bacterioplankton in a temperate freshwater system
Louis-B. Jugnia, T ´elesphore Sime-Ngando & Daniel GilbertLaboratoire de Biologie des Protistes, Universit ´e Blaise Pascal (Clermont-Ferrand II), UMR CNRS 6023, Aubie`re, France
Correspondence: Louis-B. Jugnia, Laboratoire de Biologie des Protistes, Universit ´e Blaise Pascal (Clermont-Ferrand II), UMR CNRS 6023, F-63177 Aubie`re Cedex, France. Tel.: 11 514 496 0714; fax: 11 514 496 6265; e-mail: louis.jugnia@cnrc-nrc.gc.ca
Present address: Louis-B. Jugnia, Biotechnology Research Institute, National Research Council of Canada, 6100 Royalmount Avenue, Montreal, Quebec H4P 2R2, Canada.
Present address: Daniel Gilbert, Laboratoire de Biologie et Ecophysiologie, Universit ´e de Franche Comt ´e, Place Leclerc, F-25030 Besanc¸on Cedex, France.
Received 21 September 2005, revised 18 February 2006, accepted 8 March 2006. First published online 24 May 2006.
DOI:10.1111/j.1574-6941.2006.00145.x
Editor: Riks Laanbroek
Keywords
bacterioplankton; growth rate; mortality; fresh water.
Abstract
The growth rate and losses of bacterioplankton in the epilimnion of an oligo-mesotrophic reservoir were simultaneously estimated using three different meth-ods for each process. Bacterial production was determined by means of the tritiated thymidine incorporation method, the dialysis bag method and the dilution method, while bacterial mortality was assessed with the dilution method, the disappearance of thymidine-labeled natural cells and ingestion of fluorescent bacterial tracers by heterotrophic flagellates. The different methods used to estimate bacterial growth rates yielded similar results. On the other hand, the mortality rates obtained with the dilution method were significantly lower than those obtained with the use of thymidine-labeled natural cells. The bacterial ingestion rate by flagellates accounted on average for 39% of total bacterial mortality estimated by the dilution method, but this value fell to 5% when the total mortality was measured by the thymidine-labeling method. Bacterial abundance and production varied in opposite phase to flagellate abundance and the various bacterial mortality rates. All this points to the critical importance of methodological aspects in the elaboration of quantitative models of matter and energy flows over the time through microbial trophic networks in aquatic systems, and highlights the role of bacterioplankton as a source of carbon for higher trophic levels in the studied system.
Introduction
Bacteria have the potential for rapid growth, although cell numbers generally vary by less than an order of magnitude over the course of a year in pelagic ecosystems (e.g. Cole & Caraco, 1993). Even during phytoplankton blooms, rela-tively small increases in bacterial numbers are usually observed (Ducklow & Carlson, 1992; Del Giorgio & Gasol, 1995). Research on the role of bacteria in aquatic systems is confronted with this apparent ‘paradox’ of high potential growth rates and small changes in bacterial abundance. Currently, there is a growing body of evidence that viruses, which are also important biological entities in natural waters, participate in regulating bacterial numbers. Other-wise, the most well-known causes of cell loss include cell lysis and predation, primarily by protozoa. Protozoan graz-ing activity has been identified as a major factor controllgraz-ing
bacterial abundance in pelagic systems (e.g. Vaqu´e & Pace, 1992; Strom, 2000). Owing to their bacterivorous activity, protozoa serve as a means of transferring bacterial biomass towards higher trophic levels and thus play a key role within microbial trophic networks (Azam et al., 1983). Within the protozooplankton, nanoflagellates are among the most im-portant bacterivores in pelagic systems. Although their individual predation rates are lower than those of ciliates (Pace & Bailiff, 1987), the number of flagellates is usually two to three orders of magnitude greater than that of ciliates in the plankton (Porter et al., 1985).
In light of the above observations, it is important to adequately estimate bacterial production and mortality in aquatic systems, for improved accuracy of energy flow measurements in microbial food webs. Various methods have been used to determine both bacterial secondary
production and mortality (i.e. bacterivory from phago-trophic protists) in pelagic systems. For bacterial produc-tion, techniques based on the frequency of dividing cells (Hagstrom et al., 1979), dilution of plankton samples (Landry et al., 1995), incubation in dialysis bags (Herndl et al., 1993),3H-thymidine and adenine incorporation into DNA, and3H-leucine incorporation into protein, have been developed. For bacterial mortality, methods based on the use of metabolic inhibitors of grazers (Newell et al., 1983; Sherr et al., 1986), dilution of plankton samples (Landry et al., 1995), fate of radioactivity in 3H-thymidine-labeled bacterial prey (Servais et al., 1985), and incorporation of labeled bacterial tracers such as fluorescent minicells (Holli-baugh et al., 1980; Caron et al., 1993), latex microspheres (Pace & Bailiff, 1987) or natural bacteria (Sherr & Sherr, 1993), have been used. Each of these approaches has its advantages and disadvantages. To our knowledge, there has been no seasonal study in which simultaneous estimates of bacterial production and mortality have been made using the various approaches. Such studies may significantly increase our understanding of the functional role of the microbial food web in aquatic systems. Methodological shortcomings probably represent the greatest impediment to our understanding of microbial ecology and the factors influencing the growth and fate of microorganisms in natural environments.
During this study conducted in the oligo-mesotrophic Sep Reservoir (France), bacterial production and losses were simultaneously estimated using three different methods for each process. Bacterial production was determined by means of the 3H-thymidine incorporation method, the dialysis bag method and the dilution method, while bacterial mortality was assessed with the dilution method, ingestion of fluorescent bacterial tracers by heterotrophic flagellates, and the disappearance of bacterial DNA labeled with thymi-dine. In conjunction with comparing different methodolo-gical approaches, this study was performed to examine bacterial mortality vs. production in the Sep Reservoir, with emphasis put on the functional importance of flagellates in channeling bacterial secondary production towards higher trophic levels.
Materials and methods
Study site and sampling
Experimental samples were collected with a Van Dorn bottle in the surface waters (i.e. at a depth of 1 m) of the oligo-mesotrophic Sep Reservoir, located in the French ‘Massif Central’, c. 461N, 31E. The reservoir contains 4.7 106m3of water and has a surface area of 33 ha, with mean and maximum depths of 14 and 37 m, respectively. Additional details on the site description can be found in Jugnia et al.
(1999, 2000a, b). Samples were collected biweekly from 13 June to 10 September 1998, from the deepest part of the reservoir. Temperature and dissolved oxygen concentrations were measured in situ using a digital display multiparameter apparatus YSI GRANT/3800 probe.
Bacteria and protozoa counts
Samples for determination of bacterial abundance and biomass were immediately fixed after collection by adding 2% (v/v) borate-buffered formalin (from a 37% w/v solution of commercial formaldehyde). Samples for the abundance of heterotrophic flagellates were fixed under the same conditions, using glutaraldehyde (final concentration 1% v/v). In the laboratory, subsamples for counts of bacteria (1–2 mL) and heterotrophic flagellates (10–20 mL) were filtered through 0.2 and 0.8 mm pore black polycarbonate filters, respectively, using 1.2 mm pore cellulose acetate back-ing filters to obtain a uniform distribution of cells. Sub-samples for bacterial counts were treated with 40 ,6-diamidino-2-phenylindole (DAPI) (Porter & Feig, 1980), while subsamples for flagellate counts were treated with primulin (Caron, 1983) before filtration. Filters were mounted between a slide and glass coverslip with a non-fluorescent immersion oil, prior to examination with a microscope equipped with an epifluorescence illuminator, a mercury lamp, and a neofluar objective lens 100/1.25. A blank was routinely examined as a control for contamina-tion of the equipment and reagents. At least 500 bacterial and 200 nonpigmented flagellated cells were counted in 20–50 fields of view of the microscope. The length and width of at least 200 bacterial cells per sample were estimated using an eyepiece micrometer. Bacterial mean cell volumes were calculated from cell dimensions according to the formula (p/4) W2[L (W/3)], where W is the width and L the length of the cell. For spherical bacteria, W = L (Sime-Ngando et al., 1991). Bacterial biomass was calculated with the established equation Y = 88.6X0.59, where X denotes mean cell volume (mm3) per sample and Y denotes cell protein (fg), with the C : protein ratio being 0.86 (Simon & Azam, 1989).
Bacterial production
Bacterial production was estimated using three different methods: in situ incubation in dialysis bags (Herndl et al., 1993), dilution (Landry et al., 1995), and incorporation of tritiated thymidine (Wicks & Robarts, 1987).
(1) Dialysis bags. The dialysis bags were of the Visking-spectra/POR type (made of cellulose), that allow molecules of molecular mass of between 6 and 8 kDa to pass through (equivalent to a pore size of about 1–1.3 nm). They were 4 cm in diameter when flat and 2.5 cm in diameter when filled with water. On the day before use, 25 cm-long dialysis
bags were cut from the original roll, and stored in sterile distilled water. In the field, they were rinsed in lake water filtered through a 0.2-mm-pore filter and one end was closed using a polypropylene clip, the other end being closed in the same way after filling the bag with the test sample, which was prefiltered through Whatman GF/F glass fiber (nominal pore size c. 0.7 mm) to remove grazer populations. Three replicate bags were thus prepared for each sample and were incubated in situ. Before incubation, a sample of water that was used to fill the dialysis bags was fixed with formaldehyde for bacterial counts at time t0. The samples were fixed after 6 and 24 h of incubation. The bacterial growth rate was determined using the standard exponential growth equa-tion: K(h1) = (ln Nx ln N0)/(tx t0), where K is the growth rate, Nxthe number of bacteria at time tx, N0 the number of bacteria at the starting time t0, and tx t0 the incubation time.
(2) Dilution. Just before each experiment, a freshly collected water sample was prefiltered through a Whatman GF/C filter to remove large particles and then sterilized by filtration under a low vacuum (o 50 mmHg pressure) through a 0.2-mm Nuclepore membrane 47 mm in diameter. This water was used as diluent for the dilution experiments. Three identical series (triplicates) of dilutions were prepared using the proportions [test samples ( = lake water)]/(dilu-ent) 1/0, 0.5/0.5, 0.25/0.75 and 0.10/0.90, in 500-mL trans-parent polycarbonate bottles that had been thoroughly washed, stored overnight in 10% HCl, and then thoroughly rinsed with sterile water and then with the diluent. Two 30 mL subsamples were collected from each bottle, one being fixed with borate-buffered formalin and the other with purified glutaraldehyde (Grade I, Sigma no G-7776; final concentration 1% v/v), for counting bacteria and heterotrophic flagellates, respectively, at t0. The bottles were then attached to a weighted incubation frame and sub-merged to the sampling depth (1 m), where they were kept for 24 h. Six hours later and at the end of the incubation time, further samples were collected and fixed as before incubation to measure bacterial and flagellate abundances at times t6and t24. After analysis of the samples, the growth rate and grazing rate on bacteria were calculated using the model proposed by Landry & Hassett (1982): Nt= N0e(kg)t, where Ntis the abundance of prey (bacteria) at time t24, N0 the abundance of prey at time t0, k the prey growth rate, and g the grazing rate. In practice, the grazing rate g is equivalent to the negative slope of the linear regression between the apparent growth rate of bacterial prey calculated for each dilution using the generalized exponential growth model and the proportion of lake water in the various dilutions (Landry & Hassett, 1982).
(3) Incorporation of tritiated thymidine. Bacterial produc-tion (BP) was estimated by means of [3H-methyl]thymidine (Tdr) incorporation (Wicks & Robarts, 1987). Triplicate
aliquots (10 mL) of freshly collected water and one 0.5 M NaOH-killed control were inoculated with Tdr (specific activity = 80 Ci mmol1; final concentration = 20 nM) in Pyrex glass bottles and incubated in situ and in the dark for 45–60 min. Preliminary tests indicated that the uptake rate of Tdr in the Sep plankton was linear for at least 90 min, and that saturation occurred at a final concentration of 15 nM (Jugnia et al., 1999, 2000). Tdr incorporation was stopped by adding 1 mL of 5 M NaOH. Radioactive samples were then passed through 0.2 mm cellulose nitrate filters and rinsed twice with 3 mL of ice-cold 5% trichloroacetic acid (TCA). Labeled bacterial DNA was extracted according to the method of Wicks & Robarts (1987). Filters were placed in vials, allowed to dry, and solubilized with 0.5 mL of ethyl acetate. After addition of 5 mL of a scintillation cocktail, radioactivity was counted with an LKB liquid scintillation counter, and BP, calculated as moles of Tdr incorporated into DNA, was converted into the number of bacterial cells produced by using appropriate conversion factors deter-mined empirically as previously described (cf. Jugnia et al. 2000).
Bacterial losses
Three distinct methodological approaches were undertaken simultaneously to accomplish measurements: (1) the dilu-tion method (previously described), (2) a method based on measuring the disappearance of bacterial DNA previously labeled with thymidine (Servais et al., 1985), and (3) fluorescent tracers (Pace & Bailiff, 1987; Sherr & Sherr, 1993).
(1) Dilution
This was as described previously.
(2) Disappearance of bacterial DNA labeled with Tdr (Servais et al., 1985). For this method, 2 nM Tdr was added to triplicates of natural samples in 500 mL Pyrex glass bottles, which were then incubated in the dark at in situ temperature for 4–5 h. Experimental samples in the Pyrex bottles were then individually poured into the pretreated Visking dialysis bags (as described above), which were immersed in circulating water from the sampling depth for at least 10 h, in order to eliminate free Tdr from the dissolved phase, if present. Preliminary tests showed that with the low concentration of Tdr used in these experiments (i.e. 2 nM) and after the 4–5 h incubations in Pyrex glass bottles, most of the radioactive tracer was in the particulate phase, indicating exhaustive incorporation of Tdr into the bacterial biomass. Following the elimination of free dis-solved Tdr in experimental samples, each replicate was then divided into two 250 mL subsamples, one of which was filtered through a 2 mm-pore membrane to eliminate bacter-ivorous organisms and the other of which was left unfiltered. These subsamples were then kept under in situ conditions
for 3–4 days. During this period, the decrease in particulate radioactivity was monitored as follows. At regular time intervals, 10-mL samples were collected and, after addition of 10 mL of 10% TCA, were left for 15 min at 0 1C. The mixture was then filtered through a 0.2 mm-pore membrane and rinsed twice with 5 mL of 5% TCA. The radioactivity associated with each filter was counted with the LKB counter and converted into bacterial cell numbers by using an appropriate conversion factor. For all samples, the disappearance of the particulate radioactivity generally showed first-order kinetics and, accordingly, the bacterial mortality rates (i.e. with and without bacterivores) were calculated from logarithmic plots of the decrease of radio-activity as a function of time. The bacterial mortality rate obtained from the samples without bacterivores represents the residual mortality of the bacterial communities, i.e. not related to grazing (Servais et al., 1985). The difference between the two rates was considered to be the consequence of grazing.
(3) Fluorescent tracers (Pace & Bailiff, 1987; Simek & Straskrabova, 1992). Two types of fluorescent prey were used separately, natural bacteria [i.e. fluorescently labeled bacter-ia (FLB)] (Sherr & Sherr, 1993) and latex microspheres [i.e. fluorescently labeled microspheres (FLM)] (Pace & Bailiff, 1987) of similar diameter to the mean size of bacteria in the Sep Reservoir (Jugnia et al., 1999). For the FLM method, a stock solution of tracer beads was prepared from a concen-trated suspension of 0.5 mm-diameter fluorescent micro-spheres (Osi no 15700). Before experiments were run, the bead stock suspension was treated with bovine serum albumin (Sigma no. A4378; 0.5 mg mL1) to avoid clumping (Pace & Bailiff, 1987). For the FLB method, natural bacteria were concentrated from the reservoir water following the method of Simek & Straskrabova (1992), and fluorescently labeled as described in Sherr & Sherr (1993).
For both methods, separate triplicates of the experimental samples were poured into the 500 mL clean sterile Pyrex glass bottles and acclimated at in situ temperature for at least 1 h. Fluorescent tracers were then added to the experimental samples at concentrations that were 15–25% of the abun-dance of natural bacteria, a good compromise between accuracy of measurement and the potential for altering the feeding activity of natural bacterivorous protists (Sherr & Sherr, 1993). The bottles were then incubated in situ for 30 min in the dark. This time was determined in preliminary experiments, which showed that the ingestion of FLM and FLB by flagellate protozoa in the Sep Reservoir increased linearly with time during the first 40 min of incubation. At time t0and every 10 min, 50 mL samples were collected from each replicate using a sterile syringe and were fixed with ice-cold glutaraldehyde (Grade I, Sigma no. G-7776; final concentration 2% v/v). This fixation treatment prevents protozoan egestion of food vacuole contents (Sanders et al.,
1989). The fixed samples were immediately placed in an insulated container containing crushed ice. On return to the laboratory, the uptake of fluorescent tracers (i.e. FLM and FLB) by phagotrophic flagellates was determined by passing 15–30 mL aliquots from each experimental bottle through 0.8 mm Nuclepore black filters, after cell staining with primulin as described for flagellate counts. Individuals of the grazer assemblages were observed at 1250 under UV light with an epifluorescent microscope, and ingested tracers were counted in their food vacuoles under blue light. The grazing impact (bacteria L1h1) was calculated from the mean product of the specific ingestion rate (bacteria per flagellate h1) and the abundance (L1) of flagellates of interest. For all the grazing experiments conducted, an analysis of variance comparing the two protocols used showed no significant difference for flagellate grazing rates. Because of the similarity of the two protocols, only the results concerning the uptake of the natural prey tracers, i.e. FLB, are presented in the rest of this study.
Statistical analysis
Relationships between seasonal data were tested by Pearson product moment correlation analysis. The bacterial produc-tion and mortality rates obtained when the different meth-ods were used were compared by performing a one-way analysis of variance (ANOVA) or by paired t-tests, for the significance of differences between protocols. In theANOVA, the significance of differences between the different proto-cols was tested by using a posteriori tests, the Bonferroni adjusted pairwise comparisons and the Fisher least signifi-cant difference test. The null hypothesis was that there was no difference between methods.
Results
Physicochemical variables
During the study, the water temperature varied from 18 to 24 1C (mean SD = 21.34 1.82 1C). The values increased more or less uniformly from mid-June to mid-August, after which the seasonal cooling of the surface waters started. The dissolved oxygen content was between 5.09 and 14.7 mg L1, with a mean value of 9.23 2.65 mg L1. Except for a decrease between the last two sampling dates, the values for this variable remained relatively stable.
Bacterial and heterotrophic flagellate abundance
The bacterial abundance fluctuated between 1.72 and 9.83 109cells L1, with a mean value of 5.56 3.20 109cells L1. That of heterotrophic flagellates varied from 1.21 to 5.33 106
cells L1, the mean value being
3.65 1.38 106
cells L1. This community mainly con-sisted of Chrysidalis sp., Rhodomonas sp., Cryptomonas ovata and a Monas species. The temporal changes in the density of bacteria and heterotrophic flagellates had a seesaw appear-ance, the peaks in flagellate abundance coinciding with troughs in bacterial abundance and vice versa (r = 0.8, Po 0.05) (Fig. 1a).
Bacterial growth and production rates (Table 1) The growth rate obtained with the dialysis bag method varied from 7 to 13 102
h1(mean = 10 3 102 h1).
These values were usually similar to those obtained with the dilution method, which fluctuated between 8 and 18 102h1 (mean = 12 3 102h1), and those ob-tained with Tdr incorporation, which were between 5.74 and 20.53 102h1 (mean = 13 6.22 102 h1). The temporal variations in bacterial growth rates measured simultaneously using the three different methods were also similar (r 4 0.7, Po 0.05). After a fall between the two first sampling dates, the values of this variable increased during July and then fell more or less progressively until the end of the study (Fig. 1b). Statistically, there was no significant difference (P 4 0.05, ANOVA) between the three methods. Accordingly, in order to calculate bacterial production for each sampling date, we used the mean growth rates obtained with all three methods. These values of bacterial production varied from 2.8 to 14.1 106cells L1
h1 (mean-6.8 7.9 106cells L1h1), and showed similar temporal changes to those of bacterial abundance (r = 0.8, Po 0.05) (Figs 1a and c).
Bacterial mortality and bacterivory by flagellates (Table 1)
The bacterial mortality rates measured by Tdr labeling and by the dilution method showed similar temporal changes, despite a difference of about one degree of magnitude between the two types of value (Fig. 2b). The bacterial mortality rates estimated with the dilution method varied between 5 and 14 102h1 (mean = 11 4 102
h1), whereas those obtained after Tdr labeling varied from 20 to 120 102
h1 (mean = 100 30 102
h1) and were sig-nificantly higher (ANOVAand t-test, Po 0.001) than those obtained with the dilution method.
The rates of ingestion of bacteria by flagellate protists, calculated from the ingestion rates of FLB, assuming that the flagellates consumed these natural tracers at the same rate as they consumed bacteria in the unlabeled environment, fluctuated between 0.04 and 1.00 102h1 (mean = 0.50 0.44 102h1). In temporal terms, these grazing rates varied in the same manner as the mortality rates estimated with the Tdr and dilution methods, i.e. in parallel with the abundance of flagellates but in an opposite phase bacterial abundance and production (r 4 /0.5/, Po 0.05) (Figs 1a and 2b). Bacteria (x 10 cells L ) 0 2 4 6 8 10 12 Fla g ellate (x 10 cells L ) 0 1 2 3 4 5 6 Growth rate (h ) 0.00 0.05 0.10 0.15 0.20 0.25
Sampling Dates (yy/mm/dd)
0 4 8 12 16 (a) (b) (c) BP (x 10 cells L h )
Fig. 1. Fluctuations in (a) bacterial and flagellate abundances, (b) bacterial growth rates and (c) bacterial production (BP) at the sampling depth during the study. Data points represent the average determined from replicate samples, with error bars equivalent to one standard deviation.
Table 1. Growth and mortality rates of bacteria assessed by means of different methods in the Sep Reservoir. Data are given as range and mean SD (parentheses)
Dialysis bag method Dilution method Thymidine method Flagellate grazing Growth rates ( 102h1) 7.00–13.00 (10.00 3.00) 8.00–18.00 (12.00 3.00) 5.74–20.53 (13.00 6.22) ND
Mortality rates ( 102h1) ND 5.00–14.00 (11.00 4.00) 20.00–120.00 (100.00 30.00) 0.04–1.00 (0.50 0.44) ND, not determined.
Furthermore, the residual bacterial mortality, i.e. the mortality obtained after Tdr labeling in the absence of grazers, and therefore related to causes other than predation, varied between 10 and 110 102h1 (mean-50 40 102h1) (Fig. 2c). The temporal variations in this residual mortality were apparently independent of the other variables measured in this study. The values of the ratio between the ingestion rate and bacterial production, considered as the fraction of bacterial production that was consumed by bacterivorous flagellates, varied from 5% to 74%, the mean value being 39 39% (Fig. 2c).
Discussion
Estimates of bacterial production
We did not find any significant difference in bacterial growth rates estimated by the three methods tested in this study. This suggests that there are no systematic differences be-tween these methods. The assumptions of the dilution method (Landry et al., 1995) have been discussed
compre-hensively by Tranvik (1989) for bacterioplankton. Filtration under low vacuum (as done in the present study) for the preparation of the diluent remains of great importance, because this operation could induce cell breakage and enrichment of the experimental samples, and thus increase the apparent microbial growth over that occurring in situ (Ferguson et al., 1984). Also, using the dilution method in some systems or under certain conditions may affect the interpretation of results (Murrell & Hollibaugh, 1998), e.g. under saturated grazing conditions (cf. Evans & Paranjape, 1992), when grazing response is under the threshold level (Gifford, 1988; Kamiyama, 1994), or when prey are asso-ciated with suspended particles. Bacterial abundances in the Sep Reservoir (1.13–8.48 109
cells L1) are typical of oligo-mesotrophic lakes (Jugnia et al., 1999) and, during this study, were over the range of values of 0.5 to 2 109cells L1 known to be the feeding threshold prey conditions for the bacterivory response of flagellates in pelagic systems (e.g. Tanaka et al., 1997). In addition, the bacterial community in the Sep Reservoir is largely dominated by free-living cells (Jugnia et al., 1999). It is thus likely that in our dilution experiments, bacteria were growing at similar rates as in the reservoir.
The bacterial growth rates obtained using the Tdr incor-poration method also seemed to be acceptable, two of the three sources of error inherent in using this method having been taken into account in our study. The main source of error is the isotope dilution. This is generally considered to be negligible when the experimental concentration of the radioactive tracer is high enough (as in our experiments, i.e. 20 nM) to saturate the rate of bacterial incorporation (Robarts & Zohary, 1993). A second source of error is the choice of the conversion factor between the quantity of Tdr incorporated and the bacterial abundance. This factor varies greatly in space and time from 1 to 60 1018cells per mole of Tdr (cf. Bell, 1988; Iriberri et al., 1990). In our study, a conversion factor was determined, which considerably re-duced the influence of this source of error. On the other hand, the methodological bias caused by the known inability of certain bacterial strains to take up and incorporate Tdr into their DNA (e.g. Pseudomonas spp., sulfate-reducing bacteria) was a source of error that was difficult to take into account, given that our knowledge of the species composition of pelagic bacterial assemblages is actually limited (Lee & Fuhrman, 1991; Robarts et al., 1994). As a general rule, the variations in pelagic bacterial production measured using radioactive tracers and related to differences in the metabolic affinities that various bacterial strains can have for these tracers are still unknown, because of the great diversity within and between the species of the natural bacterioplankton.
Development of bacteria on the walls of dialysis bags, the main source of error associated with the dialysis bag method, has been considered to be negligible for 24 h 0.0
0.4 0.8 1.2 1.6
Sampling dates (yy/mm/dd)
Residual mortality rate (h ) 0.0 0.2 0.4 0.6 0.8 1.0 1.2 BP (x 10 Cells L h ) 0 4 8 12 16 Flagellate ngestion rate (h ) 0.000 0.003 0.006 0.009 0.012 Grazing/BP 0.0 0.2 0.4 0.6 0.8 Mortality (Dilution) rate (h ) 0.00 0.04 0.08 0.12 0.16 (a) (b) (c) Mortality (thymidine) rate (h )
Fig. 2. Fluctuations in (a) bacterial mortality rates, (b) bacterial produc-tion (BP) and ingesproduc-tion rates and (c) residual mortality, with the ratio between the ingestion rate and bacterial production. Data points represent the average determined from replicate samples, with error bars equivalent to one standard deviation.
incubations (Herndl et al., 1993). It is undeniable that this method is the one that most closely simulates natural conditions, with very little sample processing. However, it is difficult to choose between the three methods tested in terms of the estimation of bacterial growth rates, because of the similarity of the results that we obtained.
Estimates of bacterial mortality
The bacterial mortality rates based solely on the Tdr method in this study were always higher than those obtained with the dilution method (Fig. 2a). The rates obtained with these two methods in this study, as in other studies (Vaqu´e et al., 1994), were higher than the mortality rates related solely to the bacterivorous activity of flagellates, measured using bacterial tracers observed directly in the food vacuoles of these protists (Table 1). It is well known that other plankton organisms can consume bacteria (e.g. ciliates, micrometa-zoa), and that cell lysis caused mainly by phages can be a loss factor for aquatic bacterial communities (e.g. Bettarel et al., 2004; Ram et al., 2005). In the Sep Reservoir, the minimum host density for efficient viral infection was found to be 7 108
bacteria L1 (Ram et al., 2005). Throughout this study, bacterial abundance fluctuated from 1.72 to 9.83 109
cells L1 (mean = 5.56 3.20 109
cells L1), therefore providing favourable conditions for viral infec-tion. This indicates that bacteria – virus interactions could have been a significant source of bacterial mortality during the study. In the Sep Reservoir, micrometazoa consume considerable quantities of bacteria (Thouvenot et al., 1999a, b). The method using Tdr has some limitations and uncertainties (Torreton, 1991; Servais et al., 1989; Caron et al., 1993). First, this method relies in principle on the measurable decrease of a conservative constituent of bacter-ia, not subject to internal turnover or catabolism. DNA does not fulfill this requirement. Second, when the specific labeling of the bacteria is low, this can lead to the production of free tritiated catabolites, which are likely to be reincorpo-rated into intracellular molecules during long periods of incubation (Torreton, 1991). Also, with long periods of incubation of growing bacteria with Tdr, significant amounts of label were undoubtedly incorporated into cellular components of the bacterial cell other that DNA (e.g. proteins and RNA) (Caron et al., 1993; Zubkov & Sleigh, 1996). It is therefore likely that protozoa that grazed on these bacteria accumulated Tdr from a variety of bacterial molecules rather than exclusively from DNA. Another limit of this method is the interference of the precipitable TCA fraction that is attributable to labeled bacteria consumers and not to bacteria themselves (Torre-ton, 1991). All of the above mentioned factors could explain the relatively higher mortality rates observed with the Tdr method, as well as the significant differences observed in the
estimates of bacterial mortality between the Tdr method and the dilution method.
The main advantage of the dilution technique is its ability to provide a simultaneous estimation of bacterial growth and mortality in a single experiment. However, the dilution factor is used in regression analyses of the dilution experi-ment as a proxy of the relative grazing activity of micro-zooplankton (Landry & Hassett, 1982). Using fluorescent-labeled particles (FLPs) for estimating bacterivory can provide this relative grazing activity. The last method requires short incubation times, gives unambiguous indica-tion of grazer activity, and can be considered as an indepen-dent estimate of micro-zooplankton grazing activity. Nevertheless, because surrogate prey – such as the FLMs or heat-killed nonmotile natural bacteria used in this study – can be selected for or against by protist consumers (Sherr & Sherr, 1993), FLP methods should actually be considered as providing an index of the relative grazing activity on the natural bacteria which they are designed to mimic. From this context, it therefore appears that dilution and FLP methods could complement one another in a hybrid experi-mental design, where the relative grazing activity is provided by the FLP technique (Landry et al., 1995). This would compensate for a critical and unrealistic assumption of the dilution method, i.e. that grazing activity is directly propor-tional to the dilution of the population density (Evans & Paranjape, 1992). As already suggested (Landry & Hassett, 1982), we believe that the challenge is to use existing methodologies in such a way that their respective strengths are emphasized and their weaknesses mitigated or at least minimized.
Seasonal bacterial production and loss rates The rates of grazing by flagellates on bacteria in the Sep Reservoir, and the bacterial and flagellate abundances that we report, were usually similar to those recorded using the same methodological approaches in environments of similar trophic status (Vaqu´e & Pace, 1992; Mathes & Arndt, 1994; Carrias et al., 1996,1998), including earlier studies con-ducted in the Sep Reservoir during the period of thermal stratification (Thouvenot et al., 1999a, b). In the present study, the bacterial growth and mortality rates could explain the changes in the structure that were observed in the bacterial community. When the growth rate was higher than the mortality rate, an increase in bacterial abundance was observed, and vice versa (Figs 1a and 2b). The bacterial mortality rates obtained using the three independent meth-odological approaches were all positively correlated with one another and with the abundance of heterotrophic flagellates, all these variables being also negatively correlated with bacterial abundance and production. As with the results of similar studies reported in various aquatic
environments (cf. Fenchel, 1986; Tranvik 1989; Vaqu´e & Pace, 1992; Carrias et al., 1996, 1998), such relationships suggest that the bacterivorous activity of flagellates is a major factor controlling the structure and function of bacterial communities in the Sep Reservoir.
According to Fenchel (Fenchel, 1986), greater fluctua-tions in flagellate numbers are expected in eutrophic waters, whereas more stability is expected in oligotrophic situations. Also, this author observed that the logarithm of the ratio between the maximum and the minimum flagellates popu-lation numbers may fluctuate between 4.2 in eutrophic systems and 0.4 in oligotrophic systems. This gives rise to the general interpretation that in oligotrophic systems, the flagellates just manage to keep the bacterial number slightly below their carrying capacity (the number they would reach in the absence of predation), while in more eutrophic systems, the mean number of bacteria is kept well below the carrying capacity by grazing. In this study, the logarith-mic value of the above mentioned ratio, from an oligo-mesotrophic environment, was 1.5, indicating that the impact of grazing by flagellates was substantial in determin-ing the mean number of bacteria in the Sep Reservoir.
The fraction of the bacterial production that is consumed by flagellates in the pelagic environment is rather variable. In Chesapeake Bay, MacManus and Furhman (McManus & Furhman, 1988) reported values ranging from 23% to more than 100%. In the surface waters of Lake Oglethorpe, Sanders et al. (1989) estimated that flagellates consumed between 62% and 80% of bacterial production. In the metalimnion and epilimnion of the stratified Lake Vechten, Bloem and Bar-Gillissen (Bloem & B¨ar-Gilissen, 1989) reported that bacterial consumption by flagellates accounted for 30–100% of the bacterial production. In the mesotrophic Lake Arlington, this percentage was 68% (Chrzanowski & Simek, 1993), and it averaged 86% in the oligotrophic lake Annecy (Domaizon et al., 2003). In most cases, the values were greater than 30%, and in some cases could exceed 100%. The fraction of the bacterial production that transited via the grazing activity of heterotrophic flagellates in the Sep Reservoir lies within the range generally reported for the pelagic environment. This fraction varied from 5% to 74%, the mean value being 39%. Moreover, from a recent seasonal study in Sep Reservoir, Ram et al. (2005) reported that the virus-mediated bacterial mortality ranged from 1% to 60%.
The comparison of bacterial growth rates (M = 12 3 102h1) and mortality rates (M = 11 4 102h1) obtained simultaneously by the dilution method indicated that bacterial production almost completely balanced bac-terial losses during this study. The bacbac-terial ingestion rate accounted on average for 39% (range 7–93%) of total bacterial mortality when estimated by the dilution method, but this value fell to 5% (1–11%) when the total mortality was measured by the Tdr-labeling method. These
calcula-tions suggest that sources of bacterial mortality other than just predation by flagellates were significant in the Sep Reservoir. This involvement supports the conclusions of a previous study in this reservoir (Thouvenot et al., 1999a), which showed that Cladocera and especially the species Daphnia longispina and Ceriodaphnia quadrangula had high bacterivore activity. Ciliated protozoa and micrometazoa such as rotifers are also potential grazers of bacteria in pelagic systems (Vaqu´e & Pace, 1992; Jurgens, 1994; Toth & Kato, 1997), including in the Sep Reservoir (Thouvenot et al., 1999). It is also known that lysis by phages can play an important role in the mortality of the bacterioplankton (e.g. Noble et al., 1999); this was substantiated in the Sep Reservoir by a recent study in the system indicating that viruses and flagellates contributed about equally to bacterial mortality, by each destroying 55% of the bacterial production (Ram et al., 2005).
Conclusions
This study highlighted the importance of methodological aspects related to the elaboration of quantitative models of the flows of matter and energy through aquatic trophic networks. A critical function attributed to bacteria grazers in pelagic systems is that they control the prey biomass and thus prevent their accumulation (Fenchel, 1982). Our results indicate a clear link between flagellated protozoa and their bacterial prey in the Sep Reservoir. A second function of pelagic flagellated protozoa is to serve as a trophic link between bacteria and the higher-level consu-mers. Given the functional importance of flagellate grazing in the Sep Reservoir, it is likely that these grazers form one of the pathways for carbon flows to the higher trophic levels in this environment. Other sources of bacterial losses (e.g. metazoan competition and sedimentation) in this reservoir are apparently also prevalent. Their relative quantitative importance remains unknown.
Acknowledgements
We gratefully acknowledge financial support from the European Community and from the following French national and regional organizations: ‘Ministe`re de l’Envir-onnement’, ‘Agence de l’eau Loire-Bretagne’, Conseil R´egional d’Auvergne’, ‘Conseil G´en´eral du Puy-de-Doˆme’, and ‘Syndicat des Agriculteurs Irrigants de la Haute Morge’. The two anonymous reviewers of the manuscript are also thanked for their critical comments.
References
Azam F, Fenchel T, Field JG, Gray JS, Meyer-Reil LA & Thingstad F (1983) The ecological role of water-column microbes in the sea. Mar Ecol Prog Ser 10: 257–263.
Bell RT (1988) Thymidine incorporation and estimates of bacterioplankton production: are the conversion factors valid? Arch Hydrobiol Beih Ergeb Limnol 31: 163–171.
Bettarel Y, Sime-Ngando T, Amblard C & Dolan J (2004) Viral activity in two contrasting lake ecosystems. Appl Environ Microbiol 70: 2941–2951.
Bloem J & B¨ar-Gilissen M-JB (1989) Bacterial activity and protozoan grazing potential in a stratified lake. Limnol Oceanogr 34: 297–309.
Caron DA (1983) Techniques for enumeration of heterotrophic and phototrophic nanoplankton using epifluorescent microscopy, and comparison with other procedures. Appl Environ Microbiol 46: 1922–1928.
Caron DA, Lessard EJ, Voytek M & Dennett M (1993) Use of tritiated thymidine (Tdr) to estimate rate of bacterivory: implications of label retention and release by bacterivores. Mar Microb Food Webs 7: 177–196.
Carrias JF, Amblard C & Bourdier G (1996) Protistan bacterivoy in an oligomesotrophic lake: importance of attached ciliates and flagellates. Microb Ecol 31: 249–268.
Carrias JF, Amblard C, Quiblier-Lloberas C & Bourdier G (1998) Seasonal dynamics of free living and attached heterotrophic nanoflagellates in an oligotrophic lake. Freshwater Biol 39: 91–101.
Chrzanowski TH & Simek K (1993) Bacterial growth and losses due to bacterivory in a mesotrophic lake. J Plankton Res 15: 771–785.
Cole JJ & Caraco NF (1993) The pelagic food web of oligotrophic lakes. Aquatic Microbiology (Ford TE, ed), pp. 101–111. Blackwell, Boston.
Del Giorgio PA & Gasol JM (1995) Biomass distribution in freshwater plankton communities. Am Nat 146: 135–152. Domaizon I, Viboud S & Fontvieille D (2003) Taxon-specific and
seasonal variations in flagellates grazing on heterotrophic bacteria in the oligotrophic Lake Annecy – importance of mixotrophy. FEMS Microb Ecol 46: 317–329.
Ducklow HW & Carlson CA (1992) Oceanic bacterial production. Adv Microb Ecol 12: 113–181.
Evans GT & Paranjape MA (1992) Precision of estimates of phytoplankton growth and microzooplankton grazing when the functional response of grazers may be non linear. Mar Ecol Prog Ser 80: 285–290.
Fenchel T (1982) Ecology of microflagellates II. Bioenergetics and growth. Mar Ecol Prog Ser 8: 225–231.
Fenchel T (1986) The ecology of heterotrophic microflagellates. Adv Microb Ecol 9: 57–97.
Ferguson RL, Buckley EN & Palumbo AV (1984) Response of marine bacterioplankton to differential filtration and confinement. Appl Environ Microbiol 47: 49–55.
Gifford DJ (1988) Impact of grazing by microzooplankton in the northwest arm of Halifax Harbor, Nova Scotia. Mar Ecol Prog Ser 47: 249–258.
Hagstrom A, Larsson U, Horstedt P & Normark S (1979) Frequency of dividing cells, a new approach to the
determination of bacterial growth rates in aquatic environments. Appl Environ Microbiol 37: 805–812. Herndl GJ, Kaltenbok E & Muller-Niklas G (1993) Dialysis bag
incubation as a non radiolabelling technique to estimate bacterioplankton production in situ. Handbook of Methods in Aquatic Microbial Ecology (Kemp PF, Sherr BF, Sherr EB & Cole JJ, eds), pp. 495–503. Lewis publishers, Boca Raton, FL. Hollibaugh JT, Furhman JA & Azam F (1980) Radioactive
labeling of natural assemblages of bacterioplankton for use in trophic studies. Limnol Oceanogr 25: 172–181.
Iriberri J, Unanue B, Ayo B, Barcinia I & Egea L (1990) Bacterial production and growth rate estimation from (3H)thymidine incorporation for attached and free-living bacteria in aquatic systems. Appl Environ Microbiol 56: 483–487.
Jugnia L-B, Tadonleke RD, Sime-Ngando T, D´evaux J & Andrivon C (1999) Bacterial population dynamics, production and heterotrophic activity in a recently formed reservoir. Can J Microbiol 45: 747–753.
Jugnia L-B, Richardot M, Debroas D, Sime-Ngando T & D´evaux J (2000a) Variations in the number of active bacteria in the euphotic zone of a recently flooded reservoir. Aquat Microb Ecol 22: 251–259.
Jugnia L-B, Tadonl´ek´e RD, Sime-Ngando T & D´evaux J (2000b) The microbial food web in the recently flooded Sep Reservoir: diel fluctuations in bacterial biomass and metabolic activity in relation to phytoplankton and flagellate grazers. Microb Ecol 40: 317–329.
Jurgens K (1994) Impact of daphnia on planktonic microbial food webs – a review. Mar Microb Food Webs 8: 295–324. Kamiyama T (1994) The impact of grazing by microzooplankton
in northern Hiroshima Bay, the Seto Inland Sea. Japan Mar Biol 119: 77–88.
Landry MR & Hassett RP (1982) Estimating the grazing impact of marine microzooplankton. Mar Biol 67: 283–288.
Landry MR, Kirshtein J & Constantinou J (1995) A redefined dilution technique for measuring the community grazing impact of microzooplankton, with experimental tests in the central equatorial pacific. Mar Ecol Prog Ser 120: 53–63. Lee S & Fuhrman JA (1991) Spatial and temporal variation of
natural bacterioplankton assemblages studied by genomic DNA cross-hybridization. Limnol Oceanogr 36: 1277–1287. Mathes J & Arndt H (1994) Biomass and composition of
protozooplankton in relation to lake trophy in north German lakes. Mar Microb Food Webs 8: 357–375.
McManus GB & Furhman JA (1988) Clearance of bacteria size particles by natural population of nanoplankton in the Chesapeake bay outflow plume. Mar Ecol Prog Ser 42: 199–206. Murrell MC & Hollibaugh JT (1998) Microzooplankton grazing
in northern San Francisco Bay measured by the dilution method. Aquat Microb Ecol 15: 53–63.
Newell SY, Sherr BF, Sherr EB & Fallon RD (1983) Bacterial response to presence of eucaryotic inhibitors in water from a coastal marine environment. Mar Environ Res 10: 147–157.
Noble RT, Middelboe M & Fuhrman JA (1999) Effects of viral enrichment on the mortality and growth of heterotrophic bacterioplankton. Aquat Microb Ecol 18: 1–13.
Pace ML & Bailiff MD (1987) Evaluation of a fluorescent microsphere technique for measuring grazing rates of phagotrophic microorganisms. Mar Ecol Prog Ser 40: 185–193.
Porter KG & Feig YS (1980) The use of DAPI for identifying and counting aquatic microflora. Limnol Oceanogr 25: 943–948. Porter KG, Sherr EB, Sherr BF, Pace M & Sanders RW (1985)
Protozoa in planktonic food webs. J Protozool 32: 409–415. Ram ASP, Boucher D, Sime-Ngando T, Debroas D & Romagoux
JC (2005) Phage bacteriolysis, protistan bacterivory potential, and bacterial production in a freshwater reservoir: coupling with temperature. Microb Ecol 50: 64–72.
Robarts RD & Zohary T (1993) Fact or fiction – bacterial growth rate and production as determined by [3H-methyl]thymidine? Adv Microb Ecol 13: 371–425.
Robarts RD, Arts MT, Evans MS & Waiser MJ (1994) The coupling of heterotrophic bacterial and phytoplankton production in a hypertrophic, shallow prairie lake. Can J Fish Aquat Sci 51: 2219–2226.
Sanders RW, Porter KG, Bennett SJ & Debiase AE (1989) Seasonal patterns of bacterivory by flagellates, ciliates, rotifers, and cladocerans in a freshwater planktonic community. Limnol Oceanogr 34: 673–687.
Servais P, Billen G, Martinez J & Vives-Rego J (1989) Estimating bacterial mortality by the disappearance of3H-labeled intracellular DNA. FEMS Microbiol Ecol 62: 119–126. Servais P, Billen G & Vives-Rego J (1985) Rate of bacterial
mortality in aquatic environments. Appl Environ Microbiol 49: 1448–1454.
Sherr EB & Sherr BF (1993) Protistan grazing rates via uptake of fluorescently labeled prey. Handbook of Methods in Aquatic Microbial Ecology (Kemp PF, Sherr BF, Sherr EB & Cole JJ, eds), pp. 695–702. Lewis Publishers, Boca Raton, FL.
Sherr BF, Sherr EB, Andrew TA, Fallon RD & Newell SY (1986) Trophic interaction between heterotrophic protozoa and bacterioplankton in estuarine water analysed with selective metabolic inhibitors. Mar Ecol Prog Ser 32: 169–180.
Sime-Ngando T, Bourdier G, Amblard C & Pinel-Alloul B (1991) Short-term variations in specific biovolume of different bacterial forms in aquatic ecosystems. Microb Ecol 21: 211–226.
Simek K & Straskrabova V (1992) Bacterioplankton production and protozoan bacterivory in a mesotrophic reservoir. J Plankton Res 14: 773–787.
Simon M & Azam F (1989) Protein content and protein synthesis rates of planktonic marine bacteria. Mar Ecol Prog Ser 52: 201–213.
Strom SL (2000) Bacterivory: interactions between bacteria and their grazers. Microbial Ecology of the Oceans (Kirchman DL, eds), pp. 351–386. Wiley-Liss, New York.
Tanaka T, Naoji F & Taniguchi A (1997) Predator prey eddy in heterotrophic nanoflagellate–bacteria relationships in a coastal marine environment: a new scheme for predator–prey associations. Aquat Microb Ecol 13: 249–256.
Thouvenot A, Debroas D, Richardot M & Devaux J (1999a) Bacterivory of metazooplankton, ciliates and flagellates in a newly flooded reservoir. J Plankton Res 21: 1659–1679. Thouvenot A, Debroas D, Richardot M & Devaux J (1999b)
Impact of natural metazooplankton assemblage on planktonic microbial communities in a newly flooded reservoir. J Plankton Res 21: 179–199.
Torreton JP (1991) Importance des bact´eries h´et´erotrophs a´erobies dans une lagune eutrophe tropicale (Lagune Ebri´e, Coˆte d’Lvoire): biomasse, production, exportations. Universit´e d’Aix-Marseille II, Marseille.
Toth LG & Kato K (1997) Size selective grazing of bacteria by Bosmina longisrostris – an image analysis study. J Plankton Res 19: 1477–1493.
Tranvik LJ (1989) Bacterioplankton growth, grazing mortality and quantitative relationship to primary production in a humic and clear water lake. J Plankton Res 11: 985–1000. Vaqu´e D & Pace ML (1992) Grazing on bacteria by flagellates and
cladocerans in contrasting food-web structure. J Plankton Res 14: 307–321.
Vaqu´e D, Gasol JM & Marras´e C (1994) Grazing rates on bacteria: the significance of methodology and ecological factors. Mar Ecol Prog Ser 109: 263–274.
Wicks RJ & Robarts RD (1987) The extraction and purification of DNA labeled with [methyl-3H]thymidine in aquatic bacterial production studies. J Plankton Res 9: 1159–1166.
Zubkov MV & Sleigh MA (1996) Bacterivory by the ciliate Euplotes in different states of hunger. FEMS Microbiol Ecol 20: 137–147.
