Analysis of changes over 44 years in the phytoplankton of Lake Võrtsjärv (Estonia): the effect of nutrients, climate and the investigator on phytoplankton-based water quality indices

Hydrobiologia (2010) 646:33–48 DOI 10.1007/s10750-010-0178-y SHALLOW LAKES Analysis of changes over 44 years in the phytoplankton of Lake Võrtsjärv (Estonia): the effect of nutrients, climate and the investigator on phytoplankton-based water quality indices Peeter Nõges • Ute Mischke • Reet Laugaste Angelo G. Solimini • Published online: 7 March 2010 Ó Springer Science+Business Media B.V. 2010 Abstract We analysed long-term changes in phytoplankton composition in relation to hydrological, meteorological and nutrient loading data in the large (270 km2) shallow (mean depth 2.8 m) Lake Võrtsjärv. Nutrient loads to the lake were heavy in the 1970s and 1980s and decreased considerably thereafter. The average nutrient concentrations for U. Mischke Department of Shallow Lakes and Lowland Rivers, Leibniz-Institute for Freshwater Ecology and Inland Fisheries, Müggelseedamm 310, 12587 Berlin, Germany 1985–2004 (1.6 mg l-1 of total nitrogen and 53 lg l-1 of total phosphorus) characterize the lake as a eutrophic water body. All four calculated taxonomic indices showed a unidirectional deterioration of the lake’s ecological status, despite reduced concentrations of nutrients. We focused our analysis on the PTSI index, which revealed a stepwise change between the years 1977 and 1979 that coincided with a large increase in water level, but also with a change of investigator. After correcting input data for possible investigator-induced differences, the step change remained because it was caused by major changes in the whole phytoplankton community. The previous dominant Planktolyngbya limnetica was replaced by two species of seasonally altering Limnothrix. Among phytoplankton functional groups, there was a decrease in all groups comprising smallsized phytoplankton species, such as X1, E, F, J, N and an increase in S1 and H1, both represented by filamentous cyanobacteria. Our results suggest a non-linear response of phytoplankton to changing nutrient loadings, and that the change observed between 1977 and 1979 was a regime shift triggered by water level change. High shade tolerance of the new dominants, and their ability to create shade, obviously stabilized the new status making it resistant to restoration efforts. A. G. Solimini Department of Experimental Medicine, Sapienza University of Rome, Regina Elena 324, 00161 Rome, Italy Keywords Regime shift
Lake water level
Nutrient loading
Trophic index
Phytoplankton functional groups Guest editors: M. Meerhoff, M. Beklioglu, R. Burks, F. Garcı́aRodrı́guez, N. Mazzeo & B. Moss / Structure and Function of World Shallow Lakes: Proceedings from the 6th Shallow Lakes Congress, held in Punta del Este, Uruguay, 23–28 November, 2008 P. Nõges (&)
R. Laugaste Centre for Limnology, Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Rannu, 61117 Tartumaa, Estonia e-mail: peeter.noges@emu.ee P. Nõges
A. G. Solimini European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via Enrico Fermi, 2749, 21027 Ispra, Italy 123 34 Hydrobiologia (2010) 646:33–48 Introduction Phytoplankton is widely used as an important water quality indicator because of its high species richness and sensitivity to environmental factors. Murphy et al. (2002) list the following main advantages of using phytoplankton in lake monitoring: 1. 2. 3. 4. 5. 6. As primary producers, algae are directly affected by physical and chemical factors. Changes in phytoplankton community status have direct implications for the bio-integrity of the lake ecosystem as a whole. Algae generally have high reproduction rates and very short life cycles, making them valuable indicators of short-term (scales of days—weeks) impacts. Phytoplankton provides a good indication of lake trophic state, measurable, for example, as chlorophyll a concentration, and responds quickly and predictably to changes in nutrient status. Laboratory chlorophyll a analysis is quick and cheap. Sampling is easy, inexpensive, and creates minimal impact to resident biota. Algal assemblages are sensitive to some pollutants which may not visibly affect other aquatic assemblages or may only affect other organisms at higher concentrations (e.g. for herbicides see Nyström et al., 1999; Netherland et al., 2009). Changes in community composition can provide finer-scale assessment of changes due to ecological impacts. The main difficulty of using phytoplankton in lake monitoring is the time consuming taxonomic identification and the need for qualified specialists. Some of the first phytoplankton indices (Thunmark, 1945; Nygaard, 1949) attempted to characterize plankton by the number of species present in different groups, regardless of their abundance. The strong influence of rare and occasional species to the quotient values was one of the main weaknesses of this type of index (Geelen, 1969). An alternative approach used, e.g. by Hörnström (1981), Brettum (1989) and Tremel (1996), was to develop trophic indices, based on ecological preferences of taxa weighted by abundance, that give a higher impact to dominant species. Recently, several new or modified indices of the same type have been elaborated by the 123 EU member states to serve the needs of the Water Framework Directive (WFD; Directive, 2000). For example, three phytoplankton trophic indices (PTIs) were elaborated for deep subalpine lakes (Salmaso et al., 2006) and one (MedPTI, Marchetto et al., 2009) for deep Mediterranean reservoirs. Austria uses Brettum’s (1989) index modified by Dokulil et al. (2001, 2005), which is based on trophic scores of a large number of planktonic algae. The German phytoplankton taxa lake index (PTSI, Mischke et al., 2008), which is one part of the national phytoplankton-based assessment system, first classifies the trophic status of lakes (from oligotrophic to hypertrophic) based on phytoplankton species composition, and secondly, in a WFD-compliant assessment, the PTSI is applied by comparing its value with the preset trophic reference value of the appropriate lake type. The trophic plankton indices used in Sweden (Willén, 2007) and Estonia (Maileht, 2008) are also based on trophic preferences of species. A slightly different approach is based on functional groups. Species frequently found to co-exist and to increase or decrease in number simultaneously were delimited and given association identities (Reynolds et al., 2002). Algae forming a functional group mostly have similar morphology, making the shape and dimensions of the algal cells or colonies powerful predictors of performance (Reynolds & Irish, 1997). Padisák et al. (2003, 2006) and Salmaso & Padisák (2007) developed further the functional group approach, which is applied now in the WFD monitoring scheme in Hungary (Padisák et al., 2006). At present, there are 41 functional groups described in freshwater phytoplankton, with more or less precisely defined ecological demands (Padisák et al., 2009). The functional group approach, based on species aggregations, enables isolation of similarities and common trends among different water bodies that are often difficult to show at the species level because of the high individuality in community composition of particular water bodies. However, it does not make concessions to the identification level because a high taxonomic resolution is still needed to correctly designate the species to functional groups. The general disadvantages of the taxonomy-based metrics are that they are time consuming and that the analysis of the community composition requires good identification expertise, which is not necessarily available. In long-term data sets, changes of Hydrobiologia (2010) 646:33–48 investigator contribute to the inhomogeneity and inconsistency of data that until now has been often overlooked. Intercalibration exercises carried out for freshwater phytoplankton analysis (Rott, 1981; Vuorio et al., 2007) have revealed large variability introduced in sample preparation and counting. There is accumulating field and laboratory evidence that phytoplankton reacts rapidly not only to changes in nutrient loadings but also to climate change, which either mimics eutrophication by increasing phytoplankton production and carrying capacity (Mooij et al., 2005) or contributes to it by inducing phosphorus release from sediments and accelerating nutrient cycling (Pettersson et al., 2003). For some phytoplankton groups and species, the climate-induced trends are well documented and can be used for future projections. For instance, climate change has been considered a potential catalyst for the further expansion of harmful cyanobacterial blooms, particularly in eutrophic waters (Paerl & Huisman, 2008, 2009; Jöhnk et al., 2008). Rising temperatures, enhanced stratification, increased residence time and salination in combination with high nutrient loading all favour cyanobacterial dominance. The recent success of Planktothrix rubescens in a number of lakes is most probably caused by a synergistic effect of increased transparency due to the reduction in the phosphorus loads, the deepening of the P-depleted zone and increased water column stability (Teubner et al., 2003, 2006; Anneville et al., 2005; Jacquet et al., 2005). As climate affects phytoplankton through various mechanisms (temperature increase, changes in ice regime, river runoff, water level, mixing depth, etc.), the responses differ regionally and by lake type. We hypothesize that phytoplankton-based assessment systems for water quality are sensitive to changes in hydro-meteorological variables and, hence, prone to be biased by climate change. In this article, based on 44 years of data on phytoplankton and meteorology and 33 years of data on nutrient loading, we test (i) (ii) if phytoplankton inferred water quality reflects adequately changes in nutrient loadings; if hydro-meteorological parameters have a detectable effect on phytoplankton composition (and indices); 35 (iii) to what extent the inconsistencies in taxa identifications by different analysts can affect the index values calculated for long-term data sets. Our aim was not to make a WFD-compliant assessment of the ecological status of Lake Võrtsjärv, as none of the indices tested was adapted to the particular region and lake type, but rather to follow the long-term dynamics of phytoplankton composition in relation to the potential affecting factors. Site description Võrtsjärv is a shallow, highly eutrophic lake in Central Estonia. Its area is 270 km2, mean depth is 2.8 m and maximum depth is 6 m. Eighteen rivers and streams collect their water from a 3104-km2 catchment where agricultural lands comprise 40%. The average water retention time is 1 year. A strong dependence of the water level on precipitation is characteristic of the hydrological regime of Võrtsjärv. Monthly mean water temperature reaches its maximum of 19.8°C in July. The ice cover lasts on average 135 days, from November to April. During 1985–2004, the average concentration of total nitrogen (TN) was 1.6 mg l-1 and that of total phosphorus (TP) was 53 lg l-1 (Nõges et al., 2008). The total nitrogen concentration characterizes Võrtsjärv as a eutrophic waterbody while by total phosphorus it remains close to the meso-eutrophic boundary (OECD, 1982). Despite a considerable decrease in external nutrient loadings, there have been no significant declines in lake concentrations (Nõges et al., 2007), owing to high variability of the concentrations in the turbulent lake environment, and likely also periodic internal loading from the sediment (Nõges & Kisand, 1999). The water is alkaline (pH 7.5–8.5) with a high seston content. During the ice-free period, the mean transparency does not exceed 1 m. Materials and methods Phytoplankton data (sampling site, frequency, analysis methods) We analysed phytoplankton data collected from the pelagic monitoring station in the deepest part (6 m) 123 36 Hydrobiologia (2010) 646:33–48 of the lake mostly at monthly intervals from 1964 to 2007. This series had gaps for single months that were filled with interpolated values, and data for the whole year 1978 were missing. Samples were taken with a Ruttner sampler either from the surface layer (until 1990) or integrated over the water column (since 1991). A comparison of the surface and bottom samples taken in the earlier period did not show any statistically significant differences in total biomass or biomasses of algal classes in this polymictic lake. Hence, we considered the data from surface and integral samples homogeneous and pooled them for the analysis. Before 1994, 2% neutralized formalin was used to fix the samples, which were counted in haemocytometer cells using a normal light microscope. Since 1994, the Utermöhl (1958) technique was used and formalin as fixative was replaced by acid Lugol’s iodine solution. Until the end of 1977, phytoplankton analyses were carried out by Reet Laugaste and continued by her former student, Peeter Nõges, since then. Phytoplankton indices In order to analyse the long-term dynamics of phytoplankton composition, we calculated the biomass of phytoplankton functional groups according to Reynolds et al. (2002), and the values of one of the three mandatory phytoplankton indices used in Germany for WFD [the indicator based phytoplankton taxa lake index (PTSI; Mischke et al., 2008)], and also a modification of the Swedish trophic plankton index (TPI; Willén, 2007), the Estonian trophic plankton index (Maileht, 2008) and the Hungarian assemblage index (Q-index; Padisák et al., 2006). The German PTSI index distinguishes between polymictic and stratified lowland lakes by two different indicator lists. Here the indicator list and trophic scores for polymictic lakes are used involving in total 193 taxa with 84 as species and 24 as genera. The PTSI was calculated by the automatic calculation tool PhytoSee (Mischke & Böhmer, 2008) as: PTSI ¼ RðAi Si  Fi Þ=RðAi Fi Þ; ð1Þ where Ai is abundance class of the ith taxon, Si the trophic score of the ith taxon and Fi the stenoecy factor of the ith taxon. 123 In order to assess the trophic status of the lake, the calculated German PTSI is compared to the ‘high’/ ‘good’ boundary of very shallow lakes, which is set as a PTSI value of 2.75. Each stepwise distance of 0.5 higher than 2.75 is one status class worse. The Swedish TPI index and its modification, the Estonian TPI, were calculated as: TPI ¼ RðBi  Si Þ=RBi ; ð2Þ where Bi is the biomass of the ith taxon and Si the trophic score of the ith taxon. The Swedish index is based on late summer samples and involves 57 indicator species. In the original (Willén, 2007), TPI is calculated only for the period from mid-July to end of August, and a mean value of three summers is recommended for a more reliable assessment, owing to the inter-annual weather-induced variations of plankton abundances and community structures. We modified the method by calculating the TPI values for all months. The Estonian Index, calculated in the same way as the Swedish index, was based on phosphorus ranges of 130 indicator species from Estonian small lakes. The Hungarian Q-index was calculated as: Q ¼ RBi =BSi ; ð3Þ where Bi is the biomass of the ith functional group (sensu Reynolds et al., 2002), B the total phytoplankton biomass, Si a factor number established for the ith functional group in the given lake type given by Padisák et al. (2006). We used factor numbers corresponding to large ([100 km2), shallow (3–15 m), calcareous lakes. Testing investigator-induced effects on German index As the German index revealed a major change in the index value between 1977 and 1979 coinciding with the change of investigator (see ‘‘Results’’ section), the possible inconsistencies in species identification were analysed specifically for this index. For this purpose, (1) We compared the frequencies of occurrence (percentage of samples where the species occurred) of all 99 German indicator species Hydrobiologia (2010) 646:33–48 (2) (3) (4) (5) found in Võrtsjärv before and after this breaking point; The 61 species with statistically significant differences in occurrence frequencies were analysed further; The two investigators (R. Laugaste and P. Nõges) analysed the species one-by-one to find out possible misidentification errors, and corrections were made in 26 cases. In 35 cases, the identification was unambiguous and the change could be considered real. Most of the corrections were needed for cases where a different taxonomic resolution was used in different periods. In 20 cases, the correction of the identification resolution from species to genus resulted in abandoning the indicator. In three cases, the resolution was regulated up for one of the periods, and in another three cases, there were obvious misidentifications that were corrected. After correcting the input data, new index values were calculated and the difference between uncorrected and corrected indices was taken for the investigator-induced error. The corrected data were used for all further analysis. Meteorological and nutrient loading data We received the meteorological data (daily air temperatures, amount of precipitation) and the hydrological data (daily lake levels, river flows) for the years 1964–2007 from the Estonian Institute of Hydrology and Meteorology. The values of the North Atlantic Oscillation Index were taken from the website http://www.cru.uea.ac.uk/cru/data/nao.htm. The year was divided into 3-month seasons, winter beginning with December of the previous year. We received total phosphorus (TP), total nitrogen (TN), total inorganic nitrogen (TIN) and phosphate phosphorus (PO4–P) concentrations in rivers for loading calculations from Tartu Environmental Research Ltd., which applies the EVS-EN ISO/IEC 17025:2000 standard. Nitrate was determined using ion chromatography, for nitrite, ammonium, orthophosphate and TP, spectrophotometric methods were applied, TN was measured spectrophotometrically after potassium peroxodisulphate oxidation. 37 Correlation analysis We correlated the phytoplankton data [annual mean, annual median and summer (July–August mean) of the four indices and the annual mean biomasses of functional groups] with hydrometeorological and loading data and analysed the number of significant correlations per factor as a first estimate for the major influences. For this analysis, logarithms of the loading data and biomass data were taken to achieve normal distribution, and all data series were detrended by subtracting the best-fit line (first-order trend). CCA In order to evaluate the relationships among phytoplankton species abundance composition and environmental variables, we ran a canonical correspondence analysis (CCA), using the Vegan package for R (Oksanen et al., 2009). We fitted environmental variables onto the ordination using Vegan Enfit routine which fits vectors of continuous variables and centroids of levels of class variables. The direction of the vector shows the direction of the gradient, and the length of the arrow is proportional to the correlation between the variable and the ordination. We also tested with the Vegan permutation tests the significance of constraints (Oksanen et al., 2009). Results Dynamics of taxonomic indices All four taxonomic indices showed a continuous deterioration of the ecological status of the lake (Fig. 1). The Swedish index indicated a transformation from ‘moderate’ to ‘poor’, the German index from ‘good’/‘moderate’ to ‘bad’, and the Hungarian index from ‘poor’ to ‘bad’ during the 44-year period. The indicated quality classes are, however, rather arbitrary as none of the indices was adapted to the region and/or the type of lake. The trend was more or less smooth in the Swedish, Estonian and Hungarian indices while the German Index revealed a sharp transition from the ‘good’/‘moderate’ status class to mostly ‘bad’ between the years 1977 and 1979 that coincided with the change of investigator. 123 38 Hydrobiologia (2010) 646:33–48 Fig. 1 Long-term changes in ecological status of Võrtsjärv as indicated by some phytoplankton taxonomy based indices. The full German index is based on three metrics: biomass, algal classes and an indicator taxa based index—PTSI, but here we used only the latter. The status classes H—‘high’, G—‘good’, M—‘moderate’, P—‘poor’ and B—‘bad’ are based on WFD normative definitions. No class boundaries were set for the Estonian index Investigator-induced differences (in fewer than 50 samples each), with low abundances and could not have a strong effect on the index values. Among the more frequent species, only one, Aulacoseira ambigua (Grunov) Simonsen, built up substantial biomasses and could affect significantly the index values. This species was identified altogether in 577 samples (in 80% of samples before 1978 and in 60% after) but as its seasonal occurrence overlapped partly with A. granulata (Ehrenb.) Simonsen and A. subarctica (O. Müll.) E. Y. Haw. And as all three species vary widely in their diameter, this group was not always identified to the species level. Because of the inconsistent resolution of identification, we decided to eliminate these species After correcting the input data for possible investigator-induced differences, the index values dropped slightly for both periods but the difference between the periods before and after the change increased (Table 1). From this analysis, we concluded that the big shift in index values between 1977 and 1979 reflected a real change in phytoplankton community which just coincided but was not caused by the change of investigator. Thirteen of the 20 indicator species that dropped out from the analysis while harmonizing the input data for the two periods occurred only occasionally 123 Hydrobiologia (2010) 646:33–48 Table 1 Comparison of the German index mean values for the two periods studied by different investigators (in rows) before and after correction for investigatorinduced effects (in columns) 39 Mean uncorrected Mean corrected Difference caused by correction (%) P Before 1978 3.62 3.48 3.7 0.0207 After 1978 4.52 4.42 2.3 0.0001 Difference between periods (%) P 25.0 26.8 \0.0001 from the analysis. A. ambigua has one of the highest trophic scores (5.68) in the German assessment scheme and its elimination caused most of the drop of the index values in both periods but did not change the difference between the two periods. Dynamics of functional groups and dominant species In the long term, there was a clear decrease in all functional groups comprising small-sized phytoplankton species, such as X1, E, F, J, N (some of them shown in Fig. 2) in Võrtsjärv. The functional group X1 was represented by the chlorococcalean genera Tetraedron, Monoraphidium, Ankistrodesmus, Lagerheimia and Crucigenia among others and was never very abundant. However, during the study period, their biomass dropped by nearly one order of magnitude. The functional group E, represented by the genera Mallomonas and Dinobryon, reached its maximum abundance (around 0.5 g m-3) during the low water period in the early 1970s, dropped suddenly at the end of the decade and had a slight increasing trend thereafter. Nearly, the same happened with the functional group F (Oocystis, Dictyosphaerium and Kirchneriella), not shown in the figure. In the functional group N, represented by desmid genera Staurastrum and Cosmarium, and the diatom Tabellaria, the decrease was smoother and lasted until the end of the 1980s. During the two last decades, the species of this group occurred with variable abundances but without showing clear trends. The functional group J was represented by the Chlorococcales Scenedesmus, Pediastrum and Coelastrum, which are the most common green algae in the lake. Their decrease has been rather constant from about 5 g m-3 in the early 1970s to 0.5–1 g m-3 in recent years. The two functional groups showing a clear increase were S1 and H1, both represented by filamentous \0.0001 cyanobacteria (S1 by Limnothrix and Planktolyngbya and H1 by Aphanizomenon). Filamentous cyanobacteria dominated for the whole period, but showed remarkable changes in their abundance (Fig. 3). Planktolyngbya limnetica (Lemm.) J. KomárkováLegnerová that dominated the phytoplankton community in the beginning of the study period and reached its maximum abundance in early 1970s, decreased continuously in the following decades and the present biomasses form about one tenth of its previous level. Limnothrix redekei (Goor) Meffert was present in small numbers already in the 1960s but had a rapid increase between 1975 and 1983. Since 1989, its biomass steadily decreased and it was replaced by Limnothrix planktonica (Wołosz.) Meffert. The latter appeared for the first time in the samples in 1979 and now builds up often more than 90% of the total phytoplankton biomass. The biomass of the nitrogen-fixing Aphanizomenon skujae J. Komárková-Legnerová & G. Cronberg increased continuously throughout the study period and reached maximum abundance in years of low water level (1996, 2000, and 2006). Relationships with potential driving factors The correlation analysis of detrended phytoplankton data with hydrometeorological and nutrient loading data (Table 2) showed that the water level had the greatest number of significant correlations with phytoplankton variables (12). After a long dry period in the 1960s and most of the 1970s, an extremely rainy summer and autumn in 1978 increased the average water level for the following year by more than a meter. Coinciding with this event, there was a change in the dominating cyanobacteria species and most of the functional groups and phytoplankton composition indices indicated a major change (Fig. 4). The change was persistent and not reversed by single low water years such as 1996 or 2003. 123 40 Hydrobiologia (2010) 646:33–48 Fig. 2 Long-term biomass changes of some phytoplankton functional groups sensu Reynolds et al. (2002) in Võrtsjärv 123 Hydrobiologia (2010) 646:33–48 41 Fig. 3 Long-term biomass changes of the four filamentous cyanobacteria species in Võrtsjärv Data on nutrient loading that cover the period since 1975 (Fig. 5) show a major increase both in TN and TP loadings in 1981, i.e. after the change in phytoplankton had occurred. Still the load of total nitrogen had five, the TN/TP loading ratio four, and the loads of inorganic nutrients, PO4–P and TIN, both three significant correlations with phytoplankton variables (Table 2). TN loadings remained high for about 10 years, which reflected the industrialization of agriculture in the 1980s. After the break-down of the Soviet Union, the use of fertilizers decreased and the loadings have nearly halved by now. Except a couple of years with peak loadings (1981, 1986) there has been a continuous decrease in TP loadings from more than 100 tons in the 1970s to less than 50 tons in the recent years. Despite the considerable decrease in nutrient loadings (for in-lake concentrations shorter data series exists and the decreasing trend is still non-significant), the phytoplankton-inferred water quality did not show any improvement. Among meteorological variables, winter air and water temperature, spring water temperature, and summer precipitation had three significant correlations each (Table 2). 123 42 Hydrobiologia (2010) 646:33–48 Table 2 Number of significant (P \ 0.05) correlations between the listed hydro-meteorological and loading variables with the 25 phytoplankton variables Factors in decreasing order Number of significant correlations with phytoplankton variables Water level 12 ln (TN load) 5 TN/TP load 4 ln (TIN load) 3 ln (PO4–P load) 3 Winter AT Winter WT 3 3 Spring WT 3 Summer PR 3 ln(TP load) 1 Spring AT 1 Summer AT 1 Summer WT 1 Autumn WT 1 Winter PR 1 NAOw 0 Autumn AT 0 Spring PR 0 Autumn PR 0 TN total nitrogen, TP total phosphorus, TIN total inorganic nitrogen, AT air temperature, WT water temperature, PR amount of precipitation, NAOw North Atlantic Oscillation index for winter CCA analysis The CCA analysis carried out on phytoplankton data from 1975 onwards (Fig. 6) showed significant relationships between species composition (in terms Fig. 4 Long-term dynamics of annual mean water level and the annual trophic score by phytoplankton indicators (German PTSI) in Võrtsjärv 123 of biomass of functional groups and dominant species) and the environmental variables (permutation test for CCA under full model P \ 0.005). The first two CCA axes accounted for 54.6% of the total variance, with the Axis 1 alone explaining 42.8%, indicating a relevant gradient in the data set. Some of the species–environmental variables correlations were high and statistically significant after a permutation test. The NAO winter index, water level, spring and winter air temperature and the TN/TP loading ratio were negatively related to Axis 2. Total nitrogen and phosphorus were negatively related to Axis 1. There was a rather large angle between the TN and TP arrows showing that their loadings were governed by different processes. High TP loadings were characteristic of years with low winter precipitation and cold summers. Both winter precipitation and summer air temperature were positively related to Axis 1. The years are located around the middle point of the graph forming almost a counterclockwise circle. The second half of the 1970s and the year 1980 form the most distant group from all the others in the upper left corner. In this quarter of the graph, TP loading has the strongest positive effect. As it was the period of higher species diversity, most of the phytoplankton functional groups with decreasing trends during the study period can be found in this sector. The lower left quarter shows a pattern typical for the 1980s and the first half of the 1990s, characterized by high water level and heavy nitrogen loading. The NAO winter index turned into a persistent positive phase in 1980 that has lasted until the present and reached its peak in 1992 (not shown) bringing about mild winters and springs for the whole period (the whole lower half of Hydrobiologia (2010) 646:33–48 43 Fig. 5 Long-term dynamics of annual loadings of total nitrogen (TN) and total phosphorus (TP) and the annual trophic score by phytoplankton indicators (German PTSI) in Võrtsjärv Fig. 6 Biplot of the canonical correspondence analysis results. TP total phosphorus load, TN total nitrogen load, NAO North Atlantic Oscillation index, AT air temperature, PR amount of precipitation, wi winter, sp spring, su summer, au autumn, P. lim.— Planktolyngbya limnetica, L. red.— Limnothrix redekei, L. pl.—L. planktonica, 1975–2007—years; A, C, D, E, F, H1, J, N, X1, X2, X3— phytoplankton functional groups according to Reynolds et al. (2002) the graph). The years of the 2000s located in the upper right quarter were characterized by high summer and autumn temperatures, increased precipitation in spring and low nutrient loadings. The increase in Aphanizomenon abundance (H1) was typical for this period. The functional group S1 as a whole did not show any affinity to the analysed driving factors and was located just in the zero point of both axes. However, if split into species, the picture changed: Planktolyngbya limnetica correlated with both loading factors, Limnothrix redekei with water level and L. planktonica with summer and autumn temperature. Discussion For nearly 50 years after the first expedition to Võrtsjärv in 1911–1913 (Mühlen & Schneider, 1920) phytoplankton of the lake was not investigated. Qualitative samples taken again in 1959–1962 showed almost no differences compared with the 123 44 picture from the beginning of the century (Nõges et al., 2004): the main dominant was the cyanobacterium Planktolyngbya limnetica, accompanied by diatoms Aulacoseira ambigua, Tabellaria fenestrata (Lyngb.) Kütz., Asterionella formosa Hassall and others. Among green algae, several species of Pediastrum were most numerous. Water blooms caused by Anabaena lemmermannii P. Richter in June and described already by Mühlen & Schneider (1920) recurred from year to year until the end of the 1970s. Sudden change in plankton composition between 1977 and 1979 involved most of the functional groups and brought about a persistent change in phytoplankton dominant species. Although A. lemmermannii has not disappeared from Võrtsjärv, no bloom was observed since the late 1970s. Considering the abruptness, extent and persistency of the change, we can talk about a regime shift. Numerous findings from both terrestrial and aquatic systems suggest that ecosystems can undergo regime shifts where they suddenly change from one state into another. For lakes, the best known eutrophication-related regime shifts involve switching from a clear macrophyte dominated to a turbid phytoplankton-dominated state (Jeppesen et al., 1998, Carpenter, 2003; Crossetti & de M Bicudo, 2008) and from eukaryotic phytoplankton to filamentous cyanobacteria dominance (Scheffer et al., 1997). As Mischke & Nixdorf (2003) noted, few Oscillatoriaceae species that can tolerate the combination of intermittent nutrient deficiency and low light conditions produced by the frequent but irregular mixing, are able to build up very dense populations, which increases turbidity. The steady state period is a selfinduced habitat, in which competitors fail because of low-light conditions combined with effective exploitation of nutrient resources by the dominant Cyanobacteria. Loss of resilience in an ecosystem is not necessarily a noticeable gradual degradation, but can be a sudden shift triggered by a stochastic event or when a threshold is suddenly reached (Troell et al., 2005). We suggest that the sudden large increase in water level in Võrtsjärv after an extensive low water period characterized by high nutrient loading broke the resilience of the system. Coops et al. (2003) showed that extreme water levels may cause shifts between the turbid and the clear, macrophyte-dominated state 123 Hydrobiologia (2010) 646:33–48 in shallow lakes. Most species of cyanobacteria are sensitive to sudden changes of environment. Individual cyanobacterial species have considerable specialisations and are intolerant of a high degree of environmental variability (Padisák & Reynolds, 1998). When conditions change, strong competition between Oscillatoriaceae species has been described by many authors (Rücker et al., 1997; Nixdorf et al., 2003; Köhler et al., 2005; Zebek, 2006). Our CCA analysis showed that the three dominating Oscillatoriaceae species, Planktolyngbya limnetica, Limnothrix redekei and L. planktonica, all belonging to the S1 functional group characteristic of highly light deficient conditions in turbid mixed layers (Reynolds et al., 2002), differed considerably in their ecological niches. We suppose that the shift from Planktolyngbya to Limnothrix dominance was caused by the sudden deterioration of light conditions in the deeper water (Nõges & Järvet, 1995). The consequent gradual increase of the proportion of L. planktonica could be related to climate change. L. redekei peaked usually in June while the biomass of L. planktonica increased smoothly during the growing season (Nõges, 1999) and reached its maximum in late autumn. Mild winters associated with the predominating positive phase of the North Atlantic Oscillation Index for winter during the last decades of the twentieth century (Jaagus, 2006) could give an advantage to L. planktonica. The continuous increase of the biomass of the nitrogen-fixing Aphanizomenon skujae can be related to the decrease of the N/P ratio especially accentuated at low water levels that enhance nitrogen loss through denitrification and intensify phosphorus release from the sediments (Nõges et al., 2008). Also the improvement of light conditions during low water periods gives a competitive advantage to N2-fixers against the commonly dominating shade tolerant species. All phytoplankton-based water quality indices tested on our long-term data indicated a deterioration of water quality and none of them reflected the decline of nutrient loadings to the lake at least since 1990 (Fig. 5). Internal loading and the mechanism of hysteresis (Scheffer et al., 1997) offer an explanation for the resistance of cyanobacteria dominance in shallow lakes to restoration efforts by means of nutrient load reduction. There are different views on the ability of phytoplankton-based indices to reflect changes in Hydrobiologia (2010) 646:33–48 the trophic status of lakes. Nygaard (1955), after a careful test of his Compound Index (Nygaard, 1949), found that the index gave good agreement at the extremes of the trophic range but was erratic in the middle. If quotients are reliable only at the extremes of the trophic scale, they can be of limited use since such differentiation is usually obvious without making any calculation. Similarly, Kaiblinger et al. (2009), who tested the applicability of two phytoplankton indices for trophic classification on three large peri-alpine lakes, showed that the indices were appropriate only approximately to distinguish lakes of different water quality but were not sensitive enough to analyse the change of water quality within the lakes. Another long-term study carried out in Lake Mondsee (Dokulil & Teubner, 2005) showed that a reduction in TP concentration was not accompanied by an immediate decline in total phytoplankton biovolume, and the species composition characterizing the phytoplankton community before nutrient reduction persisted. Reductions in phytoplankton biovolume were delayed by about 5 years. Several phytoplankton species differed in the timing of their responses to changing nutrient conditions. For example, while Planktothrix rubescens declined concomitantly with the decline in TP concentration, other species indicative of higher phosphorus concentrations, such as Tabellaria flocculosa var. asterionelloides, tended to increase further. Analysing the effect of reoligotrophication on phytoplankton variables in Lake Balaton, Hajnal & Padisák (2008) concluded that the responses of the phytoplankton to declining external phosphorus load was only slight and sometimes counterintuitive. Similar to Mondsee, the eukaryotic plankton flora showed little alteration but floristic changes in the dominant cyanoprokaryota were consistent with environmental changes attributable to the eutrophication and subsequent restoration. System shifts, and the corresponding mismatches between nutrient levels and phytoplankton responses due to hysteresis linked delay phenomena, can have important implications for formulation of management strategies, as undesired system characteristics (from a human perspective) may be highly resistance to restoration efforts (Troell et al., 2005). Resulting low sensitivity of phytoplankton indices is a real problem since several assessment systems, e.g. those created for the WFD monitoring programmes, are expected to be used to monitor quality changes within 45 single lakes (Kaiblinger et al., 2009). On the other hand, this mismatch between nutrients and phytoplankton indices can be considered a good property of the latter, making them more valuable than just a surrogate for phosphorus concentration, a variable that can be easily measured. The stand alone value of the new phytoplankton indices for the WFD is that they compare the present community to a welldefined reference status and that the loss of species present at near reference status is weighed as deterioration independently from nutrients. In that way, phytoplankton indices really reflect the ecological effect of human impact, which can last much longer than the direct impact itself. Obviously, a reduction of TP concentration much below 50 lg l-1 is needed in Võrtsjärv to induce a decline in cyanobacteria abundance. The reference status of very shallow lakes in Central Europe is thought to be a macrophyte-dominated state with phytoplankton chlorophyll a concentration below 12 lg l-1 (Carvalho et al., 2008) and a low risk of cyanobacteria blooms (Ptacnik et al., 2008). This state not only implies a successful competition for light and nutrients by macrophytes over phytoplankton, but also prevention of sediment resuspension by macrophyte coverage. It is questionable if such a high macrophyte coverage can establish in a large windexposed shallow lake like Võrtsjärv. In order to manifest this state perhaps a different concept of reference conditions should be considered for very large and shallow lakes with high sediment turbidity (Dokulil & Teubner, 2003). According to WFD-compliant assessment (Nõges & Nõges, 2006), the ecological status of Võrtsjärv was estimated to be ‘‘good’’. Large discrepancies with the ecological status indicated by the different phytoplankton indices in this article emphasize the importance of the lake type specific approach in status assessment and recall the discussions on the distinction between fundamental edaphic trophic state and so-called secondary or morphometric trophic state (Rawson, 1956). In the 1950s, very large, deep lakes of North America, despite tremendous nutrient loads from inflowing rivers, exhibited oligotrophic features mainly because of great depths and low temperatures, i.e. morphometric factors reduced plankton abundance. It seems that the situation in Võrtsjärv is, on the contrary, an expression of morphometric hypertrophy where species indicating 123 46 hypertrophy are favoured by strong light limitation rather than nutrients. Conclusions Our results add evidence to the likelihood of nonlinear response of phytoplankton to changing nutrient loadings. Phytoplankton showed a resilient reaction to eutrophication that was finally broken when an increase in water level caused a regime shift bringing about a change in dominant species. High shade tolerance of the new dominants and their ability to create shade are obviously the stabilizing factors for the new status, resistant to restoration efforts. Our study showed that hydro-meteorological factors may cause disturbances that force the system to switch between stable states. The results suggest also that in the long term, climatic variables may have an effect on the sensitive balance of species with close ecological demands. The different skills and habits of plankton analysts expressed in different taxonomic resolution, or errors in species identification, add inconsistencies to longterm data. Because nowadays most common plankton composition indices have indicator values weighted by abundance or biomass, the correct identification of the dominant and abundant species is most crucial. Acknowledgements The study was supported by SF 0170011508 from Estonian Ministry of Education and Research, by grant 7600 from Estonian Science Foundation and by EC FP7 project WISER. Data collection in frames of the state monitoring programme was supported by the Estonian Ministry of Environment. We acknowledge hydrological and meteorological data contributed by the Estonian Institute of Hydrology and Meteorology. References Anneville, O., S. Gammeter & D. Straile, 2005. Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri-alpine lakes. Freshwater Biology 50: 1731–1746. Brettum, P., 1989. Algen als Indikatoren für die Gewässerqualität in norwegischen Binnenseen. NIVA, Trondheim. Carpenter, S. R., 2003. Regime Shifts in Lake Ecosystems: Pattern and Variation, Vol. 15. Excellence in Ecology Series. Ecology Institute, Oldendorf/Luhe, Germany. Carvalho, L., A. G. Solimini, G. Phillips, M. Van den Berg, O.-P. Pietiläinen, A. Lyche Solheim, S. Poikane & U. Mischke, 2008. Chlorophyll reference conditions for 123 Hydrobiologia (2010) 646:33–48 European lake types used for intercalibration of ecological status. Aquatic Ecology 42: 203–211. Coops, H., M. Beklioglu & T. L. Crisman, 2003. The role of water-level fluctuations in shallow lake ecosystems – workshop conclusions. Hydrobiologia 506–509: 23–27. Crossetti, L. O. & C. E. de M Bicudo, 2008. Phytoplankton as a monitoring tool in a tropical urban shallow reservoir (Garças Pond): the assemblage index application. Hydrobiologia 610: 161–173. Directive, 2000. Directive 2000/60/EC of the European parliament and of the council of 23 October 2000 establishing a framework for community action in the field of water policy. Official Journal of the European Communities L 327: 1–72. Dokulil, M. T. & K. Teubner, 2003. Eutrophication and restoration of shallow lakes – the concept of stable equilibria revisited. Hydrobiologia 506–509: 29–35. Dokulil, M. T. & K. Teubner, 2005. Do phytoplankton communities correctly track trophic changes? An assessment using directly measured and palaeolimnological data. Freshwater Biology 50: 1589–1593. Dokulil, M., A. Hamm & J.-G. Kohl (eds), 2001. Ökologie und Schutz von Seen. UTB, Facultas-Universitäts-Verlag, Wien. Dokulil, M. T., K. Teubner & J. Greisberger, 2005. Typenspezifische Referenzbedingungen für die integrierende Bewertung des ökologischen Zustandes stehender Gewässer Österreichs Gemäß der EU-Wasserrahmenrichtlinie. Modul 1: Die Bewertung der Phytoplankton Struktur nach dem Brettum-Index. Projektstudie Phase 3, Abschlussbericht. Im Auftrag des Bundesministeriums für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft: Wien. Geelen, J. F. M., 1969. Vergelijkend planktononderzoek in twee Hatertse vennen (Comparative plankton investigation in two Hatertse fens). Proefschrift, Nijmegen. Hajnal, É. & J. Padisák, 2008. Analysis of long-term ecological status of Lake Balaton based on the ALMOBAL phytoplankton database. Hydrobiologia 599: 227–237. Hörnström, E., 1981. Trophic characterisation of lakes by means of quantitative phytoplankton analysis. Limnologica 13: 249–361. Jaagus, J., 2006. Climatic changes in Estonia during the second half of the 20th century in relationship with changes in large-scale atmospheric circulation. Theoretical and Applied Climatology 83: 77–88. Jacquet, S., J.-F. Briand, C. Leboulanger, C. Avois-Jacquet, L. Oberhaus, B. Tassin, B. Vinçon-Leite, G. Paolini, J.-C. Druart, O. Anneville & J.-F. Humbert, 2005. The proliferation of the toxic cyanobacterium Planktothrix rubescens following restoration of the largest natural French lake (Lac du Bourget). Harmful Algae 4: 651–672. Jeppesen, E., M. Søndergaard, M. Søndergaard & K. Christofferson (eds), 1998. The Structuring Role of Submerged Macrophytes in Lakes. Springer-Verlag, Berlin. Jöhnk, K. D., J. Huisman, J. Sharples, B. Sommeijer, P. M. Visser & J. Stroom, 2008. Summer heatwaves promote blooms of harmful cyanobacteria. Global Change Biology 14: 495–512. Kaiblinger, C., O. Anneville, R. Tadonleke, F. Rimet, J. C. Druart, J. Guillard & M. T. Dokulil, 2009. Central-European Water Quality indices applied to long-term data from Hydrobiologia (2010) 646:33–48 peri-alpine lakes: test and possible improvements. Hydrobiologia 633: 67–74. Köhler, J., S. Hilt, R. Adrian, H. P. Kozerski & N. Walz, 2005. Long-term response of a shallow, moderately flushed lake to reduced external phosphorus and nitrogen loading. Freshwater Biology 50: 1639–1650. Maileht, K., 2008. Phytoplankton as an indicator in lakes for EU Water Framework Directive classification. Master Thesis, Estonian University of Life Sciences, Tartu. Marchetto, A., B. M. Padedda, M. A. Mariani, A. Lugliè & N. Sechi, 2009. A numerical index for evaluating phytoplankton response to changes in nutrient levels in deep mediterranean reservoirs. Journal of Limnology 68: 106– 121. Mischke, U. & B. Nixdorf, 2003. Equilibrium phase conditions in shallow German lakes: how Cyanoprokaryota species establish a steady state phase in late summer. Hydrobiologia 502: 123–132. Mischke, U. & J. Böhmer, 2008. Software PhytoSee Version 3.0 Preliminary English Version of the calculation program for German Phyto-See-Index (PSI) according to Mischke et al. 2008 to assess natural lakes to implement the European Water Framework Directive. Free Internet Download (PhytoSee_Vers3_0_eng.zip), http://igb-berlin. de/abt2/mitarbeiter/mischke. Mischke, U., U. Riedmüller, E. Hoehn, I. Schönfelder & B. Nixdorf, 2008. Description of the German system for phytoplankton-based assessment of lakes for implementation of the EU Water Framework Directive (WFD). In Mischke, U. & B. Nixdorf (eds), Bewertung von Seen mittels Phytoplankton zur Umsetzung der EU-Wasserrahmenrichtlinie. Gewässerreport Nr. 10. Brandenburg Technical University of Cottbus, ISBN 978-3-940471-06-2, BTUC-AR 2/2008: 117–146. Mooij, W. M., S. Hülsmann, L. N. De Senerpont Domis, B. A. Nolet, P. L. E. Bodelier, P. C. M. Boers, L. M. Dionisio Pires, H. J. Gons, B. W. Ibelings, R. Noordhuis, R. Portielje, K. Wolfstein & E. H. R. R. Lammens, 2005. The impact of climate change on lakes in the Netherlands: a review. Aquatic Ecology 39: 381–400. Murphy, K. J., M. P. Kennedy, V. McCarthy, M. T. Ó’Hare, K. Irvine & C. Adams, 2002. A review of ecology based classification systems for standing freshwaters. SNIFFER Project Number: W(99)65, Environment Agency R&D Technical Report: E1-091/TR. Netherland, M. D., C. A. Lembi & A. G. Poovey, 2009. Screening of various herbicide modes of action for selective control of algae responsible for harmful blooms. Technical Note ADA494357, U.S. Army Engineer Research and Development Center, Coastal and Hydraulics Laboratory, Vicksburg: 1–12. http://handle.dtic.mil/ 100.2/ADA494357. Nixdorf, B., U. Mischke & J. Rücker, 2003. Phytoplankton assemblages and steady state in deep and shallow eutrophic lakes – an approach to differentiate the habitat properties of Oscillatoriales. Hydrobiologia 502: 111–121. Nõges, P. 1999. Seasonal variation in trichome length of cyanophytes Limnothrix planctonica and L. redekei in a large shallow lake. Archiv für Hydrobiologie Suppl. 129/ Algological Studies 94: 261–274. 47 Nõges, P. & A. Järvet, 1995. Water level control over light conditions in shallow lakes. Report Series in Geophysics, University of Helsinki 32: 81–92. Nõges, P. & A. Kisand, 1999. Forms and mobility of sediment phosphorus in shallow eutrophic Lake Võrtsjärv (Estonia). International Review of Hydrobiology 84: 255–270. Nõges, P. & T. Nõges, 2006. Indicators and criteria to assess ecological status of the large shallow temperate polymictic lakes Peipsi (Estonia/Russia) and Võrtsjärv (Estonia). Boreal Environment Research 11: 67–80. Nõges, P., R. Laugaste & T. Nõges, 2004. Phytoplankton. In Haberman, J., E. Pihu & A. Raukas (eds), Lake Võrtsjärv. Estonian Encyclopedia Publishers, Tallinn: 217–231. Nõges, T., A. Järvet, A. Kisand, R. Laugaste, E. Loigu, B. Skakalski & P. Nõges, 2007. Reaction of large and shallow lakes Peipsi and Võrtsjärv to the changes of nutrient loading. Hydrobiologia 584: 253–264. Nõges, T., R. Laugaste, P. Nõges & I. Tõnno, 2008. Critical N:P ratio for cyanobacteria and N2-fixing species in the large shallow temperate lakes Peipsi and Võrtsjärv, NorthEast Europe. Hydrobiologia 599: 77–86. Nygaard, G., 1949. Hydrobiological studies in some ponds and lakes. Part II: the quotient hypothesis and some new or little known phytoplankton organisms. Kongelige Danske Videnskabernes Selskab, Biologiske Skrifter 7: 1–293. Nygaard, G., 1955. On the productivity of five Danish waters. Proceedings of the International Association of Theoretical and Applied Limnology 12: 123–133. Nyström, B., B. Björnsäter & H. Blanck, 1999. Effects of sulfonylurea herbicides on non-target aquatic microorganisms. Growth inhibition of micro-algae and shortterm inhibition of adenine and thymidine incorporation in periphyton communities. Aquatic toxicology 47: 9–22. OECD, 1982. Eutrophication of Waters, Monitoring, Assessment and Control. OECD, Paris. 154 pp. Oksanen, J., R. Kindt, P. Legendre, B. O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens & H. Wagner, 2009. Community Ecology Package, Version 1.15-2. http:// www.cran.r-project.org/. Padisák, J. & C. S. Reynolds, 1998. Selection of phytoplankton associations in Lake Balaton, Hungary, in response to eutrophication and restoration measures, with special reference to the cyanoprokaryotes. Hydrobiologia 384: 41–53. Padisák, J., G. Borics, G. Fehér, I. Grigorszky, I. Oldal, A. Schmidt & Z. Zámbóné-Doma, 2003. Dominant species and frequency of equilibrium phases in late summer phytoplankton assemblages in Hungarian small shallow lakes. Hydrobiologia 502: 157–168. Padisák, J., G. Borics, I. Grigorszky & É. Soróczki-Pintér, 2006. Use of phytoplankton assemblages for monitoring ecological status of lakes within the Water Framework Directive: the assemblage index. Hydrobiologia 553: 1–14. Padisák, J., L. O. Crossetti & L. Naselli-Flores, 2009. Use and misuse in the application of the phytoplankton functional classification: a critical review with updates. Hydrobiologia 621: 1–19. Paerl, H. W. & J. Huisman, 2008. Blooms like it hot. Science 320: 57–58. 123 48 Paerl, H. W. & J. Huisman, 2009. Climate change: a catalyst for global expansion of harmful cyanobacterial blooms. Environmental Microbiology Reports 1: 27–37. Pettersson, K., K. Grust, G. Weyhenmeyer & T. Blenckner, 2003. Seasonality of chlorophyll and nutrients in Lake Erken – effects of weather conditions. Hydrobiologia 506–509: 75–81. Ptacnik, R., L. Lepistö, E. Willén, P. Brettum, P. T. Andersen, S. Rekolainen, A. Lyche Solheim & L. Carvalho, 2008. Quantitative responses of Lake Phytoplankton to eutrophication in Northern Europe. Aquatic Ecology 42: 227– 236. Rawson, D. S., 1956. Algal indicators of trophic lakes types. Limnology & Oceanography 1: 18–25. Reynolds, C. S. & A. E. Irish, 1997. Modelling phytoplankton dynamics in lakes and reservoirs: the problem of in situ growth rates. Hydrobiologia 349: 5–17. Reynolds, C. S., V. Huszar, C. Kruk, L. Naselli-Flores & S. Melo, 2002. Towards a functional classification of the freshwater phytoplankton. Journal of Plankton Research 24: 17–428. Rott, E., 1981. Some results from phytoplankton counting intercalibrations. Journal Aquatic Sciences 43: 34–62. Rücker, J., C. Wiedner & P. Zippel, 1997. Factors controlling the dominance of Planktothrix agardhii and Limnothrix redekei in eutrophic shallow lakes. Hydrobiologia 342(343): 107–115. Salmaso, N. & J. Padisák, 2007. Morpho-functional groups and phytoplankton development in two deep lakes (Lake Garda, Italy and Lake Stechlin, Germany). Hydrobiologia 578: 97–112. Salmaso, N., G. Morabito, F. Buzzi, L. Garibaldi, M. Simona & R. Mosello, 2006. Phytoplankton as an indicator of the water quality of the deep lakes south of the Alps. Hydrobiologia 563: 167–187. Scheffer, M., S. Rinaldi, A. Gragnani, L. R. Mur & E. H. van Nes, 1997. On the dominance of filamentous cyanobacteria in shallow, turbid lakes. Ecology 78: 272–282. 123 Hydrobiologia (2010) 646:33–48 Teubner, K., M. Tolotti, S. Greisberger, H. Morscheid, M. T. Dokulil & H. Morscheid, 2003. Steady state phytoplankton in a deep pre-alpine lake: species and pigments of epilimnetic versus metalimetic assemblages. Hydrobiologia 502: 49–64. Teubner, K., M. Tolotti, S. Greisberger, H. Morscheid, M. T. Dokulil & V. Kucklentz, 2006. Steady state of phytoplankton and implications for climatic changes in a deep pre-alpine lake: epilimnetic versus metalimnetic assemblages. Verhandlungen der internationale Vereinigung für theoretische und angewandte Limnologie 29: 1688–1692. Thunmark, S., 1945. Zur Sociologie des Süsswasserplanktons. Eine methodologisch-ökologische Studie. Folia Limnologica Scandinavica 3: 1–66. Tremel, B., 1996. Determination of the trophic state by qualitative and quantitative phytoplankton analysis in two gravel pit lakes. Hydrobiologia 323: 97–106. Troell, M., L. Pihl, P. Rönnbäck, H. Wennhage, T. Söderqvist & N. Kautsky, 2005. When resilience is undesirable: regime shifts and ecosystem service generation in Swedish coastal soft bottom habitats. Ecology and Society 10: 30. Utermöhl, H., 1958. Zur Vervollkommnung der quantitativen Phytoplankton-Methodik. Mitteilungen der internationalen Vereinigung der theoretische und angewandte Limnologie 5: 567–596. von zur Mühlen, M. & G. Schneider, 1920. Der See Wirzjerw in Livland (Lake Wirzjerw in Livland (in German). Archiv für die Naturkunde des Ostbaltikums 14: 1–156. Vuorio, K., L. Lepistö & A.-L. Holopainen, 2007. Intercalibrations of freshwater phytoplankton analyses. Boreal Environment Research 12: 561–569. Willén, E., 2007. Växtplankton i sjöar. Bedömningsgrunder. SLU, Institutionen för Miljöanalys, Rapport 5: 33 pp. Zebek, E., 2006. Quantitative changes of Planktolyngbya brevicellularis, Limnothrix redekei and Aphanizomenon gracile in the annual cycle vs. physicochemical water parameters in the urban Lake Jeziorak Mały. Oceanological and Hydrobiological Studies 35: 69–84.