Benthic diatoms of the Vrla River (Serbia) and their application in the water ecological status assessment

Statistical analyses were performed using CANOCO for Windows Version 5 (Ter Braak & Šmilauer 2012). Canonical correspondence analysis CCA was used to visualize the relationships between the identified diatom taxa and the measured physicochemical data (explanatory environmental variables) with sampling sites and sampling seasons included as supplementary variables. The presence-absence data were used as a measure, so a unimodal method was applied. Two hundred twenty six diatom taxa were included in the analysis, but due to the high number of taxa, only 47 of the best fitting taxa (which percentage of variation is best explained by the ordination space) were shown on the graphs. The redundancy analysis (RDA) was performed to examine the potential effects of the measured environmental variables on the dominant diatom taxa with sampling sites and sampling seasons used as supplementary variables. RDA was used since the gradient was 1.6 SD units long. The dominant taxa were represented as a percentage calculated based on the valve percentage representation of each taxon relative to the total of 400 valves on each permanent slide. RDA with the option center and standardize was used and response data (diatom taxa) were log transformed. The statistical significance of each variable in both analyses (for the level of 0.01) was assessed using the Monte Carlo unrestricted permutation test involving 4999 permutations.

A taxonomically diverse diatom flora was identified in 36 analyzed samples, including 227 diatom taxa belonging to 50 genera. Gomphonema (30 species), Navicula (28) and Nitzschia (26) were the most species-rich genera, followed by Pinnularia (12) and Encyonema (11). A considerable number of genera (18 or 36%) were represented by only one species. Achnanthidium minutissimum (Kützing) Czarnecki, Achnanthidium subatomus (Hustedt) Lange-Bertalot, Amphora pediculus (Kützing) Grunow, Geissleria acceptata (Hustedt) Lange-Bertalot & Metzeltin, Gomphonema elegantissimum Reichardt & Lange-Bertalot, Navicula antonii Lange-Bertalot, Nitzschia abbreviata Hustedt, Planothidium lanceolatum (Brébisson) Bukhtiyarova and Reimeria sinuata (Gregory) Kociolek & Stoermer occur in all seasons and in all samples. Amphora inariensis Krammer, Cocconeis placentula var. lineata (Ehrenberg) van Heurck, Diatoma mesodon (Ehrenberg) Kützing, Hannaea arcus (Ehrenberg) Patrick and Navicula gregaria Donkin were absent in only one sample. Thirteen taxa were defined as dominant. Their percentage contribution was 10% or more at least at one site (Table 1).

A taxon new to Serbia was found in the analyzed material – Geissleria acceptata (Hustedt) Lange-Bertalot & Metzeltin. Valves are elliptical to linear-elliptical; length 6.34-12.41 μm, breadth 3.89-4.64 μm, striae 16-18 per 10 μm which corresponds with literature data (Lange-Bertalot, 2001) (Fig. 2). It was found in all samples, in abundance ranging from 0.3 to 4.5% in the Vrla River. The pH values in the analyzed material varied from 7 to 8.8, the conductivity – from 58 to 115 μS cm^-1 and the water hardness – from 0.12 to 12.24 °dH.

Figure 2

The values of physicochemical parameters of the Vrla River are presented in Table 2. The water temperature during the sampling period ranged from 2 to 17.3°C, pH ranged from 7 to 8.8 and conductivity from 50 to 110 μS cm^-1. Water hardness varied from 0.12 to 12.24, which means that water can be described as soft, moderately hard, hard to very hard depending on the season.

Canonical Correspondence analysis (CCA) was used to evaluate the relationship between the examined environmental parameters and diatom taxa recorded in the Vrla River (Fig. 3).

Figure 3

CCA showed high statistical significance (F = 1.4, P = 0.0002). The first CCA axis shows the highest positive correlation with oxygen (r = 0.4399), and the second CCA axis showed a positive correlation with temperature (r = 0.6015). Conductivity showed a positive correlation with both CCA axes, with slightly higher correlation with the second axis (first axis: r = 0.4462, second axis: r = 0.5159). Alkalinity and water hardness were positively correlated with the third CCA axis (not shown on the ordination diagram). The temperature was dependent on the season and the lowest value was recorded in S5 and the highest one in S3 (at VR6). In general, considering average values of temperature at the sampling sites during all seasons, VR3 had the lowest, while VR6 the highest average temperature. Oxygen showed a reverse pattern and was negatively correlated with temperature (the angle between the vectors of these two variables is nearly 180 degrees indicating almost perfect negative correlation). Conductivity was low during S6 (lowest at VR1) and the highest in S3 and S5. Sampling site VR6 showed the highest values of conductivity in all seasons (70-110 μS cm^-1). High values of alkalinity and very high values of water hardness (WH) compared to other seasons were also recorded during S1 at sampling sites VR5 and VR6. Orthophosphates (OP) and NO₃ had the highest values recorded at VR3 in S1 and S2, respectively.

CCA divided the diatom taxa into three main groups. The first group is composed of diatom taxa (for example Planothidium frequentissimum (Lange-Bertalot) Lange-Bertalot, Sellaphora radiosa (Hustedt) H. Kobayasi, Fragilaria recapitellata Lange-Bertalot & Metzeltin, Psammothidium daonense (Lange-Bertalot) Lange-Bertalot) found at most of the sampling sites (in the center of the ordination diagram) (Fig. 3). The second group (the right side of the ordination diagram) is represented by taxa found only at sampling site VR6 in S5 (for example Amphora copulata (Kützing) Schoeman & Archibald, Gomphonema acuminatum Ehrenberg, Gyrosigma sciotoense (Sullivant) Cleve, Nitzschia recta Hantzsch ex Rabenhorst, Staurosirella pinnata var. intercedens (Grunow) Hamilton, Surirella brebissonii var. kuetzingii Krammer & Lange-Bertalot). These taxa apparently tend to occur at the higher values of oxygen and conductivity. The third group of diatom taxa is placed in the upper part of the ordination diagram and it is positively correlated with temperature and conductivity. These taxa, including Cymatopleura solea var. apiculata (W. Smith) Ralfs, Gomphonema gracile Ehrenberg, Navicula amphiceropsis Lange-Bertalot & Rumrich, Cymbopleura cuspidata (Kützing) Krammer, Navicula novaesiberica Lange-Bertalot are found only at sampling site VR6 in S3. Sampling site VR6 was characterized by high values of conductivity in all seasons (70-110 μS cm^-1). Apparently, some diatom taxa such as Navicula cryptotenella Lange-Bertalot, Cymbella excisa Kützing, Nitzschia acicularis (Kützing) W. Smith show a high correlation with conductivity, and they occurred in different seasons at sampling site VR6 only.

Potential effects of the measured environmental variables on the dominant diatom taxa were investigated using RDA (Fig. 4).

Figure 4

The first RDA axis is positively correlated with temperature (r = 5.050), and negatively correlated mainly with total phosphorus (r = -0.5651), alkalinity (r = -0.5106) and water hardness (r = -0.5776). The second RDA axis is negatively correlated with pH (r = -0.6019) and conductivity (r = -0.6497). The other measured environmental variables were weakly correlated (positively or negatively) with both RDA axes. Cocconeis placentula var. lineata, Cocconeis placentula var. placentula Ehrenberg and Cocconeis pseudolineata (Geitler) Lange-Bertalot (all abundant in S3), Mayamaea atomus var. permitis (Hustedt) Lange-Bertalot (abundant in S2 and S6), as well as Nitzschia abbreviata and Nitzschia archibaldii Lange-Bertalot are positively correlated with temperature and with the first RDA axis. All above-mentioned diatom taxa, except Mayamaea atomus var. permitis were found in higher percentage at sampling site VR6, so they are also correlated with conductivity. Hannaea arcus and Nitzschia pura Hustedt were the most abundant in S1 and S5 and most positively correlated with pH and variables that show a negative correlation with the first RDA axis (alkalinity, water hardness and total phosphorus). Achnanthidium minutissimum, Achnanthidium subatomus, Gomphonema elegantissimum, Gomphonema pumilum var. rigidum Reichardt & Lange-Bertalot were also correlated with TP, WH and Alc (and also with OP), but together with Planothidium lanceolatum were most abundant at VR2 where very low conductivity was determined, so they all showed a negative correlation with conductivity. Roughly speaking, RDA divided the taxa into two groups: those found on the left side of the ordination diagram correlated mainly with TP, WH and Alc (some of them with OP and pH) and those placed at the right side of the ordination diagram correlated mainly with temperature.

Values of the six diatom indices indicated a similar ecological status of the water at the studied sites (Table 3). The correlation between the selected indices is shown in Table 4. The highest correlation is between CEE, IPS and IBD (Table 4). These three indices had the highest values at sampling sites VR1 and VR2, while the lowest at VR5 or VR6 (Table 3). DESCY was slightly different and its maximum value was calculated at sampling sites VR6 and VR5, and the lowest at VR2 (Table 3). Values of the CEE, IPS and DESCY indices clearly indicated good (class II) ecological status (Table 3) of the Vrla River at all the studied sites. Values of the IBD index for sampling sites VR1 and VR2 indicated high (class I) ecological status (Table 3). Also, chemical analysis of pH, ammonia, OP and TP indicated good and high ecological status of the Vrla River (Table 2). The highest value of the diatom index SLA was determined at sampling site VR1, and the lowest one at VR3 and VR4 (Table 3). Values of the SLA index for sampling sites VR2, VR3, VR4 and VR6 indicated a moderate ecological status or class III (Table 3), although these values were close to values indicating a good ecological status. The highest value of TDI was at sampling site VR3 (Table 3). Based on the TDI index, the Vrla River was characterized by moderate concentration of nutrients (class III) (Table 3). Table 3 clearly shows the ecological status of the Vrla River which ranges from moderate, good to high. Also, the percentage contribution of species characteristic of organic pollution (%PT) was calculated. This is related to TDI. The 20% contribution of these species indicates organic pollution and very high values of PT (over 40%) indicate the risk of eutrophication (Kelly et al. 1995). %PT was over 20% at sampling sites VR3, VR4 and VR5 in season S1, at VR3 in season S3 and at VR2 in season S5.

Great diatom species richness was documented during the floristic survey of the Vrla River. Species diversity determined with the use of the software Omnidia was in the range of 3.06 to 4.77. Clean rivers with a low level of pollution, such as the Linda River (Szczepocka et al. 2014) and the Pilica River (Szulc 2007), are also characterized by high species diversity. A similar species richness was observed in rivers located (like the Vrla River) in the southeast of Serbia: the Nišava, the Jerma and the Temska River (Andrejić et al. 2012). Furthermore, the most species-rich genera were the same as in our study. Species of the genera Navicula and Nitzschia occur frequently in Central Europe (Noga et al. 2013a) and Euro-Asia (Solak et al. 2012). Similar diatom communities were recorded in other rivers of Serbia but also in some rivers in other parts of Europe and Euro-Asia (Nikitović 1998; Tomašević 2000; Jurišić 2004; Torrisi & Dell'Uomo 2006; Andrejić 2012; Solak et al. 2012; Várbíró et al. 2012; Noga et al. 2013b).

Achnanthidium minutissimum has a much wider range of tolerance to various environmental factors as compared to other Achnanthidium species. Teratological forms of Achnanthidium minutissimum, Eunotia exigua (Brébisson ex Kützing) Rabenhorst, Fragilaria rumpens (Kützing) Carlson, F. cf. crotonensis Kitton were found in streams polluted by acid mine drainage, which is related to high levels of metals (Silva et al. 2009; Luis et al. 2015). A. minutissimum is one of the most metal-tolerant species (Cantonati et al. 2014).

Achnanthidium subatomus achieves its optimum in oligotrophic, calcareous waters with moderate to very low content of electrolytes (Hofmann et al. 2013). Cocconeis placentula var. lineata and C. placentula var. euglypta (Ehrenberg) Grunow have a wider ecological range than C. pediculus Ehrenberg, especially with regard to the electrolyte content and the trophic situation. These species are found even in streams with a lower content of electrolytes, dominated by silicates (Hofmann et al. 2013).

Gomphonema elegantissimum is widespread in karst regions in the Balkan Peninsula and fairly common in other limestone areas of Central Europe (Reichardt 1997).

Geissleria acceptata is a new taxon in the diatom flora of Serbia. It is the only Geissleria species that can have or lack an isolated pore (Novais et al. 2013). It is probably a cosmopolitan species, found in meso- to eutrophic waters, but also in oligosaprobic waters with a moderate electrolyte content (Lange-Bertalot 2001), same as waters of the Vrla River. The species occurs mostly in rivers and streams of a silicate mountain range, but it was also frequently found in alkaline lakes of the lowlands (Hofmann et al. 2013). G. acceptata is generally considered to be a representative of good ecological status (Lek et al. 2005). Based on the results obtained in this study, the ecological status of the Vrla River is assessed as good, so our research confirms the occurrence of Geissleria acceptata in specific conditions.

Seasonal and spatial variations in water temperature affect the distribution of benthic diatoms in rivers. In general, the diversity of taxa increases in temperatures of 25-30°C and drops in temperatures above 30°C. In nature, however, seasonal and spatial dynamics of diatoms does not depend only on water temperature (Stevenson et al. 1996). In the Vrla River, Achnanthidium minutissimum was the most numerous at sites VR5 and VR6 in S1 and at VR5 in S4, where the measured temperature was <14°C. In our study, RDA showed that some species of the genus Cocconeis (placed on the right side of the ordination diagram) were positively correlated with temperature (C. placentula var. lineata, C. placentula var. placentula and C. pseudolineata), while diatom taxa on the left side show correlation with many environmental variables, including those related to nutrients. The trophic level of water significantly influences the structure of benthic diatom communities. The significance of different nitrogen compounds is demonstrated by multivariate analyses (Vilbaste & Truu 2003), but in our case, the nitrogen compound (ammonia) was weakly correlated with the two RDA axes, and its arrow was short, meaning that it has a lesser effect on the dominant diatom taxa presented on the ordination diagram. Regarding phosphorus, OP and TP were more significant (especially TP) and correlated with many dominant diatom taxa. According to Pringle (1990), an addition of inorganic and organic phosphorus resulted in an increase in the number of motile diatoms, mainly Nitzschia and Navicula species on sandy substrates and C. placentula var. placentula and A. minutissimum on glass slides. Our study also showed that Nitzschia pura and A. minutissimum are positively correlated with the concentration of total phosphorus and orthophosphates and they can be found at places with a higher level of phosphates. Research on the effects of nutrient enrichment on stream algae suggested that diatom species richness declines (Peterson et al. 1985) and algal biomass increases (Bothwell 1989) along with increasing phosphorus. According to Chételator et al. (1999), diatoms reach the maximum growth rate at the total phosphorus concentration of 0.5 mg l^-1. Even though there have been already many studies focusing on nutrients conducted worldwide, Bellinger et al. (2006) concluded that more controlled studies (experiments) are needed to better understand how nutrients (in addition to other factors) affect algal species.

The first group of diatom taxa on the CCA ordination diagram involves diatom taxa that occurred at most of the sampling sites in more than one season and shows correlation to one or more different variables. More than half of the recorded taxa were grouped in the center of the ordination diagram, but we used 47 best fitted taxa in CCA analysis. A similar composition of communities at most of the sampling sites was characteristic of this river in general. The second group showed correlation with the higher oxygen levels. According to Van Dam et al. (1994), all above-mentioned diatom taxa (Amphora copulata, Gomphonema acuminatum, Gyrosigma sciotoense, Nitzschia recta, Staurosirella pinnata var. intercedens, Surirella brebissoni var. kuetzingii) prefer water with a moderate to high oxygen level. Diatom taxa from the third group showed correlation with temperature and conductivity. For example, Navicula novaesiberica lives in a wide range of temperatures (Noga & Siry 2010). Some diatoms from this group prefer water with a lower oxygen level, as reported for Cymatopleura solea var. apiculata (Van Dam et al. 1994). Navicula cryptotenella is a good indicator of β-mesosaprobic or even better water quality and does not occur in waters with very low or very high electrolyte concentrations. This species is sensitive to pollution (Lange-Bertalot 2001), similarly to Nitzschia acicularis (Krammer & Lange-Bertalot 1988). Cymbella excisa occurs mostly in mesotrophic to slightly eutrophic waters with a moderate electrolyte content (Hofmann et al. 2013).

Cocconeis placentula was reported by Van Dam et al. (1994) as a species occurring in environment with a higher trophic level (eutrophic taxon), higher pH and moderate nitrogen content. Belal (2012) reported that C. placentula had a high optimum of total phosphorus (511.5 mg l^-1) in the River Nile. Values of total phosphorus in our localities were significantly lower. Nitzschia archibaldii has ecological optimum in oligo- to mesosaprobic waters with a medium electrolyte content, the same as Nitzschia abbreviata (Hofmann et al. 2013). According to RDA results obtained in our study, N. archibaldii preferred sampling sites with higher levels of conductivity. A. minutissimum was absent or occurred with low abundance in rivers polluted by nutrients in Appalachia. It was shown that higher abundance of A. minutissimum was mostly associated with low nutrient and ionic content (Ponader & Potapova 2007). This species is classified as a species with preferences for higher trophic levels (7 level, ammonium-nitrogen content) and moderate pH (3 level) (Van Dam et al. 1994). Hofmann et al. (2013) defined Hanneae arcus and Nitzschia pura as oligosaprobic species. In our study, H. arcus was positively correlated with pH, alkalinity, water hardness, total phosphorus and BOD5. Achnanthidium subatomus was mostly found in nutrient-poor rivers within a narrow pH range from 8 to 8.3, but over a wide range of conductivity levels, with a preference for higher values. Nutrients and pH were major factors affecting the distribution of this species, although the measured chemical characteristics were poor predictors of its relative abundance (Ponader & Potapova 2007).

Values of the diatom indices show that there are no major variations in the ecological status of water at the studied sites. The obtained values of the CEE, IPS and IBD indices for sampling sites VR1 and VR2 compared to VR3, VR4 and VR5 could be explained by the fact that VR3 is located 50 m below the discharge pipe of the trout pond through which the water is discharged directly into the recipient (i.e. the river). Furthermore, a high correlation was determined between IPS and IBD based on the research conducted in the Pilica River (Poland) (Szulc & Szulc 2013). Our study showed that the organic pollution indices, i.e. SLA and IPS, indicate a better ecological status as compared to the trophic index TDI. Similar results were obtained during the survey of the Matysówka (Noga et al. 2013a) and Baryczka stream (Noga et al. 2013b) in Poland. The highest TDI value was recorded at sampling site VR3 (near the pond) and the lowest one at VR5. In accordance with this, we obtained the expected results for %PT. There is a possibility of organic pollution at sampling sites VR3, VR4 and VR5. This is not surprising because these sites are located downstream of the trout fish pond.

The study supports the use of diatom indices, especially IPS, DESCY and CEE, to monitor rivers in Serbia. Values of IPS, DESCY and CEE indices were most similar indicating a good ecological status of the Vrla River which is confirmed by chemical data. Altogether 104 diatom taxa (45.81°% of the total number of species found in the Vrla River) were used in the calculation of the 17 indices using the Omnidia software. Differences in the values of particular indices were not significant. Statistical analyzes show that the biological methods are the best tool for the water ecological status assessment. They accurately show the condition of the ecosystem, contrary to physical and chemical parameters which can indicate the current state but not necessarily the true one. The ecological status assessment of rivers can be used for many purposes (e.g. warning and sustainable development of the region) (Ivanković et al. 2011). The applicability of diatom-based indices depends on the local characteristics of the river and the type of substrate (Kwandrans et al. 1998) but also the similarity in species composition in the study area and the taxa used in each index.