Abiotic and biotic factors regulating dynamics of bacterioplankton in a large shallow lake

FEMS Microbiology Ecology 50 (2004) 51–62 www.fems-microbiology.org

Abiotic and biotic factors regulating dynamics of bacterioplankton in a large shallow lake Veljo Kisand *, Tiina N~ oges V~ ortsj€ arv Limnological Station, Institute of Zoology and Botany, Estonian Agricultural University, Rannu, Tartumaa 61101, Estonia Received 22 February 2004; received in revised form 3 May 2004; accepted 27 May 2004 First published online 17 June 2004

Abstract Lake V~ ortsj€arv (270 km2 , mean depth 2.8 m, Estonia) is an eutrophic and turbid lake (Secchi depth 0.5–1 m) with a high nutrient load (total nitrogen 1–2 mg N l
1 , total phosphorus 50 lg P l
1 ) leading to a highly productive phytoplankton population (average chlorophyll a concentration 24 lg l
1 ). Seasonal dynamics of the main members of pelagic microbial loop – phyto-, bacterio- and protozooplankton – has been studied for several years (1993–1998) in lake. The most prominent characteristic of this naturally eutrophic, turbid shallow lake is an inter-annual water level fluctuation (3.2 m) which is in the range of its average depth. Bacterioplankton growth was favored by high water level whereas phytoplankton growth was favored by low water level. In autumn and sometimes in winter bacterioplankton production (BP) was unbalanced with respect to primary production (PP) (gross BP 20–90% higher than PP) suggesting to additional sources of organic carbon. The main grazers of bacteria were probably various ciliates. In contrast to most eutrophic and stratified lakes, the relative prevalence of the microbial loop over the linear phagotrophic food chain was obvious in this eutrophic shallow lake. The bacterioplankton showed a pronounced seasonal succession with low diversity at high production periods in summer and with high diversity at low biomass and production periods in winter and spring.
2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. Keywords: Bacterioplankton dynamics; Fluctuating water level; Large eutrophic and shallow lake

1. Introduction Lake V~ ortsj€ arv is a large natural lake with a limited stock of submerged macrophytes, high turbidity, and high amounts of filamentous phytoplankton. Low water transparency to a great extent is caused by resuspension while the development of submerged vegetation is greatly disturbed by waves. Since grazing on filamentous cyanobacteria and diatoms is difficult, the grazing food chain is proposed to be weak in L. V~ ortsj€ arv [1]. In addition, in the poly- and holomictic water column phytoplankton is exposed to the average light intensity because the mixing depth and the mean depth of the lake *

Corresponding author. Present address: Institute for Chemistry and Biology of the Marine Environment (ICBM), University of Oldenburg, Carl-von-Ossietzky-Strasse 9-11, P.O. Box 25 03, 26111 Oldenburg, Germany. E-mail address: [email protected] (V. Kisand).

are the same. The light conditions decline when the mean depth is greater than euphotic zone of lake resulting in light limited phytoplankton [2]. Furthermore, wind-induced resuspension keeps filamentous algae in the euphotic zone, and maintains high exchange rates of organic matter and inorganic nutrients between sediment, sediment pore water and water column. In L. V~ ortsj€arv the wind fetch is rather long (on average up to 32 km) ensuring high waves and strong resuspension. Shear stress on bottom sediments – the main physical parameter causing resuspension – was studied in L. V~ ortsj€arv and a theoretical model was calibrated and compared with real time measurements [3] to analyze the effect of resuspension on planktonic organisms. However, no correlations between real-time measurements or modeled intensity of resuspension and biotic parameters such as bacterioplankton biomass and activity (Kisand, unpublished data) could be found. At the same time, the most prominent physical parameter of L. V~ ortsj€ arv is

0168-6496/$22.00
2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.femsec.2004.05.009

52

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

the range of water level (WL) fluctuations (absolute range 3.2 m, annual mean amplitude 1.38 m), which can be greater than the mean depth of the lake (2.8 m) [4]. Consequently, the depth of the lake varies greatly and can be considered as the main physical factor determining the structure of food web in this lake. In L. V~ ortsj€ arv the heterotrophic pelagic food web is probably more rich and diverse than usually found in eutrophic lakes. Ciliates with various feeding modes play an extraordinarily important role in the heterotrophic foodweb. Their average biomass reaches 0.5 gC m
2 [5] whereas the biomass of metazooplankton is only 0.21 gC m
2 [6]. Moreover, bacterivorous ciliates seem to be the main grazers on bacteria [5] with an extremely wide food spectra including bacteria, flagellates, small ciliates and picoalgae [5]. In contrast, the grazing activity of metazooplankton is relatively low; its daily consumption in L. V~ ortsj€ arv has been estimated to be on average 2.5% of phytoplankton biomass and up to 28% of primary production (PP) [7]. Open questions for a more comprehensive understanding of shallow large lakes include: (a) what is the main cause and mechanisms of algal mortality in such large shallow lakes and (b) how is the PP channeled through the foodweb. The prevalence of algal grazing by metazooplankton and the classical food chain over the microbial loop is obviously not the case for shallow lakes such as L. V~ ortsj€ arv because filamentous algae are not accessible to herbivores. Also in contrast to deep and stratified lakes where sedimentation is an important loss of phytoplankton biomass, in L. V~ ortsj€ arv due to high levels of resuspension released dissolved organic matter (DOM) as well as particulate detrital organic matter formed after death of filamentous algal cells seem to be primarily transferred to pelagic heterotrophic bacteria [8]. Thus, free-living but also attached and sediment bacteria frequently resuspended into water column can be assumed to provide the basis of the foodweb. The objective of the present study was to analyze the perennial and long-term dynamics of heterotrophic bacterioplankton abundance and activity in a large shallow eutrophic lake with considerable WL fluctuations. Long-term studies of bacterioplankton in lakes are rare (e.g. [9,10]). There are few studies about phytoplankton dynamics in large shallow lakes (e.g. [11]) but comparable studies of bacterioplankton dynamics are missing. We hypothesize that the large-scale WL fluctuation has a significant influence on bacterioplankton biomass, activity and diversity in shallow eutrophic lakes. The prevalence of the microbial loop versus the linear food chain is discussed. In addition, the succession of structural and functional diversity of the bacterioplankton was studied over an annual period to get a better insight into relationships between bacterial diversity and bacterioplankton abundance and activity.

2. Materials and methods 2.1. Study site Lake V~ ortsj€arv is a shallow lake in Estonia at 58 050 – 0 58 25 N, and 25 550 –26 100 E. Its area is 270 km2 , with a mean depth of 2.8 m and maximum depth 6 m. The lake is highly eutrophic, mean concentrations of total nitrogen (TN) about 2 mg l
1 and total phosphorus (TP) about 50 lg l
1 . The water is alkaline (pH 7.5–8.5) with a high buffering capacity and a high content of suspended solids. During the ice-free period (230 days), Secchi depth does not exceed 1 m. The absolute long-term range of level fluctuations is 3.20 m, which corresponds to a 93 km2 difference in the lake area (35% of average area) and to a 0.874 km3 difference in its volume (85% of average volume). The absolute recorded range of the mean depth of the lake is from 1.44 m (6 September 1996) to 3.70 m (26 November 1923). Due to its shallowness, the water temperature can change rapidly, especially in spring and fall. Total number of bacteria varies between 1  106 and 6  106 cells ml
1 [12]. The range of chlorophyll a (Chla) concentration is 0–84 mg Chla m
3 [2]. 2.2. Sampling Water samples were collected weekly to bi-weekly from 1993 to 1998 at the deepest (5–6 m) area of the lake. Community structure samples were collected from 1996 to 1997 (CLPP) and from 1999 to 2000 (community DNA). Depth integrated water was obtained by mixing of the water samples taken by a 2 l Ruttner sampler [13] at depth intervals of 1 m from the lake surface to the near-bottom layer. Water was mixed in a 20 l sterile plastic tank, within 30 min transported to the onshore laboratory and sub-sampled for the various analyses. WL was measured at the outflow by national monitoring program (Estonian Institute of Hydrology and Meteorology). 2.3. Bacterial parameters The abundance of bacteria was measured by acriflavine staining [14]: 20 ml of water was preserved by glutaraldehyde (2% final concentration), 1–5 ml of this sample was filtered onto 0.22 lm pore size black isopore filters (Poretics Inc.) and stained with 0.01% acriflavine for 10 min. Filters were stored at )21 C until counting (within one month) with epifluorescence microscope Leica DMRB at 1000· magnification. Bacterial cells in 15–20 fields were counted (always n > 200 cells). At the microscope cells were sized visually by comparison to the globes of a calibration eyepiece graticule (Patterson Globe and Circle, GI, Eyepiece Graticules Ltd.). Protein biomass was calculated from a power function m ¼ CVa

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

[15], where m is protein biomass (fg cell
1 ), V is cell volume (lm3 cell
1 ), C is a conversion factor between protein biomass and volume (88.6 from [16]), and a is a scaling factor (0.72 from [15]). The protein biomass was converted into carbon biomass by using a factor of 0.86 [16]. Bacterial productivity was assessed by 3 H-leucine incorporation technique (TLI). TLI was measured in three 10 ml replicates (+2 formalin killed blanks) with L(4,5-3 H)-Leucine (59.0 Ci mmol
1 , Amersham Ltd.) [16]. The samples were incubated for 60 min at in situ temperature in the dark. The incubations were stopped with 2% formalin (final concentration). The precipitate for radioactivity assessment was achieved from hot trichloroacetic acid (TCA) treatment. Samples were boiled 100 C) for 30 min in 5% TCA (final concentration), filtered onto cellulose acetate filters (Millipore Inc.), and washed by cold 80% ethanol. Filters were radioassayed by LSC Rackbeta 1211 (LKB Wallac). The external standard ratio method was applied for quenching correction, the counting efficiency was about 40%. Net bacterioplankton production (BP) was calculated from leucine incorporation rates using empirically derived conversion factors (4  1016 cell mol
1 ) suggested by Kisand and N~ oges [12]. Gross BP was calculated from BP using the bacterial growth efficiency (BGE) values (average 20–30%) suggested by Biddanda et al. [17] and Del Giorgio and Cole [18]. Integrated values (per m2 ) of bacterial abundance and production were calculated by multiplying volumetric values (per m3 ) with the average depth of the water column. From October 1996 to September 1997 the community level physiological profiles (CLPP) were estimated on Biolog GN (BIOLOG, Inc.) microtitre plates detecting the respiration (the response of redox dye tetrazolium violet to nicotinamide adenine dinucleotide (NADH) formation) in the presence of 95 different substrates (28 carbohydrates, 2 esters, 5 polymers, 24 carboxylic acids, 2 alcohols, 3 amides, 3 phosphorylated compounds, 20 amino acids, 3 aromatic compounds, 1 brominated compound and 3 amines). Whole lake water (150 ll) and two dilutions with autoclaved tap water (untreated natural ground water well) (1:10 and 1:20) were directly inoculated into each well, microplates were incubated at 25 C for up to 96 h. The oxidation of substrates was monitored colorimetrically (ELISA reader Labsystems Multiscan MCC 340) by the conversion of tetrazolium violet to the vividly purple formazan after 24, 48, 72, 96 h incubation (until May 13, 1997), or only after 48 h (after May 13, 1997). 2.4. Molecular methods Picoplankton was collected from November 1999 to August 2000 by filtration of 1 l lake water through 0.2 lm pore sized filters (polysulfone Supor-200, Gelman

53

Inc.). Total genomic DNA was extracted by inserting the filters in 1 TE buffer (pH 8.0), followed by sonication in a Turner Sonic & Materials VCX 400W, programmed to 30% amplitude, 5 s pulse and 5 s pause for 2 min, and lysing of the cells with lysozyme (final conc. 1 mg ml
1 for 1 h at 37 C) and proteinase K (final concentration 100 lg ml
1 for 1 h at 37 C), followed by CTAB/NaCl (final concentration of 1%) treatment and chloroform–phenol extraction. DNA was precipitated in ethanol and re-suspended in TE buffer (pH 8.0) or MilliQ water. An estimate of the genetic richness and diversity of the bacterioplankton community was obtained by using denaturing gradient gel electrophoresis (DGGE). The specific 16S rDNA region at the positions of 341 and 928 (E. coli numbering) were amplified by polymerase chain reaction (PCR) using universal bacterial primers according to Teske et al. [19]. The resulting amplicons were quantified and equal amounts were run in a denaturing gradient of 20–70% urea/ formamide on a polyacrylamide gel at 200 V for up to 12 h at 60 C. Gels were stained with SybrGreen I or SybrGold (Molecular Probes) and analysed with a STORM (Molecular Dynamics) and a Fluor-S MultiImager (Bio-Rad Inc). For band detection and integrated band area intensities lanes were standardized and the images of DGGE gels were quantified using computer software Quantity One version 4.2.3 (Bio-Rad Inc.). 2.5. Phytoplankton, primary production, and ciliates Depth integrated values of the biotic parameters were measured in order to analyze their relationships with microbial parameters. For measuring Chla concentration, seston was collected on Whatman glass fiber filters (GF/C). Pigments were extracted with 90% acetone and analyzed spectrophotometrically [20,21]. Phytoplankton species composition and biomass were determined in Lugol iodine preserved samples by inverted microscopy using OLYMPUS IMT-2 microscope at 400 magnification. Every species was counted until reaching at least 500 counting units (cells, filaments). Particulate PP of phytoplankton was estimated by 14 CO2 assimilation technique. Water from 0 m, 0.25S (S – Secchi depth), 0.5S, 1S, 2S, 3S was withdrawn into 24 ml glass scintillation vials. One hundred microlitres of sterile NaH14 CO3 solution (1.3 lCi per 1 vial) was added to reach final activity 0.06 lCi ml
1 , after which the vials were exposed for 2 h at midday (usually from 11 a.m. to 1 p.m.), at the same depths of the lake from which the water was sampled. For PP assessment, water was filtered through membranes of 0.45 lm pore size (Millipore HA). The radioactivity of filters was assessed by LSC RackBeta 1211 (Wallac, Finland), and PP was calculated according to a standard formula [20]. Nonphotosynthetic carbon fixation was measured in dark

54

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

vials and subtracted from light assimilation. The trapeze integration over depth was applied for calculating values per m2 (PPint , mgC m
2 h
1 ). PPmax denotes the maximum PP value (mgC m
2 h
1 ) measured throughout the water column. Subsamples (250 ml) for ciliate analysis were preserved with acidified Lugol’s solution and stored at 4 C in the dark. Ciliate biomass and community composition were determined using the Uterm€ ohl [22] technique. Volumes of 10–50 ml were settled for at least 24 h in plankton chambers. Ciliates were enumerated and identified with an inverted microscope (Wild Heerbrug M40 and Nikon diaphot-TMD) at 200–600 magnification. Commonly the entire content of each Uterm€ ohl chamber was surveyed; if total counts were less than 150 organisms, an additional subsample was counted. Ciliates were usually identified to genus according to different keys [23,24]. Additional live subsamples were used to aid identification in some occasions. The first 20 sizable individuals encountered for each taxon were sized. Biovolumes of each taxa were estimated by assuming geometric shapes and expressed as wet weight (WW). Specific gravity was assumed to be 1.0 g ml
1 [25]. Ciliates were divided into bacterivores and herbivores according mainly to the data gathered in previous studies by grazing experiments of fluorescently labeled bacteria [7], and of fluorescent microspheres of sizes 0.5–6 lm (Zingel, unpublished data). The data on ciliate’s ecology provided by Foissner et al. [26] were considered. Integrated values (per m2 ) of abundance and biomass of ciliates were calculated by multiplying volumetric values (per m3 ) with the average depth of the water column.

2.6. Multivariate statistics – ordination Canonical correspondence analysis (CCA, [27,28]) was used to relate seasonal succession of functional (ranked readings of 95 BIOLOG GN substrates utilization) or genetic (binary coded of DGGE banding profiles) fingerprints to BP and abundance. The statistical significance of the relationship was assessed by Monte Carlo permutations test using 1000 permutations. CCA was carried out using free software ADE-4 (http://pbil.univ-lyon1.fr/ADE-4/ADE-4.html).

3. Results We observed large variation of WL during the investigation period. The lowest ever recorded value of WL was measured in 6 September 1996 when the mean depth of the lake was 1.44 m (the long-term average is 2.8 m). In the years 1993, 1994 and 1997 the WL could be considered as medium (difference from the long term average of )24, )9 and )44 cm, correspondingly), in 1995 and 1998 WL was quite high (difference of +5 and +17 cm, respectively), and in 1996 very low, only )112 cm (Fig. 1). Seasonally, WL is the highest after the spring flood (between April and May), low during summer (between July and August) and winter (between December and February) while some increase may also occur in autumn [29]. The total volume of the lake varied historically between 0.455 and 1.216 km3 , the range in 1993–1998 was 0.455–1.098 km3 .

Fig. 1. Biomass of bacteria (
– BB, dash and dotted line of moving average), Chla concentration (N – Chla, dashed line of moving average) and WL (d – line of moving average,
yt ¼ ðyt þ yt
1 þ    þ yt
n
1 Þ=n where y is the variable, t is the current time period, and n is the number of time periods in the average (n ¼ 5)) in 1993–98. Water level 0 is defined at the altitude of 33.68 m according to Baltic altitude system.

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

The algal community was represented by an association of filamentous diatoms and cyanobacteria, diatoms Aulacoseira spp. in spring, cyanobacteria Oscillatoria amphibia f. tenuis, (Anissimov) Elenkin Planktolyngbya limnetica (Lemm.) Kom. et Cronb., Limnotrix redekei and Aphanizomenon sp. in summer and autumn, accompanied by a low and variable biomass of small algae, mostly chlorophytes and chrysophytes (Fig. 2). In 1996, filamentous species were temporarily replaced by the fast growing chroococcal cyanobacteria Cyanonephron styloides Hickel, and the biomass of both diatoms and cyanobacteria was extremely high, the highest values were 52 (long term average 9.5) and 130 (long term average 28) mg WW m
2 , respectively. The amount of Chla ranged between 0.3 and 561 mg m
2 during the investigation period (Fig. 1). Throughout the season Chla was lowest in January– February, started to rise in March, and achieved the maximum or several high peaks in August–September. Thereafter Chla declined until December when the lake was covered by ice. Throughout the study period, Chla had the highest values in 1996 (average 134  134 (SD) mg m
2 ) and in 1994 Chla was lowest (72  35 mg m
2 ) while the reminder were quite similar with, annual means of Chla ranging from 82  68 to 88  54 mg m
2 . PP of phytoplankton was very low in winter (December–February) increasing largely usually in March. The first peak of PP occurred in May–June while the second peak followed in July–August (Fig. 3). In autumn (October–November), PP declined quickly. Annual average PP was relatively low (567  612–616  548 mgC m
2 day
1 ) in 1994–1996, the highest in 1998 (1164  802 mgC m
2 day
1 ). In 1993 and 1997 PP

55

reached a medium level (829  653 and 730  707 mgC m
2 day
1 , respectively). Bacterial biomass (BB) was generally high in years with high WL, 1993–94 and 1997–98 when the annual average BB ranged from 212  195 (SD) to 371  50 mgC m
2 and low (145  91 mgC m
2 ) in 1996 when the WL was lowest. However, BB was also low (122  82 mgC m
2 ) in 1995 when WL was rather high (Fig. 1). Throughout the season BB was typically low during winter months, increased in March and reached the highest values from July to August. BP was highly variable and ranged between 0.07 and 4119 mgC m
2 day
1 during the investigation period. BP was extremely high (908  1107 mgC m
2 day
1 ) in 1994 and the lowest (average 123  421 mgC m
2 day
1 ) in 1995. During the other years studied average BP (  SD) ranged between 168  191 and 198  264 mgC m
2 day
1 . BP dynamics were highly variable, and peaks appeared in winter, spring, and summer (Fig. 3). BP was consistently high every year in July and August. Despite of considerable differences between different years BB and BP showed typical seasonal dynamics. BB was low under ice (<75 mgC m
2 ) and increased continuously from April until July. During the rest of the season BB fluctuated from 100 to 500 mgC m
2 , and did not decline even in late fall until ice coverage occurred. BP was generally low (<150 mgC m
2 day
1 ) during the unproductive season from November to February. Usually there was a peak of BP (in range 350–700 mgC m
2 day
1 ) during the phytoplankton spring bloom in March or April followed by a period of very low BP in May. High BP season began usually in June and lasted until September with the strongest

Fig. 2. Seasonal dynamics of the biomass of cyanobacteria (N – dash and dotted line of moving average) and diatoms (
– dashed line of moving average), and WL (d – line of moving average) in L. V~ ortsj€arv in 1993–98.

56

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

Fig. 3. Bacterioplankton production (
– BP, dash and dotted line of moving average), primary production (N – PP, dashed line of moving average) and WL (d – line of moving average) in L. V~ ortsj€arv in 1993–98.

peaks always in July or August. BP continued to show relatively high values (around 350–500 mgC m
2 day
1 ) during the following 1–1.5 month period in September and October. BP and PP were only weakly coupled throughout the season (correlation not significant p > 0:05, see also Fig. 3). In summer (July–August), BP peak occurred later than peak of PP and BP was also occasionally high in winter while there was no measurable PP. The ratio of BP over PP (BP/PP) was highly variable. Net BP summarized over the whole study time (gC per 1993–98 period) equalled about 60% of summarized PP. As-

suming of 25% BGE, calculated yearly gross BP was about 20–90% higher than net PP. Bacterivorous ciliates were low in biomass in 1995, 1997 and 1998 (yearly averages 5.5  10.0, 7.4  12.8 and 19.4  19.4 mgC m
2 , respectively), in 1996 biomass was significantly higher (47  70 mgC m
2 ). Throughout the year biomass peaked usually in July–August (Fig. 4). Herbivore–bacterivore ciliates had a relatively high biomass in 1995 (average 67.6  78.2 mgC m
2 ), medium in 1996 and 1997 (average 48.2  48.8 and 44.9  62.4 mgC m
2 , correspondingly), and low in 1998 (30.1  43.0 mgC m
2 ). Seasonal dynamics of herbivore–

Fig. 4. Biomass of bacteriovorous (
– dashed line of moving average) and bacteri-herbivorous ciliates (N – dash and dotted line of moving average) compared with biomass of bacteria (d – line of moving average) in L. V~ ortsj€arv.

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

bacterivore ciliates varied throughout different years. In 1995–96 several peaks occurred from May to October, in 1997–98 two major peaks occurred in April–May, and July–August. Genetic (PCR-DGGE of partial 16S rDNA sequences) and functional fingerprints (CLPP by BIOLOG GN microplates) of the bacterioplankton could be only studied during one year. In total 36 different bands, which were assumed to represent separate operational taxonomic units (OTUs), were identified on DGGE gels. Lowest number of OTUs (up to 10) occurred in July and the highest number (up to 21) in November (Fig. 6). We did not identify any OTU, what was present in all samples, but 4 OTUs were present in at least 10–11 samples and 7 OTUs were rare (present only in 1–2 samples). In samples with the highest number of OTUs (between November and April) one single dominating OTU made up to about 10–13% of the total abundance of OTUs (measured as relative band intensity). In summer (May–August) when the diversity of OTUs was lower, one OTU comprised up to 25% of the total abundance. Throughout the whole season only few OTUs (2–6) were more abundant than 10% of the total lane intensity. CCA of DGGE profiles scaled with BB and BP (independent variables) revealed that the first axis (CCA1, related to BB, Fig. 4(b)) described 66% and the second axis (related to BP) 33% of the variability of OTU patterns in DGGE gels (statistical significance by randomization test p ¼ 0:009). The pronounced seasonal succession of the OTU pattern occurred with small differences between successive samples in late fall and winter while spring and summer communities clustered separately (Fig. 6(b)). OTU scores (‘species’ scores) showed that a limited number of OTU’s associated with summer samples together with high abundance of bacteria as well as characteristic community of low productive community occurred in spring. Similarly, CCA of CLPP profiles scaled with BB and BP (independent variables) described 57% of total variation by CCA1 (associated with BP) and 33% of variation by CCA2 (associated with BB) (Fig. 6(c)) with statistical significance (p < 0:001). Also CLPP succession had pronounced seasonal dynamics; winter, spring and summer communities clustered separately. During the early phytoplankton spring bloom the bacterioplankton community with higher productivity had CLPP patterns similar to the CLPP patters of summer community. Some carbohydrates/monomeric sugars (fructose, glucose, rhamnose, cellobiose) were ‘signature substrates’ (related to ‘BP axis’) for highly productive season of CLPP (Fig. 6(c)) while amino acids (phenylalanine, glycyl-L -aspartic acid) and organic acids (a-keto valeric acid, citric acid, succinic acid) involved in amino acids metabolism were ‘key substrates’ for BB axis. These substrates were preferably utilized when BP was rela-

57

tively low but BB was high, during transition periods between winter and early spring, and late summer – early autumn.

4. Discussion 4.1. Wind-induced resuspension of sediments The active resuspension of bottom sediments into the water column is a common feature of large, shallow lakes and represents a major physical factor in such ecosystems. Wind-induced water column mixing can theoretically have a negative (reduced growth/biomass) or a positive (enhanced growth/biomass) impact on bacterioplankton dynamics. A direct positive effect could be assumed due to (1) transport of active benthic bacteria or/and dormant cells into the water column where they remain/become active, (2) transport of substrates from the sediment and pore water into the water column, (3) reduction of the grazing pressure due to disturbance of the grazers (more non-edible particles in water). A negative effect can occur due to the inhibition of bacterial growth (or other planktonic organisms) by very intensive mixing and resuspension under conditions of high physical forces. In these conditions the fluxes of organic and inorganic matter between sediment and water column are presumably high, and nutrients may be effectively transported into the water column. Unfortunately, we have no data available on biomass, productivity of sediment and fluxes of bacteria between sediment and water column. However, the upper layer of the sediment is continuously resuspended into the water column, therefore the flux of sediment bacteria on sediment particles is not a rare event and the pelagic bacterial community may be a mixture of species primarily from the sediment and/or water column. 4.2. Water level as an important abiotic factor in shallow lake Since the intensity of resuspension is directly linked to water depth, WL fluctuations can be considered to be among the most important environmental factors shaping both phytoplankton [30] and bacterial community in L. V~ ortsj€arv. Throughout the period of our study WL dropped about 2 m within a one year period and even 2.5 m during 1.5 year (from May 1995 to September 1996). The different seasonal pattern of WL was not similar to and sometimes substantially deviated from the typical pattern (average of recorded values since year 1884). Only in 1994 and 1998 was the autumnal increase of WL noticeable, in 1996 the spring flood in May was exceptionally weak and WL dropped continuously until September. In 1997 WL was very

58

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

high after the spring flood and decreased afterwards continuously until the next winter. Years of high and low WL can be clearly distinguished. Analysis of long time-series (since year 1884) revealed a strong influence of the North Atlantic Oscillation (NAO) on the WL in L. V~ ortsj€ arv – yearly average WL increased with the winter NAO index, the higher is the yearly average WL [31]. WL fluctuation can have several consequences to the plankton community. Directly, the considerable changes of the lake volume can cause concentration/dilution effects. This phenomenon could explain extremely high concentration of Chla in 1996 (105 lg l
1 on 16 September), when WL was very low. In contrast to the algal abundance, bacteria seem not to be favored by the low WL, as the biomass of bacteria in 1996 was the lowest among the studied years (Fig. 1). In years of high WL (1994, 1995, 1998) the highest summer peaks of BP occurred between the two PP peaks, while in ‘low-level’ year 1996 BP increased only after the collapse of the last PP peak. Reasons for such phenomena are not clear, we can only speculate the possibility that bacteria are more disturbed in low-water years when strong resuspension of sediments occurs and neither pelagic nor sediment bacterial communities are able to develop normally. Further investigations are needed to study this hypothesis. No correlation between biomass of bacteria and BP was found in low-water year 1996 (r ¼ 0:018, p > 0:05), showing the low viability of the population whereas in high-water years relatively stronger correlation between these two parameters (r ¼ 0:35, p < 0:05) occurred (also in accordance with better coupling between bacterial and primary productivity). Another possible cause of the benefit of phytoplankton but disadvantage for bacteria due to low WL could be explained by increased competition for inorganic nutrients between bacteria and algae. However, this was not shown in a nutrient amendment study [32], therefore we reject hypothesis of nutrient competition between algae and bacteria in this nutrient rich lake. 4.3. Availability of organic substrate for bacterial production Integrated BP and PP were uncoupled during most of the investigation period (linear correlation was statistically insignificant) and seemed not directly related with each other (Fig. 3). Even when assuming that phytoplankton production could provide most of the substrate for bacteria, values of gross BP and PP are out of balance. By using an average BGE of 25% [18] in calculations of bacterial substrate requirements it is obvious that most daily integrated carbon values of gross BP in summer are often 2–3 times higher than those of PP even though phytoplankton productivity was high during the observation period. The assumed bacterial

growth efficiency of 25% may be even too high (e.g. [18,33]), and a lower BGE of <20% (total range of <4– 60%) will lead to even higher estimates of bacterial C demand in L. V~ ortsj€arv. It should be mentioned that a certain proportion of PP is immediately released as dissolved organic carbon (DOC) but, typically this proportion is less than 40% and it might decrease to less than 10% in eutrophic conditions [34]. In L. V~ ortsj€ arv the proportion of DOC in total PP was estimated to be up to 37% [35]. This contribution of DOC to the pool of bacterial growth substrates is not sufficient to cover the observed gap between gross PP and BP. On a seasonal scale and in general terms, however, bacterioplankton activity was coupled with PP, but a lag period was often evident. During the productive period BP generally followed PP by a time lag (from several weeks to month) while the winter peaks of BP were always independent form PP. This is clearly illustrated by the dynamics of the ratio of BP and PP (BP/PP), which seems to be an informative index showing rather good correlation with the dynamics of algal biomass (Fig. 5). BP/PP ratio was very low in early spring (March–April) and during the diatom bloom while after the collapse of cyanobacteria in late summer/autumn the BP/PP ratio typically increased. A common feature was that Chla peaks always matched with low values of BP/PP and vice versa. Such reciprocal dynamics of BP/PP and Chla may refer to an uncoupling of algal and bacterial productivity when PP becomes available to bacteria only after the decay of algal cells. The ability of L. V~ ortsj€ arv bacteria to benefit of the polymeric algal material released during decay was also demonstrated by Kisand and Tammert [8]. Analysis of the seasonal dynamics of productivity/biomass and CLPP (BIOLOG GN substrates) revealed that a highly productive bacterial community preferentially utilized simple carbohydrates and in terms of genetic diversity, the summer community was much less diverse as compared to the spring bloom community. Apparently, active bacteria of few different species may express similar physiological response to the same substrates. Further detailed studies would be directed to show if these few bacteria represent opportunistic populations typically bursting in response to the availability of labile substrates during short periods in summer. In contrast, bacterioplankton biomass was more closely related to genetic diversity (more variability was described by CCA1-‘biomass’ axis, Fig. 6c) than to CLPP. Considering other sources of organic carbon, L. V~ ortsj€arv is believed to have relatively low PP of littoral and bottom macrophytes since the reed belt and other macrophytes cover 18.8% of the lake area [36] and contribution of macrophytes to the total PP in this lake has been estimated to be about 15% [37]. Another allochthonous (non-pelagic) source of organic carbon are humic substances from the watershed, which are

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

transported into the lake through 18 rivers and streams. The transport of refractory organic matter is highest at spring flood (Kisand and N~ oges, unpublished data) but bacterioplankton seems to benefit only little from this source of carbon and energy because relatively fresh allochthonous material did not boost the BP in spring. Moreover, degradation of such compounds might be of less importance in freshwater than in marine systems [38], especially when the residence time is relatively short (1 year in L. V~ ortsj€ arv). However, focused studies on the importance of allochthonous organic matter are needed in future.

59

4.4. Grazing pressure on bacteria by ciliates Besides resource availability, grazing is the major known mechanism for controlling bacterioplankton dynamics. The main grazers on bacteria in Lake V~ ortsj€arv are abundant ciliates [5,39], (Zingel, unpublished data), which can control bacterioplankton biomass in this lake. The correlation between ciliate biomass, bacterial abundance and production was significant over the several studied years (Fig. 4, r ¼ 0:28 and 0.27, respectively, p < 0:05). The strongest topdown control on the bacterial community was assumed

Fig. 5. Ratio of BP/PP (d – line of moving average) compared with Chla concentration (N – Chla, dashed line of moving average) in L. V~ ortsj€arv in 1993–98.

Fig. 6. Functional and genetic fingerprints of bacterioplankton during one season in L. V~ ortsj€arv and relationship with production/biomass: (a) PCRDGGE gel, (b) CCA plot of DGGE bands and BP/BB (independent variables indicated as arrows); ( – scores of dates, dashed line for succession,  – scores of DGGE bands, OTUs not given) and (c) CCA plot of CLPP and BP/BB (independent variables indicated as arrows); (scores of dates, dashed line for succession; scores of BIOLOG ‘key substrates’: a – glucuronamide; b – a-keto valeric acid; c – phenylalanine; d – glycyl-L -aspartic acid; e – citric acid; f – succinic acid; g – a-D -glucose; h – L -rhamnose; i – cellobiose; j – L -glutamic acid; k – fructos.)

60

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

in 1995 and 1996 because of high biomass of bacterivorous ciliates. After low WL period in 1996 the ciliate community declined and the years with the high bacterial biomass followed. The lacking relationship between biomass and production of bacteria also indicates the prevalence of top-down control [40]. Comparing the ecosystems of two large shallow lakes – L. Peipsi and L. V~ ortsj€ arv [41], much higher biomass of protozooplankton, and the higher BP/PP ratio at rather low bacterial biomass was observed in L. V~ ortsj€ arv. This may indicate the more important role of microbial loop in food web of this lake which is more turbid and subjected to stronger wind driven resuspension compared to less turbid L. Peipsi. 4.5. L. V~ ortsj€ arv among other shallow lakes In a similar large eutrophic shallow lake (L. Neusiedlersee, Austria) lowest BP occurs in spring (March– May) or late fall (November–December), and summer BP values range between 100 and 300 mgC l
1 day
1 [42]. However, winter values of BP were lower compared to L. V~ ortsj€ arv where variation in BP was more pronounced. Abundance of bacteria was significantly higher in L. Neusiedlersee. In another eutrophic shallow but small lake (L. Vallentunasj€ on, Sweden) bacterial biomass was higher (147–492 lgC l
1 ) [43] than in L. V~ ortsj€ arv (the highest recorded biomass 228 lgC l
1 ). Bacterioplanton of L. V~ ortsj€ arv seems to have lower biomass than other similar shallow lakes which production is similar pointing to the higher grazing pressure in L. V~ ortsj€ arv. In L. Neusiedlersee bacterioplankton is to a large extent dependent on the PP of macrophytes [42]. There is very little data available to compare the bacterial community composition and dynamics in shallow lakes. As in L. V~ ortsj€ arv bacterioplankton community composition showed clear clustering between winter, spring, summer and autumn in the turbid hypertrophic and shallow lake (L. Blankaart, the Netherlands) [44]. Also in L. Blankaart the richness of OTU’s was highest in spring when bacterial biomass was relatively low. Seasonal differences in L. Blankaart were explained mostly by the presence or absence of macrophytes, unimportant factor in L. V~ ortsj€ arv.

5. Conclusions For further evaluation of the significance of bacterioplankton in energy flow of L. V~ ortsj€ arv, it is important to evaluate the realistic proportions of carbon entering the ecosystem from all allochthonous sources. A first glimpse into the genetic and functional diversity of planktonic bacteria in L. V~ ortsj€ arv, presented in this paper, gives a challenge to study similarities and the

differences in bacterial community structures of other habitats (e.g. sediment bacteria, attached communities on particles) of this lake. However, methodological problems in applying BIOLOG GN plates (e.g [45]) and PCR-DGGE method (e.g. [46]) have to be taken into account. Furthermore, PCR-DGGE is not a fully quantitative method, therefore care should be taken when interpreting the band intensities. Better understanding of short time dynamics of plankton communities of polymictic lakes will be important to evaluate the prevailing processes during short stratification periods when the weather is calm. To estimate properly the grazing pressure by protozooplankton, the seasonal measurements of grazing rates would allow better estimations of loss of bacterial biomass and carbon flux to other heterotrophs. In conclusion our observations support idea that (1) bacteria are important link in carbon cycle of L. V~ ortsj€arv (as an example of large, temperate lakes) introducing additional carbon into the pelagic system by utilizing ‘allochthonous’ DOM, and (2) a significant part of organic matter of generally filamentous algae of L. V~ ortsj€arv are accessible to other heterotrophs after their decay via bacteria. We concluded that differently from deep stratified eutrophic lakes, in turbid and shallow eutrophic systems the classical food chain is not prevailing. Another implication of such shallow lakes is that degradation of organic compounds in the wellmixed aerobic water column is faster and more efficient while in deep stratified lakes large pools deposited in sediments are subject to slow anaerobic degradation. Further long-term studies of bacterioplankton dynamics (and microbial loop) on other similar large shallow lakes are needed to draw more general conclusions and fit the results into general PEG-model [47] concept.

Acknowledgements We would like to thank Helen Tammert and Kristjan Piirim€ae for field and lab assistance. We are also grateful to Priit Zingel for ciliate data, to Jaak Truu for statistical advice, to Hans-Peter Grossart and Meinhard Simon for useful comments on an earlier version of the manuscript. Funding for this research was provided by the project No. 0362480s03 of the Estonian Ministry of Education, by the Estonian Science Foundation (grants 4080 & 5738), by the European Commissions’s Environment and Sustainable Development Program under contract EVK1-CT-2002-00121 (CLIME), and the Foreign Exchange Foundation of the Estonian Academy of Sciences. The data on water level were obtained from Estonian Institute of Hydrology and Meteorology. Hydro-chemical and biological data collection was partly supported by the State Monitoring Program of the Estonian Ministry of Environment, and by Finnish

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

Ministry of Environment. V. Kisand has been partly supported by the Hanse Wissenschaftskolleg (Delmenhorst, Germany). References [1] N~ oges, T. (1998) Cladoceran grazing in Lake V~ ortsj€arv. Limnologica 28, 67–74. [2] N~ oges, T. and N~ oges, P. (1998) Primary production of Lake V~ ortsj€ arv. Limnologica 28, 29–40. [3] Kivimaa, R., Huttula, T. and Podsetchine, V. (1998) Hydrodynamical studies. In: Present State and Future of Lake V~ ortsj€arv (Huttula, T. and N~ oges, T., Eds.), pp. 60–78. Pirkanmaa Regional Environment Centre, Tamepere. [4] N~ oges, T. (1999) The effect of extreme water level decrease on hydrochemistry and phytoplankton in a shallow eutrophic lake. Hydrobiologia 408/409, 277–283. [5] Zingel, P. (1999) Pelagic ciliated protozoa in a shallow eutrophic lake: community structure and seasonal dynamics. Arch. Hydrobiol. 146, 495–511. [6] P~ ollum€ ae, A. and Haberman, J. (1998) The effect of fluctuating water level on the zooplankton of lake V~ ortsj€arv, Central Estonia. Proc. Eston. Acad. Sci. 47, 259–267. [7] N~ oges, T. and K€ unnap, H. (2003) V~ ortsj€arve zooplanktoni toitumine ja roll pelaagilistes toiduahelates. In: Eesti Looduseuurijate Seltsi Aastaraamat (M€ ols, T., Haberman, J., Kongo, L., Kukk, E. and M€ ols, E., Eds.). TA Kirjastus, Tallinn. [8] Kisand, V. and Tammert, H. (2000) Bacterioplankton strategies for leucine and glucose uptake after a cyanobacterial bloom in an eutrophic shallow lake. Soil Biol. Biochem. 32, 1965–1972. [9] Jeppesen, E., Sortkjaer, O., Soendergaard, M. and Erlandsen, M. (1992) Impact of a Trophic Cascade on Heterotrophic Bacterioplankton Production in Two Shallow Fish-manipulated Lakes, pp. 219–231 Schweizerbart’sche Verlagsbuchhandlung, Stuttgart (FGR). [10] Simon, M., Bunte, C., Schulz, M., Weiss, M. and Wuensch, C. (1998) Bacterioplankton Dynamics in Lake Constance (Bodensee): Substrate Utilization, Growth Control, and Long-term Trends, pp. 195–221 Schweizerbart’sche Verlagsbuchhandlung, Stuttgart (FGR). [11] Padisak, J. (1998) Sudden and Gradual Responses of Phytoplankton to Global Climate Change: Case Studies from Two Large, Shallow lakes (Balaton, Hungary, and the Neusiedlersee, Austria/ Hungary). Kluwer Academic Publishers, Dordrecht. [12] Kisand, V. and N~ oges, T. (1998) Seasonal dynamics of bacterioand phytoplankton in large and shallow eutropic Lake V~ ortsj€arv. Estonia. Inter. Rev. Hydrobiol. 83, 205–216. [13] Wetzel, R.G. and Likens, G.E. (1991) Limnological Analyses, pp. 391 Springer Verlag, New York. [14] Bergstr€ om, I., Hein€anen, A. and Salonen, K. (1986) Comparison of acridine orange, acriflavine and bisbenzimide stains for enumeration of bacteria in clear and humic waters. Appl. Environ. Microbiol. 51, 664–667. [15] Norland, S. (1993) The relationship between biomass and volume of bacteria. In: Microbial Ecology (Kemp, P.F., Sherr, B.F., Sherr, E.B. and Cole, J.J., Eds.), pp. 303–307. Lewis Publishers, Boca Raton, Fla. [16] Simon, M. and Azam, F. (1989) Protein content and protein synthesis rates of planktonic marine bacteria. Mar. Ecol. Prog. Ser. 51, 201–213. [17] Biddanda, B., Ogdahl, M. and Cotner, J. (2001) Dominance of bacterial metabolism in oligotrophic relative to eutrophic waters. Limnol. Oceanogr. 46, 730–739. [18] del Giorgio, P.A. and Cole, J.J. (1998) Bacterial growth efficiency in natural aquatic systems. Annu. Rev. Ecol. Syst. 29, 503–541.

61

[19] Teske, A., Wawer, C., Muyzer, G. and Ramsing, N.B. (1996) Distribution of sulfate-reducing bacteria in a stratified fjord (Mariager Fjord, Denmark) as evaluated by most-probablenumber counts and denaturing gradient gel electrophoresis of PCR-amplified ribosomal DNA fragements. Appl. Environ. Microbiol. 62, 1405–1415. [20] Guidelines for the Baltic monitoring programme for the third stage Baltic marine environment protection commission. 1988. In: Helsinki: Helsinki Commission; pp. 101–111. [21] Strickland, J.D.H. and Parsons, T.R. (1972) A Practical Handbook of Seawater Analysis, pp. 310 Bulletin of the Fisheries Research Board of Canada, Ottawa. [22] Uterm€ ohl, H. (1958) Zur Vervollkommung der quaatitativen Phytoplanktonmethodik. Mitt. Int. Ver. Theor. Angew. Limnol. 9, 1–38. [23] Patterson, D.J. and Hedley, S. (1992) Free-living Freshwater Protozoa. In: A color guide, p. 223. Wolfe Publishing Ltd, Aylesbury. [24] Foissner, W. and Berger, H. (1996) A user-friendly guide to the ciliates (Protozoa, Ciliophora) commonly used by hydrobiologists as bioindicators in rivers, lakes, and waste waters, with notes on their ecology. Freshwater Biol. 35, 375–482. [25] Finlay, B.J. (1977) The dependence of reproductive rate on cell size and temperature in freshwater ciliated Protozoa. Oecologia 30, 75–81. [26] Foissner, W., Berger, H. and Schaumburg, J. (1999) Identification and Ecology of Limnetic Plankton ciliates, 793 pp. Bayerisches Landesamt f€ ur Wasserwirtschaft, M€ unchen. [27] ter Braak, C.J.F. (1986) Canonical correspondence analysis: a new eigenvector techniques for multivariate direct gradient analysis. Ecology 67, 1167–1179. [28] ter Braak, C.J.F. and Verdonschot, P.F.M. (1995) Canonical correspondence analysis and related multivariate methods in aquatic ecology. Aquat. Sci. 57, 255–289. [29] J€arvet, A. and N~ oges, P. (1998) Research area and period. In: Present State and Future Fate of Lake V~ ortsj€arv. Results from Finnish–Estonian joint project in 1993–1997 (Huttula, T. and N~ oges, T., Eds.), pp. 11–30. The Finnish Environment. [30] N~ oges, P. and Kisand, A. (1999) Horizontal distribution of sediment phosphorus in shallow eutrophic Lake V~ ortsjarv (Estonia). Hydrobiologia 409, 167–174. [31] N~ oges, T., N~ oges, P. and Laugaste, R. (2003) Water level as the mediator between climate change and phytoplankton composition in a large shallow temperate lake. Hydrobiologia 506, 257–263. [32] Kisand, V., Tuvikene, L. and N~ oges, T. (2001) Role of phosphorus and nitrogen for bacteria and phytopankton development in a large shallow lake. Hydrobiologia 457, 187–197. [33] Anderson, T.R. and Ducklow, H.W. (2001) Microbial loop carbon cycling in ocean environments studied using a simple steady-state model. Aquat. Microb. Ecol. 26, 37–49. [34] Baines, S.B. and Pace, M.L. (1991) The production of dissolved organic matter by phytoplankton and its importance to bacteria patterns across marine and freshwater systems. Limnol. Oceanogr. 36, 1078–1090. [35] N~ oges, T. (1999) Different fractions of primary production in the large shallow eutrophic Lake V~ ortsj€arv, Estonia. Proc. Eston. Acad. Sci. Biol. Ecol. 48, 91–106. [36] Feldmann, T. and N~ oges, P. (2001) Polarized distribution of aquatic macrophytes in a large shallow lake. In: Proceedings of Ninth International Conference of the Conservation and Management of Lakes, Session 4, pp. 58–61. [37] N~ oges, P., Laugaste, R., and N~ oges, T. (2003) F€ utoplankton. In: V~ ortsj€arv. Loodus, aeg, inimene (Haberman, J., Phiu, E., Raukas, A., Eds.), Eesti Ents€ uklopeediakirjastus, pp. 261–277.

62

V. Kisand, T. N~oges / FEMS Microbiology Ecology 50 (2004) 51–62

[38] Wikner, J., Cuadros, R. and Jansson, M. (1999) Differences in consumption of allochthonous DOC between a lake and an estuary in a temperate watershed. Aquat. Microb. Ecol. 17, 289– 299. [39] Kisand, V. and Zingel, P. (2000) Dominance of ciliate grazing on bacteria during spring in a shallow eutrophic lake. Aquat. Microb. Ecol. 22, 135–142. [40] Ducklow, H.W. (1992) Factors regulating bottom-up control of bacteria biomass in open ocean plankton communities. Arch. Hydrobiol. Ergebn. Limnol. 37, 207–217. [41] N~ oges, T., Haberman, J., Kisand, V., Laugaste, R. and Zingel, P. (2001) Trophic relations and food web structure of plankton community in lake Peipsi s.s. In: L. Peipsi Flora and Fauna, Tallinn (Pihu, E., Haberman, J., Eds.), pp. 74–81. [42] Reitner, B., Herzig, A. and Herndl, G.J. (1999) Dynamics in bacterioplankton production in a shallow, temperate lake (Lake Neusiedl, Austria): evidence for dependence on macrophyte production rather than on phytoplankton. Aquat. Microb. Ecol. 19, 245–254.

[43] Lindstr€ om, E.S. (2000) Bacterioplankton community composition in five lakes differing in trophic status and humic content. Microb. Ecol. 40, 104–113. [44] Muylaert, K., Van der Gucht, K., Vloemans, N., De Meester, L., Gillis, M. and Vyverman, W. (2002) Relationship between bacterial community composition and bottom-up versus topdown variables in four eutrophic shallow lakes. Appl. Environ. Microbiol. 68, 4740–4750. [45] Haack, S.K., Garchow, H., Klug, M.J. and Forney, L.J. (1995) Analysis of factors affecting the accuracy, reproducibility and interpretation of microbial community carbon source utilization patterns. Appl. Environ. Microbiol. 61, 1458–1468. [46] Suzuki, M., Rappe, M.S. and Giovannoni, S.J. (1998) Kinetic bias in estimates of coastal picoplankton community structure obtained by measuring of small-subunit rRNA gene PCR amplicon lenght heterogeneity. Appl. Environ. Microbiol. 64, 4522–4529. [47] Sommer, U., Gliwicz, W., Lampert, W. and Duncan, A. (1986) The PEG-model of seasonal succession of planktonic events in fresh waters. Arch. Hydrobiol. Beih. 106, 433–472.