Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
MICROBIOLOGY ECOLOGY ELSEVIER FEMS Microbiology Ecology 17 (1995) 257-270 Microbial and classical food webs: A visit to a hypertrophic lake R. Sommaruga * University of InnsbrucS Institute of Zoology and Limnology, Technikerstr. 25, A-6020 Innsbruck, Austria Received 8 February 1995; revised 27 April 1995; accepted 9 May 1995 Abstract Plankton community structure and fluxes of carbon for bacteria (production and bacterivory) were investigated in the urban, hypertrophic Lake Rod6 (Uruguay) using a short time interval for sampling (5-15 d) during one year. The lake sustains a high phytoplankton biomass (up to 335 pg I- ’ chlorophyll a) always dominated by the filamentous cyanobacteria PZunktot/zrti agardhii. The zooplankton community was numerically dominated by rotifers and ciliates; cladocerans were rare during most of the year. The rotifer abundance was very high (up to lo5 individual 1_ ’ ), the bacterivorous Anurueopsis fissa being the most abundant species. Predation rates of heterotrophic nanoflagellates (HNF) on bacteria (range: 31-130 bacteria HNF-1 h- ‘) were higher than those reported in the literature for field studies. A carbon budget showed that HNF can consume on average 91 and 76% of the bacterial carbon production in summer and winter, respectively. Bacterial turnover times are the lowest reported until now from field conditions (5 to 42 h). Consequently, bacterial carbon production was extremely high (72 to 1071 pg C 1-l d- ‘). Bacterial production was positively correlated to bacterial abundance but the relationship was significantly improved by the inclusion of temperature (82% variability explained). My results support the general trend for increased bacterial production with increasing trophic status, and suggest a lower energy transfer efficiency to higher trophic levels in hypertrophic lakes due to the many trophic interactions involved. Keywords: Planktonic food web; Bacterial production; Bacterivory; Carbon flow; Hypertrophic 1. Introduction Hypertrophic lakes, the last stage in the trophic state classification, are becoming common in several parts of the world as a consequence of increasing water pollution [l]. Barica [2] has characterized hypertrophic lakes as shallow systems with high retention times, unbalanced nutrient and oxygen regimes, extreme fluctuation in nutrients, high productivity and a relatively low abundance of zooplankton due * Corresponding author. Tel: +43 (512) 507-6121; (512) 507-2930; E-mail: [email protected] 0168-6496/95/$09.50 0 1995 Federation SSDI 0168-6496(95)00030-5 of European Fax: +43 Microbiological lake to non-grazeable filamentous phytoplankton. Studies on plankton from those systems have been made less frequently than in other lakes of different trophic state [3-51. The microbial food web of hypertrophic lakes has received little attention and information is scarce, available only on a small number of systems 16-91. Eutrophication produces several direct and indirect structural changes in the pelagic food web, notably to the phytoplankton community [lo]. Although the abundance and biomass of the different microbial components increases with increasing nutrient enrichment [11,12] the importance of the microbial loop in the C flux along the productivity Societies. All rights reserved 2.58 R. Sommaruga / FEMS Microbiology Ecology I7 Cl995) 25 7-270 gradient compared to the “classical” grazing food chain (phytoplankton-zooplankton link) is still uncertain. While Riemann et al. [13] supported the idea that the importance of the microbial loop increases with increasing nutrient loading, Weisse [14] stated the opposite. Recently Weisse and Stockner [15] reviewed this topic and argued that although microbial food webs dominate the carbon flow in oligatrophic systems they can be the most important component again under very eutrophic or hypertrophic conditions when Daphnia is absent. The role of Daphnia for an efficient energy transfer from the microbial food web to higher trophic levels has been demonstrated for mesotrophic and eutrophic lakes [16,17]. The exclusion of Duphnia from the plankton has been attributed to the dominance of colonial or filamentous cyanobacteria in the phytoplankton community [18] or to a strong planktivory by fish [19]. In lakes with low abundance of cladocerans the link between picoplankton and macrozooplankton is given by flagellates, ciliates and rotifers but with a lower energy transfer efficiency. Based upon data collected over one year, this paper presents the structure and dynamics of the microbial food web components in a hypertrophic lake. To understand the carbon flow in the pelagic system of this lake the structure of the phytoplankton and zooplankton was investigated as well. Furthermore, the factors controlling bacterial production and the predation impact of heterotrophic flagellates are discussed in relation to previous models. 2. Material and methods 2.1. Description of the system and sampling strategy Lake Rod6 situated in the city of Montevideo, Uruguay (34”55’S, 56”lO’W) is small and shallow (area = 1.3 ha, maximum depth = 1.1 m). There are no limnological studies on this urban lake except the knowledge of its evergreen water color and sporadic observations .of phytoplankton. Lake Rod6 has no stream inputs but receives stormwater discharge from an ill-defined urban area. Water overflows from the lake via an intermittent outlet only at high water levels. The lake contains fish, principally Cnesterodon decemmaculatus and Cichlasoma sp. Aquatic macrophyte growth is sparse due to the shading from the algal blooms but small pockets of Potumogeton sp. occur. Thirty-six samplings were made at the deepest part of the lake from January 21 to December 28, 1992, at intervals of 5-15 d. The euphotic zone corresponding to the first 50 cm of the water coIumn was sampled with a 5 liter Schindler-Patalas sampler. Due to the existence of physical gradients in the first centimeters of the water column, water was mixed 4 times before taking the subsamples always in the same order. Sampling was carried out between 10:00 and 12:00 h. Bacterial production and bacterivory experiments were made at the laboratory within 30 min after the samples were collected. On each sampling date, water transparency, photosynthetically active radiation, temperature, pH, conductivity, oxygen, total nitrogen and total phosphorus were measured to characterize the physical and chemical conditions of the water-column. 2.2. Abundance and biomass measurements Chlorophyll a was measured spectrophotometritally following extraction with hot ethanol [20]. Phytoplankton samples were preserved with acid Lugol’s Iodine and enumerated using either a Sedwick-Rafter chamber (for counting trichomes) or a sedimentation chamber of 2-5 ml at 200 X magnification (for total counts). In the former case, triplicate counts were made. In the case of Planktothrix 400 trichomes were measured at 200 x magnification from a pool made with all samples. Nauplii and rotifers were counted together with ciliates in the sedimentation chamber. When nauplii and rotifer numbers were lower than 50 individuals, between 0.2 and 1.0 1 of lake water was filtered through a 45 pm mesh (Nytex, Switzerland) and then poured into a Bogorov chamber and counted at 200 X magnification. Rotifers were identified to the genus or species level on live or fixed samples. Macrozooplankton were collected with a Patalas trap, filtered through a nylon mesh of 100 pm and fixed with sucrose-formalin (final concentration 4%). The organisms were counted in a Bogorov chamber at 40 X magnification, and identified with appropriate keys. Bacterial samples were fixed with formalin (final concentration 2% v/v) and stained with the fluo- R. Sommaruga / FEMS Microbiology Ecology 17 (1995) 257-270 259 rochrome DAPI [21]. More than 400 bacteria were counted on black membrane filters (Poretics 0.2 pm) at 1250 X magnification in a Nikon epifluorescence microscope with a BP 365, FT 395, and LP 397 filter set. The size of the bacteria was determined with an image analysis system as described in Psenner [22]. Cell volumes were computed with the following formula: ments using appropriate geometric formulas. I used a single conversion factor of 110 fg C pm-” [26] although changes in ciliate cell volume after fixation with mercuric chloride are variable for different species [27]. V=(w2x~/4)x(I-ww)+(~xw~/6) Natural bacteria from the lake were grown in sterile lake water supplemented with a barley seed and harvested during the log phase determined by following the absorbance at 650 nm. Then the heatkilled bacteria were stained with 5-([4,6-dichlorotriazin-2-yl]amino)fluorescein (DTAF) according to Sherr et al. [28]. The fluorescently-labeled bacteria (FLB) had a mean length of 0.96 k 0.34 pm and a mean width of 0.4 + 0.1 pm with an equivalent mean cell volume of 0.104 pm3. Grazing experiments were conducted on 18 of the 36 sampling dates. Experiments were done independently for flagellates and ciliates under in situ conditions. FLB at a concentration of ca. 15% of the in situ abundance of bacteria were added for ciliates and ca. 25% for flagellates. Subsamples of 20 ml were taken during 25 min at 3-5 min time intervals and fixed with alkaline Lug01 (final concentration 0.5%), followed by the immediate addition of borate-buffered formalin (final concentration 3%; [29]). The samples were kept refrigerated (5°C) and dark until analyses within 24 h. Preserved samples were stained with DAPI, filtered onto 1 pm black membrane filters (Poretics) and washed 3 times with cold particle-free tap water. DAPI-stained organisms were located under the epifluorescent microscope at 400 X (ciliates) and 1250 X (flagellates) under UV light (365/395/397 nm) and FLB inside vacuoles were counted using the blue-light filter set (450490/510/520 nm) at 1250 X magnification. Flagellates were checked for the presence of plastids as described above. Labeled food concentrations were estimated from counts of DTAF-stained cells from the to samples under blue excitation. The uptake rate was estimated from a linear regression between time and the mean number of FLB ingested per individual. Linear regressions with a r2 < 0.9 were not included in the results. Flagellates without FLB ingestion were included in the calculation. Clearance (1) where V= volume ( ,cLm3), w = width ( pm) and 1 = length (pm). Accuracy of the algorithms used was confirmed with independently calibrated microspheres (Molecular Probes, USA) of known size (M. Kriip bather pers. comm.). Biovolumes were transformed into biomass using an allometric relationship [23] expressed in the formula: c = 0.12 x v”.72 (2) where: V= cell volume ( pm31 and C = carbon (pg cell - ’ ). The abundance of active (INT-reducing) bacteria was determined according to Zimmermann [24]. Glycerin was always used as the mounting medium. Samples for counting HNF were treated in the same way as described for bacteria. Each individual was checked with a BP 490, FT 510, and LP 520 filter set for the presence of autofluorescence i.e. presence of plastids. Length and width of individuals were measured at 1600 X magnification with a calibrated ocular micrometer and used to calculate the volume assuming appropriate geometrical formulas. The mean cell volume was used to estimate biomass using a conversion factor of 220 fg C pump3 [25]. Ciliate samples were fixed immediately with 1 ml of a saturated solution of HgCl, and enumerated using the inverted microscope technique. Depending on ciliate density, 5 to 10 ml aliquots were settled for 24 h and the entire surface of the settling chamber was examined at 200 X . Two drops of Bromophenol Blue (0.04% w/v) were added to help to differentiate the individuals. Total counts always exceeded 100 individuals. Ciliate biomass was estimated by multiplying the abundance by mean cell volume in 3 different size classes: < 20 pm, 20-35 pm and > 35 pm calculated from direct volume measure- 2.3. Bacterivory ments and bacterial production experi- 260 R. Sommaruga / FEMS Microbiology rates (CR) were estimated tion: using the following CR = UR/FLB equa- (3) where: UR = uptake rate in (FLB ingested individual -’ h- ‘> and FLB = concentration of FLB nl- ‘. The [ 14C]leucine method [30,31] was used to estimate bacterial production. [ I4C]Leucine (Amersham, UK) of 310 mCi mmol-’ specific activity was added to 3 vials containing 10 ml lake water at a final concentration of 20 nM. Saturating concentrations of Leucine were determined experimentally by measuring incorporation rates at various concentrations of substrate. Except for one occasion (December: 40 nM), saturation was reached at 20 nM. All samples were corrected for abiotic incorporation by subtracting the radioactivity in formalin-killed controls (2% final concentration). Protein was extracted in cold trichloroacetic acid (final concentration 5%). Filters were washed with 80% cold ethanol [32]. The moles of leucine incorporated per hour and carbon production were calculated using the formula developed by Simon and Azam [31], assuming a 2-fold intracellular dilution. Cell production was estimated using the estimated empirical conversion factor. Bacterial turnover times were computed as the ratio of bacterial abundance (cells 1-l ) and bacterial production (cells 1-l h-l). 2.4. Empirical conversion factor An empirical was estimated in pling, 100 ml of pressure through conversion factor for [ 14Clleucine November. Immediately after samlake water were filtered under low a 1.0 pm pore size filter and mixed Table 1 Salient physico-chemical information Ecology 17 (1995) 257-270 with 400 ml of the same water filtered through a 0.2 pm filter. Bacterial numbers and [14C]leucine incorporation were followed during 42 h at 20°C (annual mean = 18.9”C) using the procedures described above. A conversion factor of 12.9 X 1016 cells mol-’ was derived using the approach of Ducklow and Hill [33]. 2.5. Carbon budget The budget was made from estimated and calculated values averaged over the summer and winter periods using the following conversions factors and assumptions: Carbon phytoplankton biomass was estimated only for Planktothrix using the formula C = 0.11 X (V) [34]; biomass of Anuraeopsis fissa -the most abundant bacterivorous rotifer -was calculated assuming a C content of 0.009 pg C individual-’ [35]; biomass of crustaceans was estimated using the formula for species or pooled equations described in Dumont et al. [36] and converted to carbon assuming a C:dry weight ratio of 1:2; bacterial biomass were estimated using the allometric approach Eq. 2. The amount of carbon ingested by flagellates was calculated from community uptake rates converted into carbon units using the respective mean bacterial cell volumes and Eq. 2. For Anuraeopsis fissa the same procedure was followed except that a mean ingestion rate (1283 bacterial individual-’ h- ‘) was derived from values given in Ooms-Wilms [37] for summer experiments made in Lake Loosdrecht (minimum: 1020 bacterial individmaximum: 1546 bacterial ual -1 h-1. individual-’ h- ‘). of Lake Rod6 Parameter Mean Range Temperature PC) Oxygen saturation 18.9 105 8.4 552 0.20 0.53 0.43 2.90 8.5 8.2-28.0 50-170 8-9 400-740 0.15-0.30 0.50-0.60 0.16-1.0 1.10-4.7 1.6-15.0 (%) PH Conductivity ( &G cm-’ ) Transparency (m) 1% Photosynthetically active radiation (m) Total phosphorus (mg 1-l ) Total nitrogen (mg 1-l ) Total nitrogen: total phosphorus (by weight) Values are from O-O.5 m of the water column (n = 36). R. Sommaruga / FEMS Microbiology Ecology 17 (1995) 257-270 261 2.6. Statistics Correlations among variables were made using the non-parametric Spearman rank test. Where necessary, the data were log,, transformed to equalize the variance over the range of observations and to meet the normality assumptions of least-squares regression analysis. Normality of the variables and residual values from each regression were assessed using the Kolmogorov-Smirnov test. For multiple standardized partial regression coeffiregressions, cients ( p coefficients) were calculated to provide an indication of the impact of the residual unexplained variance of each independent variable on the dependent variable. 300 n- ‘E 250 200 - E 150 - _m 100 - 6 50 ” J F M A M J JASON0 MONTHS 3. Results 3.1. Physical water-column and chemical characteristics of the Values for the main physical and chemical features of Lake Rode are presented in Table 1. Due to the shallowness of the lake, the water-column was mixed most of the time. However, during the day in summer it was stratified with a maximum difference of 3°C between surface and 0.60 m. During the night and depending on wind speed, the water-column was completely mixed. The lake was well oxygenated except during summer when a steep decrease in oxygen concentration was observed with anoxic conditions at the bottom. The TN:TP ratio showed that N might be the limiting nutrient most of the time. 3.2. Components of the classical food web: phytoplankton and zooplankton Of 27 species of phytoplankton found, the cyanobacteria Planktothr~ agardhii dominated all year around representing between 82 and 99% of the total abundance of phytoplankton. Euglenales, mainly Euglena, Trachelomonas and Phacus, were the second most abundant taxonomic group. A complete list of phytoplankton species is available from the author upon request. The mean number of trichomes of P. agardhii was 4.2 X lo7 1-l. The minimum biovolume was observed in winter (3 X 10” pm3 1-l) and Fig. 1. Temporal development of Planktozhrir biovolume (a) and chlorophyll a (b) in the first 50 cm of the water-column of Lake Rod& the maximum in summer (1.5 X 1011 pm3 1-r; Fig. la). The mean chlorophyll a value was 223 mg me3 with a minimum value of 99 mg me3 in autumn and a maximum of 335 mg m -’ in summer (Fig. lb). A positive correlation was found between the abundance of P. agardhii and chlorophyll a (rs = 0.527, P < 0.001). Due to space restriction figures are shown only for the main taxa. The Calanoida (Fig. 2g) were represented by one species, Notodiaptomus incompositus, with abundances ranging from 0.1 to 23.5 individuals 1- ’ for adults and from 0.1 to 44.0 individuals I- ’ for copepodites. Maximum abundances of both adults and copepodites were observed at the end of fall and in spring. The most abundant cyclopoid copepod was Tropocyclops prasinus meridionalis, ranging from 1 to 53 individuals 1-l for adults and 0.3 to 31 individuals 1-l for copepodites (Fig. 2h). Copepods, both adults and copepodites, were very often observed to be colonized by the peritrichous ciliate Epistylis sp. Nauplii were very abundant (Fig. 2f) with densities between 50 and 11800 individuals 1-l. Maximum abundance was observed in spring. Cladocerans (Fig. 2i) were represented by three species of which Moina micrura was the most abundant with a maximum den- R. Sommaruga / FEMS Microbiology Ecology 17 (19951257-270 262 600 400 _ 200 'i 8 6 a8 .E 4 0 : Ciliates > 35 pm 0) :: S 40 iii 20 0 a 200 5 100 i 0 / ri .+._%_A&, 12008 8000 i Nauplii A Calanoid copepods 200 ~-~- Cladbcerans 100 I 0 b ----~-. _ -*-- L.. JFMAMJJASONO MONTHS Fig. 2. Changes in abundance (0) and biomass (0) of components of the plankton in the first 50 cm of the water-column in Lake Rod6 sity of 194 individuals 1-l. Both Daphnia parvula and Diaphanosoma birgei were present at low abundances (< 3 individuals 1-l). All three species reached their maximum abundance in spring, disappearing by December. Rotifers were the most abundant group of metazooplankton with densities between 1200 and 129600 individuals 1-l (Fig. 2e). The most abundant rotifer was Anuraeopsis spp. mainly A. fissa who reached a maximum of 123 200 individuals 1- ’ at the end of the summer. 3.3. Components of the microbial food web The abundance of HNF ranged between 1.1 X lo6 and 3.0 X 10’ 1-l with a mean abundance of 1.2 X 10’ 1-l (Fig. 2a). HNF density reached the maximum at the beginning of fall and the minimum in winter. The size of HNF ranged between 2 and 10 pm length. Biomass ranged between 8.4 and 579 pg C 1-l with a mean value of 129 pg C 1-l (Fig. 2a). Maximum biomass was observed in summer and minimum in winter (Fig. 2a). HNF abundance was correlated with the bacterial abundance (r, = 0.518, P < 0.01, n = 36). The mean abundance of ciliates was 31700 individuals 1- ’ with a minimum of 3600 and a maximum of 84 000 individuals 1-l. Ciliates in the size class from 20 to 35 pm were the most abundant with densities ranging from 1300 to 68 600 individuals 1-l (Fig. 2~). Dominant species in this size class were Halteria sp., Askenasia sp. and Strobilidium sp. High abundances were observed in summer (February) and spring (October) while the lowest occurred in winter (July). Ciliates in the size class < 20 pm were numerous only in summer with a maximum abundance of 27 000 individuals l- ’ (Fig. 2b) and dominated by Urotricha sp. Ciliates > 35 pm were abundant at the end of summer, in winter and in spring (Fig. 2d) but present in lower densities (O-24 200 individuals l-‘). Taxa representing this size class were Coleps hirtus, Vorticella sp., Paradileptus sp., Frontonia sp., and Didinium spp. Maximum abundance at the end of the summer was made by Vorticella sp. and Didinium spp. The winter increase in numbers was due to Paradileptus sp. which together with Frontonia, Didinium spp. and a Paramecium type were responsible for the spring peak in abundance. Total ciliate biomass ranged from 4.3 to 288 pg C 1-l. With the exception of January when ciliates < 20 pm contributed 33% to the total biomass, the main contribution during the rest of the year was made by the larger size classes (Fig. 2b-d). Bacterial abundance ranged from 1.5 X lo9 1-l to 2.0 X 10” I-’ with a mean of 7 X lo9 1-l (Fig. 3a). Summer values were an order of magnitude higher than winter values. More than 95% of the bacteria were free-living. Mean cell volumes (MCV) ranged from 0.065 to 0.546 pm3. Mean cell volume was inversely correlated to temperature (r, = - 0.865, P < 0.001, n = 30). Biomass values ranged from 80.2 to 443 pug C 1-l (Fig. 3a). Very long bacteria of up to 210 pm length accounted for 4-16% of bacterial abundance. A more detailed study of these bacteria larger than their predators will be published elsewhere. The number of INT-reducing bacteria oscillated between 0.4 and 6 X lo9 1-l accounting for 16.5-100% of the total bacterial R. Sommaruga / FEMS Microbiology Ecology I7 (1995) 257-270 263 abundance (Fig. 3b). Interestingly, autotrophic picoplankton was not found in Lake Rod6 and therefore heterotrophic bacteria are the only food source in the picoplankton size range. 3.4. Bacterial $/Q?/dLyq secondary production The [‘4C]leucine incorporation follows a very clear seasonal pattern with the maximum value in summer (14.4 nmol 1-i h-i) and minimum in winter (1.1 nmol 1-i h- ‘). Accordingly, bacterial carbon production ranged from 72 to 1071 pug C 1-l d-’ with a mean of 587 ,ug C 1-i d- ’ (Fig. 3~). Temperature was positively correlated with leucine incorporation rates (r, = 0.900, P < 0.001, n = 36). Turnover times were extremely fast and highly variable (5-42 h; Fig. 3d). 3.5. Bacteriuory by protists JFMAMJJASOND MONTHS Fig. 3. Temporal development of bacterial abundance and biomass and percentage of INT-reducing bacteria (b), bacterial carbon production (c) and turnover time (d) in the first 50 cm of the water-column in Lake Rod& (a), abundance Table 2 Summary Date of estimated uptake and clearance Uptake rate (bacteria HNF-’ h- i) rates by heterotrophic A summary of estimated grazing parameters is shown in Table 2. Uptake rates by HNF ranged from 31 to 130 bacteria HNF-’ h-’ with higher values in summer. Corresponding clearance rates (3.7 to 29 nl HNF-’ hh’) were higher in winter and inversely nanoflagellates (HNFI, January 29-December 28, 1992 in Lake Rod6 Clearance rate (nl HNF-’ h-i) Water column cleared dd’ (%I Bacterial cell production consumed (%I Jan 29 Feb 12 Feb 19 Mar 19 Apr 02 May 08 May 28 Jun 25 JuI 24 Aug 20 Sep 10 Sep 24 Ott 15 Ott 27 Nov 12 Dee 01 Dee 14 Dee 28 82 39 40 104 42 38 38 34 59 40 31 58 60 40 59 112 121 130 5.4 3.7 5.1 22.1 5.1 13.9 12.9 22.5 28.9 13.2 19.4 11.5 8.9 7.8 4.8 9.2 10.3 17.6 269 95 199 415 110 289 196 131 164 252 550 245 148 531 178 294 440 624 95 35 42 53 51 31 75 19 79 122 174 61 44 100 106 112 116 107 Mean 63 12.3 285 79 264 R. Sommaruga / FEMS Microbiology correlated to temperature (r, = - 0.529, P < 0.05). The percentage of the bacterial standing stock removed per day was very high with values equivalent to 625% of the standing stock. Community uptake rates were correlated with the abundance of bacteria and temperature (I, = 0.794 and 0.781 respectively, P < 0.001). When predation impact by HNF was compared to the bacterial production the percentage consumed ranged from 19 to 174. Predation experiments with ciliates were not successful. In some cases, FLB were observed inside vacuoles but uptake was highly variable, making estimations impossible. The rotifer Anurueopsis was observed with FLB as well as with pigments (autofluorescence) inside. 4. Discussion 4.1. The dominance and its consequence of filamentous cyanobacteria for the zooplankton When lakes change from mesotrophic to eutrophic or hypertrophic conditions the diversity of the phytoplankton community decreases [38]. During this trophic shift, cyanobacteria become dominant in the phytoplankton assemblage both in terms of abundance and biomass. In some cyanobacterial lakes, Planktothrix agardhii (formerly Oscillatoria) has become dominant over the green algae, diatoms and N,-fixing cyanobacteria [39]. Virtual monocultures of this species are typical for the most hypertrophic phase of a lake and described as a ‘climax species’ with a peak in abundance in autumn and summer [40]. However, under some conditions, it can be found the whole year round [41]. This is the case in Lake Rod6 where P. agardhii is the dominant phytoplankton species representing always more than 82% of the total abundance. P. agardhii has an extremely effective light-capturing mechanism, chromatic adaptation, a low energy requirement for maintenance and a particularly efficient transfer of photosynthetically-fixed carbon to growth under lowirradiance levels [3]. Zevenboom and Mur [38] have proposed a hypothetical model to explain phytoplankton succession in lakes of increasing P-loading. The model shows conditions under which P. agardhii becomes favored, notably an increase in self Ecology 17 (1995) 257-270 shading, a lower a,,/~,,, ratio, a lower light-energy availability and N-limiting conditions. Light is limiting in Lake Rod6 as the euphotic zone is restricted to ca. 0.5 m and for most of the time (except during day in summer) the water column is totally mixed and therefore the z,,/z, ratio is low. Optimum growth conditions of this species - both in cultures and in lakes -are observed when the N:P ratio is 5-9 [41]. Forsberg et al. [42] have reported that TN:TP values < 10 are N-limiting, > 17 are P-limiting, and ratios between 10 and 17 may be either N-or P-limiting or both. In Lake Rod& TN:TP ratios show that N is the primary limiting nutrient (Table 1) and that maximum abundance of Planktothrix occurred at the time when N deficiency was very strong (TN:TP value ca. 4). An important consequence for the planktonic food web structure of Lake Rod6 is the inability of zooplankton to graze trichomes of Plunktothrix (> 100 pm long). Although the proportion of other phytoplankters is small their absolute abundance (1.3 X lo’--4.2 X lo6 I-’ represents an important food source. However, this resource may not be fully available due to the mechanical interference of Planktothrix for zooplankton grazers. Experimental and field results indicate that high concentrations of filamentous algae affect the nutrition of herbivorous zooplankton [43] and reduce drastically the reproduction and growth rates of cladocerans [18]. A noticeable characteristic of the zooplankton community of Lake Rod6 is the absence of cladocerans for most time of the year with only a short peak in spring, mainly made by Moina micrura. Infante [43] suggested that a long period of absence of Moina micrura, Ceriodaphnia and Diaphanosoma in Lake Valencia (Venezuela) may have been caused primarily by the presence of high abundances of cyanobacteria like Oscillatoria and Lyngbya. While this argument may hold for most part of the year in Lake Rod& the drastic disappearance of cladocerans in December can not be attributed solely to the interference of Planktothrix as abundance of trichomes in October and December were not markedly different. A possible explanation may be the fact that from December to March schools of the planktivorous fish Cnesterodon decemmaculatus (Poeciliidae, Cyprinodontiformes; adult size 3 cm) were observed and captured inside the water sampler (up to 10 individu- R.Sommaruga/FEMSMicrobiologyEcology 17(1995) 257-270 als). The impact may be enhanced in summer when owing to low oxygen concentrations at depth zooplankton distribution is mostly restricted to the upper 0.5 m of the water column, making them more vulnerable for fish predation. In brief, the crustacean zooplankton community of Lake Rod6 is subjected to a particularly harsh form of trophic interactions caused by a heavy predation pressure from planktivores and the dominance of non-edible algae. Rotifers became dominant among the metazooplankton and bacterivorous species like Anuraeopsis fissa reach very high abundance (10’ individuals l- ‘). 4.2. A highly dynamic microbial food web with several trophic interactions In addition to rotifers, the ciliate assemblage also dominated the zooplankton in terms of abundance. As well as rotifers, ciliates of all three size classes reached the highest abundance in summer, declining later in winter and showing the clear effect of temperature on their seasonal dynamics. The abundance found in Lake Rod6 was below the range of 90000215 000 individuals l- ’ found for a set of hypertrophic lakes in Florida with chlorophyll a values > 56 mg me3 [lo]. The mean annual abundance in Lake Rod6 was 32000 individuals 1-r and the maximum 84000 individuals 1-l. A possible explanation for this discrepancy could be a high predation pressure on the ciliate community by the zooplankton community as was found in several studies and summarized by Gifford [44]. Some indications of possible predator-prey interactions among rotifers and the smaller ciliates of different size classes are clearly visible at the end of February and in October (Fig. 2). Didinium is a taxon well known to feed on other ciliates. In Lake Rod6, the seasonal oscillation of HNF abundance was very pronounced changing by one order of magnitude between winter and summer (Fig. 2a). Compared to Priest Pot (England), a hypertrophic system but with lower phytoplankton biomass (range Chlorophyll a: 32-41 mg rnp3 [12]) mean abundance of HNF in Lake Rod6 was lower (4.3 X lo7 individuals 1-r and 1.7 X lo7 individuals l-‘, respectively). However, HNF biomass reached very high values of up to 579 pg C 1-r although the means calculated with the same conversion factor 265 were similar (Lake Rod& 129 and Priest Pot: 118 pg C I-‘). As shown above, it seems that a high predation pressure at this trophic level is exerted by ciliates. A reduction in predation pressure on HNF by nanociliates and ciliates in the size class 20-35 pm when they were concurrently preyed upon by large ciliates and possibly larger zooplankton can be observed in March and October (Fig. 2). Berninger [12] has argued that in productive systems the increased predation pressure keeps prey populations well below the carrying capacity, and my results support this idea. HNF were the main consumers of bacterial standing stock (95-625% removed per day) with very high individual uptake rates of up to 130 bacteria HNF-i hh’. The last value is similar to those found in experiments with cultures [45] and 17-times higher than the average uptake rate for HNF found in freshwaters studies [46]. Such extreme values (> 100%) have also been found by others [47]. However, this calculation gives only a rough idea or probably a wrong figure of the bacterivory impact because clearance rates [48] as well as HNF and bacterial abundances [49] differ significantly during the day. Therefore, an extrapolation of one value from a finite time measurement to 24 h can lead to under or overestimates depending on the time the bacterivory experiments were done. A comparison to bacterial production is more appropriate and with the exception of one value (174%), the percentage of bacterial production consumed in Lake Rod6 was close to or under 100%. A very important point to understand the overall impact of the HNF on bacteria is the existence of very long bacterial cells that escape to predation. Therefore, the high values of predation impact found in Lake Rod6 do not necessarily imply a collapse of the bacterial assemblage. I have also used the model developed by Vaqut et al. [46] to estimate total grazing rates (their GT) of the community (including organisms other than HNF). This model uses temperature and the abundance of HNF and bacteria as independent variables and explain 53% of the variability of GT (data set for the whole temperature range: 1.5-32°C). Using mean values of these variables for Lake Rod6 (18.9”C, 11749 HNF ml-’ and 6.9 X lo6 bacteria ml-‘) the model predicts a lower value (log GT = 5.1) than observed (5.9). The difference may be due to other 266 R. Sommaruga /FEMS Microbiology Ecology 17 (19951257-270 factors accounting for the variability included in the model. 4.3. Factors production controlling bacterial of GT and not abundance and 106 Bacterioplankton production and abundance have typically been compared with primary production, chlorophyll a, extracellular organic carbon production, and other variables in an attempt to demonstrate how growth is regulated in situ. Bird and Kalff [50] have found chlorophyll a to be correlated with bacterial abundance along a trophic gradient. Cole et al. [51] have presented bottom-up (resource control) regressions relating bacterial production to chlorophyll a and primary production for marine and freshwater systems of different trophy. However, these studies did not include systems with chlorophyll values > 200 mg me3. Introducing new data from systems of higher trophy may help to evaluate the generalization of previous models. In Lake Rod6 bacterial abundance was significantly correlated to chlorophyll a (r, = 0.55, P < 0.01) but the use of the mean chlorophyll value (223 mg mP3) to predict the number of bacteria with the equations of Bird and Kalff [50] and Cole et al. [51] (5.6 X 10’ and 1.7 X lo7 ml-‘, respectively) gave higher values than the observed mean (6.9 X lo6 bacteria ml- ‘). Although more data in the upper chlorophyll range are needed, my results and those from Robarts and Wicks [8] indicate that there is a decline in bacterial numbers relative to chlorophyll concentration along the trophic gradient. Bird and Kalff [50] suggested that this may be a necessary consequence if bacterial productivity per unit bacterial biomass increases at a faster rate with increasing trophic status than primary production per unit chlorophyll. However, Robarts and Wicks [8] did not support this idea as they found in hypertrophic Hartbeespoort Dam (South Africa) a very low bacterial production (maximum 46 pg C 1-l d- ‘> although chlorophyll and primary production are the highest recorded in the literature ( > 6500 mg mm3 and 8886 mg C mm3 h-‘, respectively). Conversely, my results indicate that there is an increase in bacterial production with increasing trophy although the model of Cole et al. [51] underestimate the observed values. In Lake Rod6 bacterial production is very high with a mean value of 587 pg C 10’ BACTERIAL 104 Planktothrix ABUNDANCE 2.104 3.104 ABUNDANCE (cells ml -1) 5.104 7.104 (thricomes 105 ml -1) Fig. 4. Bottom-up correlations in Lake Rod6 between leucine incorporation and bacterial abundance (a) and Planktothrti abundance (b). 1-l d&’ and a maximum of 1071 pg C 1-l d-l well above the highest value registered (302 pg C 1~’ d- ‘> in a survey of 275 data from 22 freshwater systems [52]. This corresponds well to the rapid bacterial turnover times (5-42 h) and the high percentage of active bacteria registered (16-100%). Moreover, considering the number of INT reducing bacteria instead of the total number, the turnover times will be in some cases even faster. The incorporation of leucine was significantly correlated with bacterial abundances as well as with Planktothrix abundances (Fig. 4a and b). A possible explanation to understand the contradictory findings in Lake Rod6 and Harbeespoort Dam may be differences in phytoplankton species dominance. In Hartbeespoort Dam Microcystis aeruginosa is the dominant autotroph producing very dense scums in the top meters. Although one would expect a high substrate supply from this high autotrophic biomass Robarts and Zohary [53] have argued that bacteria were substrate limited, based on enrichment experiments and cell size. In fact the size of bacteria was very small with a MCV = 0.013 Frn3 [8]. Beside the arguments on substrate limitation there is also some evidence that the toxin produced by Microcystis aeruginosa affects the survival of bacteria and flagellates [54] as well as bacterivory rates (Sommaruga et al., unpublished). Although Ro- R. Sommaruga / FEMS Microbiology barts and Wicks [8] did not reported abundance or uptake rates of HNF they assumed that flagellate bacterivory was low. Another important factor to explain variability in bacterial production within and among systems is temperature. The role of temperature on laboratory bacterial cultures is well known. Recently, Shiah and Duklow [55] found temperature to be the most important factor regulating bacterial production, abundance and growth rate in Chesapeake Bay. Furthermore, Pomeroy and Wiebe [S6] have stressed the important role of temperature as a main controller of bacterial growth in aquatic systems. Also, temperature is an important modulator of the bacterial cell size. Hagstriim and Larsson [57] found that at a constant bacterial growth rate higher temperatures produce smaller cells. The analysis of the relationship between bacterial production and abundance was significantly improved by the inclusion of in situ temperature as an additional independent variable. The following equation was found for Lake Rod& 1ogBP = 0.37 + 0.16 log NB + 0.042T (n=36, r* = 0.816) (4) where BP = bacterial production ( ,ug C 1-l dd’); NB = number of bacteria (cells 1-l ) and T = temperature CC>. White et al. [52] have also found a better prediction level of bacterial production when temperature was included in their freshwater data analysis. The variability explained by the independent variables was 2-times higher in Lake Rod6 than that found by White et al. [52] ( r2 = 0.37). The standardized re- Table 3 Major carbon pools for the planktonic Lake Rod6 community and fluxes of bacterial gression coefficients for Lake Rod6 indicate that the impact of temperature on production is greater than that of abundance (/3 = 0.805 and 0.035, respectively). White et al. [52] have also found a greater impact of temperature on bacterial production with the exception of marine data where the reverse was true. The inclusion of chlorophyll and Planktothrti abundance in the case of Lake Rod6 as substitutive or additional parameters did not improve the percentage of variability explained. Also the replacement of the number of bacteria by the number of active (INT) bacteria did not increase the variability explained. The results indicate that temperature is an important positive modulator of bacterial production on a seasonal as well as probably on shorter timescales and support the findings of White et al. [52] for freshwaters. 4.4. Carbon flow in the euphotic zone The carbon budget is far from being complete and the number of assumptions make it far from perfect. I am especially aware of the consequences in extrapolating production or bacterivory results from one hour to one day. Nevertheless, the budgets for summer and winter illustrates relevant features from the system (Table 3). In summer, HNF are able to crop 91% of the BP per day and have a biomass almost equivalent to the bacteria. This implies a rapid bacterial growth to supply the amount of C required by the HNF. The carbon pool of macrozooplankton is very much reduced compared to other highly eutrophic lakes like Frederiksborg Slotsso (715-1206 carbon production Summer Planktothrix Bacteria Flagellates Ciliates Anuraeopsis Macrozooplankton Values in brackets represent a n.d., not determined. 267 Ecology 17 (1995) 257-270 and bacterivory during summer and winter in Winter Pool (/.LgCI_‘) Flux (FgCl-‘d-l) Pool (pgC1-1) Flux (PgCl-‘d-l) 13700 240 220 81 246 56 n.d. a 950 865 (91%) n.d. 20 (2%) n.d. 5000 1.51 29 40 8 26 n.d. 148 113 (76%) n.d. 2.2 (1.5%) the percentage of bacterial carbon production consumed. md. See Material and methods for explanations. 268 R. Sommaruga / FEMS Microbiology Ecology 17 (1995) 257-270 pg c I-‘, [71) as a consequence of the suspected heavy fish predation and poor food availability (inedible phytoplankton). The rotifer pool, here only exemplified by its most abundant species Anuraeopsis, dominated the rnetazooplankton community not only in abundance but also in biomass. However, their estimated grazing impact on bacteria at maximum abundance accounts only for 2% of the BP compared to 91% consumed by HNF. The 7% of BP left in summer may be consumed by ciliates or lost by other pathways (e.g. viral Iysis). In winter, all the pools were reduced and the bacteria:HNF biomass ratio was 5:l. Nevertheless, in this period the HNF crop 76% of the BP per day while Anuraeopsis consumes only 1.5%. The scenarios depicted here indicate that the channeling of energy from the primary producers to higher trophic levels through the ‘classical food web’ is reduced in this hypertrophic lake. Conversely, the microbial food web is the main component for conveying energy to higher trophic levels although probably less efficiently due to the higher number of steps involved and consequent energy losses. Also the absence of keystone species like Daphnia able to graze efficiently on preys of very different size leads to energy sink rather than transfer. The data here presented indicate a highly dynamic and active microbial food web supporting the ideas of Riemann [13] and Weisse and Stockner [15]. The results gathered from this study enlarge our knowledge about food webs in systems stressed by nutrient enrichment. Acknowledgements I thank Roland Psenner, Cohn Reynolds, John Stockner and two anonymous reviewers for comments made on a earlier version of the manuscript. I am grateful to Lizet deLe6n, Daniel Fabisn, Daniel Conde and Ma. Laura Arin for helping during sampling and plankton counting and to Eugen Rott for the determination of Planktothrix. This work was supported by a grant from CSIC Univ. de la RepGblica (Uruguay), the North-South Dialogue Programme (Austria) and the Tonolli Foundation @IL). References [I] Barica, J. and Mur, L.R. (Eds.) (1980) Hypertrophic Ecosystems. Developments in Hydrobiology Vol. 2. Dr. W. Junk. [2] Barica, J. (1980) Why hypertrophic ecosystems. In: Hypertrophic Ecosystems (Barica, J. and Mm. L.R., Eds.), pp. ix-xi. Developments in Hydrobiology Vol. 2. Dr. W. Junk. [3] Reynolds, C.S. (1984) The Ecology of Freshwater Phytoplankton, 1st Edn., 384 pp. Cambridge University Press, Cambridge. [4] Carney, H.J. (1990) A general hypothesis for the strength of food web interactions in relation to trophic state. Verh. internat. Verein. Limnol. 24, 487-492. [.5] Romo, S. and Miracle, M.R. (1993) Long-term periodicity of Planktothrix agardhii, Pseudanabaena gale&a and Geitlerinema sp. in a shallow hypertrophic lagoon, the Albufera of Valencia (Spain). Arch. Hydrobiol. 126, 469-486. [6] Finlay, B.J., Clarke, K.J.. Cowling, A.J., Hindle, R.M., Rogerson, A. and Berninger, U.G. (1988) On the abundance and distribution of protozoa and their food in a productive freshwater pond. Eur. J. Protistol. 23, 205-217. [7] Christoffersen, K, Riemann, B., Hansen, L.R., Klysner, A. and Sorensen, H.B. (1990) Qualitative importance of the microbial loop and plankton community structure in a eutrophic lake during a bloom of cyanobacteria. Microb. Ecol. 20. 253-272. [S] Robarts, R.D. and Wicks. R.J. (1990) Heterotrophic bacterial production and its dependence on autotrophic production in a hypertrophic African reservoir. Can. J. Fish. Aquat. Sci. 47, 1027-1037. [9] Berninger, U.G., Wickham, S.A. and Finlay, B.J. (1993) Trophic coupling within the microbial food web: a study with fine temporal resolution in a eutrophic freshwater ecosystem. Freshw. Biol. 30, 419-432. [lo] Carney, H.J. and Elser, J.J. (1990) The strength of zooplankton-phytoplankton coupling in relation to trophic state. in: Large lakes: Ecological structure and Function (Tilzer, M.M. and Serruya, C., Eds.), pp. 616-631. Springer Verlag. [ll] Beaver. J.R. and Crisman, T.L. (1989) The role of ciliated protozoa in pelagic freshwater ecosystems. Microb. Ecol. 17, 111-136. [12] Berninger, U.G., Findlay, B.J. and Kuuppo-Leinikki, P. (1991) Protozoan control of bacterial abundances in freshwater. Limnol. Oceanogr. 36, 139-146. [13] Riemann, B., Sondergaard, M., Persson, L. and Johansson, L. (1986) Carbon metabolism and community regulation in eutrophic, temperate lakes. In: Carbon dynamics in eutrophic, temperate lakes (Riemann, B. and Sonderggard, M., Ed%), pp. 267-280. Elsevier. [141 Weisse. T. (1991) The microbial food web and its sensitivity to eutrophication and contaminant enrichment: a cross-system overview. Int. Revue. ges. Hydrobiol. 76, 327-337. [151 Weisse, T. and Stockner, J. (1994) Eutrophication: the role of microbial food webs. In: Proc. V Internat. Conf. Conservation and Management of lakes, pp. 191-194. Stresa, Italy. t161 Pace, M.L., McManus. G.B. and Finlay. SE. (1990) Plank- R. Sommaruga / FEMS Microbiology tonic community structure determines the fate of bacterial production in a temperate lake. Limnol. Oceanogr. 35, 795808. [17] Wylie, J.L. and Currie, J. (19911 The relative importance of bacteria and algae as food sources for crustacean zooplankton. Limnol. Oceanogr. 36, 708-728. [18] Sterner, R.W. (1989) The role of grazers in phytoplankton succession. In: Plankton Ecology (Sommer, U., Ed.), pp. 107-170. Springer Verlag. [19] Persson, L., Andersson, S., Hamrin, F. and Johansson, L. (1988) Predator regulation and primary production along a eutrophication gradient of temperate lake ecosystems. In: Complex interactions in lake communities (Carpenter, S.R., Ed.), pp. 45-65. Springer Verlag. [20] Nusch, EA. (1980) Comparison of different methods for chlorophyll and phaeopigments determination. Arch. Hydrobiol. Beih. Ergeb. Limnol. 17, 14-36. [21] Porter, K.G. and Feig, Y.S. (1980) The use of DAPI for identifying and counting aquatic microflora. Limnol. Oceanogr. 25, 943-948. [22] Psenner, R. (1993) Determination of size and morphology of aquatic bacteria by automated image analysis. In: Handbook of Methods in Aquatic Microbial Ecology (Kemp, P., Sherr, B., Sherr, E. and Cole, J., Eds.1, pp. 339-346. Lewis. [=I Norland, S. (1993) The relationship between biomass and volume of bacteria. In: Handbook of Methods in Aquatic Microbial Ecology (Kemp, P., Sherr, B., Sherr, E. and Cole, J., Eds.1, pp. 303-308. Lewis. [241 Zimmermann, R.R., Iturriaga, R. and Becker-Birck, J. (19781 Simultaneous determination of the total number of aquatic bacteria and the number thereof involved in respiration. Appl. Environ. Microbial. 36, 926-935. D51 Borsheim, K.Y. and Bratbak, G. (1987) Cell volume to carbon conversion factors for a bacterivorous Monas sp. enriched from sea water. Mar. Ecol. Prog. Ser. 36, 171-175. bl Turley, C.M., Newell, R.C. and Robins, D.B. (1986) Survival strategies of two small marine ciliates and their role on regulating bacterial community structure under experimental conditions. Mar. Ecol. Prog. Ser. 33, 59-70. [271Wiackoski, K., Donie, A. and Fyda, J. (1994) An empirical study of the effect of fixation on ciliate cell volume. Mar. Microb. Food Webs. 8, 59-69. [281Sherr, B., Sherr, E. and Fallon, R.D. (1987) Use of monodispersed, fluorescently labeled bacteria to estimate in situ protozoan bacterivory. Appl. Environ. Microbial. 53, 958965. [291 Sherr, B., Sherr, E. and Pedros-AIi6, C. (1989) Simultaneous measurement of bacterioplankton production and protozoan bacterivory in estuarine water. Mar. Ecol. Prog. Ser. 54, 209-219. [301Kirchman, D.L., K’Nees, E. and Hodson, R. (1985) Leucine incorporation and its potential as a measure of protein synthesis by bacteria in natural aquatic systems. Appl. Environ. Microbial. 49, 599-607. [311Simon, M. and Azam, F. (19891 Protein content and synthesis rates of planktonic marine bacteria. Mar. Ecol. Prog. Ser. 51, 201-213. Ecology I7 (1995) 257-270 269 [321 Wicks, R.J. and Robarts, R. (1988) Ethanol extraction requirement for purification of protein labeled with 13H] Leucine in aquatic bacterial production studies. Appl. Environ. Microbial. 54, 3191-3193. [331Ducklow, H.W. and Hill, S.M. (1985) Tritiated thymidine incorporation and the growth of heterotrophic bacteria in warm core rings. Limnol. Oceanogr. 40, 260-268. [341Rocha, 0. and Duncan, A. (1985) The relationship between cell carbon and cell volume in freshwater algal species used in zooplanktonic studies. J. Plankton Res. 7, 279-294. A.L., Van Tongeren, O.F.R., [351Gulati, R.D., Ooms-Wilms, Postema, G. and Siewertsen, K. (19921 The dynamics and role of limnetic zooplankton in Loosdrecht lakes (The Netherlands). Hydrobiologia 233, 69-86. 1361Dumont, H.J., Van de Velde, I. and Dumont, S. (1975) The dry weight estimate of biomass in a selection of Cladocera, Copepoda and Rotifera from the plankton, periphyton and benthos of continental waters. Oecologia 19, 75-97. 1371Ooms-Wilms, A.L. (19911 Ingestion of fluorescently labelled bacteria by rotifers and cladocerans in Lake Loosdrecht as measures of bacterivory: Preliminary results. Mem. 1st. ital. Idrobiol. 48, 269-278. [381Zevenboom, W. and Mur, R. (1980) N,-fixing cyanobacteria: why they do not become dominant in Dutch, hypertrophic lakes. In: Hypertrophic Ecosystems (Barica, J. and Mur, L.R., Eds.), pp. 123-130. Developments in Hydrobiology Vol. 2., Dr. W. Junk. of Oscilluforia agardhii [391Berger, C. (19751 Occurrence Gomont in some shallow eutrophic lakes. Verh. int. Ver. Limnol. 19, 2689-2697. [40] Chorus, I. and Schlag, G. (1993) Importance of intermediate disturbances for the species composition and diversity of phytoplankton in two very different Berlin lakes. Hydrobiologia 249, 67-92. [41] Liere van, L. and Mur, L.R. (1980) Occurrence of Oscillotoria agardhii and some related species: a survey. In: Hypertrophic Ecosystems (Barica, J. and Mur, L.R., Eds.), pp. 67-78. Developments in Hydrobiology Vol. 2. Dr. W. Junk. 1421 Forsberg, C., Ryding, S.-O., Claesson, A. and Forsberg, D.A. (1978) Water chemical analyses and/or algal assays? Sewage effluent and polluted lake water studies. Mitt. int. Ver. Limnol. 21, 352-363. [43] Infante, A. and Riehl, W. (1984) The effect of Cyanophyta upon zooplankton in a eutrophic tropical lake. Hydrobiologia 113, 293-298. [44] Gifford, D.J. (19911 The Protozoan-Metazoan trophic link in pelagic ecosystems. J. Protozool. 38, 81-86. 1451 Fenchel, T. (19821 Ecology of heterotrophic microflagellates. II. Bioenergetics and growth. Mar. Ecol. Prog. Ser. 8, 225231. [46] Vaque, D., Gasol, J.M. and Marrase, C. (1994) Grazing rates on bacteria: the significance of methodology and ecological factors. Mar. Ecol. Prog. Ser. 109, 263-274. [47] Scavia, D. and Laird, G.A. (1987) Bacterioplankton in Lake Michigan: Dynamics, controls, and significance to carbon flux. Limnol. Oceanogr. 32, 1017-1033. [48] Christoffersen, K. (1994) Do heterotrophic nanoflagellates 270 [49] [50] [51] [521 1531 R. Sommaruga / FEMS Microbiology feed on picoplankton in rhytms? Mar. Microb. Food Webs 8 (l-2), 111-124. Psenner, R. and Sommaruga, R. (1992) Are rapid changes in bacterial biomass caused by shifts from top-down to bottomup control? Limnol. Oceanogr. 37, 1092-1100. Bird, D.F. and Kalff, .I. (1984) Empirical relationships between bacterial abundance and chlorophyll concentration in fres and marine waters. Can. J. Fish. Aquat. Sci. 41, 10151023. Cole, J.J., Findlay, S. and Pace, M.L. (1988) Bacterial production in fresh and saltwater ecosystems: a cross overview. Mar. Ecol. Prog. Ser. 43, l-10. White, P.A., Kalff, J., Rasmussen, J.B. and Gasol, J.M. (1991) The effect of temperature and algal biomass on bacterial production and specific growth rate in freshwater and marine habitats. Microb. Ecol. 21, 99-118. Robarts, R. and Zohary, T. (1986) Influence of cyanobacte- Ecology 17 (1995) 257-270 [541 1551 t561 1571 rial hyperscum on heterotrophic activity of planktonic bacteria in a hypertrophic lake. Appl. Environ. Microbial. 51, 609-613. Christoffersen, K. (1994) Mycrocystin concentrations of natural populations of cyanobacteria and effect in microbial food web structure. Abstract ASLO & PSA Meeting, p. a-17. Shia, F.K. and Ducklow, H. (1994) Temperature regulation of heterotrophic bacterioplankton abundance, production, and specific growth rate in Chesapeake Bay. Limnol. Oceanogr. 39, 1243-1250. Pomeroy, L.R. and Wiebe, W.J. (1988) Energetics of microbial food webs. Hydrobiologia 159, 7-18. Hagstrom, A,. and Larsson, U. (1984) Die1 and seasonal variation in growth rates of pelagic bacteria. In: Heterotrophic activity in the sea (Hobbie, J.E. and LeB Williams, P.J. Eds.), pp. 249-262. Plenum.