Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Major changes in the composition of a Southern Ocean bacterial community in response to diatom-derived dissolved organic matter

Major changes in the composition of a Southern Ocean bacterial community in response to... Abstract In the Southern Ocean, natural iron fertilization in the wake of islands leads to annually occurring spring phytoplankton blooms associated with enhanced heterotrophic activity through the release of labile dissolved organic matter (DOM). The aim of this study was to investigate experimentally how diatom-derived DOM affects the composition of Southern Ocean winter water bacterial communities and to identify the most responsive taxa. A bacterial community collected in the naturally iron-fertilized region off Kerguelen Island (KEOPS2 October-November 2011) was grown onboard in continuous cultures, on winter water alone or amended with diatom-derived DOM supplied at identical DOC concentrations. 454 sequencing of 16S amplicons revealed that the two DOM sources sustained strikingly different bacterial communities, with higher relative abundances of Sulfitobacter, Colwellia and Methylophaga operational taxonomic units (OTUs) and lower relative abundances of Polaribacter, Marinobacter, NAC11–7 and SAR11 OTUs in diatom-DOM compared to winter water conditions. Using a modeling approach, we obtained growth rates for phylogenetically diverse taxa varying between 0.12 and 0.49 d−1 under carbon-limited conditions. Our results identify diatom DOM as a key factor shaping Southern Ocean winter water bacterial communities and suggest a role for niche partitioning and microbial interactions in organic matter utilization. bacterial community composition, Southern Ocean, dissolved organic matter, 16S rRNA gene sequencing, continuous cultures INTRODUCTION Bacteria are key players in the transformation of carbon present in marine dissolved organic matter (DOM). Understanding the mechanisms of carbon utilization by marine bacteria is presently a major challenge because both the DOM pool (Hertkorn et al. 2007) and bacterial communities (Sogin et al. 2006; Sunagawa et al. 2015) are highly complex. Advances in the fields of analytical chemistry and constantly improving high-throughput sequencing technologies have enabled progress but much remains to be learned about the composition and interactions between the chemical and microbiological components of the ocean. From an ecological point of view, understanding the factors that drive bacterial community composition would shed light on their roles in the biogeochemical cycling of carbon and other elements. Because DOM is the main source of substrates for marine bacteria, its characteristics, such as chemical composition and bioavailability, and its concentration are likely to play important roles in structuring bacterial communities. It has been suggested that individual bacterial taxa possess different metabolic properties, reflecting the vast diversity of compounds found in a given environment (Wawrik et al. 2005; Teeling et al. 2012; Satinsky et al. 2014). As a result, various taxa could be adapted to different lifestyles based on distinct resource exploitation strategies (Lauro et al. 2009). This hypothesis is supported by numerous studies showing biogeographic patterns as well as predictable bacterial community composition in recurring conditions (Fuhrman et al. 2006; Martiny et al. 2006; Nelson et al. 2008; Gilbert et al. 2012). However, there is also a large body of evidence for functional redundancy within and across bacterial communities (Langenheder, Lindström and Tranvik, 2005, 2006; Sjöstedt et al. 2013; Beier et al. 2015), as well as generalist carbon consumption (Mou et al. 2008; Landa et al. 2014; Pedler, Aluwihare and Azam 2014; Sarmento, Morana and Gasol 2016), suggesting functional traits alone do not explain taxon distribution and abundance. Taken together, these observations indicate that the factors driving microbial community structure and activity are complex and vary in nature and intensity across space and time. The extent to which characteristics of the carbon pool used by heterotrophic bacteria act as shaping force on community composition remains to be determined. Phytoplankton supplies labile DOM to heterotrophic bacteria, which results in a tight link between these components in all aquatic ecosystems (Cole, Findlay and Pace 1988). Numerous studies have investigated how pulses of bioavailable DOM released during phytoplankton blooms affect bacterial heterotrophic activity on seasonal (Church 2008 and references therein) and inter-annual scales (Kim and Ducklow 2016). The concurrent changes in the bacterial community were suggested to be driven by the quantity and quality of DOM released during bloom events (Kerkhof et al. 1999; Riemann, Steward and Azam 2000; Pinhassi et al. 2004; Rink et al. 2007; West et al. 2008; Tada et al. 2012; Delmont et al. 2014; Kim et al. 2014; Luria et al. 2016; Needham and Fuhrman 2016). However, few studies have supported their field observations with experimental studies, directly testing the effect of phytoplankton-derived DOM on bacterial community composition (Luria et al. 2017). The Southern Ocean is the largest high-nutrient, low-chlorophyll region in the world's oceans. Iron-limited phytoplankton primary production (Boyd et al. 2000; Blain et al. 2007) in combination with permanent upwelling of deep water around Antarctica result in Southern Ocean surface waters characterized by low concentrations of dissolved organic carbon (DOC, about 50 µM, Hansell 2013), of rather refractory nature. This leads to a double constraint on heterotrophic bacteria, which require both iron and carbon for growth (Church, Hutchins and Ducklow 2000; Kirchman et al. 2000). While iron was identified as the limiting nutrient for heterotrophic microbes in only a few studies, the persistent enhancement of heterotrophic bacterial metabolism upon addition of labile organic carbon compounds (see for review Obernosterer, Fourquez and Blain 2015) clearly highlights that bacterial activity in the Southern Ocean is constrained by the bioavailability of DOM. In the present study, we tested the hypothesis that the supply of labile DOM derived from diatoms, the dominant bloom forming phytoplankton in naturally iron-fertilized regions of the Southern Ocean (Blain et al. 2007; Quéguiner 2013), leads to shifts in the composition of a natural, C-limited bacterial community. Continuous cultures were implemented, to actively select bacterial members most adapted either to the background DOM present in the winter waters, or to a mixture of background and diatom-derived DOM. Diversity analyses revealed striking differences in the identity of the taxa selected by each of the tested DOM conditions. We observe that, in this environment where labile substrates are scarce, bacterial community composition is strongly affected by qualitative differences in the available DOM pool , demonstrating a tight link between the bioavailability of organic carbon and bacterial metabolism and diversity in the Southern Ocean. MATERIALS AND METHODS Experimental design The continuous culture experiment presented in this study was ran on board on the R/V Marion Dufresne during the KEOPS2 cruise conducted in the naturally iron-fertilized region around the Kerguelen Islands in the Southern Ocean in October and November 2011. Two experimental conditions were run in triplicates. In the first condition, or control, bacteria were supplied with 0.2 µm-filtered seawater, while in the second condition, or diatom-DOM treatment, bacteria were supplied with 0.2 µm-filtered seawater mixed with diatom-derived DOM. For the control medium preparation, 60 l of seawater were sampled with the trace-metal clean rosette in the upper 80 m at Station A3 (Table S1, Supporting Information) located above the Kerguelen plateau (Fig. S1, Supporting Information), in the iron-fertilized region before the onset of the bloom (Landa et al. 2016). The 60 l were collected in acid-washed (10% HCl), Milli-Q water-rinsed polycarbonate (PC) carboys and immediately 0.2 μm-filtered. This water was stored at 4°C in acid-washed, Milli-Q water-rinsed PC carboys for the entire duration of the experiment. Water for the bacterial inoculum was collected at 40 m and 4 l were filtered through a 0.8 μm PC filter (45 mm diameter, Nuclepore, Whatman, Little Chalfont, UK). The 0.8 µm filtrate was stored at 6°C for a few hours until the beginning of the experiment. Concurrently, 4.8 l of the <0.8 μm filtrate were collected on a 0.2 μm cartridge (Sterivex, Millipore, Darmstadt, Germany), 500 μl of lysis buffer (40 mM EDTA, 50 mM Tris, 0.75 M sucrose) were added, and the cartridge was stored at -80°C for further diversity analysis. The diatom-derived DOM was prepared in the laboratory before the cruise. A non-axenic strain of the diatom Chaetoceros debilis, isolated from the Southern Ocean and provided by the Alfred Wegener Institute (Germany) was cultivated in artificial seawater containing low amounts of background organic carbon (DOC ∼15 µM) and amended with nutrients according to Guillard's f/2 medium. The C. debilis cultures were grown at 4°C at a light:dark cycle of 16h:8h. Prior to the cruise, cells were removed from the culture by gentle filtration on two acid-washed, Milli-Q water-rinsed 0.2 μm PC filters (90 mm diameter, Nuclepore) placed on top of each other. The filtrate (5 l) was collected in acid-washed, Milli-Q water-rinsed PC carboys and stored at 4°C until the start of the experiment. The initial cultures (2 l final volume each) were prepared in acid-washed, Milli-Q water-rinsed 2 l PC carboys. For the control, the 0.2 μm-filtered in situ seawater was used as the medium, and for the diatom-DOM treatment the 0.2 μm-filtered in situ seawater was mixed with 0.2 μm-filtered diatom-derived DOM (see above). The volumes to mix were determined based on the DOC concentration of the diatom filtrate, so that diatom DOM contributed to 10 μM of the final DOC concentration. The triplicate cultures of each condition were inoculated with 200 ml of 0.8 μm-filtered in situ seawater, corresponding to a 1:10 dilution. Inorganic phosphate and nitrate were added to the control to match the concentrations obtained in the diatom-DOM treatment. The cultures were conducted in a cold room (6°C) in the dark. Each culture was bubbled with filter-sterilized air (Sartorious, Göttingen, Germany). A Milli-Q water purge was placed between the source and the filters to avoid potential DOC contamination by air. Cultures were homogenized by stirring. They were run as a batch for a few hours, to allow bacterial growth and avoid washing out the cells. Each medium, prepared in the same conditions without bacterial inoculum, was connected to the three replicate cultures of the corresponding condition by Teflon tubings, and the peristaltic pumps (Gilson, Middleton, WI, USA) were switched on to start the inflow of input medium and the outflow of culture. The imposed flow rate was 0.2 day−1, which was chosen based on the in situ growth rates determined during the first KEOPS cruise (Obernosterer et al. 2008), and resulted in a residence time of 5.6 days. Sampling of cultures The experiment was run for 19 days, during which the cultures were sampled every two to three days. The air purge was renewed every day. Input media were prepared every other day, after another 0.2 μm filtration of the required volume of phytoplankton exudate. Freshly prepared media were sampled for DOC concentration and inorganic nutrients. Sampling was performed without opening the cultures using glass syringes (Hamilton, Reno, NV) connected with a luer-lock to Teflon tubes and glass cubes that plunged to the bottom of the cultures. The glass syringes were extensively rinsed with Milli-Q water prior to sampling. A maximum volume of 50 ml, including syringe rinsing, was taken from the cultures for the various analyses, representing less than 3% of the volume of the cultures. DOC and inorganic nutrients For the determination of DOC concentrations, duplicate 15 ml samples were filtered through two precombusted GF/F filters (Whatman) and the samples were stored acidified (H3PO4, pH = 2) in combusted 20 ml glass ampoules in the dark at room temperature. DOC concentrations were measured in the home lab on a Shimadzu TOC-VCP analyzer with a Pt catalyst at 680°C (Benner and Strom 1993) as detailed in Tremblay et al. (2015). For the determination of nitrate (NO3−) and phosphate (PO43−) concentrations, a 10 ml subsample was fixed with mercuric chloride (HgCl2, 20 mg l−1 final concentration) and analyzed onboard using continuous flow analysis (Aminot and Kérouel 2007) with a Skalar instrument. Bacterial abundance and diversity A 1.8 ml sample was taken for bacterial abundance analysis, fixed with formaldehyde (2% final concentration), kept at 4°C for 30 min, and then stored at -80°C until analysis. Bacterial abundances were determined in the lab on a FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA) equipped with an air-cooled laser, providing 15 mW at 488 nm with the standard filter setup. Heterotrophic bacteria were stained with SYBR Green I and determined by flow cytometry as described previously (Obernosterer et al. 2008). Samples for bacterial diversity were collected in acid-washed, Milli-Q water-rinsed PC flasks from the outflow of the cultures at day 5, 10, 15 and 19. The tubes were placed in the flasks at night and between 135 and 216 ml of collected culture were filtered the next morning on 0.2 μm PC filters (Nuclepore), placed in tubes containing 500 μl of lysis buffer (40 mM EDTA, 50 mM Tris, 0.75 M sucrose) and stored at -80°C until the end of the cruise. The last day of the experiment, the volumes filtered for bacterial diversity were ∼500 ml for each culture. DNA extraction and pyrosequencing After lysis of the cells and proteins as described in Landa et al. (2013), a combined DNA-RNA extraction using the AllPrep DNA/RNA Micro Kit (Qiagen, Hilden, Germany) was performed on the 0.2 μm filters, following the manufacturer's instructions. The molecular size and purity of the DNA were analyzed using agarose gel electrophoresis (1%). In total, 25 samples of genomic DNA were sent to the Molecular Research DNA laboratory (Shallowater, TX, USA) for 454 pyrosequencing. The sequencing was performed using an FLX-Titanium platform (Roche, Wolcott et al. 2009). Universal bacterial primers 27F (5΄-AGRGTTTGATCMTGGCTCAG-3΄) and 530R (5΄-CCGCNGCNGCTGGCAC-3΄) were used, targeting hypervariable regions V1 to V3 of the 16S rRNA gene. The raw sequences for this dataset were deposited in the Sequence Read Archive database under accession number SRP041580. Sequences can be found on HUFUDAT01.sff, and the barcode corresponding to each sample can be found in Table S2 (Supporting Information). Bioinformatic analyses Raw sequences were denoised using AmpliconNoise (Quince et al. 2011) in Mothur (Schloss et al. 2009). Chimeras were removed during clustering of reads into OTUs at 97% identity with Usearch (Edgar 2010) in QIIME (Caporaso et al. 2010). The most abundant sequence in each OTU was chosen as the representative sequence and used for further analyses. Taxonomy was assigned to each OTU using the RDP classifier (Wang et al. 2007) and the May 2013 Greengenes database as a reference. A matrix was constructed with the number of reads corresponding to each OTU in each sample. Relative abundances were calculated for each OTU, after subsampling of the data to 2930 reads per sample. Phylogenetic analysis Phylogenetic analysis was used to clarify the affiliation of 21 OTUs identified as abundant . The partial 16S gene sequences were aligned using MUSCLE v3.8.31 (Edgar 2004). The phylogenetic inference was restricted to 242 sites that could be unambiguously aligned for the Proteobacteria tree, and 258 sites for the Flavobacteria and Verrucomicrobia tree. Phylogenetic analyses were performed using the Maximum Likelihood method implemented in MEGA7 (Kumar, Stecher and Tamura 2016). The Maximum Likelihood analysis was based on the General Time Reversible model, and initial trees for the heuristic search were obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood approach. Bootstrap proportions (BPs) were obtained from 1000 pseudo-replicates. Bayesian Posterior Probabilities (BPPs) were calculated with the Maximum Likelihood method from 1000 replicates for 1 million generations, until convergence (average standard deviation below 0.01), using MrBayes 3.2.6. Modeling approach To obtain estimates of the growth rates of the abundant OTUs, we used the classical mathematical model of the chemostat (Monod 1950; Novick and Szilard 1950). We consider a multi-species population, represented here by the 21 most abundant OTUs (Dumont et al. 2008). The model assumes that these OTUs are in competition on a single limiting substrate. The application of this basic hypothesis to our data requires some considerations. The total pool of DOM is a complex mixture of heterogeneous substrates, each characterized by variable biological availability. In our experiment, only a small fraction of this DOM pool, representing the most labile fraction, is consumed during the residence time period determined by the imposed dilution rate. Because the composition and the concentration of the specific compounds used by bacteria, as well as the affinities and uptake kinetics by different taxa are not known, the ensemble of these labile compounds was presented in the model as one limiting resource. It is well known that this model, which has no explicit intra- or inter-specific interaction terms between the populations of microorganisms, predicts competitive exclusionin the long term (i.e. at steady state). Here, we consider its non-equilibrium dynamics where a complex microbial consortium is present in the chemostat (Rapaport, Dochain and Harmand 2009). The mathematical analysis of this model predicts that the species exclusion is an asymptotic result. It can take a long time before being observed, notably when species have similar growth characteristics. Therefore, we consider here the transient dynamics of this model and not the steady state,   \begin{equation*} \begin{array}{@{}*{1}{c}@{}} {\ \frac{{ds}}{{dt}} = \ D\left( {{s_{\rm in}} - s} \right) - \mathop \sum \limits_i^n \frac{{{\mu _i}\left( s \right)}}{{{Y_i}}}{x_i}}\\ {\ \frac{{{\rm{d}}{x_i}}}{{dt}} = \ {\mu _i}\left( s \right){x_i} - D{x_i}\ \left( {1 \cdots n} \right)}. \end{array} \end{equation*} (1) Here $$s$ $ and $${x_i}$ $ stand for the substrate (that is the labile pool) and biomass concentrations of each OTU or ‘species’ $$i$ $ in the cultures, $${s_{\rm in}}$ $ is the concentration of labile organic carbon in the input medium, and $$D$ $ the dilution rate. Our objective is to estimate from experimental data the specific growth rates $${\mu _i}( . ):s \to $ $$${\mu _i}( s )$ $for each OTU. The available data from the experiments are the operating parameters $$D$ $ and$$\ {s_{\rm in}}$ $, and the measured values $$s( {{t_j}} )$ $ and $${x_i}( {{t_j}} )$ $ at four measurement instants $${t_j}$ $, for each replicate. As these data reveal that the four values of $$s( {{t_j}} )$ $ are quite close, we have used two different estimation procedures, which reinforce the robustness of the estimations. The first method allows the concentration of the consumed substrate to vary over time, while the second method assumes explicitly the concentration of the consumed substrate to be constant over time. Method 1 We identify the function $${\mu _i}( {\rm{s}} )$ $ from the model (M1) allowing $$s{\rm{\ }}$ $to be time-varying but close to the observed value. We also assume that the bacterial growth efficiency, referred to as yield coefficient $${Y_i}{\rm{\ }}$ $of each OTU is close to a common value Y. One can then observe from equations (M1) that under this assumption   \begin{equation*}\frac{d}{{dt}}\ \left( {Ys + b} \right) = {\rm{\ }}D\left( {Y{s_{\rm in}} - Ys - b} \right),\end{equation*} where $$b\ = \sum {x_i}{\rm{\ \ }}$ $is the total biomass. This equation predicts that the quantity $$Ys + b$ $ changes exponentially towards $${s_{\rm in}}$ $ with a decay rate $$D$ $. As the total biomass revealed only minor variation over time, we have considered the approximation b$$ = {\rm{\ }}Y( {{s_{\rm in}} - s} )$ $, allowing us to obtain an estimation of $$Y$ $ close to those found in the literature (in the present case, Y = 0.42). We have made the generally applied assumption that the specific growth function $${\mu _i}( {\rm{s}} )$ $ follows a Monod kinetics:   \begin{equation*}{\mu _i}\ \left( x \right) = \frac{{{\mu _{{\rm max},i}}\ s}}{{{K_{s,i}} + s}}.\ \end{equation*} Nevertheless, from the magnitude of the values of the parameters $${\mu _{{\rm max}}}$ $ and $${K_s}{\rm{\ }}$ $found in the literature and the measured values $$s( {{t_j}} )$ $, we expect it to be in the linear part of the Monod function. Consequently, we have identified the slopes $$\ \ \mu {'_i} = \ {\mu _{{\rm max},\ i}}/{K_{s,\ i}}$ $ of the functions $${\mu _i}( {\rm{s}} )$ $ at zero instead of two distinct parameters. Method 2 We assume $${\rm{s}}$ $ to be constantly equal to $$\bar{s}$ $ and we estimate parameters $$\overline {{\mu _i}} $ $ as $${\mu _i}( {\bar{s}} )$ $ and not the functions $${\mu _i}( {\rm{s}} )$ $. This leads to a simplification of the model (M1) where the dynamic of s is ignored:   \begin{equation*}\ \frac{{d{x_i}}}{{dt}} = {\rm{\ }}{\mu _i}{x_i} - D{x_i}\ \left( {1 \cdots n} \right).\end{equation*} As observed from the data that the total biomass does not vary much over time (about 10%), we have considered that the OTU proportions $${p_i} = {x_i}/b\ $ $ represent more reliable information than their abundances $${x_i}$ $. Then, under the assumption that the substrate concentration is constant at $$\bar{s}$ $, one can give their explicit expressions as functions of time (Rapaport, Dochain and Harmand 2009):   \begin{equation*}{p_i}\ \left( t \right) = {p_i}{\rm{\ }}\left( {{t_0}} \right)\frac{{{e^{{\mu _i}t}}}}{{\mathop \sum \nolimits_j {p_j}\left( t \right){e^{{\mu _j}t}}}}\end{equation*} (2)under the constraint of the quasi-stationarity of the total biomass, that is   \begin{equation*}\mathop \sum \limits_j {p_j}\left( t \right)\ {\mu _j} = \ D.\end{equation*} Note that this method allows an estimation of the value $${{\rm{\mu }}_{\rm{i}}}$ $ without requiring the knowledge of $$Y$ $, s or $${s_{\rm in}}$ $. Because the consumption of organic carbon in the cultures could not be quantified, we estimated the bacterial carbon demand, based on the bacterial production (BP = µx bacterial biomass) and a bacterial growth efficiency of 10%. We assumed the µ to be equal to the dilution rate (0.2 d−1) and we used the conversion factor of 12.4 fg C cell−1 (Fukuda et al. 1998) to determine bacterial biomass from abundance. This estimate revealed a DOC consumption of 1 µmol and 3 µmol organic carbon in the control and the diatom-DOM treatments, respectively. Both methods have used the least-square criterion for the estimation of the parameters. For each replicate and the average of replicates we have chosen estimations that provide the best R2. One can then compare the values $${\mu '_i}.s( {{t_j}} )$ $ estimated by model (M1) with the values $${\mu _i}$ $ estimated by model (M2). Comparison of the biomass of the individual OTUs at four time points based on observations and model simulations validated M1 (Fig. S2 and Table S3, Supporting Information). RESULTS Dissolved organic matter and inorganic nutrients DOC concentrations in the background seawater and the diatom-derived material used to prepare each culture medium were similar, resulting in equivalent DOC concentrations before and after the mixing. Measurements of DOC performed in each of the freshly prepared media over the course of the experiment were 49 ± 1 µM for the winter water DOM (n = 7), thereafter referred to as control, and 49 ± 2 µM for the diatom-DOM condition (n = 7). In the cultures, concentrations of DOC were slightly higher than in the respective media, in particular during the first few days of the experiment (range 50–56 µM) (Fig. 1). From day 7 (in the diatom-DOM treatment) and day 10 (in the control) to the end of the experiment, DOC concentrations remained relatively stable and they were significantly higher in the control (52 ± 1, n = 5) than in the diatom-DOM condition (50 ± 1, n = 6) (Fig. 1, Student test, P ≤ 0.05). A possibly small DOC contamination during the setup of the cultures cannot be excluded. However, given the excellent reproducibility of our observations between the biological triplicates (i.e. DOC concentrations, bacterial abundance and community composition) after several complete renewals of the cultures, it is unlikely that it substantially affected the results of our experiment. Concentrations of nitrate and phosphate were identical (Student test, P > 0.05) in the input medium (39 ± 2 and 41 ± 1 µM NO3− and 2 ± 0.3 and 2 ± 1 µM PO43− (n = 8) for the control and the diatom-DOM treatment, respectively) and the cultures (40 ± 2 and 42 ± 1 µM NO3− and 2 ± 0.2 and 2 ± 0.1 µM PO43− (n = 9) for the control and the diatom-DOM treatment, respectively). This indicates that bacterial consumption of these nutrients in the cultures was lower than the precision of the analytical method. C:N and C:P ratios were similar in the two treatments (1.3:1 and 26:1, respectively), and below the C:N and C:P ratios of heterotrophic bacteria (5:1 and 45:1) (Goldman, Caron and Dennett 1987), indicating that heterotrophic bacteria were C-limited in these incubations. Figure 1. View largeDownload slide Abundance of heterotrophic bacteria (circles) and DOC concentration (triangles) in the control (white) and the diatom-DOM (black) cultures over time. Arrows indicate time points sampled for bacterial diversity analyses. Mean values ± SD of biological triplicates are shown for each time point. * and ** indicate significant (P < 0.05) and highly significant (P < 0.005) differences between conditions, according to a Student test. Figure 1. View largeDownload slide Abundance of heterotrophic bacteria (circles) and DOC concentration (triangles) in the control (white) and the diatom-DOM (black) cultures over time. Arrows indicate time points sampled for bacterial diversity analyses. Mean values ± SD of biological triplicates are shown for each time point. * and ** indicate significant (P < 0.05) and highly significant (P < 0.005) differences between conditions, according to a Student test. Heterotrophic bacterial abundance In the first days of the experiment, bacterial abundances increased in all cultures. They stabilized after 5 days in the diatom-DOM treatment, and after 10 days in the control treatment (Fig. 1). Bacterial abundances were significantly higher in the diatom-DOM treatment (1.3 ± 0.1 × 106 cells ml−1, n = 7) than in the control (8.7 ± 0.3 × 105 cells ml−1, n = 5) (Student test, P < 0.0001). A 97% identity clustering of the reads obtained by 454 pyrosequencing of the V1 to V3 regions of the 16S gene yielded a total number of 544 OTUs for the 25 samples considered, after the denoising, quality check and chimera removal steps. One sample (diatom-DOM replicate 2, after 15 days) only contained 476 sequences and was excluded from further analysis. The 24 remaining samples had 2930 to 11 400 sequences, with an average of 5913 ± 2000 sequences per sample. To enable comparison, the dataset was subsampled to 2930 sequences per sample. A clustering of the 24 samples obtained from the two experimental conditions after 0, 5, 10, 15 and 19 days showed a clear separation between treatments, regardless of time (Fig. 2). A bootstrap analysis of the clustering indicated highly reproducible branching patterns. The communities growing in the control cultures were more similar to the inoculum than the communities growing in the diatom-DOM cultures. The control cultures showed little change with time, with an overall difference between all control samples smaller than 8%. By contrast, the communities growing in the diatom-DOM cultures exhibited a larger range of change over time. At the end of the experiment, two of the triplicate cultures were 22% different to the other diatom DOM cultures (Fig. 2). Figure 2. View largeDownload slide UPGMA clustering of bacterial communities in the control and the diatom-DOM cultures after 5, 10, 15 and 19 days, based on a weighted UNIFRAC similarity matrix. The biological triplicates are annotated 1, 2 and 3 for each treatment. The scale indicates % dissimilarity. Bootstrap values are indicated on each node. Figure 2. View largeDownload slide UPGMA clustering of bacterial communities in the control and the diatom-DOM cultures after 5, 10, 15 and 19 days, based on a weighted UNIFRAC similarity matrix. The biological triplicates are annotated 1, 2 and 3 for each treatment. The scale indicates % dissimilarity. Bootstrap values are indicated on each node. Diverse bacterial taxa belonging mostly to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes were observed in both experimental conditions (Fig. 3). Interestingly, distinct, sometimes closely related taxa within these broad groups showed strong association to either condition. For instance, the Roseobacter clade subgroups NAC11–7 and Sulfitobacter were comparatively abundant in the control and in the diatom-DOM cultures, respectively. In the order of Alteromonadales, the genera Marinobacter and Shewanella were found abundant in the control, whereas Colwellia was highly abundant in the diatom-DOM cultures. Most Bacteroidetes sequences belonged to the Polaribacter genus, which had high relative abundances in the control (50 ± 10% of total communities, n = 12) compared to the diatom-DOM cultures (4 ± 2% of total communities, n = 11). By comparison, the Ulvibacter genus was not abundant in the first three time points in either condition, and reached an average of 7 ± 2% of the total communities in both experimental treatments after 19 days (n = 3). The SAR11 clade, making up 51% of the inoculum, persisted in both treatments over the entire course of the experiment, with higher relative abundances in the control cultures (24 ± 7% of the total communities, n = 12) than in the diatom-DOM cultures (9 ± 4% of the total communities, n = 11). In the diatom-DOM cultures, the final time point was characterized by a striking increase in the relative abundance of the Lentimonas group in two out of three replicate cultures, making up to 51% of the total community in replicate 3 (Fig. 3). Overall, the communities developing in the two conditions were widely different. Figure 3. View largeDownload slide Bacterial community composition in the control cultures (A) and the diatom-DOM cultures (B), in % of total sequences. The natural community sampled in the winter waters on the plateau east of the Kerguelen Islands and used as an inoculum for these cultures is shown on the left in both graphs. Bacterial groups belonging to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes are shown in blue, green and red shades, respectively. Figure 3. View largeDownload slide Bacterial community composition in the control cultures (A) and the diatom-DOM cultures (B), in % of total sequences. The natural community sampled in the winter waters on the plateau east of the Kerguelen Islands and used as an inoculum for these cultures is shown on the left in both graphs. Bacterial groups belonging to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes are shown in blue, green and red shades, respectively. To determine how many and which OTUs were driving the differences observed in the composition of these communities, we examined the distribution patterns of abundant OTUs (relative abundance ≥ 1% in at least one time point in all the replicates of a treatment). We found 21 abundant OTUs (Fig. 4), together comprising 64% to 97% (average of 89% ± 8%, n = 23) of the total communities. Out of these 21 abundant OTUs, 16 had significantly different relative abundances between conditions in at least one time point (Fig. 5), including 8 for which the differences were significant throughout the entire experiment (Student test, P ≤ 0.05, Fig. 5A). Two Polaribacter (OTUs 1 and 6) and one NAC11–7 (OTU 13) OTUs had significantly higher relative abundances in the control compared to the diatom-DOM cultures at all time points. Conversely, one Sulfitobacter (OTU 1348), two Methylophaga (OTUs 44 and 202) and two Colwellia (OTUs 8 and 24) OTUs had significantly higher relative abundances in the diatom-DOM cultures at all time points (Fig. 5A). Taken together, these 8 OTUs, constituting the core differences between the two conditions, represented 22% to 89% of the total communities in the different samples, with an average of 56% ± 10% of the total communities in the control cultures (n = 12) and 60% ± 23% of the total communities in the diatom-DOM cultures (n = 11). Figure 4. View largeDownload slide Unrooted maximum likelihood distance tree based on 16S gene sequences clarifying affiliation of (A) 17 OTUs belonging to the group of Proteobacteria and (B) 4 OTUs belonging to the groups Flavobacteria and Verrucomicrobia. The sequences obtained in this study are shown in bold. The numbers at the node indicate BPs and BPPs respectively, given by ML and Bayesian analyses with 1000 replicates. The asterisks designate nodes with BPs or BPPs below 50% or 0.5. Figure 4. View largeDownload slide Unrooted maximum likelihood distance tree based on 16S gene sequences clarifying affiliation of (A) 17 OTUs belonging to the group of Proteobacteria and (B) 4 OTUs belonging to the groups Flavobacteria and Verrucomicrobia. The sequences obtained in this study are shown in bold. The numbers at the node indicate BPs and BPPs respectively, given by ML and Bayesian analyses with 1000 replicates. The asterisks designate nodes with BPs or BPPs below 50% or 0.5. Figure 5. View largeDownload slide Relative abundances of abundant OTUs with significant differences between control and diatom-DOM cultures at all time points (A). Relative abundances of abundant OTUs with significant differences between control and diatom-DOM amended cultures at certain time points, or without significant differences (B). Relative abundances in the control cultures are shown in blue on the right side of the graphs, relative abundances in the diatom-DOM cultures are shown in green on the left side of the graphs. Light to dark blue and green represent successive, early (day 5) to late (day 19) time points. * represent statistical differences (P ≤ 0.05) between the two conditions according to a Student test, except for graph A where all the OTUs have significantly different relative abundances at all time points. Figure 5. View largeDownload slide Relative abundances of abundant OTUs with significant differences between control and diatom-DOM cultures at all time points (A). Relative abundances of abundant OTUs with significant differences between control and diatom-DOM amended cultures at certain time points, or without significant differences (B). Relative abundances in the control cultures are shown in blue on the right side of the graphs, relative abundances in the diatom-DOM cultures are shown in green on the left side of the graphs. Light to dark blue and green represent successive, early (day 5) to late (day 19) time points. * represent statistical differences (P ≤ 0.05) between the two conditions according to a Student test, except for graph A where all the OTUs have significantly different relative abundances at all time points. Three other abundant OTUs, a SAR11 (OTU 1582) and two Sulfitobacter (OTUs 2 and 31) OTUs had significantly different relative abundances between treatments, but in two or three time points only (Fig. 5B). The SAR11 OTU had higher relative abundance in the control cultures, but it was significant only for 3 out of 4 time points. The Sulfitobacter OTUs had higher relative abundances in the diatom-DOM cultures, and the difference with the control was significant after 5 and 19 days for both OTUs, and also after 10 days for OTU 2 (Fig. 5B). Interestingly, the four OTUs associated with the control were also detected in the diatom-DOM cultures, but the seven OTUs with higher relative abundances in the diatom-DOM cultures were not detected in the control cultures, with some exceptions ( ≤ 0.1% of the total community in the control). Two abundant OTUs, a Marinobacter (OTU 41) and a Lentimonas (OTU 15) had higher relative abundances in the control and in the diatom-DOM amended cultures, respectively (Fig. 5B). Because of large variability among replicates, these differences were not significant according to a Student test. One Alphaproteobacteria, 2 SAR11 and 2 SAR86 OTUs only had significantly higher relative abundances in the control after 5 days (Fig. S3, Supporting Information). Finally, 3 abundant OTUs belonging to the betaproteobacterial clade OM43 (OTU 1288), the gammaproteobacterial family of Oceanospirillaceae (OTU 51) and the Bacteroidetes genus Ulvibacter (OTU 17) had similar relative abundances in the two treatments (Fig. 5B, P ≥ 0.06). Modeled growth rates of abundant OTUs Because of the small number of cells inoculated into the cultures, bacterial growth was not limited by organic carbon during the first days of the experiment. After the first 5 days, the supply of organic carbon and the growth rate were constrained by the dilution rate. Using the classical chemostat model (Monod 1950; Novick and Szilard 1950) allowed us to simulate growth rates for abundant OTUs in mixed communities under carbon-limited growth conditions. For a robust estimation, we used two methods of the model that are both based on the changes in biomasses of the abundant OTUs between 5 and 19 days, representing a transient phase towards equilibrium. The first method (method 1) allows the concentration of the consumed substrate to vary over time, while the second method (method 2) assumes explicitly the concentration of the consumed substrate to be constant over time. This second approach represents a simplification of the first method, because knowledge of the concentration of the consumed substrate is not required. The aim of considering these two methods is to verify a posteriori the hypothesis that the communities have reached a quasi-constant level of substrate consumption. The estimated growth rates for the 21 OTUs ranged between 0.12 and 0.49 d−1 for method 1 and between 0.14 and 0.33 d−1 for method 2 (Fig. 6 and Table S4, Supporting Information). Even though the two methods revealed variable absolute values for some OTUs, the ranking of OTUs from highest to lowest growth rates exhibited coherent patterns. Because all 21 OTUs were abundant in the diatom-DOM treatment, we illustrate the modeled growth rates for this treatment (Fig. 6). Our model results suggest three groups of OTUs characterized by growth rates that are high, intermediate and low. Highest growth rates (0.28–0.49 d−1) were attributed by both methods to OTUs belonging to Lentimonas, Ulvibacter, Oceanospirillaceae and the OM43 clade. Intermediate growth rates (0.21–0.31 d−1) were attributed to the NAC11–7, 2 Polaribacter, 1 Sulfitobacter, 3 SAR11 and 1 SAR86 OTUs. Lowest growth rates (0.12–0.18 d−1) were determined for OTUs belonging to Methylophaga, Colwellia and Marinobacter. Interestingly, the two methods suggested overall similar growth rates for a given OTU under different DOM resource regimes (Table S4, Supporting Information), except for Marinobacter that had higher growth rates in the control as compared to the DOM treatment. Figure 6. View largeDownload slide Growth rates of the abundant OTUs in the diatom-DOM treatment obtained by two methods of the chemostat model. OTUs are grouped according to high, intermediate and low growth rates as described in the text. Respective growth rates of the control are given in Table S1, Supporting Information. The relative abundances of the OTUs over the course of the experiment are presented in Figs. 5 and S1. OTU 1 –Polaribacter; OTU 2-Sulfitobacter; OTU 6-Polaribacter; OTU 8 –Colwellia; OTU 9-SAR86; OTU 13 –NAC11–7; OTU 15-Lentimonas; OTU 17-Ulvibacter; OTU 24-Colwellia; OTU 29-Alphaproteobacteria; OTU 41-Marinobacter; OTU 44 –Methylophaga; OTU 51-Oceanospirillaceae; OTU 202-Methylophaga; OTU 1457-SAR11; OTU 1283-SAR11; OTU 1288-OM43; OTU 1348 –Sulfitobacter; OTU 1582 – SAR11; OTU 1867-SAR86; OTU 31-Sulfitobacter. Figure 6. View largeDownload slide Growth rates of the abundant OTUs in the diatom-DOM treatment obtained by two methods of the chemostat model. OTUs are grouped according to high, intermediate and low growth rates as described in the text. Respective growth rates of the control are given in Table S1, Supporting Information. The relative abundances of the OTUs over the course of the experiment are presented in Figs. 5 and S1. OTU 1 –Polaribacter; OTU 2-Sulfitobacter; OTU 6-Polaribacter; OTU 8 –Colwellia; OTU 9-SAR86; OTU 13 –NAC11–7; OTU 15-Lentimonas; OTU 17-Ulvibacter; OTU 24-Colwellia; OTU 29-Alphaproteobacteria; OTU 41-Marinobacter; OTU 44 –Methylophaga; OTU 51-Oceanospirillaceae; OTU 202-Methylophaga; OTU 1457-SAR11; OTU 1283-SAR11; OTU 1288-OM43; OTU 1348 –Sulfitobacter; OTU 1582 – SAR11; OTU 1867-SAR86; OTU 31-Sulfitobacter. DISCUSSION We observed major differences in the identity and the abundance of the bacterial taxa selected by continuous cultures supplied with DOM from Southern Ocean winter water alone or amended with diatom-derived organic matter. Concentrations of DOC were identical in both conditions, but DOC bioavailability was higher in the diatom-DOM condition, as shown by increased bacterial abundances and DOC consumption, revealing that differences in origin and chemical composition of the organic matter sources were the main drivers of the observed patterns in bacterial community composition. The response to each of our experimental conditions was phylogenetically diverse, as it involved OTUs from various marine groups commonly observed in polar regions, including the Southern Ocean (Bano and Hollibaugh 2002; Gentile et al. 2006; West et al. 2008; Ghiglione and Murray 2012; Thiele et al. 2012; Wilkins et al. 2013; Luria et al. 2016). Differences were already detectable after 5 days, suggesting the mechanisms that allowed the successful groups to develop in each condition were fast. Our results indicate that in the carbon-limited Southern Ocean, changes in the DOM bioavailability such as induced by an input of diatom-derived organic matter can act as powerful selective constraints on bacterial communities through the rapid growth of most adapted members. Ecology and functional traits of taxa selected by winter water conditions A commontrait for taxa reaching high abundances in the cultures supplied with winter water alone is the ability to grow on DOM of overall low biological availability. The most abundant OTUs in the cultures supplied with winter water belonged to the Polaribacter genus in the class of Flavobacteria, and to the SAR11 and NAC11–7 alphaproteobacterial clades. Flavobacteria are known to be efficient degraders of complex polymers, such as polysaccharides, proteins or chitin, and tend to be less active than other bacterial groups in taking up simple sugars or amino acids (Cottrell and Kirchman 2000; González et al. 2008; Straza et al. 2010; Simon et al. 2012; Wietz et al. 2015; Fourquez et al. 2016). These characteristics are supported by proteomic and genomic studies showing that Flavobacteria possess numerous genes involved in hydrolytic activities as well as high molecular weight compounds transporters such as TonB-dependent receptors, and comparatively low numbers of genes involved in the uptake of low molecular weight compounds (González et al. 2008; Gómez‐Pereira et al. 2012; Fernández-Gómez et al. 2013; Williams et al. 2013). Such metabolic traits could have given Polaribacter OTUs an advantage in the cultures maintained with organic matter from winter water. These Polaribacter OTUs were also detected in the diatom-DOM cultures, however, in much lower abundances. While Flavobacteria have been described as associated with phytoplankton, many studies examining bacterial community succession over the course of natural or experimental phytoplankton blooms show different timing of the response from Bacteroidetes groups compared to taxa belonging to Alpha- and Gammaproteobacteria (Tada et al. 2012; Teeling et al. 2012; Luria et al. 2016). Our investigation of bacterial community composition during a five-month diatom bloom at the study site showed a late peak in Polaribacter abundances, associated with bloom decay (Landa et al. 2016). These results point to weak competitiveness on fresh, labile phytoplankton material and a preference for more complex, detrital algal products for these taxa. Polaribacter were abundant in mesoscosms performed with DOM from coastal Antarctic waters under pre-bloom conditions (Luria et al. 2017). In that study, Polaribacter dominated not only in unamended mesocosms, but also in mesocosms amended with DOM from the diatom Thalassiosira weissflogii. However, their final abundances in these mesocosms seemed to depend on their abundances in the bacterial communities used as inoculum, and were consistently lower when Colwellia OTUs became abundant (Luria et al. 2017). In our study, low initial abundances of Polaribacter coupled to higher competitiveness of other groups, such as Colwellia, on the fresh diatom DOMcollected during exponential growthare possible factors contributing to Polaribacter OTUs being outcompeted in this experimental condition. SAR11 are abundant in oligotrophic waters across the world (Morris et al. 2002; Giovannoni 2017). Recent genomic and proteomic information obtained from both cultivation-independent methods and cultivated representatives of this diverse group confirm genetic and physiological adaptation to life in nutrient-poor waters for most members of the group (Sowell et al. 2009; Thompson et al. 2013). SAR11 are characterized by unusual nutrient utilization profiles, including carbon (Carini et al. 2013; Tripp 2013). Single cell approaches have shown that SAR11 efficiently take up labile, low molecular weight compounds such as glucose and free amino acids, preferentially over high molecular weight compounds such as proteins (Malmstrom et al. 2004, 2005; Alonso and Pernthaler 2006), a pattern that was also observed in our study region (Fourquez et al. 2016). The high abundance of SAR11 in the presence of winter-water DOM is likely due to their efficient uptake of labile compounds present in low concentration (Tremblay et al. 2015) or of substrates derived from organic matter degradation performed by other organisms (Morris, Papoulis and Lenski 2014). Evidence for this type of cross-feeding involving SAR11 has been reported in the Atlantic Ocean (Wietz et al. 2015) and in the Southern Ocean (Williams et al. 2013). In this latter study, a metaproteomics approach revealed that SAR11 expressed transporters for molecules released by other microorganisms, in particular Flavobacteria. The Roseobacter clade is abundant across the world ocean and is characterized by high phylogenetic, functional and ecological diversity (Selje, Simon and Brinkhoff 2004; Brinkhoff, Giebel and Simon 2008; Newton et al. 2010; Buchan et al. 2014). The NAC11–7 cluster is detected during phytoplankton bloom events (González et al. 2000), but was also reported in other environmental conditions (Buchan, González and Moran 2005). Recent genomic evidence based on the one cultivated representative of this cluster (HTCC2255) suggests a more streamlined genome and adaptation to more oligotrophic conditions than what is commonly observed for most Roseobacter clade members (Luo et al. 2013). Such adaptation could explain the success of one NAC11–7 OTU in the cultures provided with non-amended winter water. Ecology and functional traits of taxa selected by diatom-DOM Three other Roseobacters were abundant in the diatom-DOM cultures, which is in agreement with previous work showing the opportunistic lifestyle of many members of this clade and their frequently observed associations with phytoplankton. A substantial fraction of their genomic content allows high versatility in carbon utilization compared to other abundant marine taxa, a characteristic that facilitates this ecological strategy (Buchan, González and Moran 2005; Newton et al. 2010). All three Roseobacter OTUs belonged to Sulfitobacter, a group known for their capacity to use organic sulfur compounds, such as dimethylsulfoniopropionate (Moran et al. 2007; Ankrah et al. 2014), a common osmolyte produced by phytoplankton, including the diatom species used in this study (Speeckaert et al. 2018). Roseobacters have been shown to use a variety of other sulfur compounds produced by phytoplankton (Durham et al. 2015). Sulfur compounds, among other modes of nutrition, could have contributed to maintain the high abundances of the three Sulfitobacter OTUs in the diatom-DOM amended cultures. Four OTUs belonging to the gammaproteobacterial genera Colwellia and Methylophaga were also abundant in the diatom-DOM cultures. Members of the genus Colwellia are commonly found in polar environments and are well adapted to life in cold environments through a number of extracellular enzymes that are most active at low temperatures (Methé et al. 2005). High Colwellia abundances were observed in mesocosms performed in the Western Antarctic Peninsula during winter (Luria et al. 2017) and in environments experiencing important pulses in organic matter, such as oil spills (Redmond and Valentine 2012; Kleindienst et al. 2015). In an experimental study, Colwellia became a dominant taxon at an intermediate stage of hydrocarbon degradation (Hu et al. 2017). The increase in abundance of this group was concurrent with an increased number of aromatic degradation genes as shown by metagenomic analyses and the decrease in the concentration of one—to three—ring aromatic compounds (Hu et al. 2017). These findings and the results from the present study point to a potentially widespread ability among members of the group to quickly use inputs of organic carbon. In particular, the capacity of Colwellia to degrade small aromatic compounds could have provided an ecological advantage to this group in the DOM-amended treatment. Several Methylophaga strains have been suggested to associate with marine algae (Li et al. 2007) and members of this group are known to utilize C1 compounds (Methé et al. 2005; Neufeld et al. 2007), a metabolic trait that could favor their growth in the presence of phytoplankton DOM. A recent transcriptomic study conducted in the McMurdo Sound showed utilization by Methylophaga of phytoplankton-derived C1 compounds such as methanol, methylamine and dimethylsufide (Bertrand et al. 2015). C1 compounds are common end-products of microbial metabolism and could have been present in our cultures, either directly in the diatom-derived DOM, or as a result of bacterial processing of diatom-DOM, providing a functional niche for these four OTUs (Krause et al. 2017). Ecological implications for DOM processing Our results from an experimental system point to interesting roles of niche occupation as well as interactions among members of a microbial community in DOM decomposition. Genomic data predict that Colwellia and Polaribacter share the ability to cleave polymeric substances. In our cultures, the presence of Polaribacter OTUs in both treatments suggests they were using polymers present in winter water, while Colwellia OTUs present in the diatom-DOM condition were almost undetectable in the non-amended winter water, indicating a preference for phytoplankton-derived polymers. Specialized functional niches seem to exist at a fine phylogenetic level, as suggested by the co-existence in a given treatment of closely related OTUs, such as the two OTUs in the Polaribacter, Colwellia and Methylophaga groups, or by the preferential association of the various Roseobacter OTUs within a specific treatment. These observations are consistent with recent functional, genomic and transcriptional studies demonstrating the diversity of metabolisms existing among members of the same genus or clade (Kimes et al. 2014; Neumann et al. 2015; Xing et al. 2015). The distributions of similar or complementary traits in co-occurring bacterial taxa further suggest that interactions among microbes such as competition and cooperation are important for the degradation of organic matter, and play a role in community structure. Decreasing relative abundances over time for some OTUs in the diatom-DOM cultures, as illustrated in the low growth rates obtained by the model approach, could reflect competition for the freshly supplied phytoplankton-derived material and replacement over time of initially successful members by functionally similar taxa. As discussed above, C1 compounds produced by organic matter degradation performed by some bacterial groups such as Roseobacter and Flavobacteria could have benefited other members in our cultures, such as the Colwellia, Methylophaga, OM43 and SAR11 OTUs who all share this metabolic trait (Tripp et al. 2008; Sun et al. 2011; Carini et al. 2013; Gifford et al. 2016). These interactive processes have been demonstrated in other environmental contexts (Datta et al. 2016; Krause et al. 2017). These observations highlight that interactions among taxa are driven by both the chemical nature of the available substrates and by the traits present in the microbial members, ultimately impacting community structure and organic matter decomposition. Growth rates of marine bacterial taxa We used the classical chemostat model (Monod 1950; Novick and Szilard 1950) to simulate the growth rates for abundant OTUs in a mixed community and under carbon-limited conditions. From the modeling point of view, the equilibrium in the chemostat model corresponds to the exclusion of all taxa except for the most competitive ones. However, when taxa have similar growth rates their dynamics over time in the chemostat can be slow even at stationary substrate levels (Rapaport, Dochain and Harmand 2009). As a diverse microbial community was present over the course of our experiment, we consider non-equilibrium dynamics. In this transient stage, the growth rates are not necessarily reflected in the relative abundance of taxa. Even though our model estimates are in the range of previously determined values in the oligotrophic ocean for bacterial groups at broad phylogenetic levels (see review by Kirchman 2016), we discuss our findings in relative rather than absolute terms. Interestingly, phylogenetically diverse taxa had similar growth rates in our experimental conditions. According to the model simulations, the SAR11 OTUs observed in this experimental system grew as fast as OTUs belonging to Polaribacter and Roseobacter. OTUs belonging to Colwellia and Methylophaga were among those with slowest growth rates. These observations provide a complementary view on the thus far poorly documented growth rates of bacterial groups at fine phylogenetic level (Campbell and Kirchman 2013; Lankiewicz, Cottrell and Kirchman 2016). An issue that was debated in this context is whether the growth rates of SAR11 are lower (Teira et al. 2009; Ferrera et al. 2011; Campbell and Kirchman 2013) or similar (Malmstrom et al. 2004; Laghdass et al. 2012; Salter et al. 2015; Lankiewicz, Cottrell and Kirchman 2016) as compared to other bacterial groups or the average community. Our model simulations support the idea that some SAR11 can have growth rates similar to other taxa under carbon-limited conditions, coherent with the dominance of this clade in the oligotrophic ocean. Conclusion We demonstrate here the tight link between changes in the composition of DOM and that of the bacterial community in the severely C-limited Southern Ocean. The rapid emergence in the diatom-DOM treatment of bacterial groups that were also abundant in Southern Ocean surface waters during an intense diatom bloom (Landa et al. 2016) emphasizes the relevance of our findings to help explain the ecosystem response to a supply in fresh, labile phytoplankton material. Our experimental system enabled us to simulate the growth rates of several bacterial taxa under constrained resource supply. These simulations in combination with present knowledge on the metabolic strategies of the co-occurring abundant taxa highlight important roles for ecological interplays of marine microbes in the regulation of DOM degradation. Future studies combining detailed community composition and metabolic profiling with chemical characterization of organic matter will help understand the interactions that exist between specific metabolites and microbial taxa, and how these interactions in return shape DOM processing and bacterial community structure in the ocean. SUPPLEMENTARY DATA Supplementary data are available at FEMSEC online. ACKNOWLEDGEMENTS We thank the captain and the crew of the R/V Marion Dufresne and the chief scientist B. Quéguiner for their help and support aboard. We thank L. Oriol and J. Caparros for inorganic nutrient and DOC analyses, respectively, and the trace-metal team for providing large volumes of filtered seawater. N. Fuchs (Alfred Wegener Institute, Germany) kindly provided the strain Chaetoceros debilis. We thank the handling editor and four anonymous reviewers for their constructive comments on a previous version of the manuscript. FUNDING This work was supported by the Centre National de la Recherche Scientifique (CNRS)-Institut National des Sciences de l'Univers (INSU)-Les Enveloppes Fluides et l'Environnement (LEFE)-Cycles Biogéochimiques, Environnement et Ressources (CYBER); the National Research Agency [grant ANR-10-BLAN-0614]; the Paul Emile Victor Institute and the French Ministry of Science [to M.L.]. Conflict of interest. None declared. REFERENCES Alonso C, Pernthaler J. Roseobacter and SAR11 dominate microbial glucose uptake in coastal North Sea waters. Environ Microbiol . 2006; 8: 2022– 30. Google Scholar CrossRef Search ADS PubMed  Aminot A, Kérouel R. Dosage Automatique Des Nutriments Dans Les Eaux Marines: Méthodes En Flux Continu . France: Ifremer, 2007. Ankrah NYD, Lane T, Budinoff CR et al.  . Draft genome sequence of Sulfitobacter sp. CB2047, a member of the Roseobacter clade of marine bacteria, isolated from an Emiliania huxleyi bloom. Genome Announc . 2014; 2: e01125– 14. Google Scholar PubMed  Bano N, Hollibaugh JT. Phylogenetic composition of bacterioplankton assemblages from the Arctic Ocean. Appl Environ Microbiol . 2002; 68: 505– 18. Google Scholar CrossRef Search ADS PubMed  Beier S, Rivers AR, Moran MA et al.  . The transcriptional response of prokaryotes to phytoplankton‐derived dissolved organic matter in seawater. Environ Microbiol . 2015; 17: 3466– 80. Google Scholar CrossRef Search ADS PubMed  Benner R, Strom M. A critical evaluation of the analytical blank associated with DOC measurements by high-temperature catalytic oxidation. Mar Chem . 1993; 41: 153– 60. Google Scholar CrossRef Search ADS   Bertrand EM, McCrow JP, Moustafa A et al.  . Phytoplankton–bacterial interactions mediate micronutrient colimitation at the coastal Antarctic sea ice edge. Proc Natl Acad Sci . 2015; 112: 9938– 43. Google Scholar CrossRef Search ADS PubMed  Blain S, Quéguiner B, Armand L et al.  . Effect of natural iron fertilization on carbon sequestration in the Southern Ocean. Nature . 2007; 446: 1070– 4. Google Scholar CrossRef Search ADS PubMed  Boyd PW, Watson AJ, Law CS et al.  . A mesoscale phytoplankton bloom in the polar Southern Ocean stimulated by iron fertilization. Nature . 2000; 407: 695– 702. Google Scholar CrossRef Search ADS PubMed  Brinkhoff T, Giebel HA, Simon M. Diversity, ecology, and genomics of the Roseobacter clade: a short overview. Arch Microbiol . 2008; 189: 531– 9. Google Scholar CrossRef Search ADS PubMed  Buchan A, González JM, Moran MA. Overview of the marine Roseobacter lineage. Appl Environ Microbiol . 2005; 71: 5665– 77. Google Scholar CrossRef Search ADS PubMed  Buchan A, LeCleir GR, Gulvik CA et al.  . Master recyclers: features and functions of bacteria associated with phytoplankton blooms. Nat Rev Microbiol . 2014; 12: 686– 98. Google Scholar CrossRef Search ADS PubMed  Campbell BJ, Kirchman DL. Bacterial diversity, community structure and potential growth rates along an estuarine salinity gradient. ISME J . 2013; 7: 210– 20. Google Scholar CrossRef Search ADS PubMed  Caporaso JG, Kuczynski J, Stombaugh J et al.  . QIIME allows analysis of high-throughput community sequencing data. Nat Methods . 2010; 7: 335– 6. Google Scholar CrossRef Search ADS PubMed  Carini P, Steindler L, Beszteri S et al.  . Nutrient requirements for growth of the extreme oligotroph Candidatus Pelagibacter ubique HTCC1062 on a defined medium. ISME J . 2013; 7: 592– 602. Google Scholar CrossRef Search ADS PubMed  Church MJ. Resource Control of Bacterial Dynamics in the Sea. Microbial Ecology of the Oceans . Hoboken, NJ, USA: John Wiley & Sons, Inc., 2008, 335– 82. Google Scholar CrossRef Search ADS   Church MJ, Hutchins DA, Ducklow HW. Limitation of bacterial growth by dissolved organic matter and iron in the Southern Ocean. Appl Environ Microbiol . 2000; 66: 455– 66. Google Scholar CrossRef Search ADS PubMed  Cole JJ, Findlay S, Pace ML. Bacterial production in fresh and saltwater ecosystems: a cross-system overview. Mar Ecol Prog Ser . 1988; 43: 1– 10. Google Scholar CrossRef Search ADS   Cottrell MT, Kirchman DL. Natural assemblages of marine Proteobacteria and members of the Cytophaga-Flavobacter cluster consuming low- and high-molecular-weight dissolved organic matter. Appl Environ Microbiol . 2000; 66: 1692– 7. Google Scholar CrossRef Search ADS PubMed  Datta MS, Sliwerska E, Gore J et al.  . Microbial interactions lead to rapid micro-scale successions on model marine particles. Nat Commun . 2016; 7: 11965. Google Scholar CrossRef Search ADS PubMed  Delmont TO, Hammar KM, Ducklow HW et al.  . Phaeocystis antarctica blooms strongly influence bacterial community structures in the Amundsen Sea polynya. Front Microbiol . 2014; 5: 646. Google Scholar CrossRef Search ADS PubMed  Dumont M, , Rapaport A, Harmand J et al.  . Observers for microbial ecology - How including molecular data into bioprocess modeling ? In: 2008 16th Mediterranean Conference on Control and Automation . 2008, 1381– 6., IEEE (Piscataway, NJ, USA). Durham BP, Sharma S, Luo H et al.  . Cryptic carbon and sulfur cycling between surface ocean plankton. Proc Natl Acad Sci . 2015; 112: 453– 7. Google Scholar CrossRef Search ADS PubMed  Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res . 2004; 32: 1792– 7. Google Scholar CrossRef Search ADS PubMed  Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics . 2010; 26: 2460– 1. Google Scholar CrossRef Search ADS PubMed  Fernández-Gómez B, Richter M, Schüler M et al.  . Ecology of marine Bacteroidetes: a comparative genomics approach. ISME J . 2013; 7: 1026– 37. Google Scholar CrossRef Search ADS PubMed  Ferrera I, Gasol JM, Sebastián M et al.  . Comparison of growth rates of aerobic anoxygenic phototrophic bacteria and other bacterioplankton groups in coastal Mediterranean waters. Appl Environ Microbiol . 2011; 77: 7451– 8. Google Scholar CrossRef Search ADS PubMed  Fourquez M, Beier S, Jongmans E et al.  . Uptake of leucine, chitin, and iron by prokaryotic groups during spring phytoplankton blooms induced by natural iron fertilization off Kerguelen Island (Southern Ocean). Front Mar Sci . 2016; 3: 256. Google Scholar CrossRef Search ADS   Fuhrman JA, Hewson I, Schwalbach MS et al.  . Annually reoccurring bacterial communities are predictable from ocean conditions. Proc Natl Acad Sci . 2006; 103: 13104– 9. Google Scholar CrossRef Search ADS PubMed  Fukuda R, Ogawa H, Nagata T et al.  . Direct determination of carbon and nitrogen contents of natural bacterial assemblages in marine environments. Appl Environ Microbiol . 1998; 64: 3352– 8. Google Scholar PubMed  Gentile G, Giuliano L, D'Auria G et al.  . Study of bacterial communities in Antarctic coastal waters by a combination of 16S rRNA and 16S rDNA sequencing. Environ Microbiol . 2006; 8: 2150– 61. Google Scholar CrossRef Search ADS PubMed  Ghiglione JF, Murray AE. Pronounced summer to winter differences and higher wintertime richness in coastal Antarctic marine bacterioplankton. Environ Microbiol . 2012; 14: 617– 29. Google Scholar CrossRef Search ADS PubMed  Gifford SM, Becker JW, Sosa OA et al.  . Quantitative transcriptomics reveals the growth- and nutrient-dependent response of a streamlined marine methylotroph to methanol and naturally occurring dissolved organic matter. MBio . 2016; 7: e01279– 16. Google Scholar CrossRef Search ADS PubMed  Gilbert Ja, Steele Ja, Caporaso JG et al.  . Defining seasonal marine microbial community dynamics. ISME J . 2012; 6: 298– 308. Google Scholar CrossRef Search ADS PubMed  Giovannoni S. SAR11 bacteria: the most abundant plankton in the oceans, Ann Rev Mar Sci . 2017; 9: 231– 55. Goldman JC, Caron DA, Dennett MR. Regulation of gross growth efficiency and ammonium regeneration in bacteria by substrate C: N ratio. Limnol Ocean . 1987; 32: 1239– 52. Google Scholar CrossRef Search ADS   Gómez‐Pereira PR, Schüler M, Fuchs BM et al.  . Genomic content of uncultured Bacteroidetes from contrasting oceanic provinces in the North Atlantic Ocean. Environ Microbiol . 2012; 14: 52– 66. Google Scholar CrossRef Search ADS PubMed  González JM, Fernández-Gómez B, Fernàndez-Guerra A et al.  . Genome analysis of the proteorhodopsin-containing marine bacterium Polaribacter sp. MED152 (Flavobacteria). Proc Natl Acad Sci . 2008; 105: 8724– 9. Google Scholar CrossRef Search ADS PubMed  González JM, Simó R, Massana R et al.  . Bacterial community structure associated with a dimethylsulfoniopropionate-producing North Atlantic algal bloom. Appl Environ Microbiol . 2000; 66: 4237– 46. Google Scholar CrossRef Search ADS PubMed  Hansell DA. Recalcitrant dissolved organic carbon fractions. Ann Rev Mar Sci . 2013; 5: 421– 45. Google Scholar CrossRef Search ADS PubMed  Hertkorn N, Ruecker C, Meringer M et al.  . High-precision frequency measurements: indispensable tools at the core of the molecular-level analysis of complex systems. Anal Bioanal Chem . 2007; 389: 1311– 27. Google Scholar CrossRef Search ADS PubMed  Hu P, Dubinsky EA, Probst AJ et al.  . Simulation of Deepwater Horizon oil plume reveals substrate specialization within a complex community of hydrocarbon degraders. Proc Natl Acad Sci U S A . 2017; 114: 7432– 7. Google Scholar CrossRef Search ADS PubMed  Kerkhof LJ, Voytek MA, Sherrell RM et al.  . Variability in bacterial community structure during upwelling in the coastal ocean. Hydrobiologia . 1999; 401: 139– 48. Google Scholar CrossRef Search ADS   Kim H, Ducklow HW. A Decadal (2002–2014) Analysis for dynamics of heterotrophic bacteria in an Antarctic coastal ecosystem: variability and physical and biogeochemical forcings. Front Mar Sci . 2016; 3: 214. Google Scholar CrossRef Search ADS   Kim J-G, Park S-J, Quan Z-X et al.  . Unveiling abundance and distribution of planktonic bacteria and archaea in a polynya in Amundsen Sea, Antarctica: prokaryotic communities in Antarctic polynya. Environ Microbiol . 2014; 16: 1566– 78. Google Scholar CrossRef Search ADS PubMed  Kimes NE, López-Pérez M, Ausó E et al.  . RNA sequencing provides evidence for functional variability between naturally co-existing Alteromonas macleodii lineages. BMC Genomics . 2014; 15: 938. Google Scholar CrossRef Search ADS PubMed  Kirchman DL. Growth rates of microbes in the oceans, Ann Rev Mar Sci . 2016; 8: 285– 309. Kirchman DL, Meon B, Cottrell MT et al.  . Carbon versus iron limitation of bacterial growth in the California upwelling regime. Limnol Oceanogr . 2000; 45: 1681– 8. Google Scholar CrossRef Search ADS   Kleindienst S, Seidel M, Ziervogel K et al.  . Chemical dispersants can suppress the activity of natural oil-degrading microorganisms. Proc Natl Acad Sci . 2015; 112: 14900– 5. Google Scholar CrossRef Search ADS PubMed  Krause SMB, Johnson T, Karunaratne YS et al.  . Lanthanide-dependent cross-feeding of methane-derived carbon is linked by microbial community interactions. Proc Natl Acad Sci . 2017; 114: 358– 63. Google Scholar CrossRef Search ADS PubMed  Kumar S, Stecher G, Tamura K. MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets. Mol Biol Evol . 2016; 33: 1870– 4. Google Scholar CrossRef Search ADS PubMed  Laghdass M, Catala P, Caparros J et al.  . High contribution of SAR11 to microbial activity in the North West Mediterranean Sea. Microb Ecol . 2012; 63: 324– 33. Google Scholar CrossRef Search ADS PubMed  Landa M, Blain S, Christaki U et al.  . Shifts in bacterial community composition associated with increased carbon cycling in a mosaic of phytoplankton blooms. ISME J . 2016; 10: 39– 50. Google Scholar CrossRef Search ADS PubMed  Landa M, Cottrell MT, Kirchman DL et al.  . Changes in bacterial diversity in response to dissolved organic matter supply in a continuous culture experiment. Aquat Microb Ecol . 2013; 69: 157– 68. Google Scholar CrossRef Search ADS   Landa M, Cottrell MT, Kirchman DL et al.  . Phylogenetic and structural response of heterotrophic bacteria to dissolved organic matter of different chemical composition in a continuous culture study. Environ Microbiol . 2014; 16: 1668– 81. Google Scholar CrossRef Search ADS PubMed  Langenheder S, Lindström ES, Tranvik LJ. Weak coupling between community composition and functioning of aquatic bacteria. Limnol Oceanogr . 2005; 50: 957– 67. Google Scholar CrossRef Search ADS   Langenheder S, Lindström ES, Tranvik LJ. Structure and function of bacterial communities emerging from different sources under identical conditions. Appl Environ Microbiol . 2006; 72: 212– 20. Google Scholar CrossRef Search ADS PubMed  Lankiewicz TS, Cottrell MT, Kirchman DL. Growth rates and rRNA content of four marine bacteria in pure cultures and in the Delaware estuary. ISME J . 2016; 10: 823– 32. Google Scholar CrossRef Search ADS PubMed  Lauro FM, McDougald D, Thomas T et al.  . The genomic basis of trophic strategy in marine bacteria. Proc Natl Acad Sci . 2009; 106: 15527– 33. Google Scholar CrossRef Search ADS PubMed  Li TD, Doronina NV, Ivanova EG et al.  . Vitamin B12-independent strains of Methylophaga marina isolated from Red Sea algae. Microbiology . 2007; 76: 75– 81. Google Scholar CrossRef Search ADS   Luo H, Csűros M, Hughes AL et al.  . Evolution of divergent life history strategies in marine Alphaproteobacteria. MBio . 2013; 4: e00373– 13. Google Scholar CrossRef Search ADS PubMed  Luria CM, Amaral-Zettler LA, Ducklow HW et al.  . Seasonal succession of free-living bacterial communities in coastal waters of the Western Antarctic Peninsula. Front Microbiol . 2016; 7: 1731. Google Scholar CrossRef Search ADS PubMed  Luria CM, Amaral-Zettler LA, Ducklow HW et al.  . Seasonal shifts in bacterial community responses to phytoplankton-derived dissolved organic matter in the Western Antarctic Peninsula. Front Microbiol . 2017; 8: 2117. Google Scholar CrossRef Search ADS PubMed  Malmstrom RR, Cottrell MT, Elifantz H et al.  . Biomass production and assimilation of dissolved organic matter by SAR11 bacteria in the Northwest Atlantic Ocean. Appl Environ Microbiol . 2005; 71: 2979– 86. Google Scholar CrossRef Search ADS PubMed  Malmstrom RR, Kiene RP, Cottrell MT et al.  . Contribution of SAR11 bacteria to dissolved dimethylsulfoniopropionate and amino acid uptake in the North Atlantic Ocean. Appl Environ Microbiol . 2004; 70: 4129– 35. Google Scholar CrossRef Search ADS PubMed  Martiny JBH, Bohannan BJM, Brown JH et al.  . Microbial biogeography: putting microorganisms on the map. Nat Rev Microbiol . 2006; 4: 102– 12. Google Scholar CrossRef Search ADS PubMed  Methé BA, Nelson KE, Deming JW et al.  . The psychrophilic lifestyle as revealed by the genome sequence of Colwellia psychrerythraea 34H through genomic and proteomic analyses. Proc Natl Acad Sci U S A . 2005; 102: 10913– 8. Google Scholar CrossRef Search ADS PubMed  Monod J. La technique de culture continue: Théorie et applications. Ann Inst Pasteur . 1950; 79: 390– 410. Moran MA, Belas R, Schell MA et al.  . Ecological genomics of marine roseobacters. Appl Environ Microbiol . 2007; 73: 4559– 69. Google Scholar CrossRef Search ADS PubMed  Morris JJ, Papoulis SE, Lenski RE. Coexistence of evolving bacteria stabilized by a shared Black Queen function. Evolution (N Y) . 2014; 68: 2960– 71. Morris RM, Rappé MS, Connon SA et al.  . SAR11 clade dominates ocean surface bacterioplankton communities. Nature . 2002; 420: 806– 10. Google Scholar CrossRef Search ADS PubMed  Mou X, Sun S, Edwards RA et al.  . Bacterial carbon processing by generalist species in the coastal ocean. Nature . 2008; 451: 708– 11. Google Scholar CrossRef Search ADS PubMed  Needham DM, Fuhrman JA. Pronounced daily succession of phytoplankton, archaea and bacteria following a spring bloom. Nat Microbiol . 2016; 1: 16005. Google Scholar CrossRef Search ADS PubMed  Nelson JD, Boehme SE, Reimers CE et al.  . Temporal patterns of microbial community structure in the Mid-Atlantic Bight. FEMS Microbiol Ecol . 2008; 65: 484– 93. Google Scholar CrossRef Search ADS PubMed  Neufeld JD, Schäfer H, Cox MJ et al.  . Stable-isotope probing implicates Methylophaga spp and novel Gammaproteobacteria in marine methanol and methylamine metabolism. ISME J . 2007; 1: 480– 91. Google Scholar CrossRef Search ADS PubMed  Neumann AM, Balmonte JP, Berger M et al.  . Different utilization of alginate and other algal polysaccharides by marine Alteromonas macleodii ecotypes. Environ Microbiol . 2015; 17: 3857– 68. Google Scholar CrossRef Search ADS PubMed  Newton RJ, Griffin LE, Bowles KM et al.  . Genome characteristics of a generalist marine bacterial lineage. ISME J . 2010; 4: 784– 98. Google Scholar CrossRef Search ADS PubMed  Novick A, Szilard L. Description of the chemostat. Science . 1950; 112: 715– 6. Google Scholar CrossRef Search ADS PubMed  Obernosterer I, Christaki U, Lefèvre D et al.  . Rapid bacterial mineralization of organic carbon produced during a phytoplankton bloom induced by natural iron fertilization in the Southern Ocean. Deep Sea Res Part II Top Stud Oceanogr . 2008; 55: 777– 89. Google Scholar CrossRef Search ADS   Obernosterer I, Fourquez M, Blain S. Fe and C co-limitation of heterotrophic bacteria in the naturally fertilized region off the Kerguelen Islands. Biogeosciences . 2015; 12: 1983– 92. Google Scholar CrossRef Search ADS   Pedler BE, Aluwihare LI, Azam F. Single bacterial strain capable of significant contribution to carbon cycling in the surface ocean. Proc Natl Acad Sci . 2014; 111: 7202– 7. Google Scholar CrossRef Search ADS PubMed  Pinhassi J, Havskum H, Peters F et al.  . Changes in bacterioplankton composition under different phytoplankton regimens, Appl Environ Microbiol . 2004; 70: 6753– 66. Quéguiner B. Iron fertilization and the structure of planktonic communities in high nutrient regions of the Southern Ocean. Deep Res Part II Top Stud Oceanogr . 2013; 90: 43– 54. Google Scholar CrossRef Search ADS   Quince C, Lanzen A, Davenport RJ et al.  . Removing noise from pyrosequenced amplicons. BMC Bioinformatics . 2011; 12: 38. Google Scholar CrossRef Search ADS PubMed  Rapaport A, Dochain D, Harmand J. Long run coexistence in the chemostat with multiple species. J Theor Biol . 2009; 257: 252– 9. Google Scholar CrossRef Search ADS PubMed  Redmond MC, Valentine DL. Natural gas and temperature structured a microbial community response to the Deepwater Horizon oil spill. Proc Natl Acad Sci U S A . 2012; 109: 20292– 7. Google Scholar CrossRef Search ADS PubMed  Riemann L, Steward GF, Azam F. Dynamics of bacterial community composition and activity during a mesocosm diatom bloom. Appl Environ Microbiol . 2000; 66: 578– 87. Google Scholar CrossRef Search ADS PubMed  Rink B, Seeberger S, Martens T et al.  . Effects of phytoplankton bloom in a coastal ecosystem on the composition of bacterial communities. Aquat Microb Ecol . 2007; 48: 47– 60. Google Scholar CrossRef Search ADS   Salter I, Galand PE, Fagervold SK et al.  . Seasonal dynamics of active SAR11 ecotypes in the oligotrophic Northwest Mediterranean Sea. ISME J . 2015; 9: 347– 60. Google Scholar CrossRef Search ADS PubMed  Sarmento H, Morana C, Gasol JM. Bacterioplankton niche partitioning in the use of phytoplankton-derived dissolved organic carbon: quantity is more important than quality. ISME J . 2016; 10: 2582– 92. Google Scholar CrossRef Search ADS PubMed  Satinsky BM, Crump BC, Smith CB et al.  . Microspatial gene expression patterns in the Amazon River Plume. Proc Natl Acad Sci . 2014; 111: 11085– 90. Google Scholar CrossRef Search ADS PubMed  Schloss PD, Westcott SL, Ryabin T et al.  . Introducing mothur: Open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol . 2009; 75: 7537– 41. Google Scholar CrossRef Search ADS PubMed  Selje N, Simon M, Brinkhoff T. A newly discovered Roseobacter cluster in temperate and polar oceans. Nature . 2004; 427: 445– 8. Google Scholar CrossRef Search ADS PubMed  Simon M, Billerbeck S, Kessler D et al.  . Bacterioplankton communities in the Southern Ocean: Composition and growth response to various substrate regimes. Aquat Microb Ecol . 2012; 68: 13– 28. Google Scholar CrossRef Search ADS   Sjöstedt J, Pontarp M, Tinta T et al.  . Reduced diversity and changed bacterioplankton community composition do not affect utilization of dissolved organic matter in the Adriatic Sea. Aquat Microb Ecol . 2013; 71: 15– 24. Google Scholar CrossRef Search ADS   Sogin ML, Morrison HG, Huber JA et al.  . Microbial diversity in the deep sea and the underexplored “rare biosphere”. Proc Natl Acad Sci . 2006; 103: 12115– 20. Google Scholar CrossRef Search ADS PubMed  Sowell SM, Wilhelm LJ, Norbeck AD et al.  . Transport functions dominate the SAR11 metaproteome at low-nutrient extremes in the Sargasso Sea. ISME J . 2009; 3: 93– 105. Google Scholar CrossRef Search ADS PubMed  Speeckaert G, Borges AV, Champenois W et al.  . Annual cycle of dimethylsulfoniopropionate (DMSP) and dimethylsulfoxide (DMSO) related to phytoplankton succession in the Southern North Sea. Sci Total Environ . 2018; 622–623: 362– 72. Google Scholar CrossRef Search ADS PubMed  Straza TRA, Ducklow HW, Murray AE et al.  . Abundance and single-cell activity of bacterial groups in Antarctic coastal waters. Limnol Oceanogr . 2010; 55: 2526– 36. Google Scholar CrossRef Search ADS   Sun J, Steindler L, Thrash JC et al.  . One carbon metabolism in SAR11 pelagic marine bacteria. PLoS One . 2011; 6: e23973. Google Scholar CrossRef Search ADS PubMed  Sunagawa S, Coelho LP, Chaffron S et al.  . Structure and function of the global ocean microbiome. Science . 2015; 348: 1261359. Google Scholar CrossRef Search ADS PubMed  Tada Y, Taniguchi A, Sato-Takabe Y et al.  . Growth and succession patterns of major phylogenetic groups of marine bacteria during a mesocosm diatom bloom. J Oceanogr . 2012; 68: 509– 19. Google Scholar CrossRef Search ADS   Teeling H, Fuchs BM, Becher D et al.  . Substrate-controlled succession of marine bacterioplankton populations induced by a phytoplankton bloom. Science . 2012; 336: 608– 11. Google Scholar CrossRef Search ADS PubMed  Teira E, Martínez‐García S, Lønborg C et al.  . Growth rates of different phylogenetic bacterioplankton groups in a coastal upwelling system. Environ Microbiol Rep . 2009; 1: 545– 54. Google Scholar CrossRef Search ADS PubMed  Thiele S, Fuchs BM, Ramaiah N et al.  . Microbial community response during the iron fertilization experiment LOHAFEX. Appl Environ Microbiol . 2012; 78: 8803– 12. Google Scholar CrossRef Search ADS PubMed  Thompson LR, Field C, Romanuk T et al.  . Patterns of ecological specialization among microbial populations in the Red Sea and diverse oligotrophic marine environments. Ecol Evol . 2013; 3: 1780– 97. Google Scholar CrossRef Search ADS PubMed  Tremblay L, Caparros J, Leblanc K et al.  . Origin and fate of particulate and dissolved organic matter in a naturally iron-fertilized region of the Southern Ocean. Biogeosciences . 2015; 12: 607– 21. Google Scholar CrossRef Search ADS   Tripp HJ. The unique metabolism of SAR11 aquatic bacteria. J Microbiol . 2013; 51: 147– 53. Google Scholar CrossRef Search ADS PubMed  Tripp HJ, Kitner JB, Schwalbach MS et al.  . SAR11 marine bacteria require exogenous reduced sulphur for growth. Nature . 2008; 452: 741– 4. Google Scholar CrossRef Search ADS PubMed  Wang Q, Garrity GM, Tiedje JM et al.  . Naïve bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol . 2007; 73: 5261– 7. Google Scholar CrossRef Search ADS PubMed  Wawrik B, Kerkhof L, Kukor J et al.  . Effect of different carbon sources on community composition of bacterial enrichments from soil effect of different carbon sources on community composition of bacterial enrichments from soil. Appl Environ Microbiol . 2005; 71: 6776– 83. Google Scholar CrossRef Search ADS PubMed  West NJ, Obernosterer I, Zemb O et al.  . Major differences of bacterial diversity and activity inside and outside of a natural iron-fertilized phytoplankton bloom in the Southern Ocean. Environ Microbiol . 2008; 10: 738– 56. Google Scholar CrossRef Search ADS PubMed  Wietz M, Wemheuer B, Simon H et al.  . Bacterial community dynamics during polysaccharide degradation at contrasting sites in the Southern and Atlantic Oceans. Environ Microbiol . 2015; 17: 3822– 31. Google Scholar CrossRef Search ADS PubMed  Wilkins D, Yau S, Williams TJ et al.  . Key microbial drivers in Antarctic aquatic environments. FEMS Microbiol Rev . 2013; 37: 303– 35. Google Scholar CrossRef Search ADS PubMed  Williams TJ, Wilkins D, Long E et al.  . The role of planktonic Flavobacteria in processing algal organic matter in coastal East Antarctica revealed using metagenomics and metaproteomics. Environ Microbiol . 2013; 15: 1302– 17. Google Scholar CrossRef Search ADS PubMed  Wolcott RD, Gontcharova V, Sun Y et al.  . Evaluation of the bacterial diversity among and within individual venous leg ulcers using bacterial tag-encoded FLX and Titanium amplicon pyrosequencing and metagenomic approaches. BMC Microbiol . 2009; 9: 226. Google Scholar CrossRef Search ADS PubMed  Xing P, Hahnke RL, Unfried F et al.  . Niches of two polysaccharide-degrading Polaribacter isolates from the North Sea during a spring diatom bloom. ISME J . 2015; 9: 1410– 22. Google Scholar CrossRef Search ADS PubMed  © FEMS 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png FEMS Microbiology Ecology Oxford University Press

Major changes in the composition of a Southern Ocean bacterial community in response to diatom-derived dissolved organic matter

Download PDF
16 pages

Loading next page...
1
 
/lp/ou_press/major-changes-in-the-composition-of-a-southern-ocean-bacterial-tsqAzJRMd7

References (116)

Publisher
Oxford University Press
Copyright
© FEMS 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0168-6496
eISSN
1574-6941
DOI
10.1093/femsec/fiy034
Publisher site
See Article on Publisher Site

Abstract

Abstract In the Southern Ocean, natural iron fertilization in the wake of islands leads to annually occurring spring phytoplankton blooms associated with enhanced heterotrophic activity through the release of labile dissolved organic matter (DOM). The aim of this study was to investigate experimentally how diatom-derived DOM affects the composition of Southern Ocean winter water bacterial communities and to identify the most responsive taxa. A bacterial community collected in the naturally iron-fertilized region off Kerguelen Island (KEOPS2 October-November 2011) was grown onboard in continuous cultures, on winter water alone or amended with diatom-derived DOM supplied at identical DOC concentrations. 454 sequencing of 16S amplicons revealed that the two DOM sources sustained strikingly different bacterial communities, with higher relative abundances of Sulfitobacter, Colwellia and Methylophaga operational taxonomic units (OTUs) and lower relative abundances of Polaribacter, Marinobacter, NAC11–7 and SAR11 OTUs in diatom-DOM compared to winter water conditions. Using a modeling approach, we obtained growth rates for phylogenetically diverse taxa varying between 0.12 and 0.49 d−1 under carbon-limited conditions. Our results identify diatom DOM as a key factor shaping Southern Ocean winter water bacterial communities and suggest a role for niche partitioning and microbial interactions in organic matter utilization. bacterial community composition, Southern Ocean, dissolved organic matter, 16S rRNA gene sequencing, continuous cultures INTRODUCTION Bacteria are key players in the transformation of carbon present in marine dissolved organic matter (DOM). Understanding the mechanisms of carbon utilization by marine bacteria is presently a major challenge because both the DOM pool (Hertkorn et al. 2007) and bacterial communities (Sogin et al. 2006; Sunagawa et al. 2015) are highly complex. Advances in the fields of analytical chemistry and constantly improving high-throughput sequencing technologies have enabled progress but much remains to be learned about the composition and interactions between the chemical and microbiological components of the ocean. From an ecological point of view, understanding the factors that drive bacterial community composition would shed light on their roles in the biogeochemical cycling of carbon and other elements. Because DOM is the main source of substrates for marine bacteria, its characteristics, such as chemical composition and bioavailability, and its concentration are likely to play important roles in structuring bacterial communities. It has been suggested that individual bacterial taxa possess different metabolic properties, reflecting the vast diversity of compounds found in a given environment (Wawrik et al. 2005; Teeling et al. 2012; Satinsky et al. 2014). As a result, various taxa could be adapted to different lifestyles based on distinct resource exploitation strategies (Lauro et al. 2009). This hypothesis is supported by numerous studies showing biogeographic patterns as well as predictable bacterial community composition in recurring conditions (Fuhrman et al. 2006; Martiny et al. 2006; Nelson et al. 2008; Gilbert et al. 2012). However, there is also a large body of evidence for functional redundancy within and across bacterial communities (Langenheder, Lindström and Tranvik, 2005, 2006; Sjöstedt et al. 2013; Beier et al. 2015), as well as generalist carbon consumption (Mou et al. 2008; Landa et al. 2014; Pedler, Aluwihare and Azam 2014; Sarmento, Morana and Gasol 2016), suggesting functional traits alone do not explain taxon distribution and abundance. Taken together, these observations indicate that the factors driving microbial community structure and activity are complex and vary in nature and intensity across space and time. The extent to which characteristics of the carbon pool used by heterotrophic bacteria act as shaping force on community composition remains to be determined. Phytoplankton supplies labile DOM to heterotrophic bacteria, which results in a tight link between these components in all aquatic ecosystems (Cole, Findlay and Pace 1988). Numerous studies have investigated how pulses of bioavailable DOM released during phytoplankton blooms affect bacterial heterotrophic activity on seasonal (Church 2008 and references therein) and inter-annual scales (Kim and Ducklow 2016). The concurrent changes in the bacterial community were suggested to be driven by the quantity and quality of DOM released during bloom events (Kerkhof et al. 1999; Riemann, Steward and Azam 2000; Pinhassi et al. 2004; Rink et al. 2007; West et al. 2008; Tada et al. 2012; Delmont et al. 2014; Kim et al. 2014; Luria et al. 2016; Needham and Fuhrman 2016). However, few studies have supported their field observations with experimental studies, directly testing the effect of phytoplankton-derived DOM on bacterial community composition (Luria et al. 2017). The Southern Ocean is the largest high-nutrient, low-chlorophyll region in the world's oceans. Iron-limited phytoplankton primary production (Boyd et al. 2000; Blain et al. 2007) in combination with permanent upwelling of deep water around Antarctica result in Southern Ocean surface waters characterized by low concentrations of dissolved organic carbon (DOC, about 50 µM, Hansell 2013), of rather refractory nature. This leads to a double constraint on heterotrophic bacteria, which require both iron and carbon for growth (Church, Hutchins and Ducklow 2000; Kirchman et al. 2000). While iron was identified as the limiting nutrient for heterotrophic microbes in only a few studies, the persistent enhancement of heterotrophic bacterial metabolism upon addition of labile organic carbon compounds (see for review Obernosterer, Fourquez and Blain 2015) clearly highlights that bacterial activity in the Southern Ocean is constrained by the bioavailability of DOM. In the present study, we tested the hypothesis that the supply of labile DOM derived from diatoms, the dominant bloom forming phytoplankton in naturally iron-fertilized regions of the Southern Ocean (Blain et al. 2007; Quéguiner 2013), leads to shifts in the composition of a natural, C-limited bacterial community. Continuous cultures were implemented, to actively select bacterial members most adapted either to the background DOM present in the winter waters, or to a mixture of background and diatom-derived DOM. Diversity analyses revealed striking differences in the identity of the taxa selected by each of the tested DOM conditions. We observe that, in this environment where labile substrates are scarce, bacterial community composition is strongly affected by qualitative differences in the available DOM pool , demonstrating a tight link between the bioavailability of organic carbon and bacterial metabolism and diversity in the Southern Ocean. MATERIALS AND METHODS Experimental design The continuous culture experiment presented in this study was ran on board on the R/V Marion Dufresne during the KEOPS2 cruise conducted in the naturally iron-fertilized region around the Kerguelen Islands in the Southern Ocean in October and November 2011. Two experimental conditions were run in triplicates. In the first condition, or control, bacteria were supplied with 0.2 µm-filtered seawater, while in the second condition, or diatom-DOM treatment, bacteria were supplied with 0.2 µm-filtered seawater mixed with diatom-derived DOM. For the control medium preparation, 60 l of seawater were sampled with the trace-metal clean rosette in the upper 80 m at Station A3 (Table S1, Supporting Information) located above the Kerguelen plateau (Fig. S1, Supporting Information), in the iron-fertilized region before the onset of the bloom (Landa et al. 2016). The 60 l were collected in acid-washed (10% HCl), Milli-Q water-rinsed polycarbonate (PC) carboys and immediately 0.2 μm-filtered. This water was stored at 4°C in acid-washed, Milli-Q water-rinsed PC carboys for the entire duration of the experiment. Water for the bacterial inoculum was collected at 40 m and 4 l were filtered through a 0.8 μm PC filter (45 mm diameter, Nuclepore, Whatman, Little Chalfont, UK). The 0.8 µm filtrate was stored at 6°C for a few hours until the beginning of the experiment. Concurrently, 4.8 l of the <0.8 μm filtrate were collected on a 0.2 μm cartridge (Sterivex, Millipore, Darmstadt, Germany), 500 μl of lysis buffer (40 mM EDTA, 50 mM Tris, 0.75 M sucrose) were added, and the cartridge was stored at -80°C for further diversity analysis. The diatom-derived DOM was prepared in the laboratory before the cruise. A non-axenic strain of the diatom Chaetoceros debilis, isolated from the Southern Ocean and provided by the Alfred Wegener Institute (Germany) was cultivated in artificial seawater containing low amounts of background organic carbon (DOC ∼15 µM) and amended with nutrients according to Guillard's f/2 medium. The C. debilis cultures were grown at 4°C at a light:dark cycle of 16h:8h. Prior to the cruise, cells were removed from the culture by gentle filtration on two acid-washed, Milli-Q water-rinsed 0.2 μm PC filters (90 mm diameter, Nuclepore) placed on top of each other. The filtrate (5 l) was collected in acid-washed, Milli-Q water-rinsed PC carboys and stored at 4°C until the start of the experiment. The initial cultures (2 l final volume each) were prepared in acid-washed, Milli-Q water-rinsed 2 l PC carboys. For the control, the 0.2 μm-filtered in situ seawater was used as the medium, and for the diatom-DOM treatment the 0.2 μm-filtered in situ seawater was mixed with 0.2 μm-filtered diatom-derived DOM (see above). The volumes to mix were determined based on the DOC concentration of the diatom filtrate, so that diatom DOM contributed to 10 μM of the final DOC concentration. The triplicate cultures of each condition were inoculated with 200 ml of 0.8 μm-filtered in situ seawater, corresponding to a 1:10 dilution. Inorganic phosphate and nitrate were added to the control to match the concentrations obtained in the diatom-DOM treatment. The cultures were conducted in a cold room (6°C) in the dark. Each culture was bubbled with filter-sterilized air (Sartorious, Göttingen, Germany). A Milli-Q water purge was placed between the source and the filters to avoid potential DOC contamination by air. Cultures were homogenized by stirring. They were run as a batch for a few hours, to allow bacterial growth and avoid washing out the cells. Each medium, prepared in the same conditions without bacterial inoculum, was connected to the three replicate cultures of the corresponding condition by Teflon tubings, and the peristaltic pumps (Gilson, Middleton, WI, USA) were switched on to start the inflow of input medium and the outflow of culture. The imposed flow rate was 0.2 day−1, which was chosen based on the in situ growth rates determined during the first KEOPS cruise (Obernosterer et al. 2008), and resulted in a residence time of 5.6 days. Sampling of cultures The experiment was run for 19 days, during which the cultures were sampled every two to three days. The air purge was renewed every day. Input media were prepared every other day, after another 0.2 μm filtration of the required volume of phytoplankton exudate. Freshly prepared media were sampled for DOC concentration and inorganic nutrients. Sampling was performed without opening the cultures using glass syringes (Hamilton, Reno, NV) connected with a luer-lock to Teflon tubes and glass cubes that plunged to the bottom of the cultures. The glass syringes were extensively rinsed with Milli-Q water prior to sampling. A maximum volume of 50 ml, including syringe rinsing, was taken from the cultures for the various analyses, representing less than 3% of the volume of the cultures. DOC and inorganic nutrients For the determination of DOC concentrations, duplicate 15 ml samples were filtered through two precombusted GF/F filters (Whatman) and the samples were stored acidified (H3PO4, pH = 2) in combusted 20 ml glass ampoules in the dark at room temperature. DOC concentrations were measured in the home lab on a Shimadzu TOC-VCP analyzer with a Pt catalyst at 680°C (Benner and Strom 1993) as detailed in Tremblay et al. (2015). For the determination of nitrate (NO3−) and phosphate (PO43−) concentrations, a 10 ml subsample was fixed with mercuric chloride (HgCl2, 20 mg l−1 final concentration) and analyzed onboard using continuous flow analysis (Aminot and Kérouel 2007) with a Skalar instrument. Bacterial abundance and diversity A 1.8 ml sample was taken for bacterial abundance analysis, fixed with formaldehyde (2% final concentration), kept at 4°C for 30 min, and then stored at -80°C until analysis. Bacterial abundances were determined in the lab on a FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA) equipped with an air-cooled laser, providing 15 mW at 488 nm with the standard filter setup. Heterotrophic bacteria were stained with SYBR Green I and determined by flow cytometry as described previously (Obernosterer et al. 2008). Samples for bacterial diversity were collected in acid-washed, Milli-Q water-rinsed PC flasks from the outflow of the cultures at day 5, 10, 15 and 19. The tubes were placed in the flasks at night and between 135 and 216 ml of collected culture were filtered the next morning on 0.2 μm PC filters (Nuclepore), placed in tubes containing 500 μl of lysis buffer (40 mM EDTA, 50 mM Tris, 0.75 M sucrose) and stored at -80°C until the end of the cruise. The last day of the experiment, the volumes filtered for bacterial diversity were ∼500 ml for each culture. DNA extraction and pyrosequencing After lysis of the cells and proteins as described in Landa et al. (2013), a combined DNA-RNA extraction using the AllPrep DNA/RNA Micro Kit (Qiagen, Hilden, Germany) was performed on the 0.2 μm filters, following the manufacturer's instructions. The molecular size and purity of the DNA were analyzed using agarose gel electrophoresis (1%). In total, 25 samples of genomic DNA were sent to the Molecular Research DNA laboratory (Shallowater, TX, USA) for 454 pyrosequencing. The sequencing was performed using an FLX-Titanium platform (Roche, Wolcott et al. 2009). Universal bacterial primers 27F (5΄-AGRGTTTGATCMTGGCTCAG-3΄) and 530R (5΄-CCGCNGCNGCTGGCAC-3΄) were used, targeting hypervariable regions V1 to V3 of the 16S rRNA gene. The raw sequences for this dataset were deposited in the Sequence Read Archive database under accession number SRP041580. Sequences can be found on HUFUDAT01.sff, and the barcode corresponding to each sample can be found in Table S2 (Supporting Information). Bioinformatic analyses Raw sequences were denoised using AmpliconNoise (Quince et al. 2011) in Mothur (Schloss et al. 2009). Chimeras were removed during clustering of reads into OTUs at 97% identity with Usearch (Edgar 2010) in QIIME (Caporaso et al. 2010). The most abundant sequence in each OTU was chosen as the representative sequence and used for further analyses. Taxonomy was assigned to each OTU using the RDP classifier (Wang et al. 2007) and the May 2013 Greengenes database as a reference. A matrix was constructed with the number of reads corresponding to each OTU in each sample. Relative abundances were calculated for each OTU, after subsampling of the data to 2930 reads per sample. Phylogenetic analysis Phylogenetic analysis was used to clarify the affiliation of 21 OTUs identified as abundant . The partial 16S gene sequences were aligned using MUSCLE v3.8.31 (Edgar 2004). The phylogenetic inference was restricted to 242 sites that could be unambiguously aligned for the Proteobacteria tree, and 258 sites for the Flavobacteria and Verrucomicrobia tree. Phylogenetic analyses were performed using the Maximum Likelihood method implemented in MEGA7 (Kumar, Stecher and Tamura 2016). The Maximum Likelihood analysis was based on the General Time Reversible model, and initial trees for the heuristic search were obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood approach. Bootstrap proportions (BPs) were obtained from 1000 pseudo-replicates. Bayesian Posterior Probabilities (BPPs) were calculated with the Maximum Likelihood method from 1000 replicates for 1 million generations, until convergence (average standard deviation below 0.01), using MrBayes 3.2.6. Modeling approach To obtain estimates of the growth rates of the abundant OTUs, we used the classical mathematical model of the chemostat (Monod 1950; Novick and Szilard 1950). We consider a multi-species population, represented here by the 21 most abundant OTUs (Dumont et al. 2008). The model assumes that these OTUs are in competition on a single limiting substrate. The application of this basic hypothesis to our data requires some considerations. The total pool of DOM is a complex mixture of heterogeneous substrates, each characterized by variable biological availability. In our experiment, only a small fraction of this DOM pool, representing the most labile fraction, is consumed during the residence time period determined by the imposed dilution rate. Because the composition and the concentration of the specific compounds used by bacteria, as well as the affinities and uptake kinetics by different taxa are not known, the ensemble of these labile compounds was presented in the model as one limiting resource. It is well known that this model, which has no explicit intra- or inter-specific interaction terms between the populations of microorganisms, predicts competitive exclusionin the long term (i.e. at steady state). Here, we consider its non-equilibrium dynamics where a complex microbial consortium is present in the chemostat (Rapaport, Dochain and Harmand 2009). The mathematical analysis of this model predicts that the species exclusion is an asymptotic result. It can take a long time before being observed, notably when species have similar growth characteristics. Therefore, we consider here the transient dynamics of this model and not the steady state,   \begin{equation*} \begin{array}{@{}*{1}{c}@{}} {\ \frac{{ds}}{{dt}} = \ D\left( {{s_{\rm in}} - s} \right) - \mathop \sum \limits_i^n \frac{{{\mu _i}\left( s \right)}}{{{Y_i}}}{x_i}}\\ {\ \frac{{{\rm{d}}{x_i}}}{{dt}} = \ {\mu _i}\left( s \right){x_i} - D{x_i}\ \left( {1 \cdots n} \right)}. \end{array} \end{equation*} (1) Here $$s$ $ and $${x_i}$ $ stand for the substrate (that is the labile pool) and biomass concentrations of each OTU or ‘species’ $$i$ $ in the cultures, $${s_{\rm in}}$ $ is the concentration of labile organic carbon in the input medium, and $$D$ $ the dilution rate. Our objective is to estimate from experimental data the specific growth rates $${\mu _i}( . ):s \to $ $$${\mu _i}( s )$ $for each OTU. The available data from the experiments are the operating parameters $$D$ $ and$$\ {s_{\rm in}}$ $, and the measured values $$s( {{t_j}} )$ $ and $${x_i}( {{t_j}} )$ $ at four measurement instants $${t_j}$ $, for each replicate. As these data reveal that the four values of $$s( {{t_j}} )$ $ are quite close, we have used two different estimation procedures, which reinforce the robustness of the estimations. The first method allows the concentration of the consumed substrate to vary over time, while the second method assumes explicitly the concentration of the consumed substrate to be constant over time. Method 1 We identify the function $${\mu _i}( {\rm{s}} )$ $ from the model (M1) allowing $$s{\rm{\ }}$ $to be time-varying but close to the observed value. We also assume that the bacterial growth efficiency, referred to as yield coefficient $${Y_i}{\rm{\ }}$ $of each OTU is close to a common value Y. One can then observe from equations (M1) that under this assumption   \begin{equation*}\frac{d}{{dt}}\ \left( {Ys + b} \right) = {\rm{\ }}D\left( {Y{s_{\rm in}} - Ys - b} \right),\end{equation*} where $$b\ = \sum {x_i}{\rm{\ \ }}$ $is the total biomass. This equation predicts that the quantity $$Ys + b$ $ changes exponentially towards $${s_{\rm in}}$ $ with a decay rate $$D$ $. As the total biomass revealed only minor variation over time, we have considered the approximation b$$ = {\rm{\ }}Y( {{s_{\rm in}} - s} )$ $, allowing us to obtain an estimation of $$Y$ $ close to those found in the literature (in the present case, Y = 0.42). We have made the generally applied assumption that the specific growth function $${\mu _i}( {\rm{s}} )$ $ follows a Monod kinetics:   \begin{equation*}{\mu _i}\ \left( x \right) = \frac{{{\mu _{{\rm max},i}}\ s}}{{{K_{s,i}} + s}}.\ \end{equation*} Nevertheless, from the magnitude of the values of the parameters $${\mu _{{\rm max}}}$ $ and $${K_s}{\rm{\ }}$ $found in the literature and the measured values $$s( {{t_j}} )$ $, we expect it to be in the linear part of the Monod function. Consequently, we have identified the slopes $$\ \ \mu {'_i} = \ {\mu _{{\rm max},\ i}}/{K_{s,\ i}}$ $ of the functions $${\mu _i}( {\rm{s}} )$ $ at zero instead of two distinct parameters. Method 2 We assume $${\rm{s}}$ $ to be constantly equal to $$\bar{s}$ $ and we estimate parameters $$\overline {{\mu _i}} $ $ as $${\mu _i}( {\bar{s}} )$ $ and not the functions $${\mu _i}( {\rm{s}} )$ $. This leads to a simplification of the model (M1) where the dynamic of s is ignored:   \begin{equation*}\ \frac{{d{x_i}}}{{dt}} = {\rm{\ }}{\mu _i}{x_i} - D{x_i}\ \left( {1 \cdots n} \right).\end{equation*} As observed from the data that the total biomass does not vary much over time (about 10%), we have considered that the OTU proportions $${p_i} = {x_i}/b\ $ $ represent more reliable information than their abundances $${x_i}$ $. Then, under the assumption that the substrate concentration is constant at $$\bar{s}$ $, one can give their explicit expressions as functions of time (Rapaport, Dochain and Harmand 2009):   \begin{equation*}{p_i}\ \left( t \right) = {p_i}{\rm{\ }}\left( {{t_0}} \right)\frac{{{e^{{\mu _i}t}}}}{{\mathop \sum \nolimits_j {p_j}\left( t \right){e^{{\mu _j}t}}}}\end{equation*} (2)under the constraint of the quasi-stationarity of the total biomass, that is   \begin{equation*}\mathop \sum \limits_j {p_j}\left( t \right)\ {\mu _j} = \ D.\end{equation*} Note that this method allows an estimation of the value $${{\rm{\mu }}_{\rm{i}}}$ $ without requiring the knowledge of $$Y$ $, s or $${s_{\rm in}}$ $. Because the consumption of organic carbon in the cultures could not be quantified, we estimated the bacterial carbon demand, based on the bacterial production (BP = µx bacterial biomass) and a bacterial growth efficiency of 10%. We assumed the µ to be equal to the dilution rate (0.2 d−1) and we used the conversion factor of 12.4 fg C cell−1 (Fukuda et al. 1998) to determine bacterial biomass from abundance. This estimate revealed a DOC consumption of 1 µmol and 3 µmol organic carbon in the control and the diatom-DOM treatments, respectively. Both methods have used the least-square criterion for the estimation of the parameters. For each replicate and the average of replicates we have chosen estimations that provide the best R2. One can then compare the values $${\mu '_i}.s( {{t_j}} )$ $ estimated by model (M1) with the values $${\mu _i}$ $ estimated by model (M2). Comparison of the biomass of the individual OTUs at four time points based on observations and model simulations validated M1 (Fig. S2 and Table S3, Supporting Information). RESULTS Dissolved organic matter and inorganic nutrients DOC concentrations in the background seawater and the diatom-derived material used to prepare each culture medium were similar, resulting in equivalent DOC concentrations before and after the mixing. Measurements of DOC performed in each of the freshly prepared media over the course of the experiment were 49 ± 1 µM for the winter water DOM (n = 7), thereafter referred to as control, and 49 ± 2 µM for the diatom-DOM condition (n = 7). In the cultures, concentrations of DOC were slightly higher than in the respective media, in particular during the first few days of the experiment (range 50–56 µM) (Fig. 1). From day 7 (in the diatom-DOM treatment) and day 10 (in the control) to the end of the experiment, DOC concentrations remained relatively stable and they were significantly higher in the control (52 ± 1, n = 5) than in the diatom-DOM condition (50 ± 1, n = 6) (Fig. 1, Student test, P ≤ 0.05). A possibly small DOC contamination during the setup of the cultures cannot be excluded. However, given the excellent reproducibility of our observations between the biological triplicates (i.e. DOC concentrations, bacterial abundance and community composition) after several complete renewals of the cultures, it is unlikely that it substantially affected the results of our experiment. Concentrations of nitrate and phosphate were identical (Student test, P > 0.05) in the input medium (39 ± 2 and 41 ± 1 µM NO3− and 2 ± 0.3 and 2 ± 1 µM PO43− (n = 8) for the control and the diatom-DOM treatment, respectively) and the cultures (40 ± 2 and 42 ± 1 µM NO3− and 2 ± 0.2 and 2 ± 0.1 µM PO43− (n = 9) for the control and the diatom-DOM treatment, respectively). This indicates that bacterial consumption of these nutrients in the cultures was lower than the precision of the analytical method. C:N and C:P ratios were similar in the two treatments (1.3:1 and 26:1, respectively), and below the C:N and C:P ratios of heterotrophic bacteria (5:1 and 45:1) (Goldman, Caron and Dennett 1987), indicating that heterotrophic bacteria were C-limited in these incubations. Figure 1. View largeDownload slide Abundance of heterotrophic bacteria (circles) and DOC concentration (triangles) in the control (white) and the diatom-DOM (black) cultures over time. Arrows indicate time points sampled for bacterial diversity analyses. Mean values ± SD of biological triplicates are shown for each time point. * and ** indicate significant (P < 0.05) and highly significant (P < 0.005) differences between conditions, according to a Student test. Figure 1. View largeDownload slide Abundance of heterotrophic bacteria (circles) and DOC concentration (triangles) in the control (white) and the diatom-DOM (black) cultures over time. Arrows indicate time points sampled for bacterial diversity analyses. Mean values ± SD of biological triplicates are shown for each time point. * and ** indicate significant (P < 0.05) and highly significant (P < 0.005) differences between conditions, according to a Student test. Heterotrophic bacterial abundance In the first days of the experiment, bacterial abundances increased in all cultures. They stabilized after 5 days in the diatom-DOM treatment, and after 10 days in the control treatment (Fig. 1). Bacterial abundances were significantly higher in the diatom-DOM treatment (1.3 ± 0.1 × 106 cells ml−1, n = 7) than in the control (8.7 ± 0.3 × 105 cells ml−1, n = 5) (Student test, P < 0.0001). A 97% identity clustering of the reads obtained by 454 pyrosequencing of the V1 to V3 regions of the 16S gene yielded a total number of 544 OTUs for the 25 samples considered, after the denoising, quality check and chimera removal steps. One sample (diatom-DOM replicate 2, after 15 days) only contained 476 sequences and was excluded from further analysis. The 24 remaining samples had 2930 to 11 400 sequences, with an average of 5913 ± 2000 sequences per sample. To enable comparison, the dataset was subsampled to 2930 sequences per sample. A clustering of the 24 samples obtained from the two experimental conditions after 0, 5, 10, 15 and 19 days showed a clear separation between treatments, regardless of time (Fig. 2). A bootstrap analysis of the clustering indicated highly reproducible branching patterns. The communities growing in the control cultures were more similar to the inoculum than the communities growing in the diatom-DOM cultures. The control cultures showed little change with time, with an overall difference between all control samples smaller than 8%. By contrast, the communities growing in the diatom-DOM cultures exhibited a larger range of change over time. At the end of the experiment, two of the triplicate cultures were 22% different to the other diatom DOM cultures (Fig. 2). Figure 2. View largeDownload slide UPGMA clustering of bacterial communities in the control and the diatom-DOM cultures after 5, 10, 15 and 19 days, based on a weighted UNIFRAC similarity matrix. The biological triplicates are annotated 1, 2 and 3 for each treatment. The scale indicates % dissimilarity. Bootstrap values are indicated on each node. Figure 2. View largeDownload slide UPGMA clustering of bacterial communities in the control and the diatom-DOM cultures after 5, 10, 15 and 19 days, based on a weighted UNIFRAC similarity matrix. The biological triplicates are annotated 1, 2 and 3 for each treatment. The scale indicates % dissimilarity. Bootstrap values are indicated on each node. Diverse bacterial taxa belonging mostly to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes were observed in both experimental conditions (Fig. 3). Interestingly, distinct, sometimes closely related taxa within these broad groups showed strong association to either condition. For instance, the Roseobacter clade subgroups NAC11–7 and Sulfitobacter were comparatively abundant in the control and in the diatom-DOM cultures, respectively. In the order of Alteromonadales, the genera Marinobacter and Shewanella were found abundant in the control, whereas Colwellia was highly abundant in the diatom-DOM cultures. Most Bacteroidetes sequences belonged to the Polaribacter genus, which had high relative abundances in the control (50 ± 10% of total communities, n = 12) compared to the diatom-DOM cultures (4 ± 2% of total communities, n = 11). By comparison, the Ulvibacter genus was not abundant in the first three time points in either condition, and reached an average of 7 ± 2% of the total communities in both experimental treatments after 19 days (n = 3). The SAR11 clade, making up 51% of the inoculum, persisted in both treatments over the entire course of the experiment, with higher relative abundances in the control cultures (24 ± 7% of the total communities, n = 12) than in the diatom-DOM cultures (9 ± 4% of the total communities, n = 11). In the diatom-DOM cultures, the final time point was characterized by a striking increase in the relative abundance of the Lentimonas group in two out of three replicate cultures, making up to 51% of the total community in replicate 3 (Fig. 3). Overall, the communities developing in the two conditions were widely different. Figure 3. View largeDownload slide Bacterial community composition in the control cultures (A) and the diatom-DOM cultures (B), in % of total sequences. The natural community sampled in the winter waters on the plateau east of the Kerguelen Islands and used as an inoculum for these cultures is shown on the left in both graphs. Bacterial groups belonging to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes are shown in blue, green and red shades, respectively. Figure 3. View largeDownload slide Bacterial community composition in the control cultures (A) and the diatom-DOM cultures (B), in % of total sequences. The natural community sampled in the winter waters on the plateau east of the Kerguelen Islands and used as an inoculum for these cultures is shown on the left in both graphs. Bacterial groups belonging to Alphaproteobacteria, Gammaproteobacteria and Bacteroidetes are shown in blue, green and red shades, respectively. To determine how many and which OTUs were driving the differences observed in the composition of these communities, we examined the distribution patterns of abundant OTUs (relative abundance ≥ 1% in at least one time point in all the replicates of a treatment). We found 21 abundant OTUs (Fig. 4), together comprising 64% to 97% (average of 89% ± 8%, n = 23) of the total communities. Out of these 21 abundant OTUs, 16 had significantly different relative abundances between conditions in at least one time point (Fig. 5), including 8 for which the differences were significant throughout the entire experiment (Student test, P ≤ 0.05, Fig. 5A). Two Polaribacter (OTUs 1 and 6) and one NAC11–7 (OTU 13) OTUs had significantly higher relative abundances in the control compared to the diatom-DOM cultures at all time points. Conversely, one Sulfitobacter (OTU 1348), two Methylophaga (OTUs 44 and 202) and two Colwellia (OTUs 8 and 24) OTUs had significantly higher relative abundances in the diatom-DOM cultures at all time points (Fig. 5A). Taken together, these 8 OTUs, constituting the core differences between the two conditions, represented 22% to 89% of the total communities in the different samples, with an average of 56% ± 10% of the total communities in the control cultures (n = 12) and 60% ± 23% of the total communities in the diatom-DOM cultures (n = 11). Figure 4. View largeDownload slide Unrooted maximum likelihood distance tree based on 16S gene sequences clarifying affiliation of (A) 17 OTUs belonging to the group of Proteobacteria and (B) 4 OTUs belonging to the groups Flavobacteria and Verrucomicrobia. The sequences obtained in this study are shown in bold. The numbers at the node indicate BPs and BPPs respectively, given by ML and Bayesian analyses with 1000 replicates. The asterisks designate nodes with BPs or BPPs below 50% or 0.5. Figure 4. View largeDownload slide Unrooted maximum likelihood distance tree based on 16S gene sequences clarifying affiliation of (A) 17 OTUs belonging to the group of Proteobacteria and (B) 4 OTUs belonging to the groups Flavobacteria and Verrucomicrobia. The sequences obtained in this study are shown in bold. The numbers at the node indicate BPs and BPPs respectively, given by ML and Bayesian analyses with 1000 replicates. The asterisks designate nodes with BPs or BPPs below 50% or 0.5. Figure 5. View largeDownload slide Relative abundances of abundant OTUs with significant differences between control and diatom-DOM cultures at all time points (A). Relative abundances of abundant OTUs with significant differences between control and diatom-DOM amended cultures at certain time points, or without significant differences (B). Relative abundances in the control cultures are shown in blue on the right side of the graphs, relative abundances in the diatom-DOM cultures are shown in green on the left side of the graphs. Light to dark blue and green represent successive, early (day 5) to late (day 19) time points. * represent statistical differences (P ≤ 0.05) between the two conditions according to a Student test, except for graph A where all the OTUs have significantly different relative abundances at all time points. Figure 5. View largeDownload slide Relative abundances of abundant OTUs with significant differences between control and diatom-DOM cultures at all time points (A). Relative abundances of abundant OTUs with significant differences between control and diatom-DOM amended cultures at certain time points, or without significant differences (B). Relative abundances in the control cultures are shown in blue on the right side of the graphs, relative abundances in the diatom-DOM cultures are shown in green on the left side of the graphs. Light to dark blue and green represent successive, early (day 5) to late (day 19) time points. * represent statistical differences (P ≤ 0.05) between the two conditions according to a Student test, except for graph A where all the OTUs have significantly different relative abundances at all time points. Three other abundant OTUs, a SAR11 (OTU 1582) and two Sulfitobacter (OTUs 2 and 31) OTUs had significantly different relative abundances between treatments, but in two or three time points only (Fig. 5B). The SAR11 OTU had higher relative abundance in the control cultures, but it was significant only for 3 out of 4 time points. The Sulfitobacter OTUs had higher relative abundances in the diatom-DOM cultures, and the difference with the control was significant after 5 and 19 days for both OTUs, and also after 10 days for OTU 2 (Fig. 5B). Interestingly, the four OTUs associated with the control were also detected in the diatom-DOM cultures, but the seven OTUs with higher relative abundances in the diatom-DOM cultures were not detected in the control cultures, with some exceptions ( ≤ 0.1% of the total community in the control). Two abundant OTUs, a Marinobacter (OTU 41) and a Lentimonas (OTU 15) had higher relative abundances in the control and in the diatom-DOM amended cultures, respectively (Fig. 5B). Because of large variability among replicates, these differences were not significant according to a Student test. One Alphaproteobacteria, 2 SAR11 and 2 SAR86 OTUs only had significantly higher relative abundances in the control after 5 days (Fig. S3, Supporting Information). Finally, 3 abundant OTUs belonging to the betaproteobacterial clade OM43 (OTU 1288), the gammaproteobacterial family of Oceanospirillaceae (OTU 51) and the Bacteroidetes genus Ulvibacter (OTU 17) had similar relative abundances in the two treatments (Fig. 5B, P ≥ 0.06). Modeled growth rates of abundant OTUs Because of the small number of cells inoculated into the cultures, bacterial growth was not limited by organic carbon during the first days of the experiment. After the first 5 days, the supply of organic carbon and the growth rate were constrained by the dilution rate. Using the classical chemostat model (Monod 1950; Novick and Szilard 1950) allowed us to simulate growth rates for abundant OTUs in mixed communities under carbon-limited growth conditions. For a robust estimation, we used two methods of the model that are both based on the changes in biomasses of the abundant OTUs between 5 and 19 days, representing a transient phase towards equilibrium. The first method (method 1) allows the concentration of the consumed substrate to vary over time, while the second method (method 2) assumes explicitly the concentration of the consumed substrate to be constant over time. This second approach represents a simplification of the first method, because knowledge of the concentration of the consumed substrate is not required. The aim of considering these two methods is to verify a posteriori the hypothesis that the communities have reached a quasi-constant level of substrate consumption. The estimated growth rates for the 21 OTUs ranged between 0.12 and 0.49 d−1 for method 1 and between 0.14 and 0.33 d−1 for method 2 (Fig. 6 and Table S4, Supporting Information). Even though the two methods revealed variable absolute values for some OTUs, the ranking of OTUs from highest to lowest growth rates exhibited coherent patterns. Because all 21 OTUs were abundant in the diatom-DOM treatment, we illustrate the modeled growth rates for this treatment (Fig. 6). Our model results suggest three groups of OTUs characterized by growth rates that are high, intermediate and low. Highest growth rates (0.28–0.49 d−1) were attributed by both methods to OTUs belonging to Lentimonas, Ulvibacter, Oceanospirillaceae and the OM43 clade. Intermediate growth rates (0.21–0.31 d−1) were attributed to the NAC11–7, 2 Polaribacter, 1 Sulfitobacter, 3 SAR11 and 1 SAR86 OTUs. Lowest growth rates (0.12–0.18 d−1) were determined for OTUs belonging to Methylophaga, Colwellia and Marinobacter. Interestingly, the two methods suggested overall similar growth rates for a given OTU under different DOM resource regimes (Table S4, Supporting Information), except for Marinobacter that had higher growth rates in the control as compared to the DOM treatment. Figure 6. View largeDownload slide Growth rates of the abundant OTUs in the diatom-DOM treatment obtained by two methods of the chemostat model. OTUs are grouped according to high, intermediate and low growth rates as described in the text. Respective growth rates of the control are given in Table S1, Supporting Information. The relative abundances of the OTUs over the course of the experiment are presented in Figs. 5 and S1. OTU 1 –Polaribacter; OTU 2-Sulfitobacter; OTU 6-Polaribacter; OTU 8 –Colwellia; OTU 9-SAR86; OTU 13 –NAC11–7; OTU 15-Lentimonas; OTU 17-Ulvibacter; OTU 24-Colwellia; OTU 29-Alphaproteobacteria; OTU 41-Marinobacter; OTU 44 –Methylophaga; OTU 51-Oceanospirillaceae; OTU 202-Methylophaga; OTU 1457-SAR11; OTU 1283-SAR11; OTU 1288-OM43; OTU 1348 –Sulfitobacter; OTU 1582 – SAR11; OTU 1867-SAR86; OTU 31-Sulfitobacter. Figure 6. View largeDownload slide Growth rates of the abundant OTUs in the diatom-DOM treatment obtained by two methods of the chemostat model. OTUs are grouped according to high, intermediate and low growth rates as described in the text. Respective growth rates of the control are given in Table S1, Supporting Information. The relative abundances of the OTUs over the course of the experiment are presented in Figs. 5 and S1. OTU 1 –Polaribacter; OTU 2-Sulfitobacter; OTU 6-Polaribacter; OTU 8 –Colwellia; OTU 9-SAR86; OTU 13 –NAC11–7; OTU 15-Lentimonas; OTU 17-Ulvibacter; OTU 24-Colwellia; OTU 29-Alphaproteobacteria; OTU 41-Marinobacter; OTU 44 –Methylophaga; OTU 51-Oceanospirillaceae; OTU 202-Methylophaga; OTU 1457-SAR11; OTU 1283-SAR11; OTU 1288-OM43; OTU 1348 –Sulfitobacter; OTU 1582 – SAR11; OTU 1867-SAR86; OTU 31-Sulfitobacter. DISCUSSION We observed major differences in the identity and the abundance of the bacterial taxa selected by continuous cultures supplied with DOM from Southern Ocean winter water alone or amended with diatom-derived organic matter. Concentrations of DOC were identical in both conditions, but DOC bioavailability was higher in the diatom-DOM condition, as shown by increased bacterial abundances and DOC consumption, revealing that differences in origin and chemical composition of the organic matter sources were the main drivers of the observed patterns in bacterial community composition. The response to each of our experimental conditions was phylogenetically diverse, as it involved OTUs from various marine groups commonly observed in polar regions, including the Southern Ocean (Bano and Hollibaugh 2002; Gentile et al. 2006; West et al. 2008; Ghiglione and Murray 2012; Thiele et al. 2012; Wilkins et al. 2013; Luria et al. 2016). Differences were already detectable after 5 days, suggesting the mechanisms that allowed the successful groups to develop in each condition were fast. Our results indicate that in the carbon-limited Southern Ocean, changes in the DOM bioavailability such as induced by an input of diatom-derived organic matter can act as powerful selective constraints on bacterial communities through the rapid growth of most adapted members. Ecology and functional traits of taxa selected by winter water conditions A commontrait for taxa reaching high abundances in the cultures supplied with winter water alone is the ability to grow on DOM of overall low biological availability. The most abundant OTUs in the cultures supplied with winter water belonged to the Polaribacter genus in the class of Flavobacteria, and to the SAR11 and NAC11–7 alphaproteobacterial clades. Flavobacteria are known to be efficient degraders of complex polymers, such as polysaccharides, proteins or chitin, and tend to be less active than other bacterial groups in taking up simple sugars or amino acids (Cottrell and Kirchman 2000; González et al. 2008; Straza et al. 2010; Simon et al. 2012; Wietz et al. 2015; Fourquez et al. 2016). These characteristics are supported by proteomic and genomic studies showing that Flavobacteria possess numerous genes involved in hydrolytic activities as well as high molecular weight compounds transporters such as TonB-dependent receptors, and comparatively low numbers of genes involved in the uptake of low molecular weight compounds (González et al. 2008; Gómez‐Pereira et al. 2012; Fernández-Gómez et al. 2013; Williams et al. 2013). Such metabolic traits could have given Polaribacter OTUs an advantage in the cultures maintained with organic matter from winter water. These Polaribacter OTUs were also detected in the diatom-DOM cultures, however, in much lower abundances. While Flavobacteria have been described as associated with phytoplankton, many studies examining bacterial community succession over the course of natural or experimental phytoplankton blooms show different timing of the response from Bacteroidetes groups compared to taxa belonging to Alpha- and Gammaproteobacteria (Tada et al. 2012; Teeling et al. 2012; Luria et al. 2016). Our investigation of bacterial community composition during a five-month diatom bloom at the study site showed a late peak in Polaribacter abundances, associated with bloom decay (Landa et al. 2016). These results point to weak competitiveness on fresh, labile phytoplankton material and a preference for more complex, detrital algal products for these taxa. Polaribacter were abundant in mesoscosms performed with DOM from coastal Antarctic waters under pre-bloom conditions (Luria et al. 2017). In that study, Polaribacter dominated not only in unamended mesocosms, but also in mesocosms amended with DOM from the diatom Thalassiosira weissflogii. However, their final abundances in these mesocosms seemed to depend on their abundances in the bacterial communities used as inoculum, and were consistently lower when Colwellia OTUs became abundant (Luria et al. 2017). In our study, low initial abundances of Polaribacter coupled to higher competitiveness of other groups, such as Colwellia, on the fresh diatom DOMcollected during exponential growthare possible factors contributing to Polaribacter OTUs being outcompeted in this experimental condition. SAR11 are abundant in oligotrophic waters across the world (Morris et al. 2002; Giovannoni 2017). Recent genomic and proteomic information obtained from both cultivation-independent methods and cultivated representatives of this diverse group confirm genetic and physiological adaptation to life in nutrient-poor waters for most members of the group (Sowell et al. 2009; Thompson et al. 2013). SAR11 are characterized by unusual nutrient utilization profiles, including carbon (Carini et al. 2013; Tripp 2013). Single cell approaches have shown that SAR11 efficiently take up labile, low molecular weight compounds such as glucose and free amino acids, preferentially over high molecular weight compounds such as proteins (Malmstrom et al. 2004, 2005; Alonso and Pernthaler 2006), a pattern that was also observed in our study region (Fourquez et al. 2016). The high abundance of SAR11 in the presence of winter-water DOM is likely due to their efficient uptake of labile compounds present in low concentration (Tremblay et al. 2015) or of substrates derived from organic matter degradation performed by other organisms (Morris, Papoulis and Lenski 2014). Evidence for this type of cross-feeding involving SAR11 has been reported in the Atlantic Ocean (Wietz et al. 2015) and in the Southern Ocean (Williams et al. 2013). In this latter study, a metaproteomics approach revealed that SAR11 expressed transporters for molecules released by other microorganisms, in particular Flavobacteria. The Roseobacter clade is abundant across the world ocean and is characterized by high phylogenetic, functional and ecological diversity (Selje, Simon and Brinkhoff 2004; Brinkhoff, Giebel and Simon 2008; Newton et al. 2010; Buchan et al. 2014). The NAC11–7 cluster is detected during phytoplankton bloom events (González et al. 2000), but was also reported in other environmental conditions (Buchan, González and Moran 2005). Recent genomic evidence based on the one cultivated representative of this cluster (HTCC2255) suggests a more streamlined genome and adaptation to more oligotrophic conditions than what is commonly observed for most Roseobacter clade members (Luo et al. 2013). Such adaptation could explain the success of one NAC11–7 OTU in the cultures provided with non-amended winter water. Ecology and functional traits of taxa selected by diatom-DOM Three other Roseobacters were abundant in the diatom-DOM cultures, which is in agreement with previous work showing the opportunistic lifestyle of many members of this clade and their frequently observed associations with phytoplankton. A substantial fraction of their genomic content allows high versatility in carbon utilization compared to other abundant marine taxa, a characteristic that facilitates this ecological strategy (Buchan, González and Moran 2005; Newton et al. 2010). All three Roseobacter OTUs belonged to Sulfitobacter, a group known for their capacity to use organic sulfur compounds, such as dimethylsulfoniopropionate (Moran et al. 2007; Ankrah et al. 2014), a common osmolyte produced by phytoplankton, including the diatom species used in this study (Speeckaert et al. 2018). Roseobacters have been shown to use a variety of other sulfur compounds produced by phytoplankton (Durham et al. 2015). Sulfur compounds, among other modes of nutrition, could have contributed to maintain the high abundances of the three Sulfitobacter OTUs in the diatom-DOM amended cultures. Four OTUs belonging to the gammaproteobacterial genera Colwellia and Methylophaga were also abundant in the diatom-DOM cultures. Members of the genus Colwellia are commonly found in polar environments and are well adapted to life in cold environments through a number of extracellular enzymes that are most active at low temperatures (Methé et al. 2005). High Colwellia abundances were observed in mesocosms performed in the Western Antarctic Peninsula during winter (Luria et al. 2017) and in environments experiencing important pulses in organic matter, such as oil spills (Redmond and Valentine 2012; Kleindienst et al. 2015). In an experimental study, Colwellia became a dominant taxon at an intermediate stage of hydrocarbon degradation (Hu et al. 2017). The increase in abundance of this group was concurrent with an increased number of aromatic degradation genes as shown by metagenomic analyses and the decrease in the concentration of one—to three—ring aromatic compounds (Hu et al. 2017). These findings and the results from the present study point to a potentially widespread ability among members of the group to quickly use inputs of organic carbon. In particular, the capacity of Colwellia to degrade small aromatic compounds could have provided an ecological advantage to this group in the DOM-amended treatment. Several Methylophaga strains have been suggested to associate with marine algae (Li et al. 2007) and members of this group are known to utilize C1 compounds (Methé et al. 2005; Neufeld et al. 2007), a metabolic trait that could favor their growth in the presence of phytoplankton DOM. A recent transcriptomic study conducted in the McMurdo Sound showed utilization by Methylophaga of phytoplankton-derived C1 compounds such as methanol, methylamine and dimethylsufide (Bertrand et al. 2015). C1 compounds are common end-products of microbial metabolism and could have been present in our cultures, either directly in the diatom-derived DOM, or as a result of bacterial processing of diatom-DOM, providing a functional niche for these four OTUs (Krause et al. 2017). Ecological implications for DOM processing Our results from an experimental system point to interesting roles of niche occupation as well as interactions among members of a microbial community in DOM decomposition. Genomic data predict that Colwellia and Polaribacter share the ability to cleave polymeric substances. In our cultures, the presence of Polaribacter OTUs in both treatments suggests they were using polymers present in winter water, while Colwellia OTUs present in the diatom-DOM condition were almost undetectable in the non-amended winter water, indicating a preference for phytoplankton-derived polymers. Specialized functional niches seem to exist at a fine phylogenetic level, as suggested by the co-existence in a given treatment of closely related OTUs, such as the two OTUs in the Polaribacter, Colwellia and Methylophaga groups, or by the preferential association of the various Roseobacter OTUs within a specific treatment. These observations are consistent with recent functional, genomic and transcriptional studies demonstrating the diversity of metabolisms existing among members of the same genus or clade (Kimes et al. 2014; Neumann et al. 2015; Xing et al. 2015). The distributions of similar or complementary traits in co-occurring bacterial taxa further suggest that interactions among microbes such as competition and cooperation are important for the degradation of organic matter, and play a role in community structure. Decreasing relative abundances over time for some OTUs in the diatom-DOM cultures, as illustrated in the low growth rates obtained by the model approach, could reflect competition for the freshly supplied phytoplankton-derived material and replacement over time of initially successful members by functionally similar taxa. As discussed above, C1 compounds produced by organic matter degradation performed by some bacterial groups such as Roseobacter and Flavobacteria could have benefited other members in our cultures, such as the Colwellia, Methylophaga, OM43 and SAR11 OTUs who all share this metabolic trait (Tripp et al. 2008; Sun et al. 2011; Carini et al. 2013; Gifford et al. 2016). These interactive processes have been demonstrated in other environmental contexts (Datta et al. 2016; Krause et al. 2017). These observations highlight that interactions among taxa are driven by both the chemical nature of the available substrates and by the traits present in the microbial members, ultimately impacting community structure and organic matter decomposition. Growth rates of marine bacterial taxa We used the classical chemostat model (Monod 1950; Novick and Szilard 1950) to simulate the growth rates for abundant OTUs in a mixed community and under carbon-limited conditions. From the modeling point of view, the equilibrium in the chemostat model corresponds to the exclusion of all taxa except for the most competitive ones. However, when taxa have similar growth rates their dynamics over time in the chemostat can be slow even at stationary substrate levels (Rapaport, Dochain and Harmand 2009). As a diverse microbial community was present over the course of our experiment, we consider non-equilibrium dynamics. In this transient stage, the growth rates are not necessarily reflected in the relative abundance of taxa. Even though our model estimates are in the range of previously determined values in the oligotrophic ocean for bacterial groups at broad phylogenetic levels (see review by Kirchman 2016), we discuss our findings in relative rather than absolute terms. Interestingly, phylogenetically diverse taxa had similar growth rates in our experimental conditions. According to the model simulations, the SAR11 OTUs observed in this experimental system grew as fast as OTUs belonging to Polaribacter and Roseobacter. OTUs belonging to Colwellia and Methylophaga were among those with slowest growth rates. These observations provide a complementary view on the thus far poorly documented growth rates of bacterial groups at fine phylogenetic level (Campbell and Kirchman 2013; Lankiewicz, Cottrell and Kirchman 2016). An issue that was debated in this context is whether the growth rates of SAR11 are lower (Teira et al. 2009; Ferrera et al. 2011; Campbell and Kirchman 2013) or similar (Malmstrom et al. 2004; Laghdass et al. 2012; Salter et al. 2015; Lankiewicz, Cottrell and Kirchman 2016) as compared to other bacterial groups or the average community. Our model simulations support the idea that some SAR11 can have growth rates similar to other taxa under carbon-limited conditions, coherent with the dominance of this clade in the oligotrophic ocean. Conclusion We demonstrate here the tight link between changes in the composition of DOM and that of the bacterial community in the severely C-limited Southern Ocean. The rapid emergence in the diatom-DOM treatment of bacterial groups that were also abundant in Southern Ocean surface waters during an intense diatom bloom (Landa et al. 2016) emphasizes the relevance of our findings to help explain the ecosystem response to a supply in fresh, labile phytoplankton material. Our experimental system enabled us to simulate the growth rates of several bacterial taxa under constrained resource supply. These simulations in combination with present knowledge on the metabolic strategies of the co-occurring abundant taxa highlight important roles for ecological interplays of marine microbes in the regulation of DOM degradation. Future studies combining detailed community composition and metabolic profiling with chemical characterization of organic matter will help understand the interactions that exist between specific metabolites and microbial taxa, and how these interactions in return shape DOM processing and bacterial community structure in the ocean. SUPPLEMENTARY DATA Supplementary data are available at FEMSEC online. ACKNOWLEDGEMENTS We thank the captain and the crew of the R/V Marion Dufresne and the chief scientist B. Quéguiner for their help and support aboard. We thank L. Oriol and J. Caparros for inorganic nutrient and DOC analyses, respectively, and the trace-metal team for providing large volumes of filtered seawater. N. Fuchs (Alfred Wegener Institute, Germany) kindly provided the strain Chaetoceros debilis. We thank the handling editor and four anonymous reviewers for their constructive comments on a previous version of the manuscript. FUNDING This work was supported by the Centre National de la Recherche Scientifique (CNRS)-Institut National des Sciences de l'Univers (INSU)-Les Enveloppes Fluides et l'Environnement (LEFE)-Cycles Biogéochimiques, Environnement et Ressources (CYBER); the National Research Agency [grant ANR-10-BLAN-0614]; the Paul Emile Victor Institute and the French Ministry of Science [to M.L.]. Conflict of interest. None declared. REFERENCES Alonso C, Pernthaler J. Roseobacter and SAR11 dominate microbial glucose uptake in coastal North Sea waters. Environ Microbiol . 2006; 8: 2022– 30. Google Scholar CrossRef Search ADS PubMed  Aminot A, Kérouel R. Dosage Automatique Des Nutriments Dans Les Eaux Marines: Méthodes En Flux Continu . France: Ifremer, 2007. Ankrah NYD, Lane T, Budinoff CR et al.  . Draft genome sequence of Sulfitobacter sp. CB2047, a member of the Roseobacter clade of marine bacteria, isolated from an Emiliania huxleyi bloom. Genome Announc . 2014; 2: e01125– 14. Google Scholar PubMed  Bano N, Hollibaugh JT. Phylogenetic composition of bacterioplankton assemblages from the Arctic Ocean. Appl Environ Microbiol . 2002; 68: 505– 18. Google Scholar CrossRef Search ADS PubMed  Beier S, Rivers AR, Moran MA et al.  . The transcriptional response of prokaryotes to phytoplankton‐derived dissolved organic matter in seawater. Environ Microbiol . 2015; 17: 3466– 80. Google Scholar CrossRef Search ADS PubMed  Benner R, Strom M. A critical evaluation of the analytical blank associated with DOC measurements by high-temperature catalytic oxidation. Mar Chem . 1993; 41: 153– 60. Google Scholar CrossRef Search ADS   Bertrand EM, McCrow JP, Moustafa A et al.  . Phytoplankton–bacterial interactions mediate micronutrient colimitation at the coastal Antarctic sea ice edge. Proc Natl Acad Sci . 2015; 112: 9938– 43. Google Scholar CrossRef Search ADS PubMed  Blain S, Quéguiner B, Armand L et al.  . Effect of natural iron fertilization on carbon sequestration in the Southern Ocean. Nature . 2007; 446: 1070– 4. Google Scholar CrossRef Search ADS PubMed  Boyd PW, Watson AJ, Law CS et al.  . A mesoscale phytoplankton bloom in the polar Southern Ocean stimulated by iron fertilization. Nature . 2000; 407: 695– 702. Google Scholar CrossRef Search ADS PubMed  Brinkhoff T, Giebel HA, Simon M. Diversity, ecology, and genomics of the Roseobacter clade: a short overview. Arch Microbiol . 2008; 189: 531– 9. Google Scholar CrossRef Search ADS PubMed  Buchan A, González JM, Moran MA. Overview of the marine Roseobacter lineage. Appl Environ Microbiol . 2005; 71: 5665– 77. Google Scholar CrossRef Search ADS PubMed  Buchan A, LeCleir GR, Gulvik CA et al.  . Master recyclers: features and functions of bacteria associated with phytoplankton blooms. Nat Rev Microbiol . 2014; 12: 686– 98. Google Scholar CrossRef Search ADS PubMed  Campbell BJ, Kirchman DL. Bacterial diversity, community structure and potential growth rates along an estuarine salinity gradient. ISME J . 2013; 7: 210– 20. Google Scholar CrossRef Search ADS PubMed  Caporaso JG, Kuczynski J, Stombaugh J et al.  . QIIME allows analysis of high-throughput community sequencing data. Nat Methods . 2010; 7: 335– 6. Google Scholar CrossRef Search ADS PubMed  Carini P, Steindler L, Beszteri S et al.  . Nutrient requirements for growth of the extreme oligotroph Candidatus Pelagibacter ubique HTCC1062 on a defined medium. ISME J . 2013; 7: 592– 602. Google Scholar CrossRef Search ADS PubMed  Church MJ. Resource Control of Bacterial Dynamics in the Sea. Microbial Ecology of the Oceans . Hoboken, NJ, USA: John Wiley & Sons, Inc., 2008, 335– 82. Google Scholar CrossRef Search ADS   Church MJ, Hutchins DA, Ducklow HW. Limitation of bacterial growth by dissolved organic matter and iron in the Southern Ocean. Appl Environ Microbiol . 2000; 66: 455– 66. Google Scholar CrossRef Search ADS PubMed  Cole JJ, Findlay S, Pace ML. Bacterial production in fresh and saltwater ecosystems: a cross-system overview. Mar Ecol Prog Ser . 1988; 43: 1– 10. Google Scholar CrossRef Search ADS   Cottrell MT, Kirchman DL. Natural assemblages of marine Proteobacteria and members of the Cytophaga-Flavobacter cluster consuming low- and high-molecular-weight dissolved organic matter. Appl Environ Microbiol . 2000; 66: 1692– 7. Google Scholar CrossRef Search ADS PubMed  Datta MS, Sliwerska E, Gore J et al.  . Microbial interactions lead to rapid micro-scale successions on model marine particles. Nat Commun . 2016; 7: 11965. Google Scholar CrossRef Search ADS PubMed  Delmont TO, Hammar KM, Ducklow HW et al.  . Phaeocystis antarctica blooms strongly influence bacterial community structures in the Amundsen Sea polynya. Front Microbiol . 2014; 5: 646. Google Scholar CrossRef Search ADS PubMed  Dumont M, , Rapaport A, Harmand J et al.  . Observers for microbial ecology - How including molecular data into bioprocess modeling ? In: 2008 16th Mediterranean Conference on Control and Automation . 2008, 1381– 6., IEEE (Piscataway, NJ, USA). Durham BP, Sharma S, Luo H et al.  . Cryptic carbon and sulfur cycling between surface ocean plankton. Proc Natl Acad Sci . 2015; 112: 453– 7. Google Scholar CrossRef Search ADS PubMed  Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res . 2004; 32: 1792– 7. Google Scholar CrossRef Search ADS PubMed  Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics . 2010; 26: 2460– 1. Google Scholar CrossRef Search ADS PubMed  Fernández-Gómez B, Richter M, Schüler M et al.  . Ecology of marine Bacteroidetes: a comparative genomics approach. ISME J . 2013; 7: 1026– 37. Google Scholar CrossRef Search ADS PubMed  Ferrera I, Gasol JM, Sebastián M et al.  . Comparison of growth rates of aerobic anoxygenic phototrophic bacteria and other bacterioplankton groups in coastal Mediterranean waters. Appl Environ Microbiol . 2011; 77: 7451– 8. Google Scholar CrossRef Search ADS PubMed  Fourquez M, Beier S, Jongmans E et al.  . Uptake of leucine, chitin, and iron by prokaryotic groups during spring phytoplankton blooms induced by natural iron fertilization off Kerguelen Island (Southern Ocean). Front Mar Sci . 2016; 3: 256. Google Scholar CrossRef Search ADS   Fuhrman JA, Hewson I, Schwalbach MS et al.  . Annually reoccurring bacterial communities are predictable from ocean conditions. Proc Natl Acad Sci . 2006; 103: 13104– 9. Google Scholar CrossRef Search ADS PubMed  Fukuda R, Ogawa H, Nagata T et al.  . Direct determination of carbon and nitrogen contents of natural bacterial assemblages in marine environments. Appl Environ Microbiol . 1998; 64: 3352– 8. Google Scholar PubMed  Gentile G, Giuliano L, D'Auria G et al.  . Study of bacterial communities in Antarctic coastal waters by a combination of 16S rRNA and 16S rDNA sequencing. Environ Microbiol . 2006; 8: 2150– 61. Google Scholar CrossRef Search ADS PubMed  Ghiglione JF, Murray AE. Pronounced summer to winter differences and higher wintertime richness in coastal Antarctic marine bacterioplankton. Environ Microbiol . 2012; 14: 617– 29. Google Scholar CrossRef Search ADS PubMed  Gifford SM, Becker JW, Sosa OA et al.  . Quantitative transcriptomics reveals the growth- and nutrient-dependent response of a streamlined marine methylotroph to methanol and naturally occurring dissolved organic matter. MBio . 2016; 7: e01279– 16. Google Scholar CrossRef Search ADS PubMed  Gilbert Ja, Steele Ja, Caporaso JG et al.  . Defining seasonal marine microbial community dynamics. ISME J . 2012; 6: 298– 308. Google Scholar CrossRef Search ADS PubMed  Giovannoni S. SAR11 bacteria: the most abundant plankton in the oceans, Ann Rev Mar Sci . 2017; 9: 231– 55. Goldman JC, Caron DA, Dennett MR. Regulation of gross growth efficiency and ammonium regeneration in bacteria by substrate C: N ratio. Limnol Ocean . 1987; 32: 1239– 52. Google Scholar CrossRef Search ADS   Gómez‐Pereira PR, Schüler M, Fuchs BM et al.  . Genomic content of uncultured Bacteroidetes from contrasting oceanic provinces in the North Atlantic Ocean. Environ Microbiol . 2012; 14: 52– 66. Google Scholar CrossRef Search ADS PubMed  González JM, Fernández-Gómez B, Fernàndez-Guerra A et al.  . Genome analysis of the proteorhodopsin-containing marine bacterium Polaribacter sp. MED152 (Flavobacteria). Proc Natl Acad Sci . 2008; 105: 8724– 9. Google Scholar CrossRef Search ADS PubMed  González JM, Simó R, Massana R et al.  . Bacterial community structure associated with a dimethylsulfoniopropionate-producing North Atlantic algal bloom. Appl Environ Microbiol . 2000; 66: 4237– 46. Google Scholar CrossRef Search ADS PubMed  Hansell DA. Recalcitrant dissolved organic carbon fractions. Ann Rev Mar Sci . 2013; 5: 421– 45. Google Scholar CrossRef Search ADS PubMed  Hertkorn N, Ruecker C, Meringer M et al.  . High-precision frequency measurements: indispensable tools at the core of the molecular-level analysis of complex systems. Anal Bioanal Chem . 2007; 389: 1311– 27. Google Scholar CrossRef Search ADS PubMed  Hu P, Dubinsky EA, Probst AJ et al.  . Simulation of Deepwater Horizon oil plume reveals substrate specialization within a complex community of hydrocarbon degraders. Proc Natl Acad Sci U S A . 2017; 114: 7432– 7. Google Scholar CrossRef Search ADS PubMed  Kerkhof LJ, Voytek MA, Sherrell RM et al.  . Variability in bacterial community structure during upwelling in the coastal ocean. Hydrobiologia . 1999; 401: 139– 48. Google Scholar CrossRef Search ADS   Kim H, Ducklow HW. A Decadal (2002–2014) Analysis for dynamics of heterotrophic bacteria in an Antarctic coastal ecosystem: variability and physical and biogeochemical forcings. Front Mar Sci . 2016; 3: 214. Google Scholar CrossRef Search ADS   Kim J-G, Park S-J, Quan Z-X et al.  . Unveiling abundance and distribution of planktonic bacteria and archaea in a polynya in Amundsen Sea, Antarctica: prokaryotic communities in Antarctic polynya. Environ Microbiol . 2014; 16: 1566– 78. Google Scholar CrossRef Search ADS PubMed  Kimes NE, López-Pérez M, Ausó E et al.  . RNA sequencing provides evidence for functional variability between naturally co-existing Alteromonas macleodii lineages. BMC Genomics . 2014; 15: 938. Google Scholar CrossRef Search ADS PubMed  Kirchman DL. Growth rates of microbes in the oceans, Ann Rev Mar Sci . 2016; 8: 285– 309. Kirchman DL, Meon B, Cottrell MT et al.  . Carbon versus iron limitation of bacterial growth in the California upwelling regime. Limnol Oceanogr . 2000; 45: 1681– 8. Google Scholar CrossRef Search ADS   Kleindienst S, Seidel M, Ziervogel K et al.  . Chemical dispersants can suppress the activity of natural oil-degrading microorganisms. Proc Natl Acad Sci . 2015; 112: 14900– 5. Google Scholar CrossRef Search ADS PubMed  Krause SMB, Johnson T, Karunaratne YS et al.  . Lanthanide-dependent cross-feeding of methane-derived carbon is linked by microbial community interactions. Proc Natl Acad Sci . 2017; 114: 358– 63. Google Scholar CrossRef Search ADS PubMed  Kumar S, Stecher G, Tamura K. MEGA7: molecular evolutionary genetics analysis version 7.0 for bigger datasets. Mol Biol Evol . 2016; 33: 1870– 4. Google Scholar CrossRef Search ADS PubMed  Laghdass M, Catala P, Caparros J et al.  . High contribution of SAR11 to microbial activity in the North West Mediterranean Sea. Microb Ecol . 2012; 63: 324– 33. Google Scholar CrossRef Search ADS PubMed  Landa M, Blain S, Christaki U et al.  . Shifts in bacterial community composition associated with increased carbon cycling in a mosaic of phytoplankton blooms. ISME J . 2016; 10: 39– 50. Google Scholar CrossRef Search ADS PubMed  Landa M, Cottrell MT, Kirchman DL et al.  . Changes in bacterial diversity in response to dissolved organic matter supply in a continuous culture experiment. Aquat Microb Ecol . 2013; 69: 157– 68. Google Scholar CrossRef Search ADS   Landa M, Cottrell MT, Kirchman DL et al.  . Phylogenetic and structural response of heterotrophic bacteria to dissolved organic matter of different chemical composition in a continuous culture study. Environ Microbiol . 2014; 16: 1668– 81. Google Scholar CrossRef Search ADS PubMed  Langenheder S, Lindström ES, Tranvik LJ. Weak coupling between community composition and functioning of aquatic bacteria. Limnol Oceanogr . 2005; 50: 957– 67. Google Scholar CrossRef Search ADS   Langenheder S, Lindström ES, Tranvik LJ. Structure and function of bacterial communities emerging from different sources under identical conditions. Appl Environ Microbiol . 2006; 72: 212– 20. Google Scholar CrossRef Search ADS PubMed  Lankiewicz TS, Cottrell MT, Kirchman DL. Growth rates and rRNA content of four marine bacteria in pure cultures and in the Delaware estuary. ISME J . 2016; 10: 823– 32. Google Scholar CrossRef Search ADS PubMed  Lauro FM, McDougald D, Thomas T et al.  . The genomic basis of trophic strategy in marine bacteria. Proc Natl Acad Sci . 2009; 106: 15527– 33. Google Scholar CrossRef Search ADS PubMed  Li TD, Doronina NV, Ivanova EG et al.  . Vitamin B12-independent strains of Methylophaga marina isolated from Red Sea algae. Microbiology . 2007; 76: 75– 81. Google Scholar CrossRef Search ADS   Luo H, Csűros M, Hughes AL et al.  . Evolution of divergent life history strategies in marine Alphaproteobacteria. MBio . 2013; 4: e00373– 13. Google Scholar CrossRef Search ADS PubMed  Luria CM, Amaral-Zettler LA, Ducklow HW et al.  . Seasonal succession of free-living bacterial communities in coastal waters of the Western Antarctic Peninsula. Front Microbiol . 2016; 7: 1731. Google Scholar CrossRef Search ADS PubMed  Luria CM, Amaral-Zettler LA, Ducklow HW et al.  . Seasonal shifts in bacterial community responses to phytoplankton-derived dissolved organic matter in the Western Antarctic Peninsula. Front Microbiol . 2017; 8: 2117. Google Scholar CrossRef Search ADS PubMed  Malmstrom RR, Cottrell MT, Elifantz H et al.  . Biomass production and assimilation of dissolved organic matter by SAR11 bacteria in the Northwest Atlantic Ocean. Appl Environ Microbiol . 2005; 71: 2979– 86. Google Scholar CrossRef Search ADS PubMed  Malmstrom RR, Kiene RP, Cottrell MT et al.  . Contribution of SAR11 bacteria to dissolved dimethylsulfoniopropionate and amino acid uptake in the North Atlantic Ocean. Appl Environ Microbiol . 2004; 70: 4129– 35. Google Scholar CrossRef Search ADS PubMed  Martiny JBH, Bohannan BJM, Brown JH et al.  . Microbial biogeography: putting microorganisms on the map. Nat Rev Microbiol . 2006; 4: 102– 12. Google Scholar CrossRef Search ADS PubMed  Methé BA, Nelson KE, Deming JW et al.  . The psychrophilic lifestyle as revealed by the genome sequence of Colwellia psychrerythraea 34H through genomic and proteomic analyses. Proc Natl Acad Sci U S A . 2005; 102: 10913– 8. Google Scholar CrossRef Search ADS PubMed  Monod J. La technique de culture continue: Théorie et applications. Ann Inst Pasteur . 1950; 79: 390– 410. Moran MA, Belas R, Schell MA et al.  . Ecological genomics of marine roseobacters. Appl Environ Microbiol . 2007; 73: 4559– 69. Google Scholar CrossRef Search ADS PubMed  Morris JJ, Papoulis SE, Lenski RE. Coexistence of evolving bacteria stabilized by a shared Black Queen function. Evolution (N Y) . 2014; 68: 2960– 71. Morris RM, Rappé MS, Connon SA et al.  . SAR11 clade dominates ocean surface bacterioplankton communities. Nature . 2002; 420: 806– 10. Google Scholar CrossRef Search ADS PubMed  Mou X, Sun S, Edwards RA et al.  . Bacterial carbon processing by generalist species in the coastal ocean. Nature . 2008; 451: 708– 11. Google Scholar CrossRef Search ADS PubMed  Needham DM, Fuhrman JA. Pronounced daily succession of phytoplankton, archaea and bacteria following a spring bloom. Nat Microbiol . 2016; 1: 16005. Google Scholar CrossRef Search ADS PubMed  Nelson JD, Boehme SE, Reimers CE et al.  . Temporal patterns of microbial community structure in the Mid-Atlantic Bight. FEMS Microbiol Ecol . 2008; 65: 484– 93. Google Scholar CrossRef Search ADS PubMed  Neufeld JD, Schäfer H, Cox MJ et al.  . Stable-isotope probing implicates Methylophaga spp and novel Gammaproteobacteria in marine methanol and methylamine metabolism. ISME J . 2007; 1: 480– 91. Google Scholar CrossRef Search ADS PubMed  Neumann AM, Balmonte JP, Berger M et al.  . Different utilization of alginate and other algal polysaccharides by marine Alteromonas macleodii ecotypes. Environ Microbiol . 2015; 17: 3857– 68. Google Scholar CrossRef Search ADS PubMed  Newton RJ, Griffin LE, Bowles KM et al.  . Genome characteristics of a generalist marine bacterial lineage. ISME J . 2010; 4: 784– 98. Google Scholar CrossRef Search ADS PubMed  Novick A, Szilard L. Description of the chemostat. Science . 1950; 112: 715– 6. Google Scholar CrossRef Search ADS PubMed  Obernosterer I, Christaki U, Lefèvre D et al.  . Rapid bacterial mineralization of organic carbon produced during a phytoplankton bloom induced by natural iron fertilization in the Southern Ocean. Deep Sea Res Part II Top Stud Oceanogr . 2008; 55: 777– 89. Google Scholar CrossRef Search ADS   Obernosterer I, Fourquez M, Blain S. Fe and C co-limitation of heterotrophic bacteria in the naturally fertilized region off the Kerguelen Islands. Biogeosciences . 2015; 12: 1983– 92. Google Scholar CrossRef Search ADS   Pedler BE, Aluwihare LI, Azam F. Single bacterial strain capable of significant contribution to carbon cycling in the surface ocean. Proc Natl Acad Sci . 2014; 111: 7202– 7. Google Scholar CrossRef Search ADS PubMed  Pinhassi J, Havskum H, Peters F et al.  . Changes in bacterioplankton composition under different phytoplankton regimens, Appl Environ Microbiol . 2004; 70: 6753– 66. Quéguiner B. Iron fertilization and the structure of planktonic communities in high nutrient regions of the Southern Ocean. Deep Res Part II Top Stud Oceanogr . 2013; 90: 43– 54. Google Scholar CrossRef Search ADS   Quince C, Lanzen A, Davenport RJ et al.  . Removing noise from pyrosequenced amplicons. BMC Bioinformatics . 2011; 12: 38. Google Scholar CrossRef Search ADS PubMed  Rapaport A, Dochain D, Harmand J. Long run coexistence in the chemostat with multiple species. J Theor Biol . 2009; 257: 252– 9. Google Scholar CrossRef Search ADS PubMed  Redmond MC, Valentine DL. Natural gas and temperature structured a microbial community response to the Deepwater Horizon oil spill. Proc Natl Acad Sci U S A . 2012; 109: 20292– 7. Google Scholar CrossRef Search ADS PubMed  Riemann L, Steward GF, Azam F. Dynamics of bacterial community composition and activity during a mesocosm diatom bloom. Appl Environ Microbiol . 2000; 66: 578– 87. Google Scholar CrossRef Search ADS PubMed  Rink B, Seeberger S, Martens T et al.  . Effects of phytoplankton bloom in a coastal ecosystem on the composition of bacterial communities. Aquat Microb Ecol . 2007; 48: 47– 60. Google Scholar CrossRef Search ADS   Salter I, Galand PE, Fagervold SK et al.  . Seasonal dynamics of active SAR11 ecotypes in the oligotrophic Northwest Mediterranean Sea. ISME J . 2015; 9: 347– 60. Google Scholar CrossRef Search ADS PubMed  Sarmento H, Morana C, Gasol JM. Bacterioplankton niche partitioning in the use of phytoplankton-derived dissolved organic carbon: quantity is more important than quality. ISME J . 2016; 10: 2582– 92. Google Scholar CrossRef Search ADS PubMed  Satinsky BM, Crump BC, Smith CB et al.  . Microspatial gene expression patterns in the Amazon River Plume. Proc Natl Acad Sci . 2014; 111: 11085– 90. Google Scholar CrossRef Search ADS PubMed  Schloss PD, Westcott SL, Ryabin T et al.  . Introducing mothur: Open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol . 2009; 75: 7537– 41. Google Scholar CrossRef Search ADS PubMed  Selje N, Simon M, Brinkhoff T. A newly discovered Roseobacter cluster in temperate and polar oceans. Nature . 2004; 427: 445– 8. Google Scholar CrossRef Search ADS PubMed  Simon M, Billerbeck S, Kessler D et al.  . Bacterioplankton communities in the Southern Ocean: Composition and growth response to various substrate regimes. Aquat Microb Ecol . 2012; 68: 13– 28. Google Scholar CrossRef Search ADS   Sjöstedt J, Pontarp M, Tinta T et al.  . Reduced diversity and changed bacterioplankton community composition do not affect utilization of dissolved organic matter in the Adriatic Sea. Aquat Microb Ecol . 2013; 71: 15– 24. Google Scholar CrossRef Search ADS   Sogin ML, Morrison HG, Huber JA et al.  . Microbial diversity in the deep sea and the underexplored “rare biosphere”. Proc Natl Acad Sci . 2006; 103: 12115– 20. Google Scholar CrossRef Search ADS PubMed  Sowell SM, Wilhelm LJ, Norbeck AD et al.  . Transport functions dominate the SAR11 metaproteome at low-nutrient extremes in the Sargasso Sea. ISME J . 2009; 3: 93– 105. Google Scholar CrossRef Search ADS PubMed  Speeckaert G, Borges AV, Champenois W et al.  . Annual cycle of dimethylsulfoniopropionate (DMSP) and dimethylsulfoxide (DMSO) related to phytoplankton succession in the Southern North Sea. Sci Total Environ . 2018; 622–623: 362– 72. Google Scholar CrossRef Search ADS PubMed  Straza TRA, Ducklow HW, Murray AE et al.  . Abundance and single-cell activity of bacterial groups in Antarctic coastal waters. Limnol Oceanogr . 2010; 55: 2526– 36. Google Scholar CrossRef Search ADS   Sun J, Steindler L, Thrash JC et al.  . One carbon metabolism in SAR11 pelagic marine bacteria. PLoS One . 2011; 6: e23973. Google Scholar CrossRef Search ADS PubMed  Sunagawa S, Coelho LP, Chaffron S et al.  . Structure and function of the global ocean microbiome. Science . 2015; 348: 1261359. Google Scholar CrossRef Search ADS PubMed  Tada Y, Taniguchi A, Sato-Takabe Y et al.  . Growth and succession patterns of major phylogenetic groups of marine bacteria during a mesocosm diatom bloom. J Oceanogr . 2012; 68: 509– 19. Google Scholar CrossRef Search ADS   Teeling H, Fuchs BM, Becher D et al.  . Substrate-controlled succession of marine bacterioplankton populations induced by a phytoplankton bloom. Science . 2012; 336: 608– 11. Google Scholar CrossRef Search ADS PubMed  Teira E, Martínez‐García S, Lønborg C et al.  . Growth rates of different phylogenetic bacterioplankton groups in a coastal upwelling system. Environ Microbiol Rep . 2009; 1: 545– 54. Google Scholar CrossRef Search ADS PubMed  Thiele S, Fuchs BM, Ramaiah N et al.  . Microbial community response during the iron fertilization experiment LOHAFEX. Appl Environ Microbiol . 2012; 78: 8803– 12. Google Scholar CrossRef Search ADS PubMed  Thompson LR, Field C, Romanuk T et al.  . Patterns of ecological specialization among microbial populations in the Red Sea and diverse oligotrophic marine environments. Ecol Evol . 2013; 3: 1780– 97. Google Scholar CrossRef Search ADS PubMed  Tremblay L, Caparros J, Leblanc K et al.  . Origin and fate of particulate and dissolved organic matter in a naturally iron-fertilized region of the Southern Ocean. Biogeosciences . 2015; 12: 607– 21. Google Scholar CrossRef Search ADS   Tripp HJ. The unique metabolism of SAR11 aquatic bacteria. J Microbiol . 2013; 51: 147– 53. Google Scholar CrossRef Search ADS PubMed  Tripp HJ, Kitner JB, Schwalbach MS et al.  . SAR11 marine bacteria require exogenous reduced sulphur for growth. Nature . 2008; 452: 741– 4. Google Scholar CrossRef Search ADS PubMed  Wang Q, Garrity GM, Tiedje JM et al.  . Naïve bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ Microbiol . 2007; 73: 5261– 7. Google Scholar CrossRef Search ADS PubMed  Wawrik B, Kerkhof L, Kukor J et al.  . Effect of different carbon sources on community composition of bacterial enrichments from soil effect of different carbon sources on community composition of bacterial enrichments from soil. Appl Environ Microbiol . 2005; 71: 6776– 83. Google Scholar CrossRef Search ADS PubMed  West NJ, Obernosterer I, Zemb O et al.  . Major differences of bacterial diversity and activity inside and outside of a natural iron-fertilized phytoplankton bloom in the Southern Ocean. Environ Microbiol . 2008; 10: 738– 56. Google Scholar CrossRef Search ADS PubMed  Wietz M, Wemheuer B, Simon H et al.  . Bacterial community dynamics during polysaccharide degradation at contrasting sites in the Southern and Atlantic Oceans. Environ Microbiol . 2015; 17: 3822– 31. Google Scholar CrossRef Search ADS PubMed  Wilkins D, Yau S, Williams TJ et al.  . Key microbial drivers in Antarctic aquatic environments. FEMS Microbiol Rev . 2013; 37: 303– 35. Google Scholar CrossRef Search ADS PubMed  Williams TJ, Wilkins D, Long E et al.  . The role of planktonic Flavobacteria in processing algal organic matter in coastal East Antarctica revealed using metagenomics and metaproteomics. Environ Microbiol . 2013; 15: 1302– 17. Google Scholar CrossRef Search ADS PubMed  Wolcott RD, Gontcharova V, Sun Y et al.  . Evaluation of the bacterial diversity among and within individual venous leg ulcers using bacterial tag-encoded FLX and Titanium amplicon pyrosequencing and metagenomic approaches. BMC Microbiol . 2009; 9: 226. Google Scholar CrossRef Search ADS PubMed  Xing P, Hahnke RL, Unfried F et al.  . Niches of two polysaccharide-degrading Polaribacter isolates from the North Sea during a spring diatom bloom. ISME J . 2015; 9: 1410– 22. Google Scholar CrossRef Search ADS PubMed  © FEMS 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

FEMS Microbiology Ecology – Oxford University Press

Published: Apr 1, 2018

There are no references for this article.