The distribution of phytoplankton in the Baltic Sea assessed by a prokaryotic 16S rRNA gene primer system

The distribution of phytoplankton in the Baltic Sea assessed by a prokaryotic 16S rRNA gene... Abstract Due to the evolutionary relationship between cyanobacteria and chloroplasts of all oxygenic eukaryotic photoautotrophs, both can be amplified by prokaryotic 16S rRNA gene primers. In marine water samples, chloroplast sequences often make up as much as 50% of a 16S rRNA gene library, yet because of the comparatively low phylogenetic resolution within limited chloroplast databases, they are usually removed from further analyses. However, chloroplast 16S rRNA gene databases are constantly improving and our aim was to test if the combined 16S rRNA gene sequences of phototrophic prokaryotes and eukaryotes generated by a prokaryotic primer set could be used to characterize their distribution. Using the phytoREF database, in situ 16S rRNA gene distribution characterization was performed for samples throughout the Baltic Sea (cells >0.2 μm) and compared to microscopic (cells >20 μm) as well as flow-cytometric (cells <7 μm) data. Generally, microscopic and molecular methods revealed similar distribution patterns of diatoms, chlorophytes and filamentous cyanobacteria. Although not directly comparable, flow cytometry delivered semi-quantitative patterns, on a very broad classification level, to the molecular approach. In conclusion, the combination of molecular and non-molecular analyses provided an improved overview of the phototrophic community, demonstrating its usefulness as a tool in monitoring strategies. INTRODUCTION In recent years, monitoring microalgae as a key indicator for assessment of pelagic system functioning has become more and more complex, time-consuming and expensive. Large monitoring programs have been established and rely on regular sampling campaigns and subsequent expert microscopy knowledge (Wiltshire and Manly, 2004; Marshall et al., 2009; Wasmund et al., 2015) at times when the number of taxonomic experts is decreasing. The microscopic approach is further complicated by effects of fixatives on cell structures or structural changes of species during their life cycle (Decelle et al., 2013) and hidden endosymbionts. Additionally, it is known that for morphologically indistinct organisms, in particular picoplankton with small microalgae, the taxonomic resolution of a microscopic monitoring approach is often low. In principal, molecular 16S/18 S rRNA gene analyses could improve the identification of these phototrophic prokaryotes and eukaryotes (Prokopowich et al., 2003; Pei et al., 2010; Murray et al., 2016) but the generation of two different ribosomal databases would ultimately result in higher molecular and laboratory efforts. Our aim was to test one simple, non-specific molecular tool which can (i) determine the occurrence and distribution of phototrophic prokaryotes as well as eukaryotes and (ii) improve the cost efficiency and effectiveness of current monitoring techniques. Our approach is based on using the 16S rRNA gene as a taxonomic marker for the identification of phototrophic organisms such as cyanobacteria as well as eukaryotic microalgae. The idea was based on the fact that chloroplasts also possess a core set of bacterial genes in their genome (including the 16S and 23S rRNA phylogenetic marker genes) inherited from their cyanobacterial ancestor. Plastidal 16S rRNA genes are homologous to 16S rRNA genes of the cyanobacteria (Green, 2011) and taxonomic classification always places them into this phylum. Although life stages and variable number of chloroplasts per cell as well as of plastidial genome per chloroplast across taxonomic groups should be taken into account, the 16S rRNA gene copy number in Pro- as well as Eukaryotes can be relatively low and constant in comparison to the 18S rRNA gene of eukaryotes (Prokopowich et al., 2003; Pei et al., 2010; Murray et al., 2016), which principally enables comparison of relative sequence abundances between kingdoms. While 16S rRNA gene primer systems are currently optimized for prokaryotes, 16S rRNA gene chloroplast sequences can account for 50–80% of all sequences obtained within a sample (Herlemann et al., 2011; Eiler et al., 2013). In the past, these chloroplast sequences were often excluded from further analyses because (i) the preferred marker gene to resolve the taxonomic and phylogenetic diversity of phytoplankton and other protists was the 18S rRNA gene, (ii) many plastidal 16S rRNA gene sequences were misclassified in public databases due to homology with their cyanobacterial ancestors and (iii) until recently, no appropriate database for plastidal 16S rRNA genes existed. However, the limitations expressed in point (iii) have improved since Decelle et al. (2015) constructed a plastidal 16S rRNA gene reference database of photosynthetic eukaryotes with a curated taxonomy (phytoREF; accessed July 2016). This database consists of 6490 partial and complete plastidal 16S rRNA gene sequences representing all major lineages of photosynthetic eukaryotes from freshwater, marine and terrestrial habitats (Decelle et al., 2015). The Baltic Sea was chosen as a study site because the combination of intense freshwater discharge and limited water exchange with the North Sea generates a strong salinity gradient from west to east that favours distinct gradients in species distribution (Herlemann et al., 2011). Several phototrophic prokaryote and eukaryote groups characteristically appear within this salinity gradient in different seasons. For instance, diatoms and dinoflagellates tend to form massive blooms during spring (Wasmund et al., 1998; Klais et al., 2011). In summer, filamentous cyanobacteria, such as Aphanizomenon, Dolichospermum and Nodularia, are present in high numbers in the central Baltic Sea (Wasmund et al., 2015) and specifically adapted dinoflagellates, such as Dinophysis, may form dense populations in deeper water layers between 12 and 15 m (Carpenter et al., 1995). The Baltic Sea is subjected to many large monitoring programs which eventually result in annual status reports (Wasmund et al., 2017) where these recurrent patterns of the distribution of filamentous cyanobacteria, dinoflagellates and diatoms are reported. Most of these monitoring programs include comparative (i.e. microscopy) data, based on standardized HELCOM monitoring protocols (http://www.helcom.fi/Lists/Publications/Guidelines%20for%20monitoring%20phytoplankton%20species%20composition,%20abundance%20and%20biomass.pdf.). These background data were helpful for the evaluation of the new molecular 16S rRNA gene approach used in this study. METHOD Sampling and sample procedure The sampling stations were selected based on previous monitoring expeditions. Seawater samples were taken during a Baltic Sea research cruise on the R/V Alkor for the EU-BONUS BLUEPRINT project from 4 June to 17 June 2014. For this study, water was sampled at 2–5 m depths at seven stations, which were representative of large water masses within the salinity gradient of the Baltic Sea (Fig. 1; Table I). Samples for phytoplankton community examination and microbial DNA analyses were collected using a SBE-rosette SBE32 (Sea Bird Electronics Inc., USA; for detailed information see Supplementary Material 1.1), equipped with 18× 10 L FreeFlow-PWS-samplers (HYDRO-BIOS, Kiel, Germany) and a compact CTD (conductivity, temperature and depth) SBE 911 Plus. The CTD profiles were analysed to confirm that there was no density gradient between these depths and that sampling was always performed in the well-mixed surface layer (Supplementary Fig. S1). Fig. 1. View largeDownload slide Overview of the sampled stations and the salinity gradient. (A) The map shows the natural salinity gradient throughout the Baltic Sea in June 2014. The dotted lines indicate the borders of specific Baltic Sea regions which are numbered: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The grey dots represent the sampled stations in that specific region. (B) Bar plot representing the relative abundance of 16S rRNA gene >0.2 μm analysis on class-level identification. (C) Bar plot representing the relative abundance of the phytoplankton community >20 μm in size obtained by microscopic verification at class-level identification. Dinoflagellates and Mesodinium are solely represented by microscopy since no accordant 16S rRNA gene data were available. The colour code of the bars is given at the lower end of the figure. Map creation was performed using the program ODV 4.78 (Schlitzer, 2015) and the bar plots using the program sigma plot vs. 10. Fig. 1. View largeDownload slide Overview of the sampled stations and the salinity gradient. (A) The map shows the natural salinity gradient throughout the Baltic Sea in June 2014. The dotted lines indicate the borders of specific Baltic Sea regions which are numbered: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The grey dots represent the sampled stations in that specific region. (B) Bar plot representing the relative abundance of 16S rRNA gene >0.2 μm analysis on class-level identification. (C) Bar plot representing the relative abundance of the phytoplankton community >20 μm in size obtained by microscopic verification at class-level identification. Dinoflagellates and Mesodinium are solely represented by microscopy since no accordant 16S rRNA gene data were available. The colour code of the bars is given at the lower end of the figure. Map creation was performed using the program ODV 4.78 (Schlitzer, 2015) and the bar plots using the program sigma plot vs. 10. Table I: Overview of the obtained 16S rRNA gene sequences. Numbers are displayed excluding “No Relatives” and including the number and percentage of chloroplast sequences. Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Table I: Overview of the obtained 16S rRNA gene sequences. Numbers are displayed excluding “No Relatives” and including the number and percentage of chloroplast sequences. Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Phytoplankton sampling About 2–3 L subsamples were filtered through a 20 μm mesh phytoplankton net and reduced to volumes between 42 and 65 mL for phytoplankton community analyses. The filtrate was collected in a 100 mL brown glass bottle and preserved with 0.5% acidic Lugol’s solution according to the HELCOM monitoring guidelines (http://www.helcom.fi/helcom-at-work/publications/manuals-and-guidelines). Dissolving of carbonate structures in cells by this preservative may only bias samples collected at the westernmost stations, as in the central Baltic, organisms with carbonate components (e.g. coccolithophorides) are less abundant. The volumes of filtered water and filtrates were recorded (Supplementary Table S1). In the laboratory, a subsample of this concentrate was transferred to either 10 or 25 mL sedimentation chambers and the entire bottom plate was visually surveyed under an inverted microscope (Zeiss Axiovert S 100, Carl Zeiss MicroImaging GmbH, Göttingen, Germany) according to the method of Utermöhl (1958). For very abundant species that exceeded 50 individuals on less than 10% of the bottom plate, the numbers were extrapolated to the whole bottom plate according to the proportion of the area. Therefore, the theoretical detection limit of this counting method was between 0.6 and 1.9 cells L−1. Flow cytometry For flow cytometry, 4 mL of samples were collected from the CTD-rosette immediately after their retrieval and fixed for 1 h at 4°C with a mixture of paraformaldehyde and glutaraldehyde at a final concentration of 1% and 0.05%, respectively. The fixed cells were flash frozen with liquid nitrogen and stored at −80°C. For the analysis, samples were thawed and 300 μL of sample was used for measurement with the flow cytometer FACS Calibur (Becton Dickinson, San Jose, USA), making use of the autofluorescence of the small photoautotrophs Synechococcus and pigmented picoeukaryotes and nanoflagellates. The excitation wavelength was 488 nm and the detection of the cells was achieved at a flow rate of 36 μL min−1 with an acquisition time of 180 sec using emission filters FL3 (585/42 nm) and FL2 (661/12 nm). FL2 (phycoerythrin) vs. FL3 (chlorophyll a) fluorescence was used to differentiate picocyanobacteria from the pigmented picoeukaryotes and nanoeukaryotes according to the method of Gasol (1999). These gated populations are depicted in Supplementary Fig. S2 showing a FL3 vs. SSC (side scatter) plot, where the size differences between the groups are much clearer (beads size = 2 μm). Sampling for molecular analyses For DNA analysis, 1 L of seawater was vacuum filtered directly onto a 47-mm Durapore membrane filter (pore size of 0.2 μm; GVWP04700, Merck Millipore, Darmstadt, Germany) at <300 mbar. The filters were subsequently folded, flash frozen using liquid nitrogen and stored at −80°C until further analysis. Nucleic acid extraction DNA was extracted using the QIAamp DNA mini kit (51304, Qiagen, Hilden, Germany; for detailed information see Supplementary Material 1.3) as described by Drancourt et al. (2000). An initial bead-beating step was additionally employed, along with a clean-up and concentration step using the Zymo gDNA clean and concentrator kit (D4010, Zymo Research Europe, Freiburg, Germany). The concentration and the quality of eluted DNA were confirmed by gel electrophoresis and a molecular assay using the Bioanalyzer DNA12000 kit (5067-1508, Agilent Technologies, Santa Clara, USA). 16S rRNA gene amplicon library preparation Library construction and sequencing Two PCR amplifications were made per sample. First, the amplicon PCR and the subsequent index PCR were performed according to the instructions provided by Illumina (including primer selection) for preparing 16S rRNA gene amplicons for the Illumina MiSeq system (Illumina). For detailed information on the single steps, see Supplementary Material 1.4. Primers targeting the variable V3/V4 region of the 16S rRNA were suggested by Klindworth et al. (2012) to be the most promising primer pair for detecting bacteria and were used in a limited-cycle PCR that resulted in amplicon sizes of ~460 bp. The amplicon PCR products were cleaned according to the instructions included with the Agencourt® AMPure® XP kit and quality checked using the Bioanalyzer 1000 DNA kit. The quality and the quantity of each library were checked using a Bioanalyzer (Agilent Technologies, Santa Clara, USA) and Quibit assay (Thermo Fischer Scientific, Schwerte, Germany), respectively. The libraries were pooled with equimolar amounts to obtain a final concentration of 4 nM. From each pooled library, 4 pM were sequenced using an Illumina 600 v3 reagent cycle kit with paired-end protocol. The PhiX control library was spiked into each pool to a final amount of 10%. A cluster density of ~1200 K mm−2 was reached in over 80% sequencing and index reads and a Q-score ≥30 was routinely achieved. Individual runs generated between 15 and 25 million reads. FASTQ files were converted from *bcl files and used for further sequence data processing as outlined below. Bioinformatic processing For bioinformatic analyses, including pairing the forward and reverse reads, the QIIME version 1.9.1 was used (Caporaso et al., 2012). Unmerged reads were discarded. The SILVAngs pipeline (Quast et al., 2013) was used for operational taxonomic unit (OTU) picking and for determination of taxonomic affiliation. The PhytoREF database (Decelle et al., 2015) was used for taxonomic assignment of chloroplast sequences. For detailed information on the single steps, see Supplementary Material 1.5. The chloroplast sequences were identified using the program ARB (version 6.0.2; Ludwig et al., 2004) by extracting all sequences assigned “chloroplast” and mapping them to the PhytoREF database (Decelle et al., 2015). This database was accessed online at “http://phytoref.sb-roscoff.fr/” to assign taxonomy to the 16S rRNA gene sequences of the chloroplasts. The chloroplast classification table provided by PhytoREF was combined with the bacterial 16S rRNA gene OTU table based on the SILVAngs pipeline. Since all samples contained different numbers of sequences (Table 1), QIIME (Caporaso et al., 2012) was used to rarefy the dataset to the lowest number of sequences (26 019 sequences) among all samples. The tab-delimited, combined OTU table was converted into a biological observation matrix table, which functioned as an input for the QIIME single.rarefaction.py script (for detailed information, see Supplementary Material 1.5). Data deposition All raw sequence files are available from the NCBI Short Read Archive (SRA) database (BioProject: PRJNA379767, accession numbers: SAMN06619861 and SAMN06619896). RESULTS Table 1 summarizes the total number of sequences obtained, including chloroplast sequences but excluding sequences identified as “No Relatives” (i.e. reads that did not exceed the similarity threshold of 93%, which account for 0.05–0.27% of the samples). In total, 514 OTUs were identified, of which 480 OTUs were affiliated with prokaryotes and 34 OTUs with eukaryotes. About 32 out of the 480 prokaryotic OTUs belonged to unicellular or filamentous members of the Cyanobacteria, thus indicating abundances between 23% and 68% of the total amount of sequences per station. 16S rRNA gene analysis Between 3% and 9% of the sequences were identified as chloroplasts in the 16S rRNA gene amplicon libraries (Table 1). These chloroplast sequences were further classified into 15 different classes. Five of these 15 classes were present at all stations (Dictyochophyceae, Chrysophyceae, Eustigmatophyceae, Trebouxiophyceae and Cryptophyceae; Fig. 1). Whereas, chrysophyceae showed higher sequence abundances (16–23% of chloroplast sequences) in the western Baltic Sea, Trebouxiophyceae and Eustigmatophyceae increased in abundance towards the Gotland Sea and Bothnian Sea (78% and 35.5%, respectively; Fig. 1). Dictyochophyceae and Cryptophyceae revealed similar sequence abundances in both marine waters and waters at Gulf of Bothnia. Further, the classes Bacillariophyceae and Prymnesiophyceae were present at all stations except at the Gotland Sea station, with the highest sequence abundance in marine waters (Fig. 1). The class Prasinophyceae was most abundant in the western Baltic Sea, absent in the central Baltic Sea and reappeared with low numbers in the Gulf of Bothnia (Fig. 1). Overall, among the Bacillariophyceae, 10 different families were identified, but the majority of Bacillariophyceae sequences in marine waters could not be assigned to any family and remained unclassified. At stations, Mecklenburg Bight, Arkona and Bornholm Sea, the majority of Bacillariophyceae sequences were affiliated with the family Chaetocerotaceae. The remaining nine families had sequence abundances below 1% except for Cytomatosiraceae, Hemiaulaceae and Thalassiosiraceae, which showed sequence abundances between 1% and 3% at Skagerrak. For the class Dictyochophyceae and Cryptophyceae, no sequences could be assigned to any family; they remained unclassified. All sequences of the Chrysophyceae were affiliated with the family Chromulinaceae and the class Eustigmatophyceae affiliated with the family Monodopsidaceae. Among the Prasinophyceae, the family Pyramimonadaceae were identified. Within the class Trebouxiophyceae, the two families Chlorellaceae and Coccomyxaceae were identified. The Coccomyxaceae occurred with higher sequence abundances (24% of chloroplast sequences) in the Belt Sea and increased towards maximum abundance in the Gotland Sea (60% of chloroplast sequences). Chlorellaceae first occurred with more than 16% of all chloroplast sequences at the Arkona Sea station and reached maximum abundance at Bornholm station, with 23% of the chloroplast sequences. Both families decreased in abundances towards the Gulf of Bothnia. Among the haptophytes, the four families Chrysochromulinaceae, Prymnesiaceae, Noelaerhabdaceae and Phaeocystaceae were identified. The latter two families appeared at highest sequence abundances in the Skagerrak station (8% and 23%, respectively), whereas Chrysochromulinaceae and Prymnesiaceae increased in abundance between Mecklenburg Bight and Bornholm Sea (Fig. 2). Fig. 2. View largeDownload slide Comparison of flow cytometry data and 16S rRNA gene analysis in relative abundances. Flow cytometry data (cells <7 μm) represented by bar plot and 16S rRNA gene analysis (cells >0.2 μm) by heatmap. Numbers indicate Baltic Sea regions, similar to Figs 1 and 3: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. Bar plot creation was performed using sigma plot vs. 10 and the heatmap using R package pheatmap (www.r-project.org). Names marked with * were also displayed in Fig. 1B. Total numbers of chloroplast and cyanobacteria sequences are given in Table 1. Fig. 2. View largeDownload slide Comparison of flow cytometry data and 16S rRNA gene analysis in relative abundances. Flow cytometry data (cells <7 μm) represented by bar plot and 16S rRNA gene analysis (cells >0.2 μm) by heatmap. Numbers indicate Baltic Sea regions, similar to Figs 1 and 3: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. Bar plot creation was performed using sigma plot vs. 10 and the heatmap using R package pheatmap (www.r-project.org). Names marked with * were also displayed in Fig. 1B. Total numbers of chloroplast and cyanobacteria sequences are given in Table 1. Evaluation of molecular analysis In order to compare the obtained molecular analysis, flow cytometry and microscopic investigations of the phytoplankton community were performed (Figs 1–3). It should be noted that the microscopic analyses were based on 3 L of water sample for algae >20 μm, molecular analyses were based on 1 L of the total sample and flow-cytometric analyses on only 4 mL sample of autotrophs <7 μm. Comparing samples consisting of different volumes may impact organism detection limits; however, with regard to molecular analyses the most abundant organisms have been very likely recovered from the different counting and molecular methods. This is because PCR-based investigations (and associated exponential multiplications) can theoretically detect one gene per sample, especially when based on two-step PCRs (Labrenz et al., 2004). Flow cytometry assessed pigmented microorganisms, such as the cyanobacterial clade Synechococcus, pigmented picoeukaryotes and nanoeukaryotes smaller <7 μm (Supplementary Fig. S2). The latter groups were counted as bulk and no differentiation between classes and families could be made. In contrast, molecular analysis detected nine families affiliated with the classes Bacillariophyceae, Chrysophyceae, Eustigmatophyceae, Prasinophyceae and Trebouxiophyceae. Furthermore, flow cytometry showed Synechococcus to be the dominant organism at all stations, which was also reflected in the 16S rRNA gene libraries (Fig. 2). In addition to Synechococcus, other unicellular cyanobacteria, such as Snowella, Merismopedia and Microcystis, were identified by sequencing but had relative abundances below 1%. Fig. 3. View largeDownload slide Heatmap of relative abundance of diatom (Bacillariophyceae) families compared between microscopy (cells >20 μm) and DNA reads (cells >0.2 μm) determined by 16S rRNA gene amplicon libraries. Numbers represent the Baltic Sea regions similar as in Fig. 1: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The colour intensity represents relative abundance in percentage. For the shaded parts, no data are available in the PhytoREF database. Heatmap was performed using the R package pheatmap (www.r-project.org). *Underrepresented in database, only few reference sequences available. Fig. 3. View largeDownload slide Heatmap of relative abundance of diatom (Bacillariophyceae) families compared between microscopy (cells >20 μm) and DNA reads (cells >0.2 μm) determined by 16S rRNA gene amplicon libraries. Numbers represent the Baltic Sea regions similar as in Fig. 1: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The colour intensity represents relative abundance in percentage. For the shaded parts, no data are available in the PhytoREF database. Heatmap was performed using the R package pheatmap (www.r-project.org). *Underrepresented in database, only few reference sequences available. Microscopic counts focussed on the morphologies that were easily and undoubtedly detectable under the light microscope and was therefore only done on the >20 μm size fraction. This fraction comprised diatoms (Bacillariophyceae) and dinoflagellates, found in high abundance (both >40% of the total phytoplankton community; Fig. 3) in the western Baltic Sea (Skagerrak and Belt Sea) and with lower abundances (<10% and 20% of the total phytoplankton community, respectively) in the northern Baltic Sea (Gulf of Bothnia). Bacillariophyceae included the two main families (Fig. 3) Chaetocerataceae and Rhizosoleniaceae present in the western Baltic Sea and Thalassiosiraceae and Hemidiscaceae in the Gulf of Bothnia. Cearatiaceae and Dinophysiaceae were the main families among the dinoflagellates throughout the Baltic Sea transect. The chlorophytes Trebouxiophyceae dominated in the southern Baltic Sea, with highest numbers (>50% of the total phytoplankton community) in the Bornholm Basin. This phylum was exclusively represented by the species Planktonema lauterbornii. In contrast, the filamentous cyanobacterium Aphanizomenon was very abundant (>60% of the total phytoplankton community) in the central (Gotland Sea) and northern Baltic Sea (Gulf of Bothnia) and was almost completely absent in marine waters. Based on 16S rRNA gene analyses, Aphanizomenon and Nodularia were also identified in the central and northern Baltic Sea (data not shown). DISCUSSION In this study, molecular, microscopic and flow-cytometric methods were employed to record the distribution of phototrophic prokaryotes and eukaryotes in the Baltic Sea and the use of the 16S rRNA gene as a suitable target gene for phytoplankton monitoring was evaluated. Molecular observations revealed distinct distribution patterns of diatoms (Bacillariophyceae), Chrysophyceae, silicoflagellates (Dictyochophyceae) and Eustigmatophyceae, as well as cryptophytes (Cryptophyceae) and haptophytes (Prymnesiophyceae) at different Baltic Sea stations. Verification of molecular analysis by microscopy This distribution was confirmed by microscopy, although only the >20 μm fraction could be analysed. On a broad classification level, the assignment of the phytoplankton 16S rRNA gene sequences to classes using phytoREF was successful. For the majority of stations, more than 70% of the 16S rRNA gene sequences were identified at the family level; only at the Skagerrak station were the sequences limited to class-level (Bacillariophyceae) identification. In contrast, microscopy was able to identify phytoplankton down to genus-level, which was only in few cases possible by 16S rRNA gene analysis. For Bacillariophyceae, no significant differences between microscopy and 16S rRNA gene analyses were observed (t-test, P-value >0.05) except that the diversity determined was higher for molecular analysis than for microscopy (10 families vs. 5 families by microscopy; Fig. 3). This is most probably an effect of the difference in mesh size between the filtration methods for phytoplankton (20 μm mesh) and molecular samples (0.2 μm membrane filter), as smaller species or broken chains will pass through the net but will be retained on the filter. Concerning microscopic analyses, this disadvantage was accepted for the benefit of collecting large and easily identifiable phytoplankton species in statistically robust numbers, which is not always the case in non-concentrated water samples. This explanation applies to all stations with both microscopic and molecular analysis (Fig. 1). Thus, differences are most likely due to the differences in sample preparation, rather than vertical sampling heterogeneity, as in the central Baltic, the well-mixed surface layer is always deeper than 10 m (Supplementary Fig. S1). For Chlorophyta, microscopy and molecular analyses were comparable, except that 16S rRNA gene analysis detected Chlorophyta sequences in the Skagerrak and Gulf of Bothnia (Fig. 1; 8% and 36% relative read abundance, respectively). Either these are species which were not detected by microscopy due to the pre-filtration at 20 μm as mentioned above, or the molecular approach discovered cells which were rarely present and potentially overseen by microscopy. Conversely, looking deeper into the class Trebouxiophyceae, microscopy solely identified the species Planktonema lauterbornii, which was not identified using molecular investigations. Instead, Trebouxiophyceae remained unclassified because the chloroplast sequence of Planktonema lauterbornii is currently still absent from phytoREF. Comparing microscopy and molecular analyses more technically, microscopy requires time and expert knowledge and gets more difficult the smaller the cells become. However, with the molecular approach, using user-independent standardized protocols for sampling, nucleic acid extraction and library preparation (as well as sequencing and bioinformatic processing), comparability between samples from different expeditions, years and stations can be ensured. Despite the lower taxonomic resolution of this method vs. microscopy, it was able to detect phototrophic organisms in the small size range, which are often difficult to recognize by microscopy or flow cytometry (Figs 1–3). Moreover, it is advantageous that molecular investigations are designed for high-throughput analysis so that a larger number of samples (96 and more) can be processed simultaneously. Unlike microscopy, no explicit taxonomic expert knowledge is required to apply such a tool at a comparable cost per sample. Thus, it is probable that molecular investigation will be the most important way to link very small phototrophs with taxonomic information in the future. A step potentially introducing biases in the molecular approach is the 16S rRNA gene library preparation, as each sample undergoes two PCR amplification steps, which can result in template primer mismatches or over-representation of certain sequences in mixed community samples (Sipos et al., 2007). Primers also select for distinct organism groups. The primers used in this study are designed for bacteria amplification and therefore revealed a considerably low amount of chloroplast sequences. However, this low amount of chloroplast sequences gave a broad overview of the dominant phytoplankton groups at different stations of the Baltic Sea with distinct physico-chemical profiles. Verification of molecular analysis by flow cytometry As described above, flow cytometry is limited by size and counts picoeukaryotes and nanoeukaryotes as a single group without differentiating between picophytoplankton groups (Hammes and Egli, 2010). Therefore, comparing flow-cytometric data with molecular analysis is possible solely for the cyanobacterial clade Synechococcus. Further group differentiation would require flow sorting followed by sequencing of the sorted fraction (Cellamare et al., 2010) which is beyond the scope of a fast and simple monitoring approach. At the molecular level, the typical Baltic Sea picophytoplankton group haptophytes was identified down to family level and chlorophytes remained on class-level identification. Kleptoplastidy Detection of plastidal sequences in amplicon libraries can be improved when using other primers (see, for example, Hansen et al., 2013; Needham and Fuhrman, 2016). The development of new primers targeting dinoflagellate chloroplasts (specifically toxic species) would considerably advance the performance of molecular-based analyses. However, designing new primers that would also detect the chloroplast sequence of dinoflagellates is difficult. The genomes of dinoflagellate plastids have usually been broken up into mini circles, and many genes including 16S rRNA and 23S rRNA been transferred into the nucleus of dinoflagellates (Green, 2011). This makes it almost impossible to amplify the 16S rRNA gene of dinoflagellate plastids, which is reflected by their underrepresentation in the phytoREF database (0.5% of all entries) and subsequent lack of detection in the present dataset. Instead, 18S rRNA gene amplification may well reveal a better resolution in this case. In addition, some dinoflagellates carry out phototrophy based on the retained plastids of their prey (Hansen et al., 2013; Park et al., 2014). This is mainly observed in members of the order Dinophysiales (Meyer-Harms and Pollehne, 1998). Based on our microscopic data, Ceratium was generally present in the western Baltic Sea and Dinophysis dominated towards the central and northern Baltic Sea. This is analogous to the cryptophytes distribution. Interestingly, a comparison of the Cryptophyta sequence data with the dinoflagellate count data using a paired t-test (P < 0.05) did not reveal significant differences, and the trends were similar. This suggests that Cryptophyta plastids could be incorporated into dinoflagellates, but direct evidence is still lacking. Similar to several dinoflagellates, the ciliate Mesodinium rubrum also acquires chloroplasts from their (mainly cryptophyte) prey. This annexation of plastids is referred to as kleptoplastidy (Kim et al., 2012; Hansen et al., 2013; Park et al., 2014). However, we were not able to identify if those cryptophyte 16S rRNA gene sequences truly belong to cryptophytes or if they were taken up by Dinophysis, Ceratium or Mesodinium. The 16S rRNA gene sequence of the cryptophyte which is taken up by Mesodinium is named “ciliate” in PhytoREF. However, our dataset did not reveal any 16S rRNA gene sequence identified as “ciliate”; all 16S rRNA gene sequences affiliated with cryptophytes were named “Cryptophyceae-unclassified”. Choice of primers For this study, we used primers (341 F/805 R) which are standardized for 16S analysis (Klindworth et al., 2012), although the paired reads do not fully overlap when using Illumina MiSeq sequencing technology. However, they are nevertheless recommended by Illumina as the promising primer pair for detecting prokaryotic sequences and the forward and reverse reads overlap with more than 120 bp when using the v3 kit (2 × 300 bp). Furthermore, it has been shown that this primer pair also detects chloroplast 16S rRNA sequences (Eiler et al., 2013), although the chloroplast 16S rRNA gene sequence has four mismatches between positions 783–799 (according to E. coli base positioning system) compared to the 16S rRNA gene sequence of Cyanobacteria (Hanshew et al., 2013). Three of the four mismatches are located directly in the primer binding site of the 805 R primer, which eventually results in non- or rare binding of this reverse primer to the 16S rRNA gene sequence of chloroplasts, thus minimizing the amount of detected chloroplast sequences. In the present study, 10% chloroplast sequences were detected, at best. Others (Eiler et al., 2013; for example) obtained 15% chloroplast sequences for most samples and for two sample locations they obtained as much as 50% chloroplast sequences (Eiler et al., 2013). Recently, Needham and Fuhrman (2016) published primers (515 F/926 R) which seem to neglect the mismatches between plastidal and cyanobacterial 16S rRNA gene sequences. Furthermore, their results showed that the plastidal 16S rRNA analysis was highly concordant with the 18S rRNA analysis (Needham and Fuhrman, 2016). Also only recently Milici et al. (2016) demonstrated for the Atlantic Ocean that sequences retrieved with the bacterial primers F807 and R1050 covered almost 70% of the marine eukaryotic photosynthetic lineages of the PhytoREF database (Decelle et al., 2015), leading to the conclusion that this approach is sufficient to investigate the distribution of eukaryotic microalgae. CONCLUSION In summary, 16S rRNA gene analysis based on the selected prokaryotic primer set coincided well with microscopic identification. Thus, for general information on the presence and absence of specific phototrophic clades, such as diatoms (Bacillariophyceae), green microalgae (Chlorophyta) and especially Cyanobacteria, including the unicellular fraction, molecular analyses provided a sufficient toolset when using the 16S rRNA gene sequence. Despite the lower taxonomic resolution of this method vs. microscopy for the large phytoplankton fraction, molecular analysis detected phototrophic organisms in the small size range, which are often difficult to recognize by microscopy or flow cytometry (Figs 1 and 2). In addition, phytoREF should improve in the detection of plastidal sequences with the addition of more sequences into the database; thus, taxonomic expertise is necessary to build and improve such reference database. An advantage of molecular investigations is that they are designed for high-throughput analysis, handling larger number of samples (96 and more) at the same time. Meanwhile, technical systems have been developed that can combine high-throughput 16S rRNA analysis with microscopic analysis, for instance, 3D-fluorescence imaging. With a long-term perspective, the combination of molecular methods and microscopy will be able to achieve the resolution necessary for a more comprehensive assessment of phytoplankton and thereby expand the utility of monitoring strategies. However, if costs for molecular analysis continue to decrease and the taxonomic resolution continues to increase this approach could become even more important in the future. SUPPLEMENTARY DATA Supplementary data can be found online at Journal of Plankton Research online. ACKNOWLEDGEMENTS We thank the captain and crew of the R/V Alkor for assistance in sampling during the AL439 cruise. We greatly appreciate the substantial contribution of Siegfried Krüger in handling the CTD-rosette and the AFIS sampler during the cruise. We also thank Heike Benterbusch-Brockmüller and Andreas Rogge for assistance in filtration at sea. We would like to acknowledge Annett Grüttmüller for providing flow cytometry work and Jana Normann for handling and loading the Illumina MiSeq Sequencer. Special thanks are going to Alexander Tagg for critically proofreading the manuscript. FUNDING This work resulted from the EU-BONUS BLUEPRINT project and was supported by BONUS (Art 185), funded jointly by the EU and the German Federal Ministry of Education and Research (BMBF; 03F0679A), as well as from support by the Deutsche Forschungsgemeinschaft (project LA 1466/8-1). Purchase of the Illumina MiSeq was kindly supported by the EU-EFRE (European Funds for Regional Development) program and funds from the University Medicine Rostock awarded to BK. REFERENCES Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Huntley, J., Fierer, N., Owens, S. M., Betley, J. et al.   ( 2012) Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. , 8, 1621– 1624. Google Scholar CrossRef Search ADS   Carpenter, E. J., Janson, S., Boje, R., Pollehne, F. and Chang, J. ( 1995) The dinoflagellate Dinophysis norvegica. Eur. J. Phycol. , 1, 1– 9. Google Scholar CrossRef Search ADS   Cellamare, M., Rolland, A. and Jacquet, S. ( 2010) Flow cytometry sorting of freshwater phytoplankton. J. Appl. Phycol. , 1, 87– 100. Google Scholar CrossRef Search ADS   Decelle, J., Romac, S., Stern, R. F., Bendif el, M., Zingone, A., Audic, S., Guiry, M. D., Guillou, L. et al.   ( 2015) PhytoREF: a reference database of the plastidial 16S rRNA gene of photosynthetic eukaryotes with curated taxonomy. Mol. Ecol. Resour. , 6, 1435– 1445. Google Scholar CrossRef Search ADS   Decelle, J., Martin, P., Paborstava, K., Pond, D. W., Tarling, G., Mahé, F., De Vargas, C., Lampitt, R. et al.   ( 2013) Diversity, ecology and biogeochemistry of cyst-forming Acantharia (Radiolaria) in the oceans. PLoS One , 8, e53598. Google Scholar CrossRef Search ADS PubMed  Drancourt, M., Bollet, C., Carlioz, A., Martelin, R., Gayral, J.-P. and Raoult, D. ( 2000) 16S ribosomal DNA sequence analysis of a large collection of environmental and clinical unidentifiable bacterial isolates. J. Clin. Microbiol. , 38, 3623– 3630. Google Scholar PubMed  Eiler, A., Drakare, S., Bertilsson, S., Pernthaler, J., Peura, S., Rofner, C., Simek, K., Yang, Y. et al.   ( 2013) Unveiling distribution patterns of freshwater phytoplankton by a next generation sequencing based approach. PLoS One , 1, e53516. Google Scholar CrossRef Search ADS   Gasol, J. M. ( 1999) How to Count Picoalgae and Bacteria with the FACScalibur Flow Cytometer . Departament de Biologia Marina i Oceanografia. Institut de Ciencies del Mar, CSIC, Barcelona, pp. 1– 51. Green, B. R. ( 2011) Chloroplast genomes of photosynthetic eukaryotes. Plant J. , 1, 34– 44. Google Scholar CrossRef Search ADS   Hammes, F. and Egli, T. ( 2010) Cytometric methods for measuring bacteria in water: advantages, pitfalls and applications. Anal. Bioanal. Chem. , 3, 1083– 1095. Google Scholar CrossRef Search ADS   Hansen, P. J., Nielsen, L. T., Johnson, M., Berge, T. and Flynn, K. J. ( 2013) Acquired phototrophy in Mesodinium and Dinophysis—a review of cellular organization, prey selectivity, nutrient uptake and bioenergetics. Harmful Algae , 28, 126– 139. Google Scholar CrossRef Search ADS   Hanshew, A. S., Mason, C. J., Raffa, K. F. and Currie, C. R. ( 2013) Minimization of chloroplast contamination in 16S rRNA gene pyrosequencing of insect herbivore bacterial communities. J. Microbiol. Meth. , 2, 149– 155. Google Scholar CrossRef Search ADS   Herlemann, D. P., Labrenz, M., Jürgens, K., Bertilsson, S., Waniek, J. J. and Andersson, A. F. ( 2011) Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. ISME J. , 10, 1571– 1579. Google Scholar CrossRef Search ADS   Illumina. 16S Metagenomic Sequencing Library Preparation. http://www.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/16s/16s-metagenomic-library-prep-guide-15044223-b.pdf.; [accessed: 13 October 2014]. Kim, M., Kim, S., Yih, W. and Park, M. G. ( 2012) The marine dinoflagellate genus Dinophysis can retain plastids of multiple algal origins at the same time. Harmful Algae , 13, 105– 111. Google Scholar CrossRef Search ADS   Klais, R., Tamminen, T., Kremp, A., Spilling, K. and Olli, K. ( 2011) Decadal-scale changes of dinoflagellates and diatoms in the anomalous Baltic Sea spring bloom. PLoS One , 6, e21567. Google Scholar CrossRef Search ADS PubMed  Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M. and Glöckner, F. O. ( 2012) Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res. , 1, e1– e1. Labrenz, M., Brettar, I., Christen, R., Flavier, S., Bötel, J. and Höfle, M. G. ( 2004) Development and application of a real-time PCR approach for quantification of uncultured bacteria in the central Baltic Sea. Appl. Environ. Microbiol. , 70, 4971– 4979. Google Scholar CrossRef Search ADS PubMed  Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, Buchner, A., Lai, T. et al.   ( 2004) ARB: a software environment for sequence data. Nucleic Acids Res. , 4, 1363– 1371. Google Scholar CrossRef Search ADS   Marshall, H. G., Lane, M. F., Nesius, K. K. and Burchardt, L. ( 2009) Assessment and significance of phytoplankton species composition within Chesapeake Bay and Virginia tributaries through a long-term monitoring program. Environ. Monit. Assess. , 1–4, 143– 155. Google Scholar CrossRef Search ADS   Meyer-Harms, B. and Pollehne, F. ( 1998) Alloxanthin in Dinophysis norvegica (Dinophysiales, Dinophyceae) from the Baltic Sea. J. Phycol. , 2, 280– 285. Google Scholar CrossRef Search ADS   Milici, M., Deng, Z.-L., Tomasch, J., Decelle, J., Wos-Oxley, M. L., Wang, H., Jáuregui, R., Plumeier, I. et al.   ( 2016) Co-occurrence analysis of microbial taxa in the Atlantic Oceanr reveals high connectivity in the free-living bacterioplankton. Front. Microbiol. , 7, 649. Google Scholar PubMed  Murray, S. A., Suggett, D. J., Doblin, M. A., Kohli, G. S., Seymour, J. R., Fabris, M. and Ralph, P. J. ( 2016) Unravelling the functional genetics of dinoflagellates: a review of approaches and opportunities. Persp. Phycol. , 1, 37– 52. Needham, D. M. and Fuhrman, J. A. ( 2016) Pronounced daily succession of phytoplankton, archaea and bacteria following a spring bloom. Nat. Microbiol. , 1, 1– 7. Google Scholar CrossRef Search ADS   Park, M. G., Kim, M. and Kim, S. ( 2014) The acquisition of plastids/phototrophy in heterotrophic dinoflagellates. Acta Protozool. , 53, 39– 50, Special topic issue: “Marine Heterotrophic Protists”. Pei, A. Y., Oberdorf, W. E., Nossa, C. W., Agarwal, A., Chokshi, P., Gerz, E. A., Jin, Z., Lee, P. et al.   ( 2010) Diversity of 16S rRNA genes within individual prokaryotic genomes. Appl. Environ. Microbiol. , 12, 3886– 3897. Google Scholar CrossRef Search ADS   Prokopowich, C. D., Gregory, T. R. and Crease, T. J. ( 2003) The correlation between rDNA copy number and genome size in eukaryotes. Genome , 1, 48– 50. Google Scholar CrossRef Search ADS   Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J. and Glöckner, F. O. ( 2013) The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. , 41, D590– D596. Database issue. Google Scholar CrossRef Search ADS PubMed  Schlitzer, R. ( 2015). Ocean Data View. http://odv.awi.de. Sipos, R., Székely, A. J., Palatinszky, M., Révész, S., Márialigeti, K. and Nikolausz, M. ( 2007) Effect of primer mismatch, annealing temperature and PCR cycle number on 16S rRNA gene-targetting bacterial community analysis. FEMS Microbiol. Ecol. , 60, 341– 350. Google Scholar CrossRef Search ADS PubMed  Utermöhl, H. ( 1958) Zur Vervollkommnung der quantitativen Phytoplanktonmethodik. Mitteilungen Int. Ver. Theor. Angew. Limnol. Band , Vol 9. Schweizerbart, Stuttgart, p. 38. Wasmund, N., Nausch, G. and Matthäus, W. ( 1998) Phytoplankton spring blooms in the southern Baltic Sea—spatio-temporal development and long-term trends. J. Plankton Res. , 6, 1099– 1117. Google Scholar CrossRef Search ADS   Wasmund, N., Dutz, J., Pollehne, F., Siegel, H. and Zettler, M. L. 2015. Biological assessment of the Baltic Sea 2014. Meereswiss. Ber. , Warnemünde, 98. doi:10.12754/msr-2015-0098. Wasmund, N., Dutz, J. and Zettler, M. L. 2017. Contribution to the national commentary by Germany on annual assessment of the environmental status of the Baltic Sea based on biological monitoring in 2016. Leibniz Institute for Baltic Sea Research Warnemünde (IOW). Wiltshire, K. H. and Manly, B. F. J. ( 2004) The warming trend at Helgoland Roads, North Sea. Hel. Mar. Res. , 4, 269– 273. Google Scholar CrossRef Search ADS   Author notes Corresponding Editor: John Dolan © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Plankton Research Oxford University Press

The distribution of phytoplankton in the Baltic Sea assessed by a prokaryotic 16S rRNA gene primer system

Loading next page...
 
/lp/ou_press/the-distribution-of-phytoplankton-in-the-baltic-sea-assessed-by-a-ffk0JAeJq4
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0142-7873
eISSN
1464-3774
D.O.I.
10.1093/plankt/fby008
Publisher site
See Article on Publisher Site

Abstract

Abstract Due to the evolutionary relationship between cyanobacteria and chloroplasts of all oxygenic eukaryotic photoautotrophs, both can be amplified by prokaryotic 16S rRNA gene primers. In marine water samples, chloroplast sequences often make up as much as 50% of a 16S rRNA gene library, yet because of the comparatively low phylogenetic resolution within limited chloroplast databases, they are usually removed from further analyses. However, chloroplast 16S rRNA gene databases are constantly improving and our aim was to test if the combined 16S rRNA gene sequences of phototrophic prokaryotes and eukaryotes generated by a prokaryotic primer set could be used to characterize their distribution. Using the phytoREF database, in situ 16S rRNA gene distribution characterization was performed for samples throughout the Baltic Sea (cells >0.2 μm) and compared to microscopic (cells >20 μm) as well as flow-cytometric (cells <7 μm) data. Generally, microscopic and molecular methods revealed similar distribution patterns of diatoms, chlorophytes and filamentous cyanobacteria. Although not directly comparable, flow cytometry delivered semi-quantitative patterns, on a very broad classification level, to the molecular approach. In conclusion, the combination of molecular and non-molecular analyses provided an improved overview of the phototrophic community, demonstrating its usefulness as a tool in monitoring strategies. INTRODUCTION In recent years, monitoring microalgae as a key indicator for assessment of pelagic system functioning has become more and more complex, time-consuming and expensive. Large monitoring programs have been established and rely on regular sampling campaigns and subsequent expert microscopy knowledge (Wiltshire and Manly, 2004; Marshall et al., 2009; Wasmund et al., 2015) at times when the number of taxonomic experts is decreasing. The microscopic approach is further complicated by effects of fixatives on cell structures or structural changes of species during their life cycle (Decelle et al., 2013) and hidden endosymbionts. Additionally, it is known that for morphologically indistinct organisms, in particular picoplankton with small microalgae, the taxonomic resolution of a microscopic monitoring approach is often low. In principal, molecular 16S/18 S rRNA gene analyses could improve the identification of these phototrophic prokaryotes and eukaryotes (Prokopowich et al., 2003; Pei et al., 2010; Murray et al., 2016) but the generation of two different ribosomal databases would ultimately result in higher molecular and laboratory efforts. Our aim was to test one simple, non-specific molecular tool which can (i) determine the occurrence and distribution of phototrophic prokaryotes as well as eukaryotes and (ii) improve the cost efficiency and effectiveness of current monitoring techniques. Our approach is based on using the 16S rRNA gene as a taxonomic marker for the identification of phototrophic organisms such as cyanobacteria as well as eukaryotic microalgae. The idea was based on the fact that chloroplasts also possess a core set of bacterial genes in their genome (including the 16S and 23S rRNA phylogenetic marker genes) inherited from their cyanobacterial ancestor. Plastidal 16S rRNA genes are homologous to 16S rRNA genes of the cyanobacteria (Green, 2011) and taxonomic classification always places them into this phylum. Although life stages and variable number of chloroplasts per cell as well as of plastidial genome per chloroplast across taxonomic groups should be taken into account, the 16S rRNA gene copy number in Pro- as well as Eukaryotes can be relatively low and constant in comparison to the 18S rRNA gene of eukaryotes (Prokopowich et al., 2003; Pei et al., 2010; Murray et al., 2016), which principally enables comparison of relative sequence abundances between kingdoms. While 16S rRNA gene primer systems are currently optimized for prokaryotes, 16S rRNA gene chloroplast sequences can account for 50–80% of all sequences obtained within a sample (Herlemann et al., 2011; Eiler et al., 2013). In the past, these chloroplast sequences were often excluded from further analyses because (i) the preferred marker gene to resolve the taxonomic and phylogenetic diversity of phytoplankton and other protists was the 18S rRNA gene, (ii) many plastidal 16S rRNA gene sequences were misclassified in public databases due to homology with their cyanobacterial ancestors and (iii) until recently, no appropriate database for plastidal 16S rRNA genes existed. However, the limitations expressed in point (iii) have improved since Decelle et al. (2015) constructed a plastidal 16S rRNA gene reference database of photosynthetic eukaryotes with a curated taxonomy (phytoREF; accessed July 2016). This database consists of 6490 partial and complete plastidal 16S rRNA gene sequences representing all major lineages of photosynthetic eukaryotes from freshwater, marine and terrestrial habitats (Decelle et al., 2015). The Baltic Sea was chosen as a study site because the combination of intense freshwater discharge and limited water exchange with the North Sea generates a strong salinity gradient from west to east that favours distinct gradients in species distribution (Herlemann et al., 2011). Several phototrophic prokaryote and eukaryote groups characteristically appear within this salinity gradient in different seasons. For instance, diatoms and dinoflagellates tend to form massive blooms during spring (Wasmund et al., 1998; Klais et al., 2011). In summer, filamentous cyanobacteria, such as Aphanizomenon, Dolichospermum and Nodularia, are present in high numbers in the central Baltic Sea (Wasmund et al., 2015) and specifically adapted dinoflagellates, such as Dinophysis, may form dense populations in deeper water layers between 12 and 15 m (Carpenter et al., 1995). The Baltic Sea is subjected to many large monitoring programs which eventually result in annual status reports (Wasmund et al., 2017) where these recurrent patterns of the distribution of filamentous cyanobacteria, dinoflagellates and diatoms are reported. Most of these monitoring programs include comparative (i.e. microscopy) data, based on standardized HELCOM monitoring protocols (http://www.helcom.fi/Lists/Publications/Guidelines%20for%20monitoring%20phytoplankton%20species%20composition,%20abundance%20and%20biomass.pdf.). These background data were helpful for the evaluation of the new molecular 16S rRNA gene approach used in this study. METHOD Sampling and sample procedure The sampling stations were selected based on previous monitoring expeditions. Seawater samples were taken during a Baltic Sea research cruise on the R/V Alkor for the EU-BONUS BLUEPRINT project from 4 June to 17 June 2014. For this study, water was sampled at 2–5 m depths at seven stations, which were representative of large water masses within the salinity gradient of the Baltic Sea (Fig. 1; Table I). Samples for phytoplankton community examination and microbial DNA analyses were collected using a SBE-rosette SBE32 (Sea Bird Electronics Inc., USA; for detailed information see Supplementary Material 1.1), equipped with 18× 10 L FreeFlow-PWS-samplers (HYDRO-BIOS, Kiel, Germany) and a compact CTD (conductivity, temperature and depth) SBE 911 Plus. The CTD profiles were analysed to confirm that there was no density gradient between these depths and that sampling was always performed in the well-mixed surface layer (Supplementary Fig. S1). Fig. 1. View largeDownload slide Overview of the sampled stations and the salinity gradient. (A) The map shows the natural salinity gradient throughout the Baltic Sea in June 2014. The dotted lines indicate the borders of specific Baltic Sea regions which are numbered: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The grey dots represent the sampled stations in that specific region. (B) Bar plot representing the relative abundance of 16S rRNA gene >0.2 μm analysis on class-level identification. (C) Bar plot representing the relative abundance of the phytoplankton community >20 μm in size obtained by microscopic verification at class-level identification. Dinoflagellates and Mesodinium are solely represented by microscopy since no accordant 16S rRNA gene data were available. The colour code of the bars is given at the lower end of the figure. Map creation was performed using the program ODV 4.78 (Schlitzer, 2015) and the bar plots using the program sigma plot vs. 10. Fig. 1. View largeDownload slide Overview of the sampled stations and the salinity gradient. (A) The map shows the natural salinity gradient throughout the Baltic Sea in June 2014. The dotted lines indicate the borders of specific Baltic Sea regions which are numbered: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The grey dots represent the sampled stations in that specific region. (B) Bar plot representing the relative abundance of 16S rRNA gene >0.2 μm analysis on class-level identification. (C) Bar plot representing the relative abundance of the phytoplankton community >20 μm in size obtained by microscopic verification at class-level identification. Dinoflagellates and Mesodinium are solely represented by microscopy since no accordant 16S rRNA gene data were available. The colour code of the bars is given at the lower end of the figure. Map creation was performed using the program ODV 4.78 (Schlitzer, 2015) and the bar plots using the program sigma plot vs. 10. Table I: Overview of the obtained 16S rRNA gene sequences. Numbers are displayed excluding “No Relatives” and including the number and percentage of chloroplast sequences. Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Table I: Overview of the obtained 16S rRNA gene sequences. Numbers are displayed excluding “No Relatives” and including the number and percentage of chloroplast sequences. Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Sample station  Geographical position  Sample ID  Number of total sequences  Number of chloroplast sequences  Chloroplasts (%)  Number of cyanobacteria sequences  Cyanobacteria (%)  Skagerrak (SK)  N 58° 07.9376  SK_DNA  45 971  2532  5.47  12 350  26.86  E 9° 59.9990              Belt Sea (BS)  N 56° 00.0615  BS_DNA  71 094  1902  2.56  16 035  22.55  E 11° 04.6438              Mecklenburg Bight (MB)  N 54° 16.9269  MB_DNA  26 019  2363  8.55  9105  34.99  E 11° 34.0792              Arkona Sea (AS)  N 54° 41.8740  AS_DNA  103 481  7736  7.47  57 906  55.96  E 12° 42.2815              Bornholm Sea (BoS)  N 55° 33.9539  BoS_DNA  75 941  3724  4.88  50 953  68.00  E 16° 21.9907              Gotland Sea (GS)  N 57° 18.3160  GS_DNA  66 384  2855  4.30  30 910  46.56  E 20° 04.5955              Gulf of Bothnia (GB)  N 61° 46.9727  GB_DNA  38 815  3351  8.63  18 893  48.67  E 19° 17.7102              Phytoplankton sampling About 2–3 L subsamples were filtered through a 20 μm mesh phytoplankton net and reduced to volumes between 42 and 65 mL for phytoplankton community analyses. The filtrate was collected in a 100 mL brown glass bottle and preserved with 0.5% acidic Lugol’s solution according to the HELCOM monitoring guidelines (http://www.helcom.fi/helcom-at-work/publications/manuals-and-guidelines). Dissolving of carbonate structures in cells by this preservative may only bias samples collected at the westernmost stations, as in the central Baltic, organisms with carbonate components (e.g. coccolithophorides) are less abundant. The volumes of filtered water and filtrates were recorded (Supplementary Table S1). In the laboratory, a subsample of this concentrate was transferred to either 10 or 25 mL sedimentation chambers and the entire bottom plate was visually surveyed under an inverted microscope (Zeiss Axiovert S 100, Carl Zeiss MicroImaging GmbH, Göttingen, Germany) according to the method of Utermöhl (1958). For very abundant species that exceeded 50 individuals on less than 10% of the bottom plate, the numbers were extrapolated to the whole bottom plate according to the proportion of the area. Therefore, the theoretical detection limit of this counting method was between 0.6 and 1.9 cells L−1. Flow cytometry For flow cytometry, 4 mL of samples were collected from the CTD-rosette immediately after their retrieval and fixed for 1 h at 4°C with a mixture of paraformaldehyde and glutaraldehyde at a final concentration of 1% and 0.05%, respectively. The fixed cells were flash frozen with liquid nitrogen and stored at −80°C. For the analysis, samples were thawed and 300 μL of sample was used for measurement with the flow cytometer FACS Calibur (Becton Dickinson, San Jose, USA), making use of the autofluorescence of the small photoautotrophs Synechococcus and pigmented picoeukaryotes and nanoflagellates. The excitation wavelength was 488 nm and the detection of the cells was achieved at a flow rate of 36 μL min−1 with an acquisition time of 180 sec using emission filters FL3 (585/42 nm) and FL2 (661/12 nm). FL2 (phycoerythrin) vs. FL3 (chlorophyll a) fluorescence was used to differentiate picocyanobacteria from the pigmented picoeukaryotes and nanoeukaryotes according to the method of Gasol (1999). These gated populations are depicted in Supplementary Fig. S2 showing a FL3 vs. SSC (side scatter) plot, where the size differences between the groups are much clearer (beads size = 2 μm). Sampling for molecular analyses For DNA analysis, 1 L of seawater was vacuum filtered directly onto a 47-mm Durapore membrane filter (pore size of 0.2 μm; GVWP04700, Merck Millipore, Darmstadt, Germany) at <300 mbar. The filters were subsequently folded, flash frozen using liquid nitrogen and stored at −80°C until further analysis. Nucleic acid extraction DNA was extracted using the QIAamp DNA mini kit (51304, Qiagen, Hilden, Germany; for detailed information see Supplementary Material 1.3) as described by Drancourt et al. (2000). An initial bead-beating step was additionally employed, along with a clean-up and concentration step using the Zymo gDNA clean and concentrator kit (D4010, Zymo Research Europe, Freiburg, Germany). The concentration and the quality of eluted DNA were confirmed by gel electrophoresis and a molecular assay using the Bioanalyzer DNA12000 kit (5067-1508, Agilent Technologies, Santa Clara, USA). 16S rRNA gene amplicon library preparation Library construction and sequencing Two PCR amplifications were made per sample. First, the amplicon PCR and the subsequent index PCR were performed according to the instructions provided by Illumina (including primer selection) for preparing 16S rRNA gene amplicons for the Illumina MiSeq system (Illumina). For detailed information on the single steps, see Supplementary Material 1.4. Primers targeting the variable V3/V4 region of the 16S rRNA were suggested by Klindworth et al. (2012) to be the most promising primer pair for detecting bacteria and were used in a limited-cycle PCR that resulted in amplicon sizes of ~460 bp. The amplicon PCR products were cleaned according to the instructions included with the Agencourt® AMPure® XP kit and quality checked using the Bioanalyzer 1000 DNA kit. The quality and the quantity of each library were checked using a Bioanalyzer (Agilent Technologies, Santa Clara, USA) and Quibit assay (Thermo Fischer Scientific, Schwerte, Germany), respectively. The libraries were pooled with equimolar amounts to obtain a final concentration of 4 nM. From each pooled library, 4 pM were sequenced using an Illumina 600 v3 reagent cycle kit with paired-end protocol. The PhiX control library was spiked into each pool to a final amount of 10%. A cluster density of ~1200 K mm−2 was reached in over 80% sequencing and index reads and a Q-score ≥30 was routinely achieved. Individual runs generated between 15 and 25 million reads. FASTQ files were converted from *bcl files and used for further sequence data processing as outlined below. Bioinformatic processing For bioinformatic analyses, including pairing the forward and reverse reads, the QIIME version 1.9.1 was used (Caporaso et al., 2012). Unmerged reads were discarded. The SILVAngs pipeline (Quast et al., 2013) was used for operational taxonomic unit (OTU) picking and for determination of taxonomic affiliation. The PhytoREF database (Decelle et al., 2015) was used for taxonomic assignment of chloroplast sequences. For detailed information on the single steps, see Supplementary Material 1.5. The chloroplast sequences were identified using the program ARB (version 6.0.2; Ludwig et al., 2004) by extracting all sequences assigned “chloroplast” and mapping them to the PhytoREF database (Decelle et al., 2015). This database was accessed online at “http://phytoref.sb-roscoff.fr/” to assign taxonomy to the 16S rRNA gene sequences of the chloroplasts. The chloroplast classification table provided by PhytoREF was combined with the bacterial 16S rRNA gene OTU table based on the SILVAngs pipeline. Since all samples contained different numbers of sequences (Table 1), QIIME (Caporaso et al., 2012) was used to rarefy the dataset to the lowest number of sequences (26 019 sequences) among all samples. The tab-delimited, combined OTU table was converted into a biological observation matrix table, which functioned as an input for the QIIME single.rarefaction.py script (for detailed information, see Supplementary Material 1.5). Data deposition All raw sequence files are available from the NCBI Short Read Archive (SRA) database (BioProject: PRJNA379767, accession numbers: SAMN06619861 and SAMN06619896). RESULTS Table 1 summarizes the total number of sequences obtained, including chloroplast sequences but excluding sequences identified as “No Relatives” (i.e. reads that did not exceed the similarity threshold of 93%, which account for 0.05–0.27% of the samples). In total, 514 OTUs were identified, of which 480 OTUs were affiliated with prokaryotes and 34 OTUs with eukaryotes. About 32 out of the 480 prokaryotic OTUs belonged to unicellular or filamentous members of the Cyanobacteria, thus indicating abundances between 23% and 68% of the total amount of sequences per station. 16S rRNA gene analysis Between 3% and 9% of the sequences were identified as chloroplasts in the 16S rRNA gene amplicon libraries (Table 1). These chloroplast sequences were further classified into 15 different classes. Five of these 15 classes were present at all stations (Dictyochophyceae, Chrysophyceae, Eustigmatophyceae, Trebouxiophyceae and Cryptophyceae; Fig. 1). Whereas, chrysophyceae showed higher sequence abundances (16–23% of chloroplast sequences) in the western Baltic Sea, Trebouxiophyceae and Eustigmatophyceae increased in abundance towards the Gotland Sea and Bothnian Sea (78% and 35.5%, respectively; Fig. 1). Dictyochophyceae and Cryptophyceae revealed similar sequence abundances in both marine waters and waters at Gulf of Bothnia. Further, the classes Bacillariophyceae and Prymnesiophyceae were present at all stations except at the Gotland Sea station, with the highest sequence abundance in marine waters (Fig. 1). The class Prasinophyceae was most abundant in the western Baltic Sea, absent in the central Baltic Sea and reappeared with low numbers in the Gulf of Bothnia (Fig. 1). Overall, among the Bacillariophyceae, 10 different families were identified, but the majority of Bacillariophyceae sequences in marine waters could not be assigned to any family and remained unclassified. At stations, Mecklenburg Bight, Arkona and Bornholm Sea, the majority of Bacillariophyceae sequences were affiliated with the family Chaetocerotaceae. The remaining nine families had sequence abundances below 1% except for Cytomatosiraceae, Hemiaulaceae and Thalassiosiraceae, which showed sequence abundances between 1% and 3% at Skagerrak. For the class Dictyochophyceae and Cryptophyceae, no sequences could be assigned to any family; they remained unclassified. All sequences of the Chrysophyceae were affiliated with the family Chromulinaceae and the class Eustigmatophyceae affiliated with the family Monodopsidaceae. Among the Prasinophyceae, the family Pyramimonadaceae were identified. Within the class Trebouxiophyceae, the two families Chlorellaceae and Coccomyxaceae were identified. The Coccomyxaceae occurred with higher sequence abundances (24% of chloroplast sequences) in the Belt Sea and increased towards maximum abundance in the Gotland Sea (60% of chloroplast sequences). Chlorellaceae first occurred with more than 16% of all chloroplast sequences at the Arkona Sea station and reached maximum abundance at Bornholm station, with 23% of the chloroplast sequences. Both families decreased in abundances towards the Gulf of Bothnia. Among the haptophytes, the four families Chrysochromulinaceae, Prymnesiaceae, Noelaerhabdaceae and Phaeocystaceae were identified. The latter two families appeared at highest sequence abundances in the Skagerrak station (8% and 23%, respectively), whereas Chrysochromulinaceae and Prymnesiaceae increased in abundance between Mecklenburg Bight and Bornholm Sea (Fig. 2). Fig. 2. View largeDownload slide Comparison of flow cytometry data and 16S rRNA gene analysis in relative abundances. Flow cytometry data (cells <7 μm) represented by bar plot and 16S rRNA gene analysis (cells >0.2 μm) by heatmap. Numbers indicate Baltic Sea regions, similar to Figs 1 and 3: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. Bar plot creation was performed using sigma plot vs. 10 and the heatmap using R package pheatmap (www.r-project.org). Names marked with * were also displayed in Fig. 1B. Total numbers of chloroplast and cyanobacteria sequences are given in Table 1. Fig. 2. View largeDownload slide Comparison of flow cytometry data and 16S rRNA gene analysis in relative abundances. Flow cytometry data (cells <7 μm) represented by bar plot and 16S rRNA gene analysis (cells >0.2 μm) by heatmap. Numbers indicate Baltic Sea regions, similar to Figs 1 and 3: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. Bar plot creation was performed using sigma plot vs. 10 and the heatmap using R package pheatmap (www.r-project.org). Names marked with * were also displayed in Fig. 1B. Total numbers of chloroplast and cyanobacteria sequences are given in Table 1. Evaluation of molecular analysis In order to compare the obtained molecular analysis, flow cytometry and microscopic investigations of the phytoplankton community were performed (Figs 1–3). It should be noted that the microscopic analyses were based on 3 L of water sample for algae >20 μm, molecular analyses were based on 1 L of the total sample and flow-cytometric analyses on only 4 mL sample of autotrophs <7 μm. Comparing samples consisting of different volumes may impact organism detection limits; however, with regard to molecular analyses the most abundant organisms have been very likely recovered from the different counting and molecular methods. This is because PCR-based investigations (and associated exponential multiplications) can theoretically detect one gene per sample, especially when based on two-step PCRs (Labrenz et al., 2004). Flow cytometry assessed pigmented microorganisms, such as the cyanobacterial clade Synechococcus, pigmented picoeukaryotes and nanoeukaryotes smaller <7 μm (Supplementary Fig. S2). The latter groups were counted as bulk and no differentiation between classes and families could be made. In contrast, molecular analysis detected nine families affiliated with the classes Bacillariophyceae, Chrysophyceae, Eustigmatophyceae, Prasinophyceae and Trebouxiophyceae. Furthermore, flow cytometry showed Synechococcus to be the dominant organism at all stations, which was also reflected in the 16S rRNA gene libraries (Fig. 2). In addition to Synechococcus, other unicellular cyanobacteria, such as Snowella, Merismopedia and Microcystis, were identified by sequencing but had relative abundances below 1%. Fig. 3. View largeDownload slide Heatmap of relative abundance of diatom (Bacillariophyceae) families compared between microscopy (cells >20 μm) and DNA reads (cells >0.2 μm) determined by 16S rRNA gene amplicon libraries. Numbers represent the Baltic Sea regions similar as in Fig. 1: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The colour intensity represents relative abundance in percentage. For the shaded parts, no data are available in the PhytoREF database. Heatmap was performed using the R package pheatmap (www.r-project.org). *Underrepresented in database, only few reference sequences available. Fig. 3. View largeDownload slide Heatmap of relative abundance of diatom (Bacillariophyceae) families compared between microscopy (cells >20 μm) and DNA reads (cells >0.2 μm) determined by 16S rRNA gene amplicon libraries. Numbers represent the Baltic Sea regions similar as in Fig. 1: 1 = Skagerrak, 2 = Belt Sea, 3 = Mecklenburg Bight, 4 = Arkona Sea, 5 = Bornholm Sea, 6 = Gotland Sea and 7 = Gulf of Bothnia. The colour intensity represents relative abundance in percentage. For the shaded parts, no data are available in the PhytoREF database. Heatmap was performed using the R package pheatmap (www.r-project.org). *Underrepresented in database, only few reference sequences available. Microscopic counts focussed on the morphologies that were easily and undoubtedly detectable under the light microscope and was therefore only done on the >20 μm size fraction. This fraction comprised diatoms (Bacillariophyceae) and dinoflagellates, found in high abundance (both >40% of the total phytoplankton community; Fig. 3) in the western Baltic Sea (Skagerrak and Belt Sea) and with lower abundances (<10% and 20% of the total phytoplankton community, respectively) in the northern Baltic Sea (Gulf of Bothnia). Bacillariophyceae included the two main families (Fig. 3) Chaetocerataceae and Rhizosoleniaceae present in the western Baltic Sea and Thalassiosiraceae and Hemidiscaceae in the Gulf of Bothnia. Cearatiaceae and Dinophysiaceae were the main families among the dinoflagellates throughout the Baltic Sea transect. The chlorophytes Trebouxiophyceae dominated in the southern Baltic Sea, with highest numbers (>50% of the total phytoplankton community) in the Bornholm Basin. This phylum was exclusively represented by the species Planktonema lauterbornii. In contrast, the filamentous cyanobacterium Aphanizomenon was very abundant (>60% of the total phytoplankton community) in the central (Gotland Sea) and northern Baltic Sea (Gulf of Bothnia) and was almost completely absent in marine waters. Based on 16S rRNA gene analyses, Aphanizomenon and Nodularia were also identified in the central and northern Baltic Sea (data not shown). DISCUSSION In this study, molecular, microscopic and flow-cytometric methods were employed to record the distribution of phototrophic prokaryotes and eukaryotes in the Baltic Sea and the use of the 16S rRNA gene as a suitable target gene for phytoplankton monitoring was evaluated. Molecular observations revealed distinct distribution patterns of diatoms (Bacillariophyceae), Chrysophyceae, silicoflagellates (Dictyochophyceae) and Eustigmatophyceae, as well as cryptophytes (Cryptophyceae) and haptophytes (Prymnesiophyceae) at different Baltic Sea stations. Verification of molecular analysis by microscopy This distribution was confirmed by microscopy, although only the >20 μm fraction could be analysed. On a broad classification level, the assignment of the phytoplankton 16S rRNA gene sequences to classes using phytoREF was successful. For the majority of stations, more than 70% of the 16S rRNA gene sequences were identified at the family level; only at the Skagerrak station were the sequences limited to class-level (Bacillariophyceae) identification. In contrast, microscopy was able to identify phytoplankton down to genus-level, which was only in few cases possible by 16S rRNA gene analysis. For Bacillariophyceae, no significant differences between microscopy and 16S rRNA gene analyses were observed (t-test, P-value >0.05) except that the diversity determined was higher for molecular analysis than for microscopy (10 families vs. 5 families by microscopy; Fig. 3). This is most probably an effect of the difference in mesh size between the filtration methods for phytoplankton (20 μm mesh) and molecular samples (0.2 μm membrane filter), as smaller species or broken chains will pass through the net but will be retained on the filter. Concerning microscopic analyses, this disadvantage was accepted for the benefit of collecting large and easily identifiable phytoplankton species in statistically robust numbers, which is not always the case in non-concentrated water samples. This explanation applies to all stations with both microscopic and molecular analysis (Fig. 1). Thus, differences are most likely due to the differences in sample preparation, rather than vertical sampling heterogeneity, as in the central Baltic, the well-mixed surface layer is always deeper than 10 m (Supplementary Fig. S1). For Chlorophyta, microscopy and molecular analyses were comparable, except that 16S rRNA gene analysis detected Chlorophyta sequences in the Skagerrak and Gulf of Bothnia (Fig. 1; 8% and 36% relative read abundance, respectively). Either these are species which were not detected by microscopy due to the pre-filtration at 20 μm as mentioned above, or the molecular approach discovered cells which were rarely present and potentially overseen by microscopy. Conversely, looking deeper into the class Trebouxiophyceae, microscopy solely identified the species Planktonema lauterbornii, which was not identified using molecular investigations. Instead, Trebouxiophyceae remained unclassified because the chloroplast sequence of Planktonema lauterbornii is currently still absent from phytoREF. Comparing microscopy and molecular analyses more technically, microscopy requires time and expert knowledge and gets more difficult the smaller the cells become. However, with the molecular approach, using user-independent standardized protocols for sampling, nucleic acid extraction and library preparation (as well as sequencing and bioinformatic processing), comparability between samples from different expeditions, years and stations can be ensured. Despite the lower taxonomic resolution of this method vs. microscopy, it was able to detect phototrophic organisms in the small size range, which are often difficult to recognize by microscopy or flow cytometry (Figs 1–3). Moreover, it is advantageous that molecular investigations are designed for high-throughput analysis so that a larger number of samples (96 and more) can be processed simultaneously. Unlike microscopy, no explicit taxonomic expert knowledge is required to apply such a tool at a comparable cost per sample. Thus, it is probable that molecular investigation will be the most important way to link very small phototrophs with taxonomic information in the future. A step potentially introducing biases in the molecular approach is the 16S rRNA gene library preparation, as each sample undergoes two PCR amplification steps, which can result in template primer mismatches or over-representation of certain sequences in mixed community samples (Sipos et al., 2007). Primers also select for distinct organism groups. The primers used in this study are designed for bacteria amplification and therefore revealed a considerably low amount of chloroplast sequences. However, this low amount of chloroplast sequences gave a broad overview of the dominant phytoplankton groups at different stations of the Baltic Sea with distinct physico-chemical profiles. Verification of molecular analysis by flow cytometry As described above, flow cytometry is limited by size and counts picoeukaryotes and nanoeukaryotes as a single group without differentiating between picophytoplankton groups (Hammes and Egli, 2010). Therefore, comparing flow-cytometric data with molecular analysis is possible solely for the cyanobacterial clade Synechococcus. Further group differentiation would require flow sorting followed by sequencing of the sorted fraction (Cellamare et al., 2010) which is beyond the scope of a fast and simple monitoring approach. At the molecular level, the typical Baltic Sea picophytoplankton group haptophytes was identified down to family level and chlorophytes remained on class-level identification. Kleptoplastidy Detection of plastidal sequences in amplicon libraries can be improved when using other primers (see, for example, Hansen et al., 2013; Needham and Fuhrman, 2016). The development of new primers targeting dinoflagellate chloroplasts (specifically toxic species) would considerably advance the performance of molecular-based analyses. However, designing new primers that would also detect the chloroplast sequence of dinoflagellates is difficult. The genomes of dinoflagellate plastids have usually been broken up into mini circles, and many genes including 16S rRNA and 23S rRNA been transferred into the nucleus of dinoflagellates (Green, 2011). This makes it almost impossible to amplify the 16S rRNA gene of dinoflagellate plastids, which is reflected by their underrepresentation in the phytoREF database (0.5% of all entries) and subsequent lack of detection in the present dataset. Instead, 18S rRNA gene amplification may well reveal a better resolution in this case. In addition, some dinoflagellates carry out phototrophy based on the retained plastids of their prey (Hansen et al., 2013; Park et al., 2014). This is mainly observed in members of the order Dinophysiales (Meyer-Harms and Pollehne, 1998). Based on our microscopic data, Ceratium was generally present in the western Baltic Sea and Dinophysis dominated towards the central and northern Baltic Sea. This is analogous to the cryptophytes distribution. Interestingly, a comparison of the Cryptophyta sequence data with the dinoflagellate count data using a paired t-test (P < 0.05) did not reveal significant differences, and the trends were similar. This suggests that Cryptophyta plastids could be incorporated into dinoflagellates, but direct evidence is still lacking. Similar to several dinoflagellates, the ciliate Mesodinium rubrum also acquires chloroplasts from their (mainly cryptophyte) prey. This annexation of plastids is referred to as kleptoplastidy (Kim et al., 2012; Hansen et al., 2013; Park et al., 2014). However, we were not able to identify if those cryptophyte 16S rRNA gene sequences truly belong to cryptophytes or if they were taken up by Dinophysis, Ceratium or Mesodinium. The 16S rRNA gene sequence of the cryptophyte which is taken up by Mesodinium is named “ciliate” in PhytoREF. However, our dataset did not reveal any 16S rRNA gene sequence identified as “ciliate”; all 16S rRNA gene sequences affiliated with cryptophytes were named “Cryptophyceae-unclassified”. Choice of primers For this study, we used primers (341 F/805 R) which are standardized for 16S analysis (Klindworth et al., 2012), although the paired reads do not fully overlap when using Illumina MiSeq sequencing technology. However, they are nevertheless recommended by Illumina as the promising primer pair for detecting prokaryotic sequences and the forward and reverse reads overlap with more than 120 bp when using the v3 kit (2 × 300 bp). Furthermore, it has been shown that this primer pair also detects chloroplast 16S rRNA sequences (Eiler et al., 2013), although the chloroplast 16S rRNA gene sequence has four mismatches between positions 783–799 (according to E. coli base positioning system) compared to the 16S rRNA gene sequence of Cyanobacteria (Hanshew et al., 2013). Three of the four mismatches are located directly in the primer binding site of the 805 R primer, which eventually results in non- or rare binding of this reverse primer to the 16S rRNA gene sequence of chloroplasts, thus minimizing the amount of detected chloroplast sequences. In the present study, 10% chloroplast sequences were detected, at best. Others (Eiler et al., 2013; for example) obtained 15% chloroplast sequences for most samples and for two sample locations they obtained as much as 50% chloroplast sequences (Eiler et al., 2013). Recently, Needham and Fuhrman (2016) published primers (515 F/926 R) which seem to neglect the mismatches between plastidal and cyanobacterial 16S rRNA gene sequences. Furthermore, their results showed that the plastidal 16S rRNA analysis was highly concordant with the 18S rRNA analysis (Needham and Fuhrman, 2016). Also only recently Milici et al. (2016) demonstrated for the Atlantic Ocean that sequences retrieved with the bacterial primers F807 and R1050 covered almost 70% of the marine eukaryotic photosynthetic lineages of the PhytoREF database (Decelle et al., 2015), leading to the conclusion that this approach is sufficient to investigate the distribution of eukaryotic microalgae. CONCLUSION In summary, 16S rRNA gene analysis based on the selected prokaryotic primer set coincided well with microscopic identification. Thus, for general information on the presence and absence of specific phototrophic clades, such as diatoms (Bacillariophyceae), green microalgae (Chlorophyta) and especially Cyanobacteria, including the unicellular fraction, molecular analyses provided a sufficient toolset when using the 16S rRNA gene sequence. Despite the lower taxonomic resolution of this method vs. microscopy for the large phytoplankton fraction, molecular analysis detected phototrophic organisms in the small size range, which are often difficult to recognize by microscopy or flow cytometry (Figs 1 and 2). In addition, phytoREF should improve in the detection of plastidal sequences with the addition of more sequences into the database; thus, taxonomic expertise is necessary to build and improve such reference database. An advantage of molecular investigations is that they are designed for high-throughput analysis, handling larger number of samples (96 and more) at the same time. Meanwhile, technical systems have been developed that can combine high-throughput 16S rRNA analysis with microscopic analysis, for instance, 3D-fluorescence imaging. With a long-term perspective, the combination of molecular methods and microscopy will be able to achieve the resolution necessary for a more comprehensive assessment of phytoplankton and thereby expand the utility of monitoring strategies. However, if costs for molecular analysis continue to decrease and the taxonomic resolution continues to increase this approach could become even more important in the future. SUPPLEMENTARY DATA Supplementary data can be found online at Journal of Plankton Research online. ACKNOWLEDGEMENTS We thank the captain and crew of the R/V Alkor for assistance in sampling during the AL439 cruise. We greatly appreciate the substantial contribution of Siegfried Krüger in handling the CTD-rosette and the AFIS sampler during the cruise. We also thank Heike Benterbusch-Brockmüller and Andreas Rogge for assistance in filtration at sea. We would like to acknowledge Annett Grüttmüller for providing flow cytometry work and Jana Normann for handling and loading the Illumina MiSeq Sequencer. Special thanks are going to Alexander Tagg for critically proofreading the manuscript. FUNDING This work resulted from the EU-BONUS BLUEPRINT project and was supported by BONUS (Art 185), funded jointly by the EU and the German Federal Ministry of Education and Research (BMBF; 03F0679A), as well as from support by the Deutsche Forschungsgemeinschaft (project LA 1466/8-1). Purchase of the Illumina MiSeq was kindly supported by the EU-EFRE (European Funds for Regional Development) program and funds from the University Medicine Rostock awarded to BK. REFERENCES Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Huntley, J., Fierer, N., Owens, S. M., Betley, J. et al.   ( 2012) Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. , 8, 1621– 1624. Google Scholar CrossRef Search ADS   Carpenter, E. J., Janson, S., Boje, R., Pollehne, F. and Chang, J. ( 1995) The dinoflagellate Dinophysis norvegica. Eur. J. Phycol. , 1, 1– 9. Google Scholar CrossRef Search ADS   Cellamare, M., Rolland, A. and Jacquet, S. ( 2010) Flow cytometry sorting of freshwater phytoplankton. J. Appl. Phycol. , 1, 87– 100. Google Scholar CrossRef Search ADS   Decelle, J., Romac, S., Stern, R. F., Bendif el, M., Zingone, A., Audic, S., Guiry, M. D., Guillou, L. et al.   ( 2015) PhytoREF: a reference database of the plastidial 16S rRNA gene of photosynthetic eukaryotes with curated taxonomy. Mol. Ecol. Resour. , 6, 1435– 1445. Google Scholar CrossRef Search ADS   Decelle, J., Martin, P., Paborstava, K., Pond, D. W., Tarling, G., Mahé, F., De Vargas, C., Lampitt, R. et al.   ( 2013) Diversity, ecology and biogeochemistry of cyst-forming Acantharia (Radiolaria) in the oceans. PLoS One , 8, e53598. Google Scholar CrossRef Search ADS PubMed  Drancourt, M., Bollet, C., Carlioz, A., Martelin, R., Gayral, J.-P. and Raoult, D. ( 2000) 16S ribosomal DNA sequence analysis of a large collection of environmental and clinical unidentifiable bacterial isolates. J. Clin. Microbiol. , 38, 3623– 3630. Google Scholar PubMed  Eiler, A., Drakare, S., Bertilsson, S., Pernthaler, J., Peura, S., Rofner, C., Simek, K., Yang, Y. et al.   ( 2013) Unveiling distribution patterns of freshwater phytoplankton by a next generation sequencing based approach. PLoS One , 1, e53516. Google Scholar CrossRef Search ADS   Gasol, J. M. ( 1999) How to Count Picoalgae and Bacteria with the FACScalibur Flow Cytometer . Departament de Biologia Marina i Oceanografia. Institut de Ciencies del Mar, CSIC, Barcelona, pp. 1– 51. Green, B. R. ( 2011) Chloroplast genomes of photosynthetic eukaryotes. Plant J. , 1, 34– 44. Google Scholar CrossRef Search ADS   Hammes, F. and Egli, T. ( 2010) Cytometric methods for measuring bacteria in water: advantages, pitfalls and applications. Anal. Bioanal. Chem. , 3, 1083– 1095. Google Scholar CrossRef Search ADS   Hansen, P. J., Nielsen, L. T., Johnson, M., Berge, T. and Flynn, K. J. ( 2013) Acquired phototrophy in Mesodinium and Dinophysis—a review of cellular organization, prey selectivity, nutrient uptake and bioenergetics. Harmful Algae , 28, 126– 139. Google Scholar CrossRef Search ADS   Hanshew, A. S., Mason, C. J., Raffa, K. F. and Currie, C. R. ( 2013) Minimization of chloroplast contamination in 16S rRNA gene pyrosequencing of insect herbivore bacterial communities. J. Microbiol. Meth. , 2, 149– 155. Google Scholar CrossRef Search ADS   Herlemann, D. P., Labrenz, M., Jürgens, K., Bertilsson, S., Waniek, J. J. and Andersson, A. F. ( 2011) Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. ISME J. , 10, 1571– 1579. Google Scholar CrossRef Search ADS   Illumina. 16S Metagenomic Sequencing Library Preparation. http://www.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/16s/16s-metagenomic-library-prep-guide-15044223-b.pdf.; [accessed: 13 October 2014]. Kim, M., Kim, S., Yih, W. and Park, M. G. ( 2012) The marine dinoflagellate genus Dinophysis can retain plastids of multiple algal origins at the same time. Harmful Algae , 13, 105– 111. Google Scholar CrossRef Search ADS   Klais, R., Tamminen, T., Kremp, A., Spilling, K. and Olli, K. ( 2011) Decadal-scale changes of dinoflagellates and diatoms in the anomalous Baltic Sea spring bloom. PLoS One , 6, e21567. Google Scholar CrossRef Search ADS PubMed  Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M. and Glöckner, F. O. ( 2012) Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res. , 1, e1– e1. Labrenz, M., Brettar, I., Christen, R., Flavier, S., Bötel, J. and Höfle, M. G. ( 2004) Development and application of a real-time PCR approach for quantification of uncultured bacteria in the central Baltic Sea. Appl. Environ. Microbiol. , 70, 4971– 4979. Google Scholar CrossRef Search ADS PubMed  Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Yadhukumar, Buchner, A., Lai, T. et al.   ( 2004) ARB: a software environment for sequence data. Nucleic Acids Res. , 4, 1363– 1371. Google Scholar CrossRef Search ADS   Marshall, H. G., Lane, M. F., Nesius, K. K. and Burchardt, L. ( 2009) Assessment and significance of phytoplankton species composition within Chesapeake Bay and Virginia tributaries through a long-term monitoring program. Environ. Monit. Assess. , 1–4, 143– 155. Google Scholar CrossRef Search ADS   Meyer-Harms, B. and Pollehne, F. ( 1998) Alloxanthin in Dinophysis norvegica (Dinophysiales, Dinophyceae) from the Baltic Sea. J. Phycol. , 2, 280– 285. Google Scholar CrossRef Search ADS   Milici, M., Deng, Z.-L., Tomasch, J., Decelle, J., Wos-Oxley, M. L., Wang, H., Jáuregui, R., Plumeier, I. et al.   ( 2016) Co-occurrence analysis of microbial taxa in the Atlantic Oceanr reveals high connectivity in the free-living bacterioplankton. Front. Microbiol. , 7, 649. Google Scholar PubMed  Murray, S. A., Suggett, D. J., Doblin, M. A., Kohli, G. S., Seymour, J. R., Fabris, M. and Ralph, P. J. ( 2016) Unravelling the functional genetics of dinoflagellates: a review of approaches and opportunities. Persp. Phycol. , 1, 37– 52. Needham, D. M. and Fuhrman, J. A. ( 2016) Pronounced daily succession of phytoplankton, archaea and bacteria following a spring bloom. Nat. Microbiol. , 1, 1– 7. Google Scholar CrossRef Search ADS   Park, M. G., Kim, M. and Kim, S. ( 2014) The acquisition of plastids/phototrophy in heterotrophic dinoflagellates. Acta Protozool. , 53, 39– 50, Special topic issue: “Marine Heterotrophic Protists”. Pei, A. Y., Oberdorf, W. E., Nossa, C. W., Agarwal, A., Chokshi, P., Gerz, E. A., Jin, Z., Lee, P. et al.   ( 2010) Diversity of 16S rRNA genes within individual prokaryotic genomes. Appl. Environ. Microbiol. , 12, 3886– 3897. Google Scholar CrossRef Search ADS   Prokopowich, C. D., Gregory, T. R. and Crease, T. J. ( 2003) The correlation between rDNA copy number and genome size in eukaryotes. Genome , 1, 48– 50. Google Scholar CrossRef Search ADS   Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J. and Glöckner, F. O. ( 2013) The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. , 41, D590– D596. Database issue. Google Scholar CrossRef Search ADS PubMed  Schlitzer, R. ( 2015). Ocean Data View. http://odv.awi.de. Sipos, R., Székely, A. J., Palatinszky, M., Révész, S., Márialigeti, K. and Nikolausz, M. ( 2007) Effect of primer mismatch, annealing temperature and PCR cycle number on 16S rRNA gene-targetting bacterial community analysis. FEMS Microbiol. Ecol. , 60, 341– 350. Google Scholar CrossRef Search ADS PubMed  Utermöhl, H. ( 1958) Zur Vervollkommnung der quantitativen Phytoplanktonmethodik. Mitteilungen Int. Ver. Theor. Angew. Limnol. Band , Vol 9. Schweizerbart, Stuttgart, p. 38. Wasmund, N., Nausch, G. and Matthäus, W. ( 1998) Phytoplankton spring blooms in the southern Baltic Sea—spatio-temporal development and long-term trends. J. Plankton Res. , 6, 1099– 1117. Google Scholar CrossRef Search ADS   Wasmund, N., Dutz, J., Pollehne, F., Siegel, H. and Zettler, M. L. 2015. Biological assessment of the Baltic Sea 2014. Meereswiss. Ber. , Warnemünde, 98. doi:10.12754/msr-2015-0098. Wasmund, N., Dutz, J. and Zettler, M. L. 2017. Contribution to the national commentary by Germany on annual assessment of the environmental status of the Baltic Sea based on biological monitoring in 2016. Leibniz Institute for Baltic Sea Research Warnemünde (IOW). Wiltshire, K. H. and Manly, B. F. J. ( 2004) The warming trend at Helgoland Roads, North Sea. Hel. Mar. Res. , 4, 269– 273. Google Scholar CrossRef Search ADS   Author notes Corresponding Editor: John Dolan © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Plankton ResearchOxford University Press

Published: Apr 7, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off