables (Lepš & Šmilauer 2003). All variables with a significance level of p>0.05 were removed. Chloride optima and tolerance ranges for diatom species were obtained using the C2 software (Juggins 2007), including the model weighted averaging (WA).

Most sites studied from Lake Resko Przymorskie were rich in nutrients, especially NH₄⁺ (mean 0.3-1.1 mg l^-1) and PO₄^3- (mean 0.05-0.1 mg l^-1), and had high Cl^- concentration (mean 1401-3160 mg l^-1). High Cl^- concentration is a significant habitat feature of this coastal lake. During the sampling period, water temperature varied from 3.2°C to 24.4°C (average 16.5°C). Table 2 presents the summary of hydrochemical features of all sites.

Table 2

Average, maximum and minimum values of water quality variables recorded during sampling events

Individual study sites in Lake Resko Przymorskie showed a wide range of nutrient and chloride concentrations. Ammonium (NH₄⁺) occurred throughout the measurement period. Its mean concentration was 0.5 mg l^-1. The lowest ammonium concentration was observed in May 2013 and July 2013, and the highest ammonium concentration was observed in July 2014.

Increased nitrate concentration (NO₃^-) was observed only in samples from May 2013. The average concentration of phosphates (PO₄^3-) was very low and varied from 0.05 mg l^-1 to 0.1 mg l^-1. In July 2013 and May 2014, phosphates were not recorded. The mean concentration of total phosphorous (P_tot) was 58 μg l^-1. The lowest concentration was recorded in May 2013 (RP5), and the highest in September 2013 (sample RP4). The chloride concentrations during the study period showed considerable temporal variability. The highest values were determined in November 2013 (sample RP4, and sample RP5), and the lowest in May 2013 (sample RP4). The highest salinity of waters in Lake Resko Przymorskie was observed in November 2013 and the lowest in May 2013. The oxygenation of near-surface waters (DO) varied from 6.6 ppm in September 2012 (sample RP5) to 13.9 ppm in November 2013 (sample RP4). Water pH was rather high and ranged from 7.7 in September 2012 (sample RP5) to 9.9 in November 2013 (sample RP4).

A total of 82 diatom taxa were identified in the analyzed samples. Tychoplanktonic (random planktonic) forms dominated (Fig. 2). From September 2012 to July 2014, the contribution of tychoplanktonic species (e.g., Pseudostaurosiropsis geocollegarum, Staurosira construens and Pseudostaurosira brevistriata) was relatively high (from 72.6% to 99.0% average abundance). The abundance of typical planktonic (euplanktonic) species was rather low, from 0.8% to 24.1% average abundance in all samples. An increased contribution (mainly Cyclotella atomus and Cyclotella meneghiniana) was observed in July 2013 (Fig. 3).

Figure 2

Figure 3

The diatom assemblages were dominated by species with brackish, brackish-fresh and fresh-brackish salinity preferences. In September 2012, May 2013, and September 2013, brackish diatoms dominated. Their average contribution amounted to over 55%. In subsequent measurement seasons, the contribution of brackish taxa amounted to 34.1% average abundance in July 2013, 38.0% in November 2013, and more than 40.0% in May and July 2014. The dominant brackish species was Pseudostaurosiropsis geocollegarum. Brackish-fresh diatoms were most abundant in July 2013 (26.8% average abundance) and May 2014 (23.2% average abundance). They were the uncommon in September 2012 (3.1%) and May 2013 (5.1%). Brackish-fresh species in Lake Resko Przymorskie included: Cyclotella atomus, Cyclotella meneghiniana, Pseudostaurosira subsalina and Diatoma tenuis. The highest number of fresh-brackish taxa occurred in November 2013 (53.2% average abundance). In the remaining research seasons, their contribution amounted to more than 32.0%. Fresh-brackish species included: Amphora pediculus, Fragilaria sopotensis, Hippodonta hungarica, Pseudostaurosira brevistriata and Staurosira construens. Freshwater diatoms accounted for less than 1% of the assemblge.

The pH spectrum was dominated by alkalibiontic (from 37.7% to 60.6% average abundance) and alkaliphilous (from 37.8% to 60.4% average abundance) taxa. Diatoms with neutral water pH preferences were considerably less abundant (0.0-1.5% average abundance).

Among all taxa with the determined trophic preferences, eutraphentic taxa were the most abundant group (from 45.3% to 69.0% average abundance). Eutraphentic taxa were represented by Pseudostaurosira brevistriata and Staurosira construens.

The results of RDA analysis reflecting correlations between the frequencies of predominant diatoms and physicochemical water parameters are shown in Fig. 4. In the Monte Carlo test, 4 out of 10 environmental parameters appeared significant (Table 3), including: water temperature, pH, PO₄^3- and Cl^- concentrations. These 4 parameters together accounted for 23% of the variance in the diatom diversity. Temperature is particularly positively correlated with the first RDA axis, and chloride concentrations with the second axis.

Figure 4

Table 3

Results of the forward selection of environmental parameters (Monte Carlo permutation test in RDA, p<0.05 are statistically significant and given in bold)

RDA analysis revealed environmental requirements of different ecological groups. Figure 4 illustrates that the occurrence of Pseudostaurosiropsis geocollegarum (the most abundant species), Navicula cincta and Cocconeis placentula var. lineata was correlated with higher salinity (Sal), while Navicula vulpina and Cyclotella comensis were correlated with high chloride content. Caloneis amphisbaena showed a high correlation with NH₄⁺, and relatively low pH values. The occurrence of Navicula tripunctata, Navicula radiosa, Cyclotella meneghiniana and Stephanodiscus medius was related to the abundance of phosphorus and well-oxygenated waters (P_tot, DO). The occurrence of Amphora ovalis, Cavinula scuttelloides, Cyclotella atomus, Belonastrum berolinensis, Fragilaria nanana, Fragilaria sopotensis, Martyana martyi and Stephanodiscus parvus was strongly determined by high water temperature. Staurosirella pinnata, Amphora libyca, Amphora inariensis, Aulacoseira granulata and Ellerbeckia arenaria showed a strong correlation with pH. The occurrence of Fragilaria inflata var. istvanffyi, Rhoicosphenia abbreviata, Actinocyclus normanii, Amphora pediculus, Navicula rhynchocephala, Diatoma tenuis, Staurosira construens, Pseudostaurosira brevistriata, Pseudostaurosira subsalina, Placoneis placentula, Opephora olsenii and Hippodonta hungarica was related to high concentrations of NO^3- and a phosphate-rich environment.

The strongest correlation with salinity (Cl^-, salinity) occurred for samples from September 2012 (RP1, RP2, RP3, RP5) and September 2013 (RP1, RP2, RP4) (Fig. 5), but sample RP5 from September 2013 was correlated with high values of total phosphorus (P_tot). Summer samples from July 2013 (RP1, RP2, RP3, RP5) and July 2014 (RP4, RP5) showed the strongest correlations with temperature, although in July 2014, pH also had a considerable impact (in addition to temperature). The strongest correlations with phosphates (PO₄^3-) and nitrate ions (NO₃^-) were found in September 2012 (sample RP4), May 2013 (samples RP1, RP2, and RP4), November 2013 (samples RP4 and RP5) and May 2014 (sample RP5).

Figure 5

We determined salinity preferences (Cl^- concentration) for 49 diatom species. The optimum chloride concentration for the identified diatoms ranged from 1471 mg Cl^-1^-1 to 2961 mg Cl^-1^-1 (Table 4). Thirty-nine taxa showed an optimum chloride concentration above 2000 mg l^-1. Four of the species (Epithemia turgida, Staurosira binodis, Encymoena silesiacum and Amphora marina) showed particularly high optima, exceeding 2600 mg Cl^-1^-1. Moreover, E. turgida and S. binodis were distinguished by narrow tolerance ranges (E. turgida had an optimum of 2961 mg Cl^-1^-1 with a tolerance of 385 mg Cl^- l^-1, while S. binodis had an optimum of 2749 mg Cl^- l^-1 with a tolerance of 324 mg Cl^- l^-1). Particularly high tolerance to chloride concentration was revealed for the following species: A. marina (1159 mg Cl^- l^-1), Fragilaria capucina var. mesolepta (1159 mg Cl^- l^-1), Planothidium lanceolatum (1015 mg Cl^- l^-1) and Placoneis placentula (964 mg Cl^- l^-1). Species with optima below 2000 mg Cl l^-1 belonged to 10 taxa. Parlibellus crucicula and Planothidium delicatulum were found to be indicator species for very low salinity (below 1600 mg Cl^- l^-1) within the range of Lake Resko Przymorskie (Table 4).

Table 4

Optimum (weighted average) and tolerance (SD) of particular species compared to Cl^- gradient

There are many studies devoted to the general ecology of diatoms in salt waters in Poland (e.g. Andrén 1999; Bogaczewicz-Adamczak et al. 2001; Bogaczewicz-Adamczak & Dziengo 2003), however, in-depth research into relationships between populations of diatoms and various environmental variables is scarce (Zgrundo & Bogaczewicz-Adamczak 2004; Dziengo-Czaja et al. 2008). Diatoms found in Polish coastal lakes affected by saline waters are an important component of phytoplankton, though their ecology has not yet been carefully studied. What is more, relatively little is known about seasonal changes in diatom composition in coastal lakes. The influence of saline waters and salinity gradients are of crucial importance to the functioning of these ecosystems. Environmental reconstructions based on diatoms may be useful for constructing salinity variability models, for example diatom-based salinity models from lakes in British Columbia (Cumming & Smol 1994) or the Thames Estuary (Juggins 1992). The salinity of Lake Resko Przymorskie is very diverse. The lake has the highest chloride levels compared to other Polish coastal lakes. Cieslinski (2011) showed that chloride concentration levels are on average above 1500 mg Cl^- l^-1, and the highest levels were determined in the western part of the lake, i.e. over 2200 mg Cl^- l^-1. Our study demonstrated that the chloride concentrations in the lake may be even higher. In November 2013, we measured concentrations ranging from 3090 mg Cl^- l^-1 to 3231 mg Cl^- l^-1. Such high concentrations of Cl^-unequivocally indicate that the lake is strongly saline.

According to the literature (Andrén 1999; Clarke et al. 2003; Weckström et al. 2002; Weckström et al. 2004; Kauppila et al. 2005; Weckström & Juggins 2005; and Weckström 2006), the distribution patterns of diatoms in the coastal area are also determined by nutrients concentration. The Gulf of Finland shows the effect of total nitrogen and total phosphorus on the structure of diatom assemblages dominated by small planktonic taxa (Weckström & Juggins 2005). In this study, Rhoicosphenia curvata, Staurosirella pinnata, Opephora mutabilis, Navicula phyllepta, Bacillaria paxillifer and Pauliella taeniata preferred lower nitrogen concentrations. Small centric planktonic taxa, such as Cyclotella atomus, Cyclotella meneghiniana and Thalassiosira pseudonana, were identified as indicators of highly elevated nutrient concentrations. These taxa could be used as potential indicators of the quality of coastal waters in the Gulf of Finland (Weckström & Juggins 2005; Weckström 2006). Hydrochemical and diatom analyses of Roskilde Fjord in Denmark and Laajalahti Bay in Finland demonstrated changes in nitrogen concentration and diatom abundance (mainly Cyclotella choctawhatcheeana) related to, among other factors, human activity (Kauppila et al. 2005; Clarke et al. 2006).

The nutrient status (especially NH₄⁺ and PO₄^3-) and heterogeneity of the habitat in this coastal lake is very specific. More than half of the studied sites have high values of these parameters. This is due to both the anthropogenic pollution of the lake and the impact and transformation of organic deposits into biogenic substances. Despite the high salinity of the lake, density stratification may not occur, because waters are completely mixed in shallow-water conditions. Findings of the research on coastal lakes (S⊘ndergaard et al. 2001; Abesser & Robinson 2010; Cieśliński 2011; Woszczyk et al. 2014) indicate the possibility of mixing of the surface bottom sediment layer. This is reflected in the high trophic status of the lake. One indication of this phenomenon observed during the study was the significant proportion of tychoplankton species in the pelagic zone of the lake.

We have shown that the species composition of diatoms is a sensitive indicator of salinity in Lake Resko Przymorskie, which suggests that diatom species data may be applied in paleoenvironmental studies. Findings related to a coastal lake have been presented for the first time, and our research involves the analysis of the diatoms’ response to salinity gradients and the determination of chloride concentration preferences of selected diatom taxa.

The dominant diatom species was Pseudostaurosiropsis geocollegarum (31.7-69.4% of the entire assemblage). It is typically a marine species (Morales 2002). According to the salinity classification by van Dam et al. (1984), P. geocollegarum can be included in a group of brackish diatoms, tolerating salinity levels of 1.8-9.0 PSU. The species occurs, however, in waters with salinity not higher than 7-8 PSU (Witkowski et al. 2000; Seddon et al. 2011), as confirmed by our results. Our research showed the optimum chloride level of more than 2300 mg Cl^- l^-1.

Epithemia turgida had the highest chloride optimum. According to the observations by Hofmann et al (2011), it is a common species in waters with a considerable concentration of electrolytes, and has been recorded not only in the brackish waters of Lake Resko Przymorskie but throughout the entire Baltic Sea (Hällfors 2004), with decreasing abundance or even absence toward the south (Snoeijs & Potapowa 1995), but identified in the Gulf of Gdańsk by Witkowski (1994), and in the brackish sediments of Kooser Wiesen, Körkwitz by Lampe & Janke (2002).

Staurosira binodis is a species distinguished by high vitality in a wide range of trophic states and electrolyte levels (Hoffmann et al. 2011). Moreover, S. binodis is also reported as a freshwater species that does not tolerate the full salinity gradient of the area from the Southern Baltic Proper, the Gulf of Gdańsk, the Gulf of Riga, and the Gulf of Finland (Hällfors 2004; Zgrundo et al. 2009). Staurosira binodis was also found in sediments of coastal lakes and their vicinity (Lutyńska 2008a; Mazurek et al. 2008; Witkowski et al. 2012; Sydor et al. 2015). Our studies have contributed to the knowledge about the ecology of this species.

Particularly high tolerance to chloride concentrations in the species structure of a given coastal lake was found for such species as: Amphora marina, Fragilaria capucina var. mesolepta, Placoneis placentula and Planothidium lanceolatum. Besides its presence in Lake Resko Przymorskie, A. marina is widespread in the marine littoral zone (Witkowski et al. 2000). Fragilaria capucina var. mesolepta is a variety occurring in clean, slightly eutrophic waters (Krammer & Lange-Bertalot 1991a; Arrhenius et al. 2014) and was reported from the Gulf of Riga (Hällfors 2004). Planothidium lanceolatum is a species reported from the Southern Baltic, the Gulf of Riga, the Gulf of Finland (Hällfors 2004) and the Gulf of Gdańsk (Witkowski 1994).

Indicator species of the lowest salinity (below 1600 mg Cl^- l^-1) within the range of Lake Resko Przymorskie included Parlibellus crucicula and Planothidium delicatulum. Parlibellus crucicula occurs on marine coasts, in river estuaries, and in saline coastal marshes (Krammer & Lange-Bertalot 1997). It mostly occurs in the littoral zone and is reported from the Gulf of Riga and the Gulf of Finland (Hällfors 2004) and the Gulf of Gdańsk (Witkowski 1994; Zgrundo et al. 2009). Planothidium delicatulum is a species occurring in brackish inland waters and on the seashores (Witkowski et al. 2004; Krammer & Lange-Bertalot 1991b). It is reported from the Southern Baltic, the Gulf of Riga, the Gulf of Finland and the Gulf of Bothnia (Hällfors 2004). Moreover, it is abundant in the littoral and sub-littoral zone of the Gulf of Gdańsk (Witkowski 1994).

Turbid nutrient-rich areas of Lake Resko Przymorskie were dominated by tychoplanktonic species, classified as tychoplanktonic of epontic origin and benthic origin. In July 2013, an increased contribution of planktonic species was observed, including mainly Cyclotella meneghiniana and Cyclotella atomus, which were also observed in high-nutrient waters of Chesapeake Bay (Cooper 1995), Florida Bay (Wachnicka et al. 2010) and the Baltic Sea (Andrén et al. 2000; Weckström & Juggins 2005; Zgrundo et al. 2009). Our research showed an increase in the abundance of Cyclotella atomus in July 2013 samples, which correlates with temperature, whereas Cyclotella meneghiniana occurs in samples with higher concentrations of phosphorous. Small planktonic taxa such as Cyclotella atomus and Cyclotella meneghiniana may be useful indicators of highly elevated concentrations of nutrients (Weckström & Juggins 2005).