AQUATIC MICROBIAL ECOLOGY Vol. 14: 19-28.1998 Published January 2 l Aquat Microb Ecol l

Dynamics of heterotrophic bacteria attached to Microcystis spp. (Cyanobacteria)

Jakob worrnlr*, Morten S0ndergaard2

'Section of Genetics and Microbiology, Department of Ecology and Molecular Biology. The Royal Veterinary and Agricultural University, Thorvaldsensvej 40, DK-1871 Frederiksberg C,Denmark 2Freshwater Biological Laboratory, University of Copenhagen, Helsingersgade 51, DK-3400 Hillered, Denmark

ABSTRACT. We studied the ecology of heterotrophic bacteria attached to the mucilaginous colonies of Microcystis spp. (Cyanobacteria) in the eutrophic lake Frederiksborg Slotsse, Denmark. The succession in bacterial abundance, production and potential aminopeptidase activity in 20 pm fractionated sam- ples was followed during periods in which Microcystis dominated the phytoplankton. We operationally defined that nets of 20 pm mesh-width segregated bacteria associated with Mjcrocystis (Microcystis- associated bacteria, MB; s20 pm size fraction) from the mainly free-living bacteria in the filtrate (FB, <20 pm size fraction). According to this definition, the contribution of MB during summer 1995 and autumn 1994, respectively, averaged 10 i 4 (i standard deviation) and 37 i 12% of total bacterial bio- mass and 25 i 13 and 43 i 16% of total bacterial production, as estimated from thymidine (TdR) incor- poration. During summer, MB further contributed 55 i 18% of total leucine incorporation measured at 600 nM leucine and 53 i 12% of total potential aminopeptidase activity. Although 20 pm mesh-width nets also retained particles other than Microcystis, our results indicate that Microcystis was a 'hotspot' for bacterial activity, comparable to larger aggregates known as marine or lake snow. During summer, growth rate and specific aminopeptidase activity of MB generally exceeded those of FB, which points to diversified microenvironments or species compositions. In order to balance gain and loss rates within the community of MB, we hypothesize that a large fraction of MB produced were exported from Micro- cystis to the surrounding water, only modified by the loss due to viral lysis. This idea arose from reported low loss rates of Microcystis and continuous measures indicating that a surplus of more than 70% of MB production (TdR) was not reflected as biomass increases within the community of MB. According to this hypothesis Microcystis may be considered as a bacterial 'incubator' for the surround- ing water.

KEYWORDS: Attached bacteria Mjcrocystis . Eutrophic lake . Microhabitat

INTRODUCTION In general, large particles in aquatic systems are colonized by bacteria at densities exceeding the sur- The cyanobacterium Microcystis is a common genus rounding water on a per volume basis; however, in eutrophic lakes and often dominates the phyto- attached bacteria usually contribute less than 10% of plankton during summer and autumn. Microcystis total bacterial abundance (Kirchman 1993). Particles forms large (>l00 pm) mucilaginous colonies usually are argued to be favourable microenvironments com- colonized by (heterotrophic) bacteria (Whitton 1973). pared to the surrounding water, because availability of Bacteria associated with Microcystis are hypothesized potential substrates to bacterial growth is facilitated to live in a microhabitat distinct from the surrounding by organic components of particles, adsorbed matter, water, because the microscale physicochemical envi- and, possibly, an increased flux of dissolved nutrients ronment is influenced by the particulate nature of during sinking (reviewed by Fletcher 1991, Kirchman Microcystis and the metabolism of the autotrophic and 1993). Further, Herndl (1988) suggested that short- heterotrophic organisms present. circuit nutrient cycles may form when autotrophic and heterotrophic organisms are closely associated in par- 'E-mail: [email protected] ticulate matter (marine snow), thus increasing the

O Inter-Research 1998 20 Aquat Microb Ec

regenerative power and, potentially, the overall micro- ria, FB). We operationally defined that 20 pm mesh- bial activity. width nets separated MB (>20 pm) and FB (c20 pm) to Microcystis is a distinct bacterial microhabitat com- make it feasible to obtain continuous measures of bac- pared to a detrital particle, because the attached bac- terial abundance and activity. Based upon this, the teria are closely associated with the principal producer microhabitat of MB is characterized as a 'hotspot' for of organic carbon for bacterial growth. The potential bacterial activity with overall significance in. bacterial for microenvironmental conditions is demonstrated as carbon processing and population dynamics during elevated pH within the mucilage of axenic photosyn- blooms of Microcystis. thesizing Microcystis colonies (Richardson & Stolzen- bach 1995). Respiring bacteria counteract the chemical changes in 02, CO2 and pH induced by photosynthesis MATERIALS AND METHODS and a mutualistic relationship between cyanobacteria an'd associated (heteroti-ophic)bacteria has frequently Study site. The study was carried out in Frederiks- been suggested (Whitton 1973), but to our knowledge borg Slotsss, Denmark, during autumn 1994 and no studies have yet evaluated the diffusive regime of summer 1995. The lake (mean depth: 3.1 m) is highly Microcystis including attached bacteria. Intracellular eutrophic with a dense bloom of colony forming cyano- gas vacuoles enable Microcystis to ascend the water bacteria during summer and autumn, when chloro- column at speeds up to 3.2 m d-' (Reynolds 1973, Rey- phyll concentrations may exceed 100 pg 1-' (Andersen nolds & Rogers 1976), which prevcfits sedimentation & Jacobsen 1979, Christoffersen et al. 1990). and increases the exchange of nutrients with the water Sampling. Water from 0 and 0.5 m depth was taken surrounding the colonies. The buoyancy is counter- from the middle of the lake and mixed in equal balanced by accumulated carbohydrates during photo- volumes either at noon (autumn) or in the morning synthesis, and to some extent the diel light cycle drives (summer). vertical up and down migrations (Thomas & Walsby Subsamples of 10 m1 for bacterial abundance and 1986). Loss of Microcystis from the water column is activity were fractionated with 20 pm mesh-width also reduced, because grazing by zooplankton is in- 25 mm nets (Nitex). We operationally defined that hibited by the colony size and a potential toxicity from bacteria attached to Microcystis (MB) dominated the intracellular microcystin (DeMott & Dhawale 1995). >20 pm size fraction in abundance and activity and Similarly, mesozooplankton has been shown not to that bacteria in the filtrate (FB, <20 pm size fraction) ingest bacteria attached to Microcystis (Bern 1987) and mainly originated from the water phase. The 20 pm bacteria embedded in mucilage may be protected from mesh-width net retained 98% of the Microcystis protozoan ingestion (Jiirgens & Giide 1994).Gradually colonies. No further attempts were made to quantify during autumn, however, Microcystis disappears from detachment of MB during fractionation and the contri- the pelagic waters and sinks to the sediment (Bostrom bution of bacteria attached to other particles than et al. 1989), because low temperature inhibits the Microcystis in the >20 pm size fraction. ability to regain buoyancy (Thomas & Walsby 1986). Bacterial biomass. MB collected on 20 pm mesh- A few studies have estimated the relative contribu- width nets were transferred to 0.2 pm prefiltered Milli- tion of bacteria attached to Microcystis in eutrophic Q water and processed as the filtrate. Samples were lakes. Bacteria attached to Microcystis accounted for fixed with glutaraldehyde (1.2% final concentration) 6 to 40% of total bactenal abundance during 3 sam- and stored at 4°C for 3 to 4 mo (autumn) or 0.5 to 2 mo pling days and from 7 to 30% of total bacterial pro- (summer). Recent studies have demonstrated that stor- duction during a diel study in Lake Vallentjunasjon, age for several weeks can reduce bacterial number Sweden (Brunberg 1993). Less specific for Microcystis- and cell volume distribution (Turley & Hughes 1992, associated bacteria, 15 to 33 % of the bacterioplankton Gundersen et al. 1996). Effects of storage were not was retained in the >5 um wore-size fraction dilring a ;recounted for in the prcscnt study. Bacteiia associaieci 2 yr study in a Japanese pond (Konda 1984) and >50% with particles and the walls of storage containers were of bacterial production was associated with >3 pm detached by 2 min of sonication with a Branson Soni- pore-size fractions in Lake Norrviken, Sweden (Bell fier 250 adjusted to 50% duty cycle and 306 pm ampli- et al. 1983). Thus, bacteria attached to Microcystis tude (25 W) (Velji & Albright 1993). No effect of soni- may contribute significantly to pelagic bacterial carbon cation was observed on bacterial abundance in 3 pm dynamics. pore-size filtrates (data not shown). Bacteria were We aimed to study the ecology of Microcystis-asso- qu.antified by direct microscopic counting of DAPI- ciated bacteria (MB) based on successive and physio- stained (4',6-diamidino-2-phenylindole, Sigma; final logical parameters in comparison with the rest of the concentration 0.1 pg ml-') bacteria collected on 0.2 pm pelagic bacterial community (mainly free-living bacte- pore-size black polycarbonate filters (Porter & Feig Worm & Ssndergaard: Mic :rocystjs-associated bacteria 21

1980). Bacterial cell volumes were calculated from quantified subsequently with a spectrofluorophotome- measured length/widths or diameters from enlarged ter (Shimadzu RF-5001PC). From August 3 to 17 sam- micrographs. Cell volumes were converted to cell car- ples had to be stored until later quantification (less bon by the factor 105 fg C pm-3 (Theil-Nielsen & Son- than 3 wk at -18'C). After incubation and Leu-MCA dergaard in press), as the bacterial volumes used to addition to the blanks, samples were filter sterilized calculate this factor were estimated with similar equip- (0.45 pm cellulose-nitrate membrane filters, llicro Fil- ment and methods. tration System) to minimize the hydrolysis from parti- Bacterial production. Bacterial production was esti- cle bound enzymes during freezing and thawing and mated from the incorporation of [Q]-thymidine (TdR) stored at -18OC. Two later experiments showed that into DNA (Fuhrman & Azam 1980) and during summer sterile filtration upon incubation reduced fluorescence was also from [3H]-leucine (Leu) incorporation into by 26 and 62 %. Stored samples are therefore corrected proteins (Kirchman et al. 1985). Incorporation rates by an average factor of 1.99 and excluded from statisti- were measured at substrate saturation concentrations cal analysis and other concluding calculations. estimated from kinetic experiments before each sam- Chlorophyll, POC, DOC. Chlorophyll was collected pling for TdR incorporation and 4 times during summer on Whatman GF/C filters, extracted with 96 91, ethanol for Leu incorporation. overnight and measured spectrophotometrically (Jes- Triplicate samples and 1 blank were incubated at persen & Christoffersen 1987). Particulate organic car- in situconditions for 1 h with [methyl-3H]-thymidine bon (POC) and dissolved organic carbon (DOC) were (Amersham) diluted to 20 nM (10 Ci mmol-') and separated by filtration with precolnbusted (550°C) L-[4,5-3H]-leucine (Amersham) diluted to 600 nM 13 mm GF/F filters. POC was measured as CO2 emitted (1 Ci mmol-') final concentrations. Incorporation was from combustion of filters (600°C) by use of an infrared stopped by adding 100% trichloroacetic acid (TCA) to gas analyser (Sondergaard & Middelboe 1993) and a final concentration of 5%. TCA was added to the DOC was measured by a Pt-catalytic carbon analyser blanks prior to the incubations. Radiolabeled macro- (Shimadzu TOC-5000). molecules were extracted for 1 h at 4OC before the Statistical analysis. Equivalence of MB and FB in samples were fractionated with 20 l-lm mesh-width growth rates and specific aminopeptidase activity were nets. The 20 pm mesh-width nets, retaining 3H incor- tested nonparametrically with the Wilcoxon signed porated by MB, were placed on 0.4 pm pore-size cellu- rank test. Spearman rank correlation coefficients (r,) lose-nitrate membrane filters (Micro Filtration System), were calculated to test patterns among variables, similar to the filters used to collect bacteria from the fil- unless other calculations are mentioned. In the kinetic trates. Filters were rinsed with cold 5% TCA (x4) and experiments, leucine incorporations at the respective 80% ethanol (x2) (Wicks & Robarts 1988, Hollibaugh & concentrations were fitted to the Michaelis-Menten Wong 1992) and radioactivity was quantified with an equation of enzyme kinetics by the least squares esti- LKB Wallac 1219 Rackbeta liquid scintillation counter. mation of the parameters V,,, and K,. Randomness of Thymidine incorporation was converted to cell pro- data points relative to fitted lines were confirmed with duction with the factor 2 X 1018 cells mol-', which has Runs test and Sign test and equivalence of K,,, values previously been estimated for Frederiksborg Slotsse were compared with t-tests. The level of significance (Smits & Riemann 1988) and another eutrophic lake was 0.05. (Bell et al. 1983). Carbon production was derived from cell production by the average cell carbon specified for each fraction. Leucine incorporation was converted RESULTS to carbon production using the factor 3.1 kg C mol-' (Simon & Azam 1989). Exponential growth rates were Kinetic experiments showed that thymidine (TdR) calculated to relate bacterial production and biomass. incorporation was saturated at 20 nM, whereas leucine Aminopeptidase activity. The potential aminopepti- (Leu) incorporation never saturated below 200 nM. dase activity was measured with the fluorogenic model The Michaelis-Menten equation of enzyme kinetics substrate L-leucine-4-methyl-coumarinylarnid hydro- was fitted to the saturation curves of Leu incorporation chloride (Leu-MCA) according to Hoppe (1993). with values of r2 between 0.78 and 0.99. All values of MB collected on 20 pm mesh-width nets were resus- K, were higher in the >20 pm size fraction compared pended in 0.2 pm sterile filtered lake water and to the <20 pm size fraction, although this difference processed as the filtrate. Triplicate samples were incu- was statistically significant in only 2 of 3 comparisons, bated with Leu-MCA (Sigma) at 0.2 mM final concen- i.e. on July 26 and August 1. Leu incorporation in the tration (saturation level) in the dark at in situtempera- >20 pm size fraction, assigned to MB, therefore ture. After 1 h incubation, Leu-MCA was added to showed less affinity and saturated at higher concen- blank samples and fluorescing hydrolysates were trations compared to the <20 pm size fraction, domi- 22 Aquat Microb Ecol 14: 19-28, 1998

nated by FB. Leu incorporation at ? 600 nM, the concentration used to esti- mate bacterial production, tended to sat- l urate (Fig. 1). Most estimates of K, 2.0 A no fractionation 1 ranged from 21 to 164 nM, correspond- >20 pm (MB) ing to 97 and 79% saturation at 600 nM o <20 pm (FB) i according to the Michaelis-Menten equation. The exception was MB on June 26, where K,, = 1016 nM, corre- sponding to only 37 % saturation. Results from summer 1995 and autumn 1994 are presented in succes- sion according to seasons rather than chronology, because general patterns of bacterial growth dynamics are assumed to be reflected (Fig. 2). Chlorophyll and POC were positively correlated (r, > 0.75, p < 0.007) and Microcystis spp. dominated the phytoplankton during both seasons. MB increased in biomass from 8 yg C 1-* in early July to a maximum value of 15.9 pg C 1-' on August 7 and afterwards declined gradually to 5.3 pg C 1-' (Fig. 2).In parallel, MB production (TdR) increased from ca 0.25 to 3.77 pg C l-' h-' on August 3 and decreased steadily to 0.79 pg C I-' h-' (r, = 0.80, p = 0.003, 0 200 400 600 800 1000 n = 15). Similar trends were measured Leucine(nM) from leucine incorporation at 600 nM Fig. 1. Kinetics of leucine incorporation fitted to the Michaelis-Menten equa- (r, = 0.73, p = 0.008, n = 14) and con- t~onof enzyme kinetics with values of r2 from 0.78 to 0.99. (A) No fractionation, (B-D) MB designates >20 pm size fractions, presumed to be dominated by verted to production (Leu) the estimates Microcystis-associated bacteria, and FB designates <20 I.lm size fractions were 4-fold higher than calculated from dominated by mainly free-living bacteria the TdR method. Production also corre-

Table 1. Spearman rank correlation coefficients (I,) between parameters for Microcystis-associated bacteria (MB, slze fractlon >20 pm) and bacteria living in the surrounding water (FB, size fraction <20 pm). Levels of significance are ~ndlcated with asterisks: "'p < 0.001, "p < 0.01, 'p < 0.05. -: p >0.05

MB versus FB Temperature Chlorophyll POC DOC

Summer 1995 Bacterial biomass MB 0.65' - - FB - - - - Bacterial cell production 0.65' MB 0.80"' 0.59' 0.63' - (thymidine method) FE 9.55' 0.65' - - Bacterial biomass production - MB 0.81 "' - - (leucine method) FB - - - Aminopeptidase activity - MB - 0.89" 0.81 ' - FB 0.86' - - Autumn 1994 Bacterial biomass 0.60' MB 0 90"' 0.86" 0.94 "' - FB 0.60' 0.59' - Bacterial production 0.59' MB 0.69' 0.55 ' 0 83" - (thyrmdlne method) FB 0 55' - 0.56' - 'Data from August 3 to 17 are excluded (n = 10) Worm & S~ndergaard Microcystls-associated bactena 23

Fig 2 Continuous measures of ab~otic,algal and bactenal para- meters, Fredenksborg Slotss8, summer 1995 and autumn 1994 (W)Microcystis-assoc~ated bacte- na (h4B > 20 pm mesh-wldth size fraction) and (0) bacteria in the surrounding water (FB <20 pm mesh-width size fract~on)Error bars ind~catestandard deviation of triplicate samples of dissolved organic carbon (DOC) partlcu- late organic carbon (POC),bac- tenal production and amnopep- t~daseactlvity Aminopeptidase from August 3 to 17 are multi- pl~edby 1 99 to correct for fluo- ---ci';""'! rescence lost dunng stenle filtra- l 8 24 31 7 14 21 28 23 30 7 14 21 28 tlon Jul. Aug. Sep. Oct. lated significantly with aminopeptidase activity (TdR: FB biomass was rather constant and ranged mostly r,= 0.89, p=0.008, n= 10; Leu: r,=0.87, p = 0.014, n= between 75 and 125 1-19 C 1-' during both seasons, 9). During summer neither biomass nor production of although it tended to decrease during autumn (r, = MB correlated significantly with chlorophyll, whereas -0.57, p = 0.040, n = 14) (Fig. 2). During summer FB the decline in chlorophyll during autumn correlated production (TdR) increased from ca 1 1-19 C 1-' h-' in significantly with decreases in biomass and production early July with 2 peak values on July 27 at 16.2 pg C 1-' of MB (Table 1).From late September to late October, h-' and on August 21 at 8.2 pg C 1-' h-', but declined MB biomass was reduced from about 100 to 25 pg C 1-' afterwards to about 3 pg C 1-' h-'. The latter peak and production (TdR) decreased from a level between of production followed the dynamics of DOC from 0.71 and 2.68 pg C 1-' h-' to ca 0.25 pg C 1-' h-' (r, = August 7. During autumn FB production (TdR) fluc- 0.73, p = 0.009, n = 14). tuated between 0.91 and 1.82, but decreased to 0.6 pg 24 Aquat Microb Ecol 14: 19-28, 1998

Table 2. Relative contribution of Microcystis-associated bacteria (MB, size fraction >20 pm) in Frederiksborg Slotss0 during summer 1995 and autumn 1994. Values are calculated as MB/(MB+FB] x 100% and listed as average * standard deviation with the range in parentheses

Period Abundance Biomass Production Potential aminopepti- (X) (%l Thymidine (%) Leucine (%) dase activity (%la - - Summer 1995 8*3 10*4 25 13 55 i 18 (5-15) (6-19) (9-61) (20-78) Autumn 1994 32+ 11 37 * 12 43 * 16 (15-46) (19-52) (24-70) dData from August 3 to 17 are excluded

C I-' h-' in late October, significantly correlated with sured in the <20 pm size fraction after TCA extraction the production of MB (Table 1). and decreases amounted to 6.5 and 7.7 % for TdR and Consistently, MB comprised a smaller fraction of 2.1 and 6.7 % for Leu, but within the limits of the stan- total abundance or biomass as compared to total dard deviations. The measures are biased because production and potential aminopeptidase activity, recovered radiolabel averaged only 75 & 14.5 % (range which corresponds to higher specific activities of MB 58 to 92%) of unfractionated totals (Fig. 4). Anyway, (Table 2). During summer 1995 cell growth rates of MB Leu:TdR ratios in c20 pm size fractions were un- were significantly higher than those of FB (p < 0.002), affected by the order of TCA-extraction and fractiona- in contrast to autumn 1994 where averages (2 standard tion (Expt l: 15.8 2 0.34; Expt 2: 10.8 + 0.34), whereas deviation) of 0.40 * 0.19 and 0.30 & 0.10 d-' for MB and Leu:TdR ratios in >20 pm size fractions were higher FB, respectively, were not statistically different (p = and more variable without any consistent trends. On 0.069). Further, specific aminopeptidase activity of MB this basis we conclude that the relative disagreement was significantly higher than that of FB when tested between Leu:TdR ratios and cell volumes measured nonparametrically (TdR: p = 0.006, n = 10; Leu: p = during summer are robust to methodological errors. 0.018, n = 9) and values of MB and FB averaged 1.60 + Taken at face values, calculated growth rates indicate 0.67 and 0.41 * 0.32 m01 Leu-MCA hydrolysed per m01 that the leucine method overestimated the MB produc- C produced (TdR), respectively. and 0.027 * 0.010 and 0.018 + 0.006 m01 Leu-MCA hydrolysed per m01 C pro- duced (Leu), respectively, assuming negligible algal leucine incorporation. Production derived from the thymidine and leucine methods correl.ated significantly for MB (r, = 0.90, p = 0.001, n = 14), as did summed rates (MB + FB), repre- senting total production from each method (r, = 0.78, p = 0.005, n = 14) (Fig. 3). The molar ratios of incorpo- rated Leu and TdR (Leu:TdR ratio) were consistently larger for MB than for FB and averaged 23.2 and 3.81, respectively, which reflects that MB produce more bio- mass per cell division than FR. This result is qualita- tively in accordance with the higher cell volumes mea- sured for MB (average: 0.095 pm%ell-') as compared o 4 to FB (0.071 pm3 cell-'), but the small difference in G 5 10 i 5 20 volumes is insufficient to explain the larger difference Production (TdR) (pg C I-' h-') in Leu:TdR molar ratios based on reported linear or exponential relationships between cell volume and cell Fig. 3. Relation.ship between bacterial product~onof biomass carbon (Simon & Azarn 1989, Theil-Nielsen & S~nder- estimated with the thyrmdine (TdR] and leucine (Leu) method.^, gaard in press). In 2 later experiments, we tested Frederiksborg Slotsse, summer 1995. (0) Bacteria attached to whether this difference reflected that samples were Microcystis(MB, >20 pm mesh-width size fraction), (0)mainly extracted for 1 h with 5 % TCA before fractionat~on. It free-living bacteria (FB, <20 pm mesh-width size fraction) and (A) summed rates (MB + FBI,corresponding to unfractionated was the hypothesis that macromolecules labelled wlth samples. Lines are fitted with linear regressions: MB, Leu = either 3H-Leu or 3H-TdR distributed differently upon TdR X 2.992 + 1.052 (r2= 0.81); FB. Leu = TdR x 0.315 + 2.646 precipitation with TCA. Less radioactivity was mea- (r7= 0.53);and (MB + FB), Leu = TdR x 0.641 + 5.735 (r2= 0.56) Worrn & Sandergaard: A4icrocystis-associated bactel-id 25

width nets retained MB is not accurate: bacteria Expt l Fraction.+TCA-extrac. attached to large non-Microcystis particles were also TCA-extrac.+fraction. included in the >20 pin size fraction and loss of MB to ... (unfraction. - c20pm) g the flltrate during fractionation cannot be neglected (Bern 1987). We did not attempt to quantify these methodological errors, but point out that they work in opposite directions regarding the accuracy of MB in abundance and activity. We consider that our method gives a fair description of the dynamics of MB. In a parallel study during suminer 1995, particles retained by 3 pm pore-size filters were predominated in number by transparent exopolymer particles (TEP, of maximum length 3.2 to 162.0 pm), colonized by 7% of the total bacterial community (Worm & Ssndergaard in press). The relative contribution of MB was comparable (Table 2), but it is unlikely that TEP-bacteria domi- nated bacteiial abundance in the >20 pm size fraction regularly, because TEP is deformable and less effi- ciently retained by 20 pm mesh-width nets compared to Microcystis colonies. Leucine incorporation was included in the sampling programme as an independent supplement to the thymidine method to reflect bacterial growth in terms of both biomass (Leu) and cells (TdR). Previous studies with the leucine method have used maximally 100 nM leucine. We incubated at 600 nM leucine, because the kinetic experiments clearly demonstrated that more than 200 nM leucine was needed to approximate the saturation level, in agreement with previous kinetic studies in eutrophic systems (Jerrgensen 1992, Rie- Fig. 4. Two experiments showing the distribution of TdR and mann & Azam 1992, van Looij & Riemann 1993). At Leu incorporation in samples fractionated with 20 pm mesh- saturation level, intracellular isotope dilution of incor- width nets either before (open bars) or after (solid bars) 1 h porated radiolabel is minimized and independent of extraction with 5% TCA. Unrecovered radiolabel (dotted affinity, which per se optimizes the coupling to bac- bars) are added to bars of >20 pm size fractions terial production. However, high concentrations of leucine increase the risk that bacterial growth is tion. Growth rates of MB averaged 3.2 5 1.8 d-' (TdR) stimulated and/or algae incorporate leucine. Indirect and the unrealistically high value of 10.6 * 4.1 d-' evidence of minor algal leucine incorporation at (Leu),whereas estimates of FB agreed better, averag- 600 nM includes that the kinetics of leucine incorpora- ing 1.2 + 0.62-' (TdR) and 1.4 * 1.3 d-' (Leu). However, tion was hyperbolically rather than linearly related to this conclusion is sensitive to both inaccuracies of the the respective concentrations, thus indicating the pre- conversion factors used and storage effects on esti- dominance of bacterial leucine incorporation by high mates of bacterial biomass and cell volume (Turley & affinity uptake systems rather than algal low affinity Hughes 1992, Gundersen et al. 1996). uptake systems and/or passive diffusion. In addition, incorporated leucine retained on 20 pm mesh-width nets (algal fraction) was highly significantly related to DISCUSSION thymidine incorporation (r, = 0.90, p = 0.001, n = 14) and thymidine at 20 nM is neither expected to stimu- MB are not easily segregated from the FB. Manual late bacterial growth nor to be taken up by algae (Bern collection of Microcystis colonies with micropipette is 1985) (Fig. 3). The significant correlation between a specific but laborious method (Bostrom et al. 1989, summed rates of thymidine and leucine derived pro- Brunberg 1993). Our purpose, to study short-term duction agrees with previous findings (Kirchman 1992, dynamics of MB and FB in abundance, production and Servais 1992, Simon 1994). Based on these arguments aminopeptidase activity, demanded a rapid segrega- bacterial production was calculated from leucine in- tion procedure, but the definition that 20 pm mesh- corporation at 600 nM using a conversion factor, where 2 6 Aquat Microb Ecc

overall bacterial carbon production is calculated from dynamics during blooms of Microcystis (Table 2), as the direct incorporation of exogenous radiolabelled evident from previous studies (Bell et al. 1983, Konda 1euci.ne into bacterial proteins, a constant ratio be- 1984, Brunberg 19931. tween leucine in bacterial proteins and overall cell car- Obviously, Microcystis colonies occupied a negligible bon and a 2-fold intracellular isotopic dilution (Simon volume of the bulk water. Nevertheless, the abun- & Azam 1989). Whether these presumptions are ful- dance and activity of MB were indeed significant, filled for MB, including a quantification of algal which designates this microenvironment for hetero- leucine incorporation, are difficult to measure, because trophic bacteria as an intense 'hotspot'. Accordingly, MB cells then have to be separated from cells and bacterial activity of pelagic waters may be heteroge- mucilage of Microcystis. The unrealistically high neously distributed on the macroscopic scale during growth rates of MB, as calculated from leucine incor- blooms of Microcystis, as evident for larger particles poration (mean 10.6 d-'), indicate significant algal known as marine or lake snow (Smith et al. 1992, incorporation and speaks in favour of the thymidine Grossart & Simon 1993). If the measured growth rates method, but we emphasize that the conclusion is reflect, at least, qualitative differences between dependent on our choices of conversion factors. The actively growing cells, the higher growth rates of MB volume to carbon conversion factor is in the lower during summer indicate that the microenvironment of range reported (Theil-Nielsen & Ssndergaard in press) Microcystis was enriched with labile substrates rela- and lower leucine conversion factors of 0.900 and tive to the surrounding water. In general, high sub- 1.080 kg C mol-l have been reported (Servais & Gar- strate availability selects for low affinity/high capacity nier 1993, Servais & Lavandier 1993). Increasing the uptake systems (Sundergaard & Middelboe 1995). volume to carbon conversion factor and decreasing Consistently, such uptake systems characterized leu- that of leucine improves the balance between esti- cine incorporation of MB compared to FB (Fig. 1). The mates of production for MB at the expense of FB. The higher specific activity of MB may reflect the combined conversion factor for thymidine incorporation was effects of (1) higher steady-state concentration of exu- determined in the studied lake (Smits & Riemann 1988) dates within the mucilage of Microcystis, (2) increased and corresponds to values often found in other sys- exchange of nutrients with the surrounding water, as tems. Another possibility for biased growth rate cal- MB move through the water column driven by the den- culations is that the storage of samples with glutar- sity of Microcystis (Thomas & Walsby 1986), (3) adsorp- aldehyde for several weeks may have reduced the tion of labile organic matter to the mucilage of Micro- estimates of bacterial abundance and mean cell carbon cystis (Fletcher 1991), and (4) diversified species (Turley & Hughes 1992, Gundersen et al. 1996), which composition (Brunberg 1993). During autumn, how- leads to overestimated growth rates and underesti- ever, growth conditions were rather similar according mated carbon content and biomass of DAPI-stained to the statistical similarity between measured growth bacteria. However, tight correlation between bacterial rates. An opposite characterization of MB was given by biomass and production indicate that effects due to Brunberg (1993), who found that thymidine incorpora- storage were similarly distributed between the sam- tion per cell of MB was below that of FB in late summer ples. Accordingly, we have no obvious reasons to reject and early autumn. Low mortality, nutrient limitation or relative differences in either bacterial biomass or spe- accumulated algal metabolites were proposed to sus- cific activities, although both parameters are probably tain MB abundance near 'carrying-capacity'. generally somewhat biased. characterized by relatively inactive bacteria, long During summer 1995, MB averaged only 10 0/0 of total turnover times and low specific growth rates. Most bacterial biomass, but due to the high specific activity li.kely, the activity of MB may be inhibited., if the MB contributed on average elther 25% (TdR) or 55% microenvironment of Microcystis becomes alkaline (Leu) of total bacterial production. MB also showed a during photosynthesis, as evident from axenic colonies high hydrolytic capacity and contributed on average of .Microcystis (Richardson & Stolzenbach 1935). How- 53% of the total potential aminopeptidase activity ever, no general pattern about the chemical microenvi- (Table 2). Specific aminopeptidase activity of MB also ronment of Microcystis is obvious from the measured exceeded that of FB significantly. In autumn the colo- growth rates of DAPI-stained bacteria. nization of Microcystis was more intense and peak val- MB production (TdR) ranged from 0.23 to 3.77 pg C ues of MB biomass and production exceeded FB I-' h-', whereas the calculated increases in biomass episodically. However, the relative contributi.on of MB could be accounted for by a maximum of 0.067 pg C I-' declined to about 15 to 20% in late October, as Micro- h'. Therefore, more than 70% of the MB production cystis gradually disappeared from the surface water was not recovered as increased biomass, which indi- (Fig. 2). All measures of bacterial activity thus point to cates that most gain of MB, i.e. growth and coloniza- an important ecological role of MB in bacterial carbon tion, balanced the loss from grazing, lysis, sedimenta- Worm & Sondergaard: Mic .rocyst~s-associatedbacteria 27

tion and shedding (Pedros-Alio & Brock 1983). We Bern L (1987) Zooplankton grazing on [methyl-3H]thyrnidine- hypothesize that a large fraction of h4B production was labelled natural partlcle assemblages. determination of shedded to the surrounding water. This characterizes filtering rates and food selectivity. Freshwat Biol 17(1): 151-159 Adicrocystis as a bacterial 'incubator'. The idea is based Bostrom B, Petterson AK. Ahlgren l(1989) Seasonal dynamics on reported low loss rates of Microcystis from sedi- of a cyanobacteria-dominated microbial community in mentation and grazing during summer (Reynolds 1973, surface sediment of a shallow, eutrophic lake. Aquat Sci DeMott & Dhawale 1995). Further, we did not observe 51(2):153-178 Brunberg AK (1993) Microcystis in lake sediments. PhD any protozoa grazing on MB, and Bern (1987) could not thesis, Uppsala University measure any grazing of MB by mesozooplankton. Caldwell DE, Korber DR, Lawrence JK (1992) Confocal laser However, loss of MB due to vlral lysis remains microscopy and dlgltal image analysis in rnlcrobial ecol- unsolved. In one marine study virus infection was ogy. Adv Microb Ecol 12:l-67 Christoffersen K, Riemann B, Hansen LR. Klysner A, detected in 2 to 37% of the bacteria associated with Sorensen HB (1990) Qualitative importance of the micro- particles (Proctor & Fuhrman 1991). Applying this bial loop and plankton community in a eutrophic lake dur- range to our results, however, shedding of MB is likely ing a bloom of Cyanobacteria. Microb Ecol20:253-272 to be an important loss factor. Attached bacteria shed- DeMott MrR, Dhawale S (1995) Inhibition of in vitro protein ding daughter cells to the surrounding water is evident phosphatase activity 'in three zooplankton species by microcvstin-LR, a toxin from cyanobacteria Arch Hydro- from studies of biofilm formation (Caldwell et al. 1992), biol 134(4):417-424 and recognized from various types of natural particles Fletcher M (1991) The physiological activity of bacteria including copepod fecal pellets (Jacobsen & Azam attached to solid surfaces. Adv Microb Physlol 32:53-85 1984), marine snow (Azam & Smith 1991) and diatom- Fuhrman JA, Azarn F (1980) Bacterioplankton secondary pro- mucus aggregates retained by 1 pm pore-size filters duction estimates for coastal waters of British Columbia, Antarctica and California. Appl Environ Microbiol 39: (Smith et al. 1995). 1085- 1095 In conclusion, Microc)7stis offers attachment sites for Grossart HP, Simon M (1993) Limnetic macroscopic organic a very dynamic community of heterotrophic bacteria, aggregates (lake snow): occurrence, characteristics, and which contributes importantly to the carbon metabo- microbial dynamics in Lake Constance Limnol Oceanogr 38(3):532-546 lism of pelagic bacteria. Up to half of total bacterial Gundersen K,Bratbak G, Heldal M (1996) Factors influencing production was assigned to this microenvironment the loss of bacteria in preserved seawater samples. Mar during summer and autumn. As Microcystis often Ecol Prog Ser 137:305-310 dominates the phytoplankton of eutrophic lakes, the Herndl GJ (1988) Ecology of amorphous aggregations (marine snow) in the Northern Adriatic Sea. 11. Microbial density pelagic microenvironment of these systems may be and activity in marine snow and its implication to overall considered heterogeneous with intense 'hotspots' of pelagic processes. Mar Ecol Prog Ser 48:265-275 bacterial activity in association with the macroscopic Hollibaugh JT, Wong PS (1992) Ethanol-extractable substrate colonies of Microcystis. pools and the incorporation of thymidine, L-leucine, and other substrates by bacterioplankton. Can J Mlcrobiol 38: 605-613 Hoppe H (1993) Use of flourogenic model substrates for extra- Acknoruledyements. We appreciate the comments and sug- cellular enzyme activity (EEA) measurement of bacteria. gestions by Melnhard Simon and 3 anonymous reviewers. In: Kemp PF, Sherr BF, Sherr EB, Cole JJ (eds) Handbook The study was supported by The Danlsh Nat'ural Sciences of methods in aquatic microbial ecology. Lewis Publ, Boca Research Council and The Danish Environmental Research Raton, p 423-431 Programme. Jacobsen TR, Azam F (1984) Role of bacteria in copepod fecal pellet decomposition: colonization, growth rates and mineralization. Bull Mar Sci 35(3):495-502 LITERATURE CITED Jespersen AM, Christoffersen K (1987) Measurements of chlorophyll-a from phytoplankton using ethanol as extrac- Andersen JM, Jacobsen OS (1979) Production and decom- tion solvent. Arch Hydrobiol 109(3):445-454 position of organic matter in eutrophic Frederiksborg Jorgensen NOG (1992) Incorporation of [3H]leucine and Slotsso, Denmark. Arch Hydrobiol85(4):511-542 [3H]valine into protein of freshwater bacteria: field appli- Azam F, Smith DC (1991) Bacterial influence on the variability cations. Appl Environ Microbiol58(11):364?-3653 in the ocean's biogeochemical state: a mechanistic view. Jiirgens K, Gude H (1994) The potential importance of graz- In: Demers S (ed) Particle analysis in oceanography. ing-resistant bacteria in planktonic systems. Mar Ecol Springer-Verlag, Berlin, p 213-236 Prog Ser 112:169188 Bell RT. Ahlgren GM, Ahlgren I (1983) Estimating bac- Kirchman DL (1992) Incorporation of thymldine and leucine terioplankton production by measuring [3H]thymidine in the subarctic Pacific: application to estimating bacterial incorporation in a eutrophic Swedish lake. Appl Environ production. Mar Ecol Prog Ser 82:301-309 Microblol 45(6):1?09-1721 lrchrnan DL (1993) Particulate detritus and bacteria in Bern L (1985) Autoradiographic studies of [methyl-3H]thymi- marine environments In: Ford TE (ed) Aquatic micro- dine incorporation in a Cyanobacteriurn (Microcystis biology-an ecological approach. Blackwell. Oxford, wesenbergj1)-bacterium association and in selected algae p 321-341 and bacteria. Appl Environ Microbiol 49(1):232-233 Kirchman D, K'Nees E, Hodson R (1985) Leucine incorpora- Aquat Microb Ecol 14: 19-28, 1998

tion and its potential as a measure of protein synthesis by Smith DC, Simon M, Alldredge AL, Azam F (1992) Intense bacteria in natural aquatic systems. Appl Environ Micro- hydrolytic enzyme activity on marine aggregates and biol49(3)-599-607 implications for rapid particle dissolution. Nature 359: Konda T (1984) Seasonal variations in four bacterial size frac- 139-142 tions from a hypertrophic pond in Tokyo, Japan. Int Rev Smith DC, Steward GF, Long RA, Azam F (1995) Bacterial Ges Hydrobiol 69(6):843-858 mediation of carbon fluxes during a diatom bloom in a Pedros-Ali6 C, Brock TD (1983) The importance of attach- mesocosm. Deep Sea Res 42(1):75-97 ment to particles for planktonic bacteria. Arch Hydrobiol Smits JD, Riemann B (1988) Calculation of cell production 98(3):351-379 from [3H]thymidine incorporation with freshwater bacte- Porter KG, Feig YS (1980) The use of DAPI for identifying ria. Appl Envlron M1crobiol54(9):2213-2219 and counting aquatic microflora. Limnol Oceanogr 25(5]: Ssndergaard M. Middelboe M (1993) Measurements of par- 943-948 ticulate organic carbon: a note on the use of glass fiber Proctor LM, Fuhrman JA (1991). Roles of viral infection in (GF/F) and f no disc"filters. Arch Hydrobiol 127(1):73-85 organic particle flux. Mar Ecol Prog Ser 69:133-142 Ssndergaard M, Middelboe M (1995) A cross-system analysis Reynolds CS (1973) Growth and buoyancy of Microcystis of labile dissolved organic carbon Mar Ecol Prog Ser 118: aeruginosa Kiitz. emend. Elenkin in a shallow eutrophic 283-294 lake. Proc R Soc Lond B 184:29-50 Theil-Nielsen J, Ssndergaard M (in press) A note on bacterial Reynolds CS, Rogers DA (1976) Seasonal variations in the carbon biomass calculated from biovolumes. Arch Hydro- vertical distribution and buoyancy of Microcystis aerugi- biol nosa Kutz. Emend. Elenkin in Rostherne Mere, England Thomas RH, Walsby AE (1986) The effect of temperature on Hydrobiologia 48(1):17-23 recovery of buoyancy by M~crocystis.J Gen Microbiol 132: Richardson LL, Stolzenbach KD (1995) Phytoplankton cell 1665-1672 size and the development of microenvironments. FEMS Turley CM, Hughes DJ (1992) Effects of storage on direct Microbiol Ecol 16~185-192 estimates of bacterial numbers of preserved seawater Riemann B, Azam F (1992) Measurements of bacterial protein samples. Deep Sea Res 39(3/4):375-394 synthesis in aquatic envi.ronments by means of leucine van Looi] A, Riemann B (1993) Measurements of bacterial incorporation. Mar Microb Food Webs 6(2):91-105 production in coastal marine environments using leucine: Servais P (1992) Bacterial production measured by W-thymi- application of a kinetic approach to correct for isotope dine and 3H-leucine incorporation in various aquatic eco- dilution. Mar Ecol Prog Ser 102:97-104 systems. Arch Hydrobiol Beih Ergeb Limnol37:73-81 Velji MI, Albright U (1993) Improved samples preparation for Servais P, Garnier J (1993) Contribution of heterotrophic bac- enumeration of aggregated aquatic substrate bacteria In: terial production to the carbon budget of the River Seine Kemp PF, Sherr BF, Sherr EB, Cole JJ (eds) Handbook of (France). Microb Ecol25:19-33 methods in aquatic microbial ecology. Lewis Publ, Boca Servais P, Lavandier P (1993) Coherence des productions bac- Raton, p 139-142 teriennes estimees a partir de l'incorporation de 3H-thymi- Whitton BA (1973) interactions with other organisms. In: Carr dine et de 3H-leucine dans les eaux douces naturelles. CR NG, Whitton BA (eds) The biology of blue-green algae. Acad Sci Ser 111 Sci Vie 316:642-646 Blackwell, Oxford, p 415-433 Simon M (1994) Die1 variability of bacterioplankton biomass Wicks RJ, Robarts RD (1988) Ethanol extraction requirement production and cell multiplication in Lake Constance. for purification of protein labeled with [3H]leucine in Arch Hydrobiol 130(3):283-302 aquatic bacterial product~onstudies. Appl Environ Micro- Simon M, Azam F (1989) Protein content and protein synthe- bio154(12):3191-3193 sis rates of planktonic marine bacteria. Mar Ecol Prog Ser Worm J, Ssndergaard M (in press) Alcian Blue stained parti- 51:201-213 cles in a eutrophic lake. J Plankton Res

Editorial responsib~lity:John Dolan, Submitted: November 15, 1996; Accepted: June 16, 1997 Villefranche-sur- Mer, France Proofs received from author(s): September 19, 1997