Bacterioplankton community composition in a boreal forest lake

Bacterioplankton community composition in a boreal forest lake

FEMS Microbiology Ecology 27 (1998) 163^174 Bacterioplankton community composition in a boreal forest lake Eva S. Lindstroëm * Department of Limnolog...

313KB Sizes 4 Downloads 54 Views

FEMS Microbiology Ecology 27 (1998) 163^174

Bacterioplankton community composition in a boreal forest lake Eva S. Lindstroëm * Department of Limnology, Uppsala University, Norbyvaëgen 20, SE-752 36 Uppsala, Sweden Received 13 March 1998; revised 25 May 1998; accepted 24 June 1998

Abstract The composition of the dominating populations within a bacterioplankton community was investigated in a mesotrophic, boreal forest lake. Composite samples were collected monthly throughout the lake for two years. The community composition was determined by denaturing gradient gel electrophoresis (DGGE) of a polymerase chain reaction (PCR)-amplified part of 16S rDNA, extracted from organisms smaller than 1 Wm. Temporal patterns of occurrence in the lake differed among populations. There was no clear seasonal pattern of variation, but there was a gradual change. The results suggest that variation in the amount of water flowing into the lake could explain some of the changes in the bacterioplankton community. z 1998 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. Keywords : Bacterioplankton; Dimictic lake; 16S rDNA; Denaturing gradient gel electrophoresis; Temporal variation

1. Introduction Due to di¤culties in identifying the organisms, the community composition of bacterioplankton in lakes and oceans is almost unknown. However, recently molecular methods for identifying microorganisms in nature have been developed. Most of these methods are based on studies of 16S rRNA nucleotide sequences. Analysis of 16S rRNA is a powerful tool for investigations of evolutionary relationships, since this molecule possesses a so-called `clock-like behaviour', whereby its nucleotide sequence re£ects the evolutionary history of the organism in focus [1]. Since each taxon has a unique sequence of 16S

* Tel.: +46 (18) 4712717; Fax: +46 (18) 531134; E-mail: [email protected]

rRNA this molecule can also be used for identi¢cation. A number of studies using this approach for taxonomic identi¢cation of bacterioplankton has been performed in marine environments (e.g. [2,3]) showing that bacterioplankton communities consist of a large number of previously unrecognised organisms. Thus, the application of molecular methods to marine bacterioplankton communities has o¡ered new insight into a previously unstudied aspect of bacterioplankton taxonomy. The studies of freshwater bacterioplankton community composition are not as numerous as those of marine bacterioplankton. The results from the few studies conducted (e.g. [4,5]) indicate that freshwater bacterioplankton communities can resemble marine communities in some ways, and that there probably are a large number of `new' taxa to be discovered in freshwater communities as well. Thus,

0168-6496 / 98 / $19.00 ß 1998 Federation of European Microbiological Societies. Published by Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 6 4 9 6 ( 9 8 ) 0 0 0 6 5 - 8

FEMSEC 951 5-10-98

164

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

the taxonomy of freshwater bacterioplankton should also attract attention. Most studies of the community composition of marine and freshwater bacterioplankton have been performed sporadically in both time and space, resulting in limited information about population dynamics. Studies that have focused on the di¡erences in community compositions have led to the suggestion that variations are related to di¡erences in phytoplankton community compositions [6], salinity [7], acidi¢cation [8], or oxygen concentrations [9]. However, extensive studies that integrate the taxonomic perspective of bacterioplankton into bacterioplankton ecology, for instance in relation to physical, biological, and chemical factors in the environment, are still needed. In this study the community composition of bacterioplankton was followed in a mesotrophic, boreal forest lake. The aim was to determine the monthly variation of the dominating populations in the lake during two years, and to analyse these variations in relation to other parameters. The community composition was studied by use of denaturing gradient gel electrophoresis (DGGE) of a polymerase chain reaction (PCR)-ampli¢ed region of 16S rDNA (16S rRNA-encoding DNA). DGGE can separate DNA-fragments of the same size according to their nucleotide sequences, owing to differences in their melting temperatures, so that the di¡erent populations in the sample will give rise to bands at di¡erent positions on the gel [10]. Thus, the gel patterns obtained provide a picture of the dominant populations, that can be used for comparisons between communities in diverse habitats, e.g. bio¢lms [10], hot-spring microbial mats (e.g. [11]), and bacterioplankton in estuaries [12] and in a meromictic lake [9].

2. Material and methods 2.1. Study sites The main sampling program was performed in Lake Siggeforasjoën, which is a small lake with a surface area of 0.76 km2 , and a maximum depth of 11 m. The character of the lake is mesotrophic, since it has a total phosphorus content of 8^27 Wg l31 . It is

also moderately humic, with a water colour of 50^ 125 mg Pt l31 . The site is described in detail by Blomqvist et al. [13]. Additionally, one sample was taken from Lake Haîlsjoën, with a surface area of 0.22 km2 and a maximum depth of 10 m. The total phosphorus content of the lake is 30^50 Wg l31 , and the water colour is 30^40 mg Pt l31 (Lindstroëm, unpublished data). Lake Haîlsjoën, in contrast to Lake Siggeforasjoën, occasionally becomes anoxic in deeper strata owing to its wind-sheltered location. Both lakes are dimictic and situated near Uppsala in the southern part of central Sweden. 2.2. Lake sampling In Lake Siggeforasjoën, water samples were taken monthly between March 1994 and April 1996, with the exception of December 1994, when the ice-conditions on the lake prevented sampling. Water was sampled from 12 randomly chosen sites in the lake. Samples representing 0^8 m were taken with a 2-mlong tube sampler. The number of samples taken from each 2-m-interval was proportional to the volume of water within each interval. The total volume of the integrated sample was 50 l. From this sample, two subsamples of approximately 20 l each were taken for taxonomic analysis of the bacterioplankton community. All equipment used for sampling and ¢ltration of the water used for the taxonomic analysis of bacterioplankton was autoclaved or rinsed with 70% ethanol and/or lake water to avoid contamination. Upon arrival at the laboratory, following 1 h of transport, the samples were stored at lake temperature until ¢ltration. Subsamples were taken for bacterioplankton production measurements and determination of bacterioplankton abundance. The latter were immediately preserved with formaldehyde (¢nal concentration 4%). Subsamples for analysis of chemical parameters were analysed immediately upon arrival at the laboratory. Subsamples used for biomass determination and analyses of the taxonomic composition of zooplankton and phytoplankton were preserved with acidi¢ed Lugol's solution [14]. The zooplankton in 3 l of water were concentrated in a 40-Wm net before preservation. Water temperature was measured at 1-m intervals from 0^8 m depth using the thermometer in a Ruttner water sampler. Samples for analysis of dissolved

FEMSEC 951 5-10-98

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

oxygen (O2 ) were taken at 2-m intervals and analysed by Winkler titration [15]. 2.3. Collection of bacterioplankton cells and DNA extraction DNA was extracted from 3^15 l of the composite sample according to the protocol by Fuhrman et al. [16]. Brie£y, the cells were ¢ltered onto a Durapore 142-mm, 0.22-Wm ¢lter (Millipore) after pre-¢ltration through an A/E glass-¢bre ¢lter (nominal pore size 1 Wm, Gelman Sciences) within 6^8 h of sampling. The ¢lters were immediately frozen and stored at 320³C or 380³C until DNA extraction was performed according to [16]. Because humic compounds have been reported to a¡ect the PCR-reaction [17], an additional puri¢cation step was added to remove these substances. This step involved gel ¢ltration through spin columns made of Sephadex G-50 (Pharmacia) packed in sterile syringes [18]. The concentration of DNA was determined spectrophotometrically at 260 nm. Samples for determination of bacterioplankton abundance were taken from the water ¢ltered through the A/E glass-¢bre ¢lter (to estimate the retention of cells by the glass-¢bre ¢lter) and from the water that passed through the Durapore ¢lter, to make sure that the bacterioplankton had been e¤ciently collected. These samples were analysed as described below. Overnight cultures of Serratia marcescens, Escherichia coli, Bacillus subtilis and Bacillus sp. (possibly B. circulans) were used to test the separating capability of the DGGE. DNA was extracted from these cultures using a protocol that was the same as the one described above except that the ¢ltration and gel ¢ltration steps were excluded. 2.4. Thermal ampli¢cation of rDNA Thermal ampli¢cation was applied to certain parts of the 16S rDNA of the bacterioplankton cells from which the extracted DNA originated. `Universal' primers corresponding to E. coli positions 907^926 (primer 1) and 1392^1406 (primer 2) were used [19]. This part of the 16S rRNA includes the highly variable V3 and V9 regions [20]. Additionally, there was a GC-clamp (a sequence 40 base pairs long and rich

165

in G+C) added to the 5P-end of primer 1 as recommended by Myers et al. [21]. The high melting temperature of the GC-clamp enhances the accuracy with which di¡erences in the melting temperature of the DNA-fragment of interest can be detected. The sequence of the GC-clamp was published previously [22]. The base sequences of the primers were as follows : primer 1: 5P-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCC CCG CCC GAA ACT T/CAA AT/GG AAT TGA CGG-3P (GCclamp underlined); and primer 2: 5P-ACG GGC GGT GTG TA/GC-3P (where `T/C' indicates that the primer contains T or C and so on). For thermal ampli¢cation of 1^1.5 ng genomic DNA, 50 pmol of each primer was used together with 0.25 units of AmpliTaq DNA Polymerase LD (Perkin Elmer), 1.5^3 mM (¢nal concentration) of GeneAmp MgCl2 (Perkin Elmer), and 200 WM of GeneAmp dNTPs (¢nal concentration of each dNTP) (Perkin Elmer), in a total volume of 50 Wl GeneAmp PCR Bu¡er (Perkin Elmer). The solution was then overlaid with two drops of Mineral Oil (Sigma). The thermal cycling was performed on a Minicycler (MJ Research). An initial step at 94³C for 3 min was followed by 20 cycles of 1 min at 94³C, annealing at 65^55³C (in the ¢rst cycle annealing was performed at 65³C; the temperature was then lowered by 0.5³C/ cycle, `touchdown') for 1 min, and primer extension at 72³C for 3 min. This was followed by 10 cycles of 1 min each at 94³C, 1 min at 55³C and 3 min at 72³C. Finally, a primer extension at 72³C for 7 min was performed. To make sure that the products from the thermal ampli¢cation were of correct size they were analysed on a 1% agarose gel (Sigma) and compared with a commercial base pair-ladder (Pharmacia). To screen for contamination, a negative control was always run together with the samples in the reaction. This control was prepared in the same way as the samples except that the DNA was excluded. 2.5. Melting behaviour of the 16S rDNA fragment To determine whether the chosen fragment of the 16S rDNA was suitable for DGGE analysis [21], the theoretical melting behaviour was analysed using the software program MacMelt 1.0 (Bio-Rad). The nucleotide sequence of the Escherichia coli K12 rrnB

FEMSEC 951 5-10-98

166

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

operon was obtained from the Ribosomal Database Project version 3.1.1 [23] and analysed for positions 907^1406. The corresponding GenBank number for the sequence is J01695. The melting behaviour of E. coli 16S rDNA from the chosen region was also analysed by using a perpendicular gradient gel, in which the gradient of denaturant was perpendicular to the electrophoresis direction [21]. This analysis was used to determine the concentration of the gradient of the denaturant for the DGGE-analysis, as described by Myers et al. [21]. 2.6. DGGE-analysis The gel was prepared and run as described by Myers et al. [21]. The PCR-products (5^30 Wl) were loaded on a 40^70 percentage denaturant, 1.0-mm gradient gel, and run at 200 V for 6 h at 60³C, using the gel-system DGGE 2000 (CBS Scienti¢c). When lake samples were analysed, a mixture of DNA from S. marcescens and B. subtilis was always analysed in parallel to the samples. Positions of the bands representing these two taxa were used as standards. Since these bands always should be formed at the same denaturant concentration in the gel, their position was used to compare the patterns formed in di¡erent gels. Band positions were determined using a plastic ruler placed on the gel during documentation. To test the separating capabilities of the gels, a mixture of equal amounts of DNA from E. coli, Bacillus sp., B. subtilis and S. marcescens was analysed. To determine whether the results obtained were reproducible, one sample from Lake Siggeforasjoën was analysed twice. Additionally, one sample collected from the anoxic layer of Lake Haîlsjoën was analysed to test the ability to detect di¡erences between lake samples. This sample was collected in September 1995, at a depth of 9 m, using a Ruttner water sampler. The sample was treated in the same way as the samples from Lake Siggeforasjoën. The gels were stained with ethidium bromide and visualised in UV-light. The documentation of the gel images was performed with the gel documentation system 66-1000 (Techtum LAB) including a CCDvideo camera (Sony) and the image capturer Image-BW Plus (version 4.2). The gel images were then further processed using the software program

Adobe Photoshop 3.0 to maximise the contrast of the images. 2.7. Similarity coe¤cient calculation The pairwise similarity of the gel-patterns of the di¡erent samples was calculated using Sorensen's index, Cs = 2j/(a+b) [24] where j is the number of bands common to both samples, a is the number of bands in sample A, and b is the number of bands in sample B. This number was then multiplied by 100 to obtain the percentage similarity. A value of 0 indicates that the samples were completely di¡erent, while a value of 100 indicates that they were identical. 2.8. Bacterioplankton community abundance and speci¢c growth rate To determine bacterioplankton abundance, the formaldehyde-preserved cells were stained with acridine orange and counted using an epi£uorescence microscope [25]. The mean cell volume (MCV) was determined by measuring the size of 100 cells per sample. To calculate bacterioplankton biomass, the bacterioplankton abundance was multiplied by MCV and 0.308 pg C Wm33 [26]. The analysis was performed in duplicates. Bacterioplankton growth rate was measured as thymidine incorporation [27], with modi¢cations according to Vrede [28]. Triplicates of 10-ml subsamples were incubated at in situ temperature for 60^135 min. The ¢nal concentration of [3 H]thymidine (Amersham) in the samples was 4 nM. The growth rate of the bacterioplankton community (cells l31 h31 ) was calculated according to Bell [29]. The speci¢c growth rate (W) was calculated as the growth rate divided by the bacterioplankton abundance in the lake sample. 2.9. Water chemistry analyses, zooplankton and phytoplankton counts Standard chemical parameters (i.e. dissolved and particulate forms of nitrogen, phosphorus and organic carbon, inorganic and organic forms of phosphorus and nitrogen, alkalinity, pH, conductivity and suspended particles) were analysed according to Goedkoop and Sonesten [15]. The water colour

FEMSEC 951 5-10-98

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

167

was determined as the absorbance at 420 nm in a 5-cm cuvette. Qualitative changes in the organic matter were estimated by calculating the ratio between the absorbance of the water at 250 nm and its absorbance at 365 nm [30]. Zooplankton and phytoplankton samples were analysed in an inverted microscope after sedimentation to determine their taxonomic composition and biomass [14]. 2.10. Estimation of water £ow Since no data on the water £ow of Lake Siggeforasjoën were available, water £ow data from a nearby stream (River Vattholmaaîn) were obtained from the Swedish Meteorological and Hydrological Institute [31]. These values were assumed to be proportional to the water £ow in Lake Siggeforasjoën. 3. Results 3.1. Reliability of the DGGE Analysis of the theoretical melting behaviour of the 16S rDNA of E. coli K12 (positions 907^1406) showed that the DNA-fragment consists of one domain with a mean melting temperature of about 75³C (Fig. 1a). Adding the GC-clamp alters the

Fig. 1. The theoretical melting behaviour of Escherichia coli 16S rDNA, positions 907^1406, determined by MacMelt analysis. a: Without GC-clamp. b: With a 40-bp GC-clamp at the 5P-end.

Fig. 2. Negative image of an ethidium bromide stained DGGEgel. Lane 1: Negative control; lane 2: sample from the anoxic layer of Lake Haîlsjoën collected on 15 September 1995; lane 3: sample from Lake Siggeforasjoën collected on 15 September 1995; lane 4: mixture of equal amounts of DNA from Escherichia coli, Bacillus subtilis, Serratia marcescens and Bacillus sp.; lane 5: E. coli alone ; lane 6: Bacillus subtilis; lane 7: Serratia marcescens ; lane 8: Bacillus sp. The image was processed using Adobe Photoshop 3.0.

DNA-fragment melting behaviour somewhat. The GC-clamp melts at a temperature of about 95³C, while the rest of the sequence melts in one domain at about 75³C (Fig. 1b). This is the desirable melting behaviour of a DNA-fragment to be analysed by DGGE [21]. The electrophoresis of E. coli 16S rDNA on a perpendicular gradient gel also supported these results (results not shown). The analysis of DNA originating from the four pooled pure cultures of bacteria (Fig. 2) showed that the gel conditions used were appropriate to simultaneously distinguish between the DNA-fragments of closely and more distantly related bacteria. The analysis of samples from the two di¡erent bacterioplankton communities (from Lake Siggeforasjoën and from the anoxic layer of Lake Haîlsjoën) revealed clearly di¡erent gel patterns (Fig. 2). One of the samples from Lake Siggeforasjoën (collected in September 1995) was analysed twice (Figs. 2 and 3). The two gel pro¢les obtained show essentially the same patterns. Hence, the results were reproducible, and it is possible to detect di¡erences between samples. The negative control (thermal ampli¢cation without addition of genomic DNA) did not give rise to any visible bands (Fig. 2); i.e. no contamination of ex-

FEMSEC 951 5-10-98

168

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

ternal DNA during the ampli¢cation process could be detected. 3.2. DNA yield As determined by microscope counts, the number of cells passing through the pre-¢ltering glass-¢bre ¢lter, and thereby available for DNA-extraction, was 63^119% (mean 93%) of the number of cells in the original sample (Fig. 4). The number of cells that passed through the Durapore ¢lter was less than 10% of the number in the original sample (results not shown). Thus, virtually all bacterioplankton cells in the samples were collected. The amount of DNA extracted from the samples was 1.1^19.8 Wg DNA/l ¢ltered lake water (mean 4.7). 3.3. Community composition of Lake Siggeforasjoën bacterioplankton The 25 samples were analysed on three parallel DGGE-gels. The image from one of these gels is shown in Fig. 3, and the gel patterns from all gels are shown in Fig. 5. On each occasion 6^15 (mean 9.7) bands were visible, and in total 32 bands were formed at di¡erent positions on the gels. Two bands were present in every sample. The other bands showed di¡erent patterns in their appearance. Seven

Fig. 3. Negative image of gel patterns obtained from analysis of samples from Lake Siggeforasjoën. Lane 1: Sample collected 18 August 1995; lane 2: 15 September 1995; lane 3: 13 October 1995; lane 4: 23 November 1995; lane 5: 20 December 1995; lane 6: 25 January 1996; lane 7: 22 February 1996; lane 8: 29 March 1996; lane 9: 19 April 1996. The image was processed using Adobe Photoshop 3.0.

Fig. 4. Bacterioplankton abundance in Lake Siggeforasjoën between March 1994 and April 1996 (mean of duplicates). Open circles : original lake sample. Filled circles: sample ¢ltered through glass-¢bre ¢lter type A/E (nominal pore size 1 Wm).

bands were consistently found in consecutive samplings for 2^8 months, but were not detected on any other occasion. Sixteen bands were found on several occasions (2^5) and lasted for 1^14 months. Seven bands were detected on just one occasion. The similarity analysis shows that the resemblance between samples was highest for those collected close to each other in time (Table 1). Thus, out of the 25 samples, 19 were most similar to the sample collected one month earlier or to the sample collected one month later. The other six samples were most similar to samples collected 2^5 months earlier or later. The highest degree of similarity with another sample ranged between 67 and 96% (mean 86%); i.e. every sample had at least two thirds of the detected taxa in common with one other sample, but no two samples ever had the same taxonomic composition. The lowest similarity value obtained between any two samples was 18%, showing that no samples were completely di¡erent from each other, owing to the two consistently occurring bands. Samples collected during the same season in di¡erent years did not show any tendency to increased similarities to each other. Nor was there any apparent connection between the degree of similarity between samples collected close to each other in time and season, indicating no clear connection to seasonally determined dramatic changes in the lake, such as circulation and coverage with ice and snow (Fig. 5).

FEMSEC 951 5-10-98

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

169

Fig. 5. Gel patterns obtained from analysis of samples from Lake Siggeforasjoën on 25 occasions. The samples were collected between March 1994 and April 1996. Solid bands represent bands that were clearly visible, while grey bands represent bands that were more di¤cult to detect, either because of low intensities or because they were in a position with a high level of background noise. The arrows represent occasions on which there was a circulation in the lake. The boxes at the top of the picture represent the duration of snow cover (white) and ice cover (grey) on the lake. Table 1 The di¡erent gel patterns obtained from Lake Siggeforasjoën compared by Sorensen's index, Cs Date

1994 M

M A M J J A S O N J F M A M J J A S O N D J F M A

100

a

1995

1996

A

M

J

J

A

S

O

N

J

F

M

A

M

J

J

A

S

O

N

D

J

F

M

A

75 100

59 67 100

63 71 80 100

59 80 63 80 100

63 57 80 86 67 100

63 71 78 82 67 82 100

56 75 71 75 71 75 95 100

42 47 33 47 56 47 40 42 100

25 29 27 43 40 43 35 38 82 100

59 53 38 53 50 53 56 47 56 53 100

50 56 32 44 53 44 48 50 86 78 63 100

50 56 32 44 53 44 48 50 86 78 63 91 100

33 50 47 50 47 38 42 44 74 75 35 70 80 100

32 35 33 47 44 47 40 42 80 82 44 76 76 74 100

53 62 57 62 57 62 50 53 63 62 43 59 59 67 63 100

56 50 59 50 47 63 53 56 53 50 24 50 50 56 53 67 100

53 47 56 47 44 59 50 53 50 47 22 48 48 53 50 63 95 100

48 53 60 53 50 53 55 57 45 42 20 43 43 57 45 56 86 91 100

40 44 53 44 42 44 48 50 38 44 21 36 36 50 38 47 80 86 96 100

43 48 55 48 45 48 50 52 42 38 18 40 40 52 42 50 78 83 92 88 100

38 42 40 32 40 42 36 38 45 42 30 43 43 38 36 44 67 73 75 78 85 100

40 44 42 33 42 44 38 40 48 44 32 55 45 40 38 47 70 67 61 64 72 87 100

38 32 40 32 30 42 36 38 27 32 20 35 26 29 27 44 67 64 67 70 77 75 78 100

50 36 52 45 35 55 48 42 32 27 26 38 31 33 32 48 67 64 67 62 76 67 69 89 100

a

Similarity per de¢nition 100%. For every sample the highest Cs obtained is in bold.

FEMSEC 951 5-10-98

170

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

Table 2 Spearman rank correlation coe¤cient values and P-values showing the correlations between the similarity in gel patterns (Cs ) and changes in some characters of Lake Siggeforasjoën, between consecutive sampling occasions

Cs

v Water £ow (m3 s31 )

v Water colour (Abs 420 nm)

v NH4 (Wg l31 )

v Alkalinity (meq l31 )

30.45 P 6 0.05

30.52 P 6 0.05 0.40 P = 0.05

30.44 P 6 0.05 0.51 P 6 0.05 0.05 P = 0.82

30.42 P 6 0.05 0.31 P = 0.14 0.54 P 6 0.05 0.17 P = 0.42

v Water £ow v Water colour v NH4

3.4. Water chemistry and biota Among the other biota, only the phytoplankton showed a clear seasonality in taxonomic composition with diatoms dominating the community (on a carbon basis) in October of both years, and cryptophytes and chrysophytes on all other occasions (results not shown). This shift in the phytoplankton community did not coincide with a speci¢cally large change in the bacterioplankton community (Table 1). The meta- and proto-zooplankton did not show a seasonal pattern regarding the taxonomic composition of the communities. The metazooplankton were always dominated by copepodes, and cladocerans were always the second most abundant group. Among the protozooplankton ciliates dominated, and the unpigmented chrysophytes were always the second most dominating group. Dissolved oxygen was never completely depleted in the lake. The lowest value obtained (0.74 mg l31 ) was from the deepest layer of the lake (8 m), in August 1994. For every lake parameter (except oxygen concentrations, and the taxonomic composition of phytoand zooplankton) that was determined, the di¡erences, in absolute values, between the values obtained at adjacent dates were calculated. Correlations between these di¡erences and the similarity coe¤cient (Cs ) were determined by Spearman rank correlation, in order to assess the extent to which changes in these parameters a¡ected the dynamics of the bacterioplankton community (Table 2). For every one of the parameters, the correlation with Cs was consistently low, indicating that no single parameter determined the taxonomic changes in this community.

Additionally, these results showed that there were no connections between £uctuations in the taxonomic composition of bacterioplankton, and the £uctuations in bacterioplankton biomass (b = 30.022, P = 0.92) or growth rate (b = 30.063, P = 0.76). The only correlations with Cs that were statistically signi¢cant (P 6 0.05) were negative and obtained for changes in water colour, water £ow, ammonium concentration, and alkalinity. The negative correlation implicates that the similarity in gel-patterns between two samples was low (i.e. the rate of change was high) when the change in the lake parameter of interest was large. The four most important lake parameters mentioned above were also correlated (positively) with each other. However, these correlations were not always statistically signi¢cant (Table 2).

4. Discussion The computer analysis in combination with the perpendicular electrophoresis showed that the chosen fragment of 16S rDNA was suitable for DGGEanalysis. Results of the tests involving bacterial cultures and lake bacterioplankton demonstrate that the DGGE-pattern can be assumed to give a reasonable view of the taxonomic composition of a bacterioplankton community, assuming that each band represents a population in the lake sample. The DNA extracted can also be assumed to be representative for the bacterioplankton community since most cells in the lake sample were collected for DNA-extraction, and the amount of DNA extracted was high (theoretically 2.8 fg DNA cell31 , if the amount of extracted DNA is divided by the number of cells

FEMSEC 951 5-10-98

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

per litre). This is slightly lower than the values obtained (6.4 fg DNA cell31 ) in an oligotrophic humic lake [32]. Since the technique used in that study is similar to that in the present one, i.e. lysis of the cells with a detergent, it can be assumed that the amount of DNA extracted in both studies should be similar. However, it has previously been noted that additional DNA puri¢cation steps cause reductions in the DNA yield [16]. Thus, it can be assumed that the puri¢cation steps performed in this study led to a lower DNA yield compared to the other lake study [32] where no additional puri¢cation of the DNA was done. Since the loss of DNA during puri¢cation should occur randomly regardless of its taxonomic origin, this loss should not have a¡ected the genomic composition of the samples. Thus, it can be assumed that the gel-patterns obtained give a reasonable view of the bacterioplankton community. It can be argued that the number of bands on the DGGE-gels can overestimate the actual diversity of the sample due to the formation of heteroduplex [33] or chimeric molecules [34], and also because of sequence heterogeneities of di¡erent copies of the 16S rRNA gene [35]. An actual overestimation can also be seen in Fig. 2 where DNA from E. coli gave rise to two bands instead of one. It can also be argued that the number of bands can underestimate the actual diversity since the PCR may not amplify all sequences equally well [36], and since DNA-fragments of di¡erent nucleotide sequence may have the same melting temperature, and thereby are not possible to separate by DGGE [37]. In fact, the number of bands formed per sample on the gel (6^15, mean 9.7) was low compared to the number of taxa usually discovered, when the taxonomic composition of a bacterioplankton community is investigated by use of a cloning-sequencing approach (e.g. [2]). However, this is not surprising, since there must be a large number of taxa in the sample which were not abundant enough to give rise to visible bands on the gel. It has previously been shown in marine bacterioplankton communities, that a few species dominated on several occasions during a year (e.g. [38,39]). At one of these occasions it was found that the 13 most numerous taxa accounted for 86% of the nucleoid-containing cells [38]. Thus, if it is assumed that the bands on the DGGE-gels represent the dominant taxa in the sample, the number of

171

these in this study were in the same order of magnitude as in the previous one. Consequently, it can be assumed that the DGGE-patterns do not re£ect the composition of the whole community, instead they are estimates of the dominant populations in the lake, and it can be assumed that the presence or absence of individual bands should re£ect true changes in the occurrence of that particular population in the lake. Based on the gel-patterns obtained by DGGE, it should therefore be possible to estimate general similarities and di¡erences between communities. However, an estimate of the diversity of the sample, expressed for example as number of taxa per litre, is di¤cult to make. Additionally, no attempt to quantify the abundance of the members of each population was made since band intensity may not entirely re£ect the abundance of an organism in the sample, due to biases in the PCR-reaction [36], and since the number of copies of the 16S rRNA gene di¡ers between taxa [40]. Instead, all visible bands were included when calculating the degree of similarity between samples (Table 1). Two of the detected bacterioplankton populations were present at every sampling occasion, indicating that there was an omnipresent component of the system, causing stability. The gel-pattern obtained from analysis of the bacterioplankton community of the anoxic layer of Lake Haîlsjoën was clearly different, indicating that this component was not the result of artefacts associated with the sampling or processing of the lake samples. Other populations exhibited di¡erent patterns of occurrence, showing up in the lake samples for shorter or longer periods. It appears that the taxa studied di¡er in terms of their growth and survival strategies as well as in their success in colonising the lake. The results of the pairwise similarity comparisons showed that the majority of the samples were most similar to other samples collected at adjacent sampling occasions, and that there was no trend towards a higher similarity between samples collected at the same season in different years. Thus, there were no dramatic shifts in the composition of the bacterioplankton community, and there were no obvious relationships between the £uctuations and season or taxonomic composition of other biota. Instead, the community appeared to be relatively stable, showing slow, gradual changes.

FEMSEC 951 5-10-98

172

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

This successional pattern is di¡erent from what has previously been suggested regarding the taxonomic composition of marine bacterioplankton communities, where season-dependent `blooms' of specific strains of bacterioplankton [39] and couplings to the taxonomic composition of phytoplankton [6] have been suggested. Discrepancies in the results between the marine studies and the present one may have several explanations. One reason could, at least in one of the studies [39], be attributed to di¡erences in methods. In the present study, a survey of the dominant components of the community in the lake was made, whereas in the former study [39] the abundance of only a few strains was followed. Thus, if strains that were not studied showed another successional pattern, this could not be detected. A more basic di¡erence between marine and freshwater systems could also have contributed to the disparate results. For example, freshwater systems should be more stable than marine systems owing to di¡erences in the composition and origin of the organic material present [41]. In the ocean, labile organic matter excreted by phytoplankton and other biota dominates and may therefore be present only sporadically. In freshwaters, a large pool of recalcitrant allochthonous material is always present, which could cause stability in contrast to seasonality. The allochthonous material should also lead to that the organic material provided from phytoplankton is quantitatively less important for the bacterioplankton, leading to a lack of coupling between these two communities. This could certainly be true in a humic lake like Lake Siggeforasjoën. Since there were only weak correlations between the similarity coe¤cient (Cs ) and the changes of some of the characters of the lake, no single parameter caused major changes in the community. However, the strongest correlations obtained suggest that the changes in the amount of water that entered the lake from the drainage area altered the composition of the dominant bacterioplankton populations. Thus, the organisms brought to the lake from the drainage area may, at least occasionally, be important components of the pelagic bacterioplankton community. The water in Lake Siggeforasjoën has a relatively short residence time. The average turnover time of the lake has been determined to be 209 days, with a minimum during spring of about 24 days [13].

Thus, the ecosystem of this lake should be strongly in£uenced by transport of material from the drainage basin. This means that it may not be possible to explain £uctuations in the composition of bacterioplankton in lakes without considering a coupling between the drainage basin and the bacterioplankton community. To summarise, the bacterioplankton community of Lake Siggeforasjoën, as investigated by the DGGE-analysis, consisted of a number of populations di¡ering in their patterns of occurrence in the lake. Although it was not possible to identify the factors in£uencing the occurrence of individual taxa, it can be concluded that several factors were probably involved since the populations di¡ered so greatly in their patterns of occurrence. Additionally, it was found that the changes in the bacterioplankton community did not appear to be occurring due to season, instead there was a continuous, gradual change. To be able to explain which factors are important in controlling the community composition of lake bacterioplankton, the results obtained in this study lead to the suggestion that the following questions should be addressed: (1) Is the import of microorganisms from other ecosystems such as the drainage area, the littoral zone, and the lake sediments large enough to a¡ect the composition of bacterioplankton communities in lakes? If so, are these microorganisms active in their new environment, and are they thereby contributing to ecosystem functioning? (2) Is there a general di¡erence between the successional patterns of the composition of bacterioplankton communities in aquatic systems with di¡erent dominating carbon sources? If this is the case, it could be expected that in the ocean, or in a clearwater lake, the taxonomic succession would be more subject to season, for instance following the taxonomic succession of the phytoplankton community. In contrast, in humic lakes, the successional pattern could be di¡erent, with no apparent coupling to other pelagic biota or season of the year.

Acknowledgments I thank Peter Blomqvist, Jan Johansson, Kjell Hellstroëm and Raul Figueroa for help with sampling and plankton and chemical analysis. Karin Carlson

FEMSEC 951 5-10-98

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

and Dan Andersson at the Department of Microbiology, Uppsala University, provided me with cultures. Peter Blomqvist, Katarina Vrede, Lars Tranvik, and John Stockner are acknowledged for comments on the manuscript. David Tilles corrected the English. This work was supported by grants from the Swedish Environmental Protection Board (SNV) to Ingemar Ahlgren, and from the Olsson-Borgh Foundation to the author.

[12]

[13]

[14]

[15]

References [1] Woese, C.R. (1987) Bacterial evolution. Microbiol. Rev. 51, 221^271. [2] Fuhrman, J.A., McCallum, K. and Davis, A.A. (1993) Phylogenetic diversity of subsurface marine microbial communities from the Atlantic and Paci¢c oceans. Appl. Environ. Microbiol. 59, 1294^1302. [3] Giovannoni, S.J., Britschgi, T.B., Moyer, C.L. and Field, K.G. (1990) Genetic diversity in Sargasso Sea bacterioplankton. Nature 345, 60^62. [4] Bahr, M., Hobbie, J.E. and Sogin, M.L. (1996) Bacterial diversity in an arctic lake: a freshwater SAR11 cluster. Aquatic Microb. Ecol. 11, 271^277. [5] Alfreider, A., Pernthaler, J., Amann, R., Sattler, B., Gloëckner, F.-O., Wille, A. and Psenner, R. (1996) Community analysis of the bacterial assemblages in the winter cover and pelagic layers of a high mountain lake by in situ hybridization. Appl. Environ. Microbiol. 62, 2138^2144. [6] Acinas, S.G., Rodr|èguez-Valera, F. and Pedroès-Alioè, C. (1997) Spatial and temporal variation in marine bacterioplankton diversity as shown by RFLP ¢ngerprinting of PCR ampli¢ed 16S rDNA. FEMS Microbiol. Ecol. 24, 27^40. [7] Gonzaèlez, J.M. and Moran, M.A. (1997) Numerical dominance of a group of marine bacteria in the K-subclass of the class Proteobacteria in coastal seawater. Appl. Environ. Microbiol. 63, 4237^4242. [8] Hiorns, W.D., Metheè, B.A., Nierzwicki-Bauer, S.A. and Zehr, J.P. (1997) Bacterial diversity in Adirondack mountain lakes as revealed by 16S rRNA gene sequences. Appl. Environ. Microbiol. 63, 2957^2960. [9] Òvreaîs, L., Forney, L., Daae, F.L. and Torsvik, V. (1997) Distribution of bacterioplankton in meromictic Lake S×lenvannet, as determined by denaturing gradient gel electrophoresis of PCR-ampli¢ed gene fragments coding for 16S rRNA. Appl. Environ. Microbiol. 63, 3367^3373. [10] Muyzer, G., De Waal, E.C. and Uitterlinden, A.G. (1993) Pro¢ling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-ampli¢ed genes coding for 16S rRNA. Appl. Environ. Microbiol. 59, 695^700. [11] Ferris, M.J., Muyzer, G. and Ward, D.M. (1996) Denaturing gradient gel electrophoresis pro¢les of 16S rRNA-de¢ned pop-

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

173

ulations inhabiting a hot spring microbial mat community. Appl. Environ. Microbiol. 62, 340^346. Murray, A.E., Hollibaugh, J.T. and Orrego, C. (1996) Phylogenetic composition of bacterioplankton from two California estuaries compared by denaturing gradient gel electrophoresis of 16S rDNA fragments. Appl. Environ. Microbiol. 62, 2676^ 2680. Blomqvist, P., Heyman, U. and Grundstroëm, R. (1981) The structure of the pelagic ecosystem in Lake Siggeforasjoën. Scr. Limnol. Uppsal. A, 522. Utermoëhl, H. (1958) Zur Vervollkummnung der quantitativen Phytoplankton-metodic. Mitt. Int. Ver. Theor. Angew. Limnol. 9, 38. Goedkoop, W. and Sonesten, L. (1995) Laborationsmanual foër kemiska och fysikaliska analyser av inlandsvatten och sediment. Scr. Limnol. Uppsal. B, 12. î . and Chan, A.M. Fuhrman, J.A., Comeau, D.E., Hagstroëm, A (1988) Extraction from natural planktonic microorganisms of DNA suitable for molecular biological studies. Appl. Environ. Microbiol. 54, 1426^1429. Tsai, Y.-L. and Olson, B.H. (1992) Rapid method for separation of bacterial DNA from humic substances in sediments for polymerase chain reaction. Appl. Environ. Microbiol. 58, 2292^2295. Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning, a Laboratory Manual, 1st Edn. Cold Spring Harbor, NY. Lane, D.J., Pace, B., Olsen, G.J., Stahl, D.A., Sogin, M.L. and Pace, N.R. (1985) Rapid determination of 16S ribosomal RNA sequences for phylogenetic analyses. Proc. Natl. Acad. Sci. USA 82, 6955^6959. Gray, M.W., Sanko¡, D. and Cedergren, R.J. (1984) On the evolutionary descent of organisms and organelles: a global phylogeny based on a highly conserved structural core in small subunit ribosomal RNA. Nucleic Acids Res. 12, 5837^ 5852. Myers, R.M., She¤eld, V.C. and Cox, D.R. (1988) Detection of single base changes in DNA: ribonuclease cleavage and denaturing gradient gel electrophoresis. In: Genome Analysis, A Practical Approach (Davies, K.E., Ed.) pp. 95^139. IRL Press, Oxford. Wawer, C. and Muyzer, G. (1995) Genetic diversity of Desulfovibrio spp. in environmental samples analyzed by denaturing gradient gel electrophoresis of [NiFe] hydrogenase gene fragments. Appl. Environ. Microbiol. 61, 2203^2210. Maidak, B.L., Larsen, N., McCaughey, M.J., Overbeek, R., Olsen, G.J., Fogel, K., Blandy, J. and Woese, C.R. (1994) The ribosomal database project. Nucleic Acids Res. 22, 3485^3487. SÖrensen, T. (1948) A method of establishing groups of equal amplitude in plant sociology based on similarity of species content. Biol. Skr. 5, 1^34. Hobbie, J.E., Daley, R.J. and Jasper, S. (1977) Use of nuclepore ¢lters for counting bacteria by £uorescence microscopy. Appl. Environ. Microbiol. 33, 1225^1228. Fry, J.C. (1988) Determination of biomass. In: Methods in Aquatic Bacteriology (Austin, B., Ed.). Wiley, New York, NY.

FEMSEC 951 5-10-98

174

E.S. Lindstroëm / FEMS Microbiology Ecology 27 (1998) 163^174

[27] Bell, R.T. (1993) Estimating production of heterotrophic bacterioplankton via incorporation of tritiated thymidine. In: Handbook in Methods in Aquatic Microbial Ecology (Kemp, P.F., Sherr, B.F., Sherr, E.B. and Cole, J.J., Eds.), pp. 495^503. Lewis, Boca Raton, FL. [28] Vrede, K. (1996) Regulation of bacterioplankton production and biomass in an oligotrophic clearwater lake ^ the importance of the phytoplankton community. J. Plankton Res. 18, 1009^1032. [29] Bell, R.T. (1990) An explanation for the variability in the conversion factor deriving bacterial cell production from incorporation of [3 H]thymidine. Limnol. Oceanogr. 35, 910^915. [30] Strome, D.J. and Miller, M.C. (1978) Photolytic changes in dissolved humic substances. Verh. Intern. Ver. Limnol. 20, 1248^1254. [31] SMHI (1994^1996) Vaëder och Vatten. Swedish Meteorological and Hydrological Institute. [32] Tranvik, L.J. (1997) Rapid £uorometric assay of bacterial density in lake water and seawater. Limnol. Oceanogr. 42, 1629^1634. [33] Ferris, M.J. and Ward, D.M. (1997) Seasonal distribution of dominant 16S rRNA-de¢ned populations in a hot spring microbial mat examined by denaturing gradient gel electrophoresis. Appl. Environ. Microbiol. 63, 1375^1381. [34] Liesack, W., Weyland, H. and Stackebrandt, E. (1991) Potential risks of gene ampli¢cation by PCR as determined by 16S rDNA analysis of a mixed-culture of strict barophilic bacteria. Microb. Ecol. 21, 191^198.

[35] Cilia, V., Lafay, B. and Christen, R. (1996) Sequence heterogeneities among 16S ribosomal RNA sequences, and their e¡ect on phylogenetic analyses at the species level. Mol. Biol. Evol. 13, 451^461. [36] Suzuki, M.T. and Giovannoni, S.J. (1996) Bias caused by template annealing in the ampli¢cation of mixtures of 16S rRNA genes by PCR. Appl. Environ. Microbiol. 62, 625^630. [37] Kowalchuk, G.A., Stephen, J.R., de Boer, W., Prosser, J.I., Embley, T.M. and Woldendorp, J.W. (1997) Analysis of ammonia-oxidizing bacteria of the L subdivision of the class Proteobacteria in coastal sand dunes by denaturing gradient gel electrophoresis and sequencing of PCR-ampli¢ed 16S ribosomal DNA fragments. Appl. Environ. Microbiol. 63, 1489^ 1497. î . (1997) Dominant [38] Pinhassi, J., Zweifel, U.L. and Hagstroëm, A marine bacterioplankton species found among colony-forming bacteria. Appl. Environ. Microbiol. 63, 3359^3366. [39] Rehnstam, A.-S., Baëckman, S., Smith, D.C., Azam, F. and î . (1993) Blooms of sequence-speci¢c culturable Hagstroëm, A bacteria in the sea. FEMS Microbiol. Ecol. 102, 161^166. [40] Farrely, V., Rainey, F.A. and Stackebrandt, E. (1995) E¡ect of genome size and rrn gene copy number on PCR ampli¢cation of 16S rRNA genes from a mixture of bacterial species. Appl. Environ. Microbiol. 61, 2798^2801. [41] Hobbie, J.E. (1988) A comparison of the ecology of planktonic bacteria in fresh and salt water. Limnol. Oceanogr. 33, 750^764.

FEMSEC 951 5-10-98