OUP user menu

Bacterial community dynamics during the winter–spring transition in the North Sea

Melanie Sapp , Antje Wichels , Karen H. Wiltshire , Gunnar Gerdts
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00238.x 622-637 First published online: 1 March 2007

Abstract

Bacterioplankton dynamics at Helgoland Roads (54°11.3′N, 7°54.0′E) in the North Sea over the winter–spring transition were investigated. The bacterial community was analyzed and correlated with phytoplankton community data and abiotic parameters. The community structure was analyzed by ribosomal intergenic spacer analysis (RISA) and by denaturing gradient gel electrophoresis (DGGE) of 16S rRNA genes followed by DNA sequence analysis. The linkage of abiotic and biotic environmental factors and bacterial community as well as phylotypes (sequenced DGGE bands) was analyzed by the ordination technique of canonical correspondence analysis (CCA). Generally, an influence of temperature and phytoplankton on the bacterial community during the sampling period was observed. Additionally, multivariate analysis by factors revealed an influence on specific bacterial phylotypes of these factors. Overall, results indicate that changes in the bacterial community were caused not only by abiotic factors but also by the phytoplankton community.

Keywords
  • Helgoland roads
  • bacterioplankton
  • phytoplankton
  • RISA
  • DGGE
  • CCA

Introduction

Bacterioplankton dynamics are governed by seasonal changes in abiotic and biotic factors, especially phytoplankton dynamics. Little is known about the controlling factors and their effects either on the bacterial community or on bacterial species. There are growing numbers of studies dealing with the seasonality of bacterioplankton community composition (Shiah & Ducklow, 1994; Pinhassi & Hagström, 2000; Gerdts et al., 2004). One of the major controlling factors resulting in seasonality of the bacterial community composition was elucidated by Shiah & Ducklow (1994). Their investigation of the control of the whole bacterial community revealed temperature to be the major controlling factor in winter, autumn and spring, whereas limitation of inorganic nutrients and substrates was regarded to be the controlling factor in summer. Recent studies, including a mesocosm study, have supported this, indicating a general limitation of bacterioplankton in the summer due to organic carbon and inorganic nutrients in the natural environment (Rivkin & Anderson, 1997; Øvreås et al., 2003). A direct relationship between temperature and bacterial production (Pinhassi & Hagström, 2000) as well as seasonal succession of the community structure (Gerdts et al., 2004), was shown for marine bacterioplankton. Generally, seasonality of the bacterioplankton community structure has also been shown in freshwater ecosystems (Kent et al., 2004). Furthermore, it has been observed that seasonal forces play a structuring role in bacterial communities in lakes, as determined by a study conducted over two consecutive years (Yannarell et al., 2004). This was also shown for an Antarctic lake by Pearce (2005).

An influence of temperature on the community was particularly observed for lakes located in northern Sweden (Lindström, 2001). Additionally, several authors have observed changes in bacterial community composition during natural blooms or mesocosm phytoplankton experiments (Middelboe et al., 1995; Riemann et al., 2000; Fandino et al., 2001; Arrieta & Herndl, 2002; Pinhassi et al., 2004; Brussaard et al., 2005; Rooney-Varga et al., 2005), indicating close coupling of phytoplankton and bacterial community composition. A close link between bacterioplankton (especially attached bacteria) and phytoplankton dynamics has already been shown by Rooney-Varga et al. (2005). Consequently, the phytoplankton community seems to have a direct effect on the bacterioplankton, especially attached bacteria, at the phylogenetic level.

Approximately 20% of marine bacteria can currently be cultivated by traditional techniques or by dilution culturing (Selje et al., 2005). Therefore, the analysis of factors controlling the bacterioplankton must include culture-independent methods. Using this approach, Pinhassi et al. (2004) worked with mesocosms with different phytoplankton regimes and found that shifts in the bacterial community could be correlated with the phytoplankton composition. Brussaard et al.(2005) demonstrated that the breakdown of a Phaeocystis globosa bloom in a mesocosm study was accompanied by changes in bacterial community composition.

Here we try to elucidate the driving forces for shifts in bacterial community structure over the winter–spring transition in the North Sea. Additionally, the influence of specific abiotic and biotic factors on the phylotypes was studied. It was hypothesized that temperature as well as nutrients and phytoplankton dynamics account for the seasonality of bacterioplankton over the winter–spring transition. We investigated the changes in the free-living and attached bacterial community with ribosomal intergenic spacer analysis (RISA) and denaturing gradient gel electrophoresis (DGGE) of 16S rRNA genes, followed by DNA sequence analysis. The linkage of abiotic and biotic environmental factors and community composition was analyzed by the multivariate ordination method of canonical correspondence analysis (CCA).

Materials and methods

Study site, sample collection, abiotic and biotic factors

Samples were collected twice weekly from a depth of 1 m from February to May 2004 at Helgoland Roads (54°11.3′N, 7°54.0′E), North Sea by the motor boat Aade. The sampling period covered the change of seasons from winter to spring, and included a phytoplankton bloom consisting mainly of Phaeocystis spp.

Water temperature was measured immediately after sampling. Determination of salinity was performed using an inductive salinometer (GDT Autosal8400B Salinometer, Guildline, Ontario, Canada) followed by conversion to a salinity value using UNESCO tables (Cox, 1966; Grasshoff et al., 1999).

In order to monitor the concentration of nutrients, ammonium, nitrite, nitrate, silicate and phosphate were measured photometrically (Grasshoff & Johannsen, 1974; Grasshoff et al., 1999).

The samples for the enumeration of phytoplankton cells were preserved with Lugols' solution before algal cell numbers were determined. Twenty-five-milliliter samples were counted using the Uthermöhl method and an inverted microscope (Wiltshire & Manly, 2004).

For the enumeration of bacteria, the samples were prefiltered through 10-μm gauze filters.

Direct counting was performed as described above, using the stain Acridine Orange (Gerdts et al., 2004).

Sampling of biomass and extraction of nucleic acid

In order to collect the biomass of attached and free-living bacteria, 1 L of the seawater was filtered through 3-μm and 0.2-μm membrane filters (Millipore, Schwalbach, Germany) in succession. Filters were stored at −20°C until DNA extraction.

DNA of attached and free-living bacteria was extracted from cut filters by a modified standard protocol of Anderson & McKay (1983), omitting the NaOH step. This method is comparable to the method of Somerville et al. (1989), who efficiently extracted DNA from the pelagic environment. Briefly, cell lysis was performed by adding lysozyme (1 mg mL−1) and sodium dodecyl sulfate (1%). DNA extraction was carried out using phenol/chloroform/isoamyl alcohol (25:24:1). After precipitation of the DNA with isopropanol, all DNA extracts were eluted in sterile water and stored at −20°C until further analyses.

Amplification of ribosomal intergenic spacer and RISA

For amplification of the intergenic spacers between the 16S and 23S subunits of ribosomal sequences, we used the primers S-D-Bact-1522-b-S-20 (5′-TGCGGCTGGATCCCCTCCTT-3′) and L-D-Bact-132-a-A-18 (5′-CCG GGT TTC CCC ATT CGG-3′) (Normand et al., 1996; Ranjard et al., 2000a, b). PCR reaction mixtures with a volume of 100 μL contained 10 μL of 10 × Taq buffer (Eppendorf, Hamburg, Germany), 20 μL of 5 × Master Enhancer (Eppendorf), 300 μM each dNTP (Perkin Elmer, Rodgau-Jügesheim, Germany), 0.5 μM each primer, 2 U of Taq DNA polymerase (Eppendorf), and 0.5 μL of DNA of 0.2 μm filters and 0.5 μL of 3 μm filters, respectively [Correction added after publication 8 January 2007: in the preceding sentence 0.5 μL of DNA of 0.2 μM filters and 0.5 μL of 3 μm filters was corrected to 5 μL of DNA of 0.2 μm filters and 0.5 μL of 3-μm filters]. The amplification protocol was as follows: a denaturing step at 95°C for 3 min, 25 cycles at 95°C for 1 min, 53°C for 1 min and 72°C for 1 min, and a final extension step of 72°C for 5 min. PCR reactions were performed in an Eppendorf Mastercycler. Amplification of PCR products was confirmed by electrophoresis on a 1.4% (w/v) agarose gel. Fragments were resolved on 8% polyacrylamide gels (Qbiogene, Heidelberg, Germany) in 0.5 × (20 mM TrisHCL, 10 mM acetic acid, 0.5 mM EDTA) (TAE) buffer. Three lanes were used for 0.1 μg of a 100-bp ladder (Invitrogen, Karlsruhe, Germany) in order to achieve comparability. Electrophoresis was performed at 20°C and 50 V for 18 h, using a DCode system (BioRad, München, Germany). Gels were stained with SYBRGold as recommended by Molecular Probes (Invitrogen). Imaging was performed with the ChemiDoc XRS System (BioRad).

Amplification of 16S rRNA genes and DGGE

PCR amplification of 16S rRNA gene fragments was performed using the primers 341f with a 40-bp GC-rich sequence at the 5′ end (5′-CGC CCG CCG CGC CCC GCG CCC GGC CCG CCG CCC CCG CCC CCC TAC GGG AGG CAG CAG-3′) and modified 907rm (5′-CCG TCA ATT CMT TTR AGT TT-3′) (Teske et al., 1996). PCR reaction mixtures with a volume of 100 μL contained 10 μL of 10 × Taq buffer (Eppendorf), 20 μL of 5 × Master Enhancer (Eppendorf), 300 μM each dNTP (Perkin Elmer), 0.2 μM each primer, 2 U of Taq DNA polymerase (Eppendorf) and 0.5 μL of DNA of 0.2 μm filters and 0.5 μL of 3-μm filters, respectively [Correction added after publication 8 January 2007: in the preceding sentence 0.5 μL of DNA of 0.2 μM filters and 0.5 μL of 3 μm filters was corrected to 5 μL of DNA of 0.2 μm filters and 0.5 μL of 3-μm filters]. The ‘touchdown’ PCR started with a denaturing step at 94°C for 5 min. Every cycle consisted of three steps, each for 1 min: 94°C, annealing temperature, and 72°C. The initial annealing temperature of 65°C was decreased by 0.5°C per cycle until a touchdown of 55°C, at which temperature 12 additional cycles were carried out. Final primer extension was performed at 72°C for 10 min, and this was followed by 22 cycles starting at 71°C and decreasing by 1°C per cycle. A slow decrease of the temperature leads to more accurate reannealing of homologous DNA strands and therefore prevents the formation of heteroduplexes. PCR reactions were performed in an Eppendorf Mastercycler. PCR products were inspected on 1.2% (w/v) agarose gels. These served for estimation of the amount of PCR product used for DGGE analyses, which were performed with a BioRad DCode system (see above). Fragments were resolved on 6% (w/v) polyacrylamide gels in 0.5% TAE buffer with denaturing gradients of 15–55% urea/formamide (100% denaturant contains 7 M urea and 40% formamide). Electrophoresis was performed at 60°C and 150 V for 10 h (Sigler et al., 2004). DGGE gels were stained with SYBRGold (see ‘Amplification of ribosomal intergenic spacer and RISA’). Imaging was performed with a ChemiDoc XRS System (BioRad).

DNA sequencing

Prominent DGGE bands that connected or separated samples were excised, eluted (Sambrook et al., 1989), repeatedly cleaned with additional DGGE gels, and reamplified using the primers 341f without GC-clamp and 907rm. DNA was purified with the Qiaquick PCR purification kit (Qiagen, Hilden, Germany), following the manufacturer's protocol. Products were checked by electrophoresis in 1.2% (w/v) agarose gels. Sequencing was performed by Qiagen GmbH using an ABI PRISM 3700 DNA Analyzer (Applied Biosystems, Foster City, CA). Sequencing primers were 907rm and 344f (5′-ACG GGA GGC AGC AG-3′). Nearest relatives were searched for using blast (http://www.ncbi.nlm.nih.gov).

Phylogenetic analysis

Sequence data were checked for the presence of PCR-amplified chimeric sequences with the check_chimera program (Cole et al., 2003). The arb software package (http://www.arb-home.de) was used for phylogenetic analysis (Ludwig et al., 2004). After addition of sequences to the arb 16S rRNA gene sequences database (released June 2002), alignment was carried out with the Fast Aligner integrated in the program and refined by comparison of closest relatives retrieved by blast. Sequences with more than 1300 nucleotides were used to calculate phylogenetic trees. The arb‘parsimony interactive’ tool was used to add partial sequences to respective trees. Phylogenetic relationships were deduced by the neighbor-joining method with the correction algorithm of Felsenstein (1993).

Nucleotide sequence accession numbers

The sequences obtained in this study are available from GenBank under accession numbers DQ289508DQ289544.

Statistical analysis of RISA and DGGE profiles

We used two different fingerprinting methods to analyze the bacterial community during the sampling period. A general overview of bacterioplankton dynamics was achieved by RISA. Ordination techniques based on RISA fingerprints were performed to elucidate the factors affecting the whole bacterial community, whereas ordination techniques based on DGGE fingerprints were used to analyze the bacterial community at the phylotype level and the factors affecting specific bacterial phylotypes.

Analyses of RISA and DGGE fingerprints were carried out with the bionumerics 4.0 software package (Applied Maths, BVBA, Belgium). Normalization of RISA gels was performed by bionumerics software, using 100-bp ladders as references in every profile. Normalization of DGGE gels was performed using a specific sample including seven bands covering a broad area of positions as reference in addition to internal references in every profile. For sample comparison, band-matching analysis was performed. Bands were assigned to classes of common bands within all profiles. In the band-matching table based on DGGE fingerprints, sequenced DGGE bands were assigned to corresponding band classes. We omitted DGGE fingerprints of attached bacteria from band-matching because of bias due to plastid rRNA gene (see Table 4). The procedure resulted in band-matching tables that included densitometric values of fingerprints for both community analyses (Muylaert et al., 2002). These band-matching tables formed the basis for community ordination analysis.

View this table:
4

Relatedness of bacteria to known organisms

DGGE bandFractionPhylogenetic groupClosest relativeSimilarity (%)Base positions comparedAccession number of closest relative
F067*FreeAlphaproteobacteriaUncultured marine bacterium, D01598530AF177555
F068FreeAlphaproteobacteriaSulfitobacter sp., KMM 600681518AY682196
F105FreeAlphaproteobacteriaUncultured Alphaproteobacterium, PI_RT284100530AY580529
F111FreeAlphaproteobacteriaUncultured Alphaproteobacterium, DC11-80-2100532AY145625
F122FreeAlphaproteobacteriaUncultured Alphaproteobacterium, DC11-80-2100531AY145625
F125FreeAlphaproteobacteriaUncultured SAR116 Alphaproteobacterium, EF100-93A0689535AY627368
F130FreeAlphaproteobacteriaUncultured Rhodobacteraceae bacterium, F4C7482517AY697922
F132*FreeAlphaproteobacteriaRhodobacteraceae bacterium, AP-2799521AY145564
F182FreeAlphaproteobacteriaUncultured SAR116 Alphaproteobacterium, EF100-93A0698528AY627368
F198FreeAlphaproteobacteriaRhodobacteraceae bacterium, AP-27100530AY145564
F202FreeAlphaproteobacteriaUncultured Alphaproteobacterium, PI_4k2g97551AY580512
F203FreeAlphaproteobacteriaUncultured Alphaproteobacterium, PI_RT28491537AY580529
F291FreeAlphaproteobacteriaUncultured Alphaproteobacterium, PI_4z10f100534AY580535
F077AttAlphaproteobacteriaRhodobacteraceae bacterium, AP-2799536AY145564
F089AttAlphaproteobacteriaUncultured Alphaproteobacterium, DC11-80-290493AY145625
F098AttAlphaproteobacteriaUncultured Alphaproteobacterium, DC11-80-299562AY145625
F160AttAlphaproteobacteriaUncultured Alphaproteobacterium, DC11-80-299551AY145625
F310AttAlphaproteobacteriaRhodobacteraceae bacterium, AP-27100563AY145564
F120FreeBetaproteobacteriaUncultured bacterium, BN_3298528AY550846
F200FreeBetaproteobacteriaUncultured bacterium, BN_3295534AY550846
F069FreeGammaproteobacteriaMarine Gammaproteobacterium, HTCC2121100561AY386341
F070FreeGammaproteobacteriaMarine Gammaproteobacterium, HTCC212181541AY386341
F141FreeGammaproteobacteriaUncultured Gammaproteobacterium, PI_4r8d100560AY580742
F309FreeGammaproteobacteriaMarine Gammaproteobacterium, HTCC212198576AY386341
F074FreeFlavobacteriaUncultured Bacteroidetes bacterium, PI_4j12f99550AY580583
F114FreeFlavobacteriaUncultured Bacteroidetes bacterium, PI_RT30299544AY580649
F128FreeFlavobacteriaUncultured Bacteroidetes bacterium, AB-488551AY353556
F137FreeFlavobacteriaUncultured Bacteroidetes bacterium, PI_4d7b94569AY580580
F188FreeFlavobacteriaUncultured bacterium, BN_3494517AY550843
F197FreeFlavobacteriaUncultured Bacteroidetes bacterium, GWS-c7-FL89525DQ080954
F306FreeFlavobacteriaUncultured Bacteroidetes bacterium, B1294529AF466917
F086AttFlavobacteriaMarine psychrophile Bacteroidetes bacterium, SW1797554AF001368
F162AttFlavobacteriaUncultured Bacteroidetes bacterium, AB-495555AY353556
F172AttFlavobacteriaUncultured Bacteroidetes bacterium, PI_4d7b99546AY580580
F270AttFlavobacteriaUncultured Bacteroidetes bacterium, PI_RT30299530AY580649
F289AttFlavobacteriaUncultured Bacteroidetes bacterium, PI_4e10g98575AY580688
F184FreeActinobacteriaUncultured bacterium, ARKIA-4399538AF468297
F196FreeActinobacteriaUncultured Actinobacterium, PI_RT340100539AY580357
F296FreeActinobacteriaUncultured Actinobacterium, PI_RT34098474AY580357
F075FreeUnknown
F124FreeUnknown
F187FreeUnknown
F192FreeUnknown
F01neFreeNot excised
F02neFreeNot excised
F097AttPlastidTeleaulax amphioxeia100567AY453067
F127FreePlastidUncultured Prasinophyte LUR798526AY960282
F146AttPlastidEnvironmental clone OCS5498543AF001657
F151aAttPlastidTeleaulax amphioxeia99534AY453067
F167AttPlastidUncultured phototrophic eukaryote, ANT18/2_2599532AY135677
F169AttPlastidUncultured phototrophic eukaryote, JL-WNPG-T3699532AY664132
F278AttPlastidThalassiosira eccentrica98542AJ536458
F279AttPlastidRhizosolenia setigera, p11298555AJ536461
F326AttPlastidTeleaulax amphioxeia100566AY453067
  • * Phylotypes with same position in DGGE profile but different sequence data within the Alphaproteobacteria.

  • Phylotypes with same position in DGGE profile but different sequence data within the Bacteroidetes.

  • ne, nonexcised bands.

  • DGGE, denaturing gradient gel electrophoresis.

To test whether weighted-averaging techniques or linear methods were appropriate, detrended correspondence analysis (DCA) was performed using canoco for Windows 4.53 (Biometris, the Netherlands). The longest gradients resulting from DCA were 2.694 for the analysis based on RISA profiles and 2.218 for the analysis based on DGGE profiles. Those values did not indicate a clear linear or unimodal relationship (Lepš & Šmilauer, 2003), so we performed redundancy analysis (RDA) as well as CCA to compare species–environment correlations.

The data were not transformed prior to RDA or CCA. Explanatory variables included temperature, salinity, concentrations of inorganic nutrients, namely ammonium, nitrate, nitrite, phosphate and silicate, as well as cell numbers of phytoplankton species. To differentiate the attached and the free-living community in the ordination analyses based on RISA profiles, a categorical variable was introduced (filter). This variable was set to 0 and 1 for the free-living and for the attached community respectively. Using this method, we were able to analyze the variation of the community also with respect to the influence of the fraction (Rooney-Varga et al., 2005). Generally, RDA and CCA were performed as described by Lepš & Šmilauer (2003).

An automated forward selection was used to analyze intersample distances for both RISA and DGGE profiles. First, the variance inflation factor (VIF) of environmental variables was calculated. Variables displaying a value greater than 20 of this factor were excluded from RDA and CCA analyses, assuming collinearity of the respective variable with other variables included in the examined dataset.

The null hypothesis that species composition is independent of the measured variables was tested using constrained ordination with manual forward selection and a permutation test. The analysis was performed without transformation of data with focus scaling on intersample distances and manual selection of environmental variables, applying a partial Monte Carlo permutation test (499 permutations) including unrestricted permutation. The marginal effects of environmental variables were selected according to their significance level (P<0.05) prior to permutation. The partial Monte Carlo permutation test provided the conditional effect of each variable. To estimate the influence of the measured variables on specific phylotypes, interspecies distances were calculated from the dataset derived from sequence data.

For all community ordination analyses, biplot scaling was used.

Results

Environmental parameters, and phytoplankton and bacterial cell counts

Temperature, salinity and the concentrations of phosphate, nitrate, nitrite, ammonium and silicate were determined as abiotic factors, whereas phytoplankton and bacterial cell counts were determined as biotic environmental parameters (Fig. 1). The water temperature increased constantly until the end of May, from 4.9 to 10.5°C, despite slight variations. The values of salinity ranged from 29.33 to 33.51, and the values of the five major nutrients showed their lowest values in May (Fig. 1). Bacterial cell counts ranged from 3.69 × 106 to 0.84 × 106 cells mL−1, with substantial variation during the sampling period. Generally, a decreasing trend could be observed from February to May (Fig. 1).

1

Environmental parameters, phytoplankton and bacterial cell counts, including phytoplankton species. T. nitz, Thalassionema nitzschioides; P. sul, Paralia sulcata; O. aur, Odontella aurita; T. nor, Thalassiosira nordenskiöldii; T. dec, Thalassiosira decipiens; G. del, Guinardia delicatula; Chat. spp., Chattonella spp.; Ph. spp., Phaeocystis spp.

Phytoplankton counts revealed the appearance of six main diatom species, namely Thalassionema nitzschioides (Grunow) Grunow ex Hustedt, Paralia sulcata (Ehrenberg) Cleve, Odontella aurita (Lyngbye) C.A. Agardh, Thalassiosira nordenskiöldii Cleve, Thalassiosira decipiens (Grunow) Jørgensen, and Guinardia delicatula (Cleve) Hasle. Additionally, Chattonella spp., belonging to the family Raphidophyceae, Phaeocystis spp., belonging to the Heterokontophyta, and unclassified flagellates were counted (Fig. 1).

At the beginning of the sampling period, mainly Thalassionema nitzschioides, Par. sulcata and flagellates comprised the phytoplankton community. Thalassionema nitzschioides was observed at the beginning of March until the end of April, and Par. sulcata was observed until the end of March, whereas flagellates were apparent during the whole sampling period. Odontella aurita and Thalassiosira nordenskiöldii appeared in mid-April but disappeared at the end of that month. Thalassiosira decipiens could only be detected in mid-April for 2 weeks, whereas G. delicatula occurred first at the end of April, which was also true for Chattonella spp. Guinardia delicatula was observed until the end of the sampling period, whereas Chattonella spp. disappeared at the beginning of May. In addition, Phaeocystis spp. occurred in May and were observed until the end of the sampling period.

Community ordination analysis

Generally, we used two different fingerprinting methods to monitor changes in the bacterial community structure. RISA was performed to determine general differences between the attached and free-living bacterial communities during the sampling period (Fisher & Triplett, 1999; Ranjard et al., 2000a, b, 2001) whereas DGGE of the free-living bacterial community was used to study the changes in the bacterial community based on specific phylotypes. As these fingerprinting methods have certain limitations, both were used to analyze the influence of environmental factors on the bacterioplankton.

Community ordination analysis based on RISA profiles

Generally, ordination analysis of the bacterial community was carried out using phytoplankton species cell counts, salinity, temperature, the nutrients ammonium, phosphate and silicate (Fig. 1), and the variable filter differentiating free-living and attached bacteria. Because of indistinct DCA results, both RDA and CCA were performed according to Lepš & Šmilauer (2003) to compare species–environment correlations. Lower values of species–environment correlations were obtained by RDA than by CCA (Table 1). Although the difference was not pronounced, it was assumed that unimodal methods would be more appropriate for analysis of the large dataset. Generally, nonlinear models are required for analysis of ecological data collected over a large range of habitat variation (Jongman et al., 1987). Hence, we decided that ordination techniques based on weighted averaging would be more suitable, assuming a unimodal response of species to the environment, to elucidate the influence of the measured variables on the variation of the bacterial community.

View this table:
1

Eigenvalues and variance decomposition for CCA and RDA

AxesEigenvaluesSpecies– environment correlationsCumulative percentage variance of species dataCumulative percentage variance of species– environment correlation
Community analysis, CCA
RISA intersample distancesAxis 10.4240.94420.344.5
Axis 20.1860.90829.264.1
Axis 30.1070.79134.375.4
Axis 40.0630.59637.382.0
DGGE intersample and interspecies distancesAxis 10.4200.98147.058.2
Axis 20.1270.91161.275.9
Axis 30.0570.86067.683.9
Axis 40.0320.76071.288.3
Community analysis, RDA
RISA intersample distancesAxis 10.2910.91829.154.9
Axis 20.0990.87139.073.6
Axis 30.0490.75443.982.8
Axis 40.0240.70746.387.2
DGGE intersample and interspecies distancesAxis 10.4830.96848.362.1
Axis 20.1030.82958.675.4
Axis 30.0790.80266.585.5
Axis 40.0420.84870.790.9
  • CCA, canonical correspondence analysis; RDA, redundancy analysis; RISA, ribosomal intergenic spacer analysis; DGGE, denaturing gradient gel electrophoresis.

The constrained ordination revealed high values of the VIF (>20) for the variables nitrate and nitrite, which indicated collinearity with other variables. These factors were excluded from the final CCA.

The eigenvalues of the ordination analyses are presented in Table 1. The sum of all unconstrained eigenvalues indicated an overall variance in the dataset of 2.090. Total variation that could be explained by environmental variation accounted for 0.951, as indicated by the sum of all canonical eigenvalues. Concerning the variance of species data, the first axis explained 20.3% of the total variation, the first and the second axes explained 29.2%, and all four axes explained 37.3% (Table 1). Species–environment correlations were high, especially for axes 1 and 2 (0.944 and 0.908), indicating a relationship between species and environmental variables.

Biplot scaling of CCA is shown in Fig. 2. Canonical axes 1 and 2 are shown in Fig. 2a, demonstrating a strong influence of the nominal variable filter, which is equivalent to the respective fraction of the bacterial community, indicating a distinct differentiation of free-living and attached bacteria. This is supported by a high eigenvalue of canonical axis 1. Axis 1 represented a strong gradient caused by the nominal variable filter, indicated by the intraset correlation coefficient of 0.6052, which is also applicable for axis 2, which displayed an intraset correlation coefficient of – 0.6477 (Table 2). The influence of this variable was determined to be significant by a permutation test (Table 3). It became apparent that the variable filter had the highest value of lambda A of the conditional effects (Table 3), representing the highest additional variance explained by this variable at the time it was included in the permutation test (ter Braak & Šmilauer, 2002).

2

CCA of RISA profiles showing attached and free-living bacterial communities, obtained using phytoplankton species cell counts of flagellates (flag, Thalassionema nitzschioides (T. nitz), Paralia sulcata (P. sul), Odontella aurita (O. aur), Thalassiosira nordenskiöldii (T. nor), Thalassiosira decipiens (T. dec), Guinardia delicatula (G. del), Chattonella spp. (Chat. sp.) and Phaeocystis spp. (Ph. sp.), salinity (S), temperature (T), the nutrients ammonium (NH4), phosphate (PO4) and silicate (SiO4), as well as the variable ‘filter’ differentiating free-living and attached bacteria. Circles indicate free-living communities, and filled squares indicate attached communities. Numbers near the symbols indicate the month of sampling (2, February; 3, March; 4, April; 5, May). Arrows indicate the direction of increasing values of the respective variable, the length of arrows indicates the degree of correlation of the variable with community data, significant variables are indicated by bold arrows, and groups I, II and III of communities are indicated by a gray background. (a) Axes 1 and 2 of CCA biplot. (b) Axes 2 and 3 of CCA biplot.

View this table:
2

Intraset correlation coefficients of forwardly selected environmental variables with the four significant axes produced by CCA of RISA and DGGE fingerprints of the bacterial community

Community analysis, CCAEnvironmental factorsAxis 1Axis 2Axis 3Axis 4
RISA intersample distancesFilter0.6052−0.6477−0.11210.1065
Phaeocystis spp.0.67170.37630.13220.0380
Guinardia delicatula0.58690.4754−0.0911−0.1107
Chattonella spp.0.06070.1546−0.37040.2612
Temperature0.61640.37900.1727−0.0561
DGGE intersample and interspecies distancesTemperature0.90270.01690.0679−0.0456
Phaeocystis spp.0.6730−0.5602−0.04140.0229
Nitrite−0.7178−0.1004−0.48130.0075
  • CCA, canonical correspondence analysis; RISA, ribosomal intergenic spacer analysis; DGGE, denaturing gradient gel electrophoresis.

View this table:
3

Marginal and conditional effects of forwardly selected environmental variables produced by CCA

Community analysisEnvironmental variableMarginal effectsConditional effects
Lambda 1Lambda AP-valueF-factor
RISA intersample distancesFilter0.270.270.0027.53
Phaeocystis spp.0.260.240.0027.31
Guinardia delicatula0.230.120.0023.87
Chattonella spp.0.050.050.0661.73
Temperature0.230.050.081.6
DGGE intersample distances and interspecies distancesTemperature0.360.360.00216.67
Phaeocystis spp.0.250.090.0025.18
Nitrite0.250.050.0202.43
  • CCA, canonical correspondence analysis; RISA, ribosomal intergenic spacer analysis; DGGE, denaturing gradient gel electrophoresis.

The influence of the factors G. delicatula, temperature, Phaeocystis spp., Chattonella spp., flagellates and salinity on the bacterial communities deriving from May samples is also shown (Fig. 2a). The variables Phaeocystis spp., G. delicatula and temperature contributed particularly to the gradient, as indicated by the intraset correlation coefficients (Table 2). Significance was retrieved for the factors Phaeocystis spp. and G. delicatula but not for the factor temperature, applying the 5% significance level (P<0.08). The variable Chattonella spp. displayed minor influences on the gradient, as indicated by the intraset correlation coefficients shown in Table 2. The permutation test also showed no significance of this variable at the 5% level (P<0.08).

However, temperature and the phytoplankton Chattonella spp. contributed to the environmental variables explaining variation, as shown by biplot scaling of canonical axes 2 and 3 (Fig. 2b), indicated by the lengths of the respective arrows.

Biplot scaling of canonical axes 2 and 3 (Fig. 2b) showed that the attached bacterial communities deriving from samples of February, March and April are grouped together (group I, Fig. 2b). This group showed little influence of nutrients and phytoplankton species. Furthermore, some free-living and attached communities were combined (group II, Fig. 2b). They were influenced by temperature, Phaeocystis spp., G. delicatula and Chattonella spp. In this group, some free-living communities from the May, April, March and February samples were mainly influenced by temperature, Phaeocystis spp. and G. delicatula, whereas Chattonella spp. influenced some free-living communities from the April and March samples and attached communities from the May samples (group III, Fig. 2b). It is obvious that communities from the May samples were particularly influenced by several factors. Summarizing the effects of environmental variables, it became apparent that the nominal variable filter had the strongest conditional effect, followed by Phaeocystis spp. and G. delicatula. Note that temperature had a strong marginal influence but a minor conditional influence compared with the variables filter, Phaeocystis spp., G. delicatula and Chattonella spp.

Generally, the first two axes together explained 64% of the variation that could be explained by the variables, whereas all four axes explained 82% of the variation (Table 1).

Phylogenetic analysis

The most prominent DGGE bands that separated or connected the bacterial community were sequenced from excised bands. Sequence data of 45 excised bands could be retrieved, representing 36 different phylotypes (Table 4). Sequence data revealed the presence of three phyla of the Bacteria (Fig. 3). Most sequences were related to members of the Proteobacteria and members of the Bacteroidetes phylum. Within the Proteobacteria, the Alphaproteobacteria and Gammaproteobacteria were most abundant. Additionally, we found one member of the Betaproteobacteria and members of the Actinobacteria. Furthermore, several chloroplast sequences were detected. The closest relatives of the sequenced bands derived from blast analyses are listed in Table 4. The results revealed many close matches with 98–100% similarity to bacterial 16S rRNA gene sequences in GenBank.

3

Phylogenetic tree of Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, Actinobacteria and members of the Bacteroidetes. GenBank accession numbers are given in parentheses. Bootstrap values above 50% are displayed. Numbers in square brackets indicate the main appearance of the phylotype (2, February; 3, March; 4, April; 5, May).

Comparison of sequence data of excised bands appearing at the same position in DGGE gels revealed identical closest relatives in most cases (e.g. bands F111 and F122, F120 and F200, and F196 and F296). Moreover, bands F067 and F132, as well as F074 and F306, did not result in the same sequence, although the respective bands were retrieved from the same positions in DGGE gels (Table 4).

Proteobacteria

A neighbor-joining tree of the Alphaproteobacteria revealed that the majority of sequences belonged to the Roseobacter clade (65%). Thirty-five percent belonged to clusters of the SAR116 and SAR1 phylotypes. Gene sequences of the Gammaproteobacteria were all assigned to the Oceanospirillales, whereas the phylotype classified as a member of the Betaproteobacteria belonged to the Burkholderiales (Fig. 3). Members of the Alphaproteobacteria contributed to both the attached and free-living bacterial fractions, whereas the members of the Betaproteobacteria and of the Gammaproteobacteria detected in this study belonged solely to the free-living bacteria (Table 4).

Bacteroidetes and Actinobacteria

Within the Bacteroidetes phylum, sequenced phylotypes clustered with the Flavobacteria (Fig. 3). These phylotypes formed part of both the attached and free-living bacterial communities.

Two phylotypes of the free-living fraction were classified as Actinobacteria, with substantial similarities to already described uncultured bacteria, namely ‘uncultured bacterium ARKIA-43’ and ‘uncultured actinobacterium PI_RT340’ (Fig. 3).

Succession of free-living bacterial species and community ordination analysis based on 16S rRNA gene sequence data

Generally, 28 phylotypes were identified at different positions on the DGGE gels of the free-living bacterial community. The succession of free-living bacterial phylotypes is shown in Table 5. At the beginning of the sampling period, the community was mainly composed of members of the Alphaproteobacteria. Additionally, members of the Flavobacteria, Actinobacteria, Betaproteobacteria and Gammaproteobacteria, as well as unknown phylotypes, were detected.

View this table:

Some phylotypes were predominant during the whole sampling period, such as an alphaproteobacterium (F111), whereas some phylotypes disappeared in March or April. In April, additional members of the Alphaproteobacteria and Flavobacteria were detected, whereas a member of the Gammaproteobacteria (F069) and two unknown phylotypes [F070, F01ne (not excised)] were first detected in May samples. During this period, Actinobacterium F196, a specific Flavobacterium (F197) and Betaproteobacterium F200 were no longer found.

In order to achieve a detailed analysis of the free-living bacterial community and the factors influencing distinct phylotypes, RDA and CCA of bacterial phylotypes and environmental factors were performed. Bands F070, F01ne, F124 and F291 were omitted from ordination, as they appeared only once or twice in the dataset. Because of lower values of species–environment correlations in RDA (Table 1), we used weighted averaging to analyze the influence of environmental factors on the bacterial community and phylotypes.

The ordination analysis of the free-living bacterial community was performed using phytoplankton species cell counts, salinity, temperature and the nutrients ammonium, nitrate, nitrite and phosphate as explanatory variables (Fig. 1). CCA revealed high values of the variance inflation factor (>20) for the variable silicate, indicating collinearity with other variables. Therefore, it was excluded from CCA.

The eigenvalue of axis 1 of CCA (0.420, Table 1) represents a gradient due to the environmental factors temperature, nitrite and Phaeocystis spp., as indicated by their intraset correlation coefficients (Table 2). The overall variance in the dataset accounted for 0.894, whereas the total variation explained by environmental variation accounted for 0.721. Additionally, the first axis explained 47.0% of the total variation, the first and second axes together explained 61.2%, and all four axes explained 71.2% (Table 1).

Phylotype–environment correlations were especially high for axes 1 and 2 (0.981 and 0.911), indicating a relationship between phylotype and environmental variables. The first two axes together explained 75.9% of the variation, whereas all four axes explained 88.3% of the variation (Table 1). Biplot scaling of CCA of DGGE fingerprints is shown in Fig. 4. The analysis based on intersample distances revealed a group of free-living communities in February and March that were mainly influenced by nutritional factors and the phytoplankton species Par. sulcata (group I, Fig. 4). The free-living communities of April and May appeared in separate groups. Salinity and Chattonella spp. displayed a minor correlation with the communities retrieved from April samples, whereas the phytoplankton species O. aurita, Thalassiosira decipiens, Thalassiosira nordenskiöldii and Thalassionema nitzschioides were correlated with this group to a greater extent (group II, Fig. 4). The communities retrieved from May samples were also grouped together and were generally related to Phaeocystis spp. and G. delicatula (group III, Fig. 4). A correlation with temperature was also observed.

4

CCA biplot of intersample distances of DGGE fingerprints of the free-living bacterial community using phytoplankton species cell counts of flagellates (flag, Thalassionema nitzschioides (T. nitz), Paralia sulcata (P. sul), Odontella aurita (O. aur), Thalassiosira nordenskiöldii (T. nor), Thalassiosira decipiens (T. dec), Guinardia delicatula (G. del), Chattonella spp. (Chat. sp.) and Phaeocystis spp. (Ph. sp.), salinity (S), temperature (T), and the nutrients nitrate (NO3), nitrite (NO2), ammonium (NH4) and phosphate (PO4). Circles indicate free-living communities, and numbers near the symbols indicate the month of sampling (2, February; 3, March; 4, April; 5, May). Arrows indicate the direction of increasing values of the respective variable, and the length of arrows indicates the degree of correlation of the variable with community data. Significant variables are indicated by bold arrows, and groups I, II and III of communities are indicated by a gray background.

Significant correlation with the sample variation was retrieved by three factors, namely temperature, nitrite and Phaeocystis spp., based on the 5% level in a partial Monte Carlo permutation test (Table 3). Both marginal and conditional effects are shown in Table 3. It is apparent that the factor temperature had the strongest influence, followed by Phaeocystis spp. and nitrite. CCA with interspecies distances was performed to calculate the influence of the environmental variables on specific phylotypes. Biplot scaling revealed four groups of phylotypes and two bands that could not be grouped with the other bands (Fig. 5). Group I was mainly influenced by nutrients (nitrite) and Par. sulcata. In this dataset, the influence of nutrients could be separated from the effect of temperature, as the analysis of the VIF revealed no collinearity for both factors. Group II did not show any correlation with a variable included in our dataset. Little correlation was displayed by salinity and the phytoplankton species O. aurita and Chattonella spp. in group III. Phylotypes positively correlating with the phytoplankton species Phaeocystis spp. and G. delicatula as well as temperature were grouped within group IV. The factors affecting bacterial species were those retrieved in CCA based on DGGE fingerprints of the free-living community (Fig. 4). However, the phylotype F069 (Gammaproteobacteria) appearing at the beginning of May (Table 5) was strongly correlated with the occurrence of Phaeocystis spp., as indicated by CCA (Fig. 5).

5

Biplot of interspecies distances; CCA of DGGE fingerprints of the free-living bacterial community using phytoplankton species cell counts of flagellates (flag, Thalassionema nitzschioides (T. nitz), Paralia sulcata (P. sul), Odontella aurita (O. aur), Thalassiosira nordenskiöldii (T. nor), Thalassiosira decipiens (T. dec), Guinardia delicatula (G. del), Chattonella spp. (Chat. sp.) and Phaeocystis spp. (Ph. sp.), salinity (S), temperature (T), and the nutrients nitrate (NO3), nitrite (NO2), ammonium (NH4) and phosphate (PO4). Triangles with numbers indicate sequenced bands, and the suffix ‘ne’ indicates nonsequenced bands. Arrows indicate the direction of increasing values of the respective variable, and the length of arrows indicates the degree of correlation of the variable with community data. Significant variables are indicated by bold arrows, and groups I, II, III and IV of phylotypes are indicated by a gray background.

Discussion

The linkage of phytoplankton and bacterioplankton dynamics demonstrated by several investigations (Brussaard et al., 2005; Rooney-Varga et al., 2005) is still not well understood. Until now, it has been assumed that there are specific factors and effects controlling specific bacterial populations. The findings of Shiah & Ducklow (1994), Pinhassi & Hagström (2000), Gerdts et al. (2004) and Kent et al. (2004) were supported by our study, as we also observed seasonal succession of bacterioplankton in the winter–spring transition. Like other investigators, we found temperature to be the most important factor influencing the bacterioplankton composition of Helgoland Roads over the winter–spring transition in 2004–2005. Additionally, the phytoplankton species Phaeocystis spp. and G. delicatula displayed strong effects on the bacterial community during the winter–spring transition, leading to a decrease in the influence of abiotic factors and an increase in the influence of biotic factors, as indicated by CCA (Fig. 2b).

Community ordination analysis

Although the DCA results showed a linear or unimodal relationship, the results of the analysis are not clear. CCA was performed to analyze species–environment correlations, as the results of RDA showed lower correlation values. In general, the separation of the attached and free-living communities became evident as a strong effect on the analysis of RISA profiles. Distinct differences between attached and free-living bacteria have already been observed by DeLong et al. (1993) and Fandino et al. (2001), and in a multivariate analysis by Rooney-Varga et al. (2005).

Generally, the correlation of the measured environmental variables with the free-living fraction of the bacterial community appeared to be stronger than with the attached community, except for some attached communities in May that were strongly correlated with G. delicatula or Chattonella spp. (Fig. 2b). The correlation of Phaeocystis spp. with the bacterial community was determined to be significant for CCA of RISA and DGGE profiles (Table 3), indicating a structuring role for this phytoplankton species. This has already been described for culture experiments (Janse et al., 2000). The authors stated that bacteria with specific enzyme activities might be favored by a bloom of this Heterokontophyte. Furthermore, Brussaard et al. (2005) observed changes in the bacterial community during the breakdown of a Phaeocystis bloom in a mesocosm study that might support the finding that Phaeocystis spp. has a structuring role. It has to be considered that probably not only the presence of Phaeocystis spp. but also the increase of algal-derived organic matter due to the high total abundance of phytoplankton in May had a strong effect on the bacterioplankton community.

Besides the occurrence of Phaeocystis spp., it can also be assumed that temperature contributed to the shifts in the bacterial community, although the significance test did not reach the 5% level (P=0.08) in CCA of RISA profiles. It was correlated mainly with free-living bacterial communities in April and May, as determined by CCA of DGGE fingerprints. This aspect has already been described by Pinhassi & Hagström (2000), who observed a relationship between temperature and bacterial production. Furthermore, an influence of temperature on bacteria has been described by Shiah & Ducklow (1994) and Šestanović (2004).

However, we could not ascribe a significant factor that shaped the attached communities in samples obtained in February to April.

Generally, a single band in a RISA profile might contain several species, and a species could result in several bands (Ranjard et al., 2001), so we focused on differences between the bacterial communities determined by RISA. In addition, the primer set used in this study might have resulted in amplification of plastid DNA. Nevertheless, it is concluded from our results that this bias did not harm our analyses, as the biotic factors did not correlate excessively with the attached bacterial communities (Fig. 2a and b).

In contrast to CCA of RISA profiles, CCA of DGGE fingerprints showed a significant conditional effect of nutrients (nitrite), indicating an important effect on the free-living community. This finding is supported by recent studies showing that limitation of the bacterial community due to inorganic nutrients occurs in periods when temperature is not limiting (Rivkin & Anderson, 1997; Øvreås et al., 2003).

Generally, succession of communities was obvious from both fingerprinting methods, RISA and DGGE.

In addition, the seasonal dynamics of phytoplankton are reflected in bacterioplankton dynamics based on the free-living bacterial community in general. In the first part of the sampling period, the phytoplankton community was stable, as was the bacterioplankton community, despite slight variations. In March, flagellate abundance increased, accompanied by changes in the bacterial community composition. In April, further changes were not directly linked with bacterioplankton dynamics. After this, a strong correlation of the bacterioplankton community with the appearance of Phaeocystis spp. was seen. This finding supports the hypothesis that, among other factors, phytoplankton accounts for seasonality of bacterioplankton dynamics.

The analysis based on interspecies distances revealed the influence of different factors on specific phylotypes. An influence of nutrients was observed for phylotypes detected in the first part of the sampling period. It is suggested that they contributed to a specific ‘winter’ community. Group II was not influenced by any of the variables. It can be assumed that other factors influencing this group were not included in our dataset. Additionally, it should be considered that the Betaproteobacteria (F200) are rare in marine pelagic environments and are found predominantly in freshwater and coastal areas (Rappé, 1997; Fuhrmann & Ouverney, 1998; Giovannoni & Stingl, 2005). It cannot be excluded that this phylotype was detected due to water mass displacement at the sampling station, as the Elbe river has a strong impact when there are eastern winds. Further studies are needed to determine whether such impacts can be correlated with detection of specific bacterial species.

The further development of the community resulted in a group separated from the earlier community, indicating a strong shift caused by the appearance of Phaeocystis spp. and the increase in temperature. In particular, a member of the Gammaproteobacteria belonging to the Oceanospirillales (F069) occurred at the beginning of May. Its appearance is directly correlated with the bloom of Phaeocystis spp., indicating a strong effect of this alga on this phylotype. Additionally, it became apparent that two members of the Flavobacteria (group IV, Fig. 5) were strongly correlated with temperature and the alga Phaeocystis spp.

It has to be taken into account that the fingerprinting method (DGGE) based on PCR amplification has potential biases, which have been discussed elsewhere (Suzuki & Giovannoni, 1996; von Wintzingerode et al., 1997; Bidle & Azam, 2001). The primer set used in this study resulted in amplification of plastid DNA, especially in the fraction>3 μm (Table 4). These profiles of attached bacterial communities were not analyzed by ordination. Therefore, we could not consider the factors influencing those bacteria that appeared to be attached to the phytoplankton at the level of phylotypes. Factors influencing those bacteria, especially concerning their appearance as attached or free-living bacteria, remain unclear. Furthermore, one plastid phylotype was observed in the fraction of the free-living community (<3 μm and >0.2 μm). Its influence on the ordination was estimated to be low (Fig. 5). The sequence was similar to that of the 16S rRNA gene of a plastid of the Prasinophyceae (similarity of 98%) and was grouped with those phylotypes that were correlated with nutrients (nitrite) and Par. sulcata. As already shown in other studies (Bonin et al., 2002) nearly all excised DGGE bands were sequenced without interference, suggesting that each band represented one unique 16S rRNA gene sequence. Furthermore, sequencing of the majority of the DGGE bands at the same position in the gels revealed identical sequences, except for bands F067/F132 and F074/F306. Accordingly, we presume that each band corresponded to one ‘phylotype’.

The phylotypes retrieved in this study belonged mainly to the Alphaproteobacteria. Within this group, phylotype F068 clustered with sequence ATAM 173a_51, associated with the toxic dinoflagellate Alexandrium spp. (Hold et al., 2001), whereas bands F111 and F130 clustered with the sequence DC11-80-2 retrieved from the Weser estuary (Selje et al., 2004) and Roseobacter sp. (AF353235) retrieved from the Arctic Ocean (Bano & Hollibaugh, 2002). Members of the Roseobacter clade are thought to play an important role in the degradation of organic matter, colonizing a broad range of particles under algal bloom conditions (Pinhassi et al., 2004). Some members of the Alphaproteobacteria were detected, both free-living and attached. These were closely related to a member of the Rhodobacteraceae, namely AP-27 (AY145564), and the uncultured alphaproteobacterium DC11-80-2, both observed in the Weser estuary (Selje et al., 2004). In addition, a phylotype clustered with a sequence belonging to the SAR1 cluster (Fig. 3), which was retrieved in a study of the bacterial community of the Plum Island Sound estuary by Acinas et al. (2004). It might be that this phylotype is globally distributed. This can also be assumed for members of the Gammaproteobacteria clustering with the sequence HTCC2121 retrieved from the Pacific Ocean (Cho & Giovannoni, 2004) or with members of the Flavobacteria clustering with a sequence found in the Plum Island Sound estuary (Acinas et al., 2004). It has to be considered that the Flavobacteria, as particle colonizers, are thought to play an especially prominent role in the degradation of organic matter (Pinhassi et al., 2004; Abell & Bowman, 2005).

The factors measured in our study were able to explain a large part of the variation occurring in the bacterial community. However, among others, the factors of grazing by ciliates or nanoflagellates as well as control by viruses could not be considered. The importance of these factors has been demonstrated in different studies. Šestanović (2004) showed that heterotrophic nanoflagellates controlled planktonic bacteria of the Adriatic Sea in the period spring to summer, replacing the control by temperature, which was considered to be the controlling factor in the other seasons of the year in their study. The role of ciliates in the microbial food web has been investigated by Sherr & Sherr (1987), who stated that the observed clearance rates might be high enough to support the idea that ciliates contribute to a high extent to the microbial food web. Additionally, del Giorgio et al. (1996) demonstrated the control of total number of bacteria by heterotrophic nanoflagellates in dialysis experiments. A study of food selection showed that specific bacteria were selected by bacterivorous protists, indicating control by heterotrophic flagellates of the community composition of bacterioplankton (Jezbera et al., 2005). The control of bacterial abundance by viruses has been shown by several studies (Weinbauer & Peduzzi, 1995; Winter et al., 2004), and it has been suggested that viruses might have a stronger effect on bacterial abundance under certain conditions than grazing by heterotrophic nanoflagellates (Weinbauer & Peduzzi, 1995).

Conclusion

This study showed seasonal succession and dynamics of bacterioplankton in the winter–spring transition, supporting the findings of Gerdts et al. (2004). Additionally, relationships of several factors with the bacterioplankton were shown, especially for temperature, emphasizing its important role in the winter–spring transition. Also, the phytoplankton species Phaeocystis spp., G. delicatula and Chattonella spp., as well as nutrients (nitrite), contributed to shifts in the bacterial community. We also showed a correlation of factors with specific bacterial phylotypes. In particular, the strong relationship of Phaeocystis spp. with a member of the Gammaproteobacteria has been shown, indicating a strong influence of this alga on specific phylotypes of the bacterial community.

Acknowledgements

We would like to thank the crew members of the research vessel Aade, Silvia Janisch and Kristine Carstens, as well as K.-W. Klings from the Biologische Anstalt Helgoland, for their assistance. We are grateful for useful criticism by three anonymous reviewers. This work is part of the Helgoland Foodweb project in the Coastal Diversity program of the Alfred Wegener Institute.

Footnotes

  • Present address: Ecosystem Interaction, Pakefield Road, Lowestoft, Suffolk, NR33 OHT, UK.

  • Editor: Patricia Sobecky

References

View Abstract