Patterns in marine microbial community structure

Patterns in marine microbial community structure (Patrones de estructura de las comunidades de microorganismos marinos ) Thomas Lefort Tesis Doctoral presentada por Thomas Lefort para obtener el grado de Doctor por la Universidad de las Palmas de Gran Canaria, Departamento de Biología, Programa en Oceanografía, bieno (2006-2008) Director: Dr. Josep M. Gasol Universidad de Las Palmas de Gran Canaria Institut de Ciències del Mar (ICM-CSIC) En Barcelona, a de de El Doctorando El Director Thomas Lefort Josep M. Gasol à mes parents à Juliette 6 Contents 7 CONTENTS List of abbreviations Summary/Resumen/Resum/Résumé 00 General Introduction Aims and Outline of the Thesis 01 Chapter I Direct determination of carbon conversion factors for ecologically relevant small photosynthetic eukaryotes 02 Chapter II Short-time scale coupling of picoplankton community structure and heterotrophic activity in winter coastal NW Mediterranean Sea 03 Chapter III Patterns in picoplankton community structure: Multi-scale spatial and temporal variability in the NW Mediterranean Sea during late summer 04 Chapter IV Patterns in marine bacterial group distribution, as measured by FISH, in relation to chlorophyll, temperature and salinity 05 Chapter V Synthesis of results and general discussion 06 Thesis summary (Spanish version) References Acknowledgments 17 30 35 63 97 139 179 197 247 260 11 9 8 Abbreviations 9 List of the most common abbreviations and acronyms used in this thesis: Alpha: Alphaproteobacteria BBMO: Blanes Bay Microbial Observatory Beta: Betaproteobacteria BHP: Bacterial Heterotrophic Production BCS: Bacterial Community Structure CARD-FISH: Catalyzed Reporter DepositionFluorescence In Situ Hybridization CF: Conversion Factor CHLA: chlorophyll a CTC: 5-cyano-2,3-ditolyl tetrazolium chloride CTD: Conductivity, Temperature and Depth sensors CV: Coefficient of Variation C/V: Ratio of carbon to volum DCM: Deep Chlorophyll Maximum DOC: Dissolved Organic Carbon EUB: Eubacteria FSC: Forward Scatter Gamma: Gammaproteobacteria HNF: Heterotrophic Nanoflagellates HPLC: High Performance Liquid Chromatography OTU: Operational Taxonomic Unit PCS: Picoplankton Community Structure pPeuk: photosynthetic Picoeukaryotes POC: Particulate Organic Carbon PON: Particulate organic Nitrogen Pro: Prochlorococcus Ros: Rhodobacteraceae SSC: Side Scatter Syn: Synechococcus Temp: Temperature Summary 11 Understanding the distribution of the different picoplankton groups represents a central tenet of marine microbial ecology. Centering our study on the three major groups constituting the bulk picoplankton community (size 0.2-3 mm), we sought to analyze the distribution of autotrophic bacteria (Synechococcus and Prochlorococcus), photosynthetic Picoeukaryotes pPeuk, and heterotrophic bacteria. For that objective, two different strategies were used, the first one was based on flow cytometry for determining ataxonomic patterns in picoplankton distribution, and the second a comparative analysis approach for identifying broad patterns in bacterial phylogenetic community structure. Given that conversion factors (CFs) were necessary to translate group cell abundance into carbon biomass, but that large discrepancies for CF values of pPeuk had been reported in the literature, we first (re-) evaluated the CF for small phototrophic picoeukaryotes (<5 mm). On one hand, as the set of cultures of Peuk used for that purpose were maintained in non-axenic conditions, we compared two different methods for correcting errors in biomass estimation due to presence of bacteria. Secondly, a relatively higher CF value for pPeuk than those previously reported to date was found, with implications on the role generally attributed to pPeuk in the carbon cycling and other ecosystem processes. Applying this CF, we could identify patterns of variability in picoplankton group distribution at different spatio-temporal scales during winter in a NW Mediterranean coastal station and during a cruise performed in summer from coast to offshore off the Catalan coast. By focusing on the variability at the short time scale, our work showed not only evidences of coupling between picophytoplankton variability and the single-cell bacterial activities but also highlighted how a relatively small variation in meteorology changed considerably the structure of the microbial community. Different trends of variability were observed between the different picoplankton groups, pPeuk cell numbers exhibiting the highest spatio-temporal variability, and bacterial abundance the lowest. Opposite patterns between picoplankton community structure and chlorophyll a levels were observed not only spatially, but also at both the short-term and large temporal scale, suggesting that picoplankton group distribution are useful indicators of the ecosystem state. Finally, we assessed the biogeography of the bacterial phylogenetic groups along a continuum of environmental parameters such as chlorophyll a, temperature and salinity, and identified different patterns in bacterial community structure as related to phytoplankton biomass among coastal and open ocean ecosystems, suggesting unequal metabolic aptitudes of the different bacterial groups for utilizing algal-derived DOC. Initial approaches 18 Introduction 19 biogeography and microbial ecology: initial approaches Understanding the patterns and processes involved in the distribution of life forms constitutes the central tenet of biogeography. Rooting its origins with Carl Linnaeus’s (1707-1778) taxonomical classification that was based on both the differences and similarities shared among types of plants and animals (Species Plantarum 1753), biogeography began with the Comte de Buffon’s theory (1707-1788) who stated that similar but geographically separated environments presented distinct plants and animals (Histoire naturelle, générale et particulière, 1749-1788). In the late 1800’s, contradictory theories emerged, the earth-history explanations of the endemism of species distributions proposed by Charles Lyell’s (1797-1875) contrasted with the global dispersal theory proposed by the botanist Alphonse de Condolle (1806-1893) who formulated that “the lower the organization of the body is, the more generally it is distributed” implying that microbial life was everywhere and would proliferate under appropriate conditions. This latter theory was quickly considered as one “fundamental law” of biogeography, fueled by the emergence of microbiology. The study of patterns in marine microbe’s distribution has been constrained by the technological limits in observing and identifying tiny organisms. At first, measurement of microbial abundance was based on pure culture isolates and optical microscopy (Certes 1884; Zobell 1946) that greatly underestimated bacterial abundance by several orders of magnitude (Jannash and Jones 1959). The developments of epifluorescence microscopy during the 70´s (Hobbie et al. 1977; Zimmerman 1977; Porter and Feig 1980), as well as the introduction of automated cell counting by Coulter Counter and Flow Cytometry (Sheldon and Parsons 1967; Sheldon 1978) enabled the detection and enumeration of different microbial groups in field samples (Olson et al. 1985). Whether distinct distribution patterns in microbes can be identified depends on the criteria used for the definition and classification of the different microbial groups. These criteria range from size categories, function and activity categories to more phylogenetic-based categorization. Identification of microbial groups by flow cytometry Based on the detection by scatter and fluorescence sensors of fluorescently pigmented cells passing through a laser beam, flow cytometry can be used routinely (Marie and Partensky 2006) to identify and quantify cells within at least three ecological distinct groups: eukaryotic algae, cyanobacteria (Synechococcus and Prochlorococcus) and heterotrophic bacteria (Olson et al. 1993; Gasol and del Giorgio 2000). Chlorophyll a pigment concentration constitutes the principal factor that is used to discriminate phytoplankton and photosynthetic picoeukaryotic cells from other particles (Yentsch and Yentsch 1979; Li et al. 1995). Other photo-pigments, Initial approaches 20 such as phycoerythrin (emitting from 550 to 590 nm once excited by blue light), are used to distinguish between most Synechococcus (Johnson and Sieburth 1979; Waterbury et al. 1979; Wood et al. 1985) and Prochlorococcus species (Chisholm et al. 1988) (e.g. Figure 1). Nonphotosynthetic bacteria are too small for being detected by size alone. And even Prochlorococcus autofluorescence in oligotrophic ecosystems is often too low for being accurately detected by the cytometer sensors. Then, DNA-based fluorescent dyes can be used to stain the sample and this allows enumeration of heterotrophic bacteria and viruses (Li et al. 1995; Marie et al. 1999) and to distinguish Prochlorococcus from nonphotosynthetic bacteria (Monger and Landry 1993). Figure 1. Cytograms showing the main picoplankton groups discriminated by their different scatter and pigments or nucleic acid dye fluorescence. (Syn for Synechococcus, Pro for Prochlorococcus, Peuk for photosynthetic Picoeukaryotes, B for reference 1 µm Beads, H and L for bacteria with High and Low Nucleic Acid content). Images: J.M. Gasol. Introduction 21 Heterotrophic bacteria in the microbial food web Accounting for a total of 1029 cells at the global ocean scale, heterotrophic bacteria are now considered as the most abundant living organisms on Earth (typically found at around 109 cells l-1) (Whitman et al. 1998) and play a key role in the oceanic carbon cycling through the microbial loop (Azam et al., 1983). Heterotrophic bacteria in aquatic ecosystems were soon recognized for their role in the decomposition of organic material and the remineralization of inorganic nutrients, but Pomeroy in 1974 showed that their functions could be more diverse than it was previously thought. Indeed, Hagström et al. (1979) and Fuhrman and Azam (1980) showed that an important proportion of bacteria were not dormant but actively growing through the utilization of oceanic organic matter, bacterial heterotrophic production later estimated as accounting between 20-30% of primary production on average (Cole et al. 1988). The microbial food web, summarized in Figure 2, assumes that heterotrophic bacteria are loosely coupled to phytoplankton (Bird and Kalff 1984; Cole et al. 1988; Fuhrman and Azam 1980) and balanced not only by grazing from ciliates and flagellates (Fenchel 1982; Sherr and Sherr 1984; 1987) but also by crustacean and rotifers (Pedrós-Alió and Brock 1983). Figure 2. Schematic Illustration of the marine microbial food web and some of the biogeochemical fluxes involved (from Azam and Malfatti 2007) Initial approaches 22 The link between autotrophs and heterotrophs assumes that bacteria use the dissolved organic matter produced by phytoplankton and grazers to support growth as bacterial secondary production (Nagata et al. 2000; Morán et al. 2002). Studies based on large data set comparisons have demonstrated such a link and described a positive relationship with a log-log slope <1 between heterotrophic and autotrophic biomass (for instance: Gasol et al. 1997), showing that bacterial biomass vary less than chlorophyll a along a gradient of trophy and suggesting pronounced heterotrophy in low production ecosystems and a less relevant role of the microbial food web in the most eutrophic sites. Single cell level activity Whether all bacterial cells contribute equally to the bulk metabolic activity of bacterioplankton or only a few key players are involved, has long been a central question of microbial ecology. Complex aquatic bacterial assemblages present a variety of different metabolic states. Allocating a wide range of different activities to distinct groups of cells by the use of single cell techniques, bulk bacterial activity can be represented as a continuum of different physiological states (Smith and del Giorgio 2003; del Giorgio and Gasol 2008). Microscopy was first used for measuring the division rate of bacteria (Hagström et al. 1979) and microautoradiography is still extensively applied for measuring the substrate uptake activity in aquatic ecosystems (Parsons and Strickland 1961; Wright and Hobbie 1965; Hoppe 1976). To date, the application of specific molecular probes in combination with flow cytometry appears as a suitable set of tools for characterizing rapidly and with statistical significance the physiological community structure (del Giorgio and Gasol 2008). Physiological probes can be used for detecting a variety of cellular states, either measuring cellular death by screening the membrane-damaged cells (e.g. Grégori et al. 2001, Chapter II), or quantifying different metabolic processes such as the percentage of actively respiring cells estimated by the quantification of CTC positive cells (e.g. Gasol and Arístegui 2007) (e.g. Figure 3). If the availability of organic carbon represents perhaps the most important factor influencing bulk heterotrophic bacterial activity in marine ecosystems (Azam 1998; del Giorgio and Gasol 2008; Church 2008), the factors regulating the different single cell activities at different spatio-temporal scales still remain poorly studied. Introduction 23 Picophytoplankton in the microbial food web While bacterioplankton community processes have been modeled as being solely heterotrophic, and early oceanographic models understated the importance of oxygenic photoautotrophic picoplankton, Richardson and Jackson (2007) noted that picophytoplankton constitutes also an important source of organic carbon for large zooplankton and are also contributing to the flux of particles sinking to the deep ocean. By being of a similar size range than heterotrophic bacteria, are subject to similar (but not identical) loss and growth processes. They are consumed by protists, particularly nano-sized protists (Caron et al. 1991; Dolan and Simek 1998; Guillou et al. 2001), are subject to viral lyses (Proctor and Fuhrman 1991) and compete for nutrients (Hall and Vincent 1990; Li 1994; Vaulot et al. 1996). Quantitative cell counts and flow cytometry have revealed autotrophic picoplankton as ubiquitous players dominating photosynthetic activities in open-ocean gyres (e.g. Partensky et al. 1999). Ubiquitously found at around 105 - 106 cells ml-1 in a variety of ecosystems, Synechococcus most likely dominate picophytoplankton in nutrient rich well mixed waters, (Partensky et al. 1999) while Prochlorococcus prevalence is observed between 40ºN and 40ºS latitude, peaking in well stratified and nutrient poor deep waters as well as present Figure 3. Cytograms showing different physiological groups of bacteria discriminated by the presence of the fluorogenic tetrazolium dye CTC indicator of actively respiring cells (A) and by the action of the cellpermanent nucleic acid stain SybrGreen I and the cell-impermeant propidium iodine (B), C for CTC positive cells, B for reference 1 µm Beads, L for Live cells and D for Dead cells as labelled by the NADS protocol. Image: J.M. Gasol. Initial approaches 24 in deeply mixed and nutrient rich spring or winter waters (Campbell et al. 1997; Durand et al. 2001; Partensky et al. 1999) where they appear to contribute to up to 30% of the biomass in the oligotrophic North Pacific ecosystem (Campbell et al. 1994). Photosynthetic Picoeukaryotes and Synechococcus abundances have been shown to covary in a variety of ecosystems (Campbell et al. 1998; Shalapyonok et al. 2001; Durand et al. 2001; Worden et al. 2004). In comparison with Cyanobacteria, less importance had been given to Picoeukaryotes due to their lower abundances. However, the calculation of their contribution in terms of biomass and primary production reveal a much higher importance in the oceanic carbon cycling that it was expected before (Li et al. 1995; Worden et al. 2004). Carbon conversion factors for biomass estimation Underestimation of the Picoeukaryotes’ relevance compared to other picoplankton members might have stemmed from the lack of well-defined carbon conversion factors. To convert the different microbial group abundances (as estimated for example from flow cytometry) into carbon biomass, cell size and cellular carbon content are two necessary parameters. To date, estimations of picoeukaryote carbon content have been mostly made by converting cell size or cell volume into carbon using empirically derived linear relationships (Mullin et al. 1966; Strathmann 1967) established from the study of cultures of larger algal species. However, in comparison with large phytoplankton cells, small eukaryotes (e.g. Figure 4) have relatively smaller vacuoles (similarly than small bacterial cells contain less cellular water (Simon and Azam 1989) and relatively higher cellular carbon content per unit of volume, as revealed by negative relationships between the cellular carbon content per unit of volume (fgC mm-3) with increasing cell volume (Verity et al. 1992). Figure 4. Image of Micromonas: T. Deerinck, M. Terada, J. Obiyashi, M. Ellisman (all National Center for Microscopy and Imaging Research) and A. Z. Worden (MBARI). Introduction 25 When no direct cell size measurements are available, the only possible course of action for estimating the cellular carbon content is to assume an average cell size. Within the picoplanktonic size category (0.2 - 3 mm), the cell size averages of Synechococcus and Prochlorococcus appear as relatively stable when compared to Picoeukaryotes (Durand et al. 2001; Worden et al. 2004). Prochlorococcus are the smallest photosynthetic prokaryotes with a cell diameter estimated at 0.7 mm (Shalapyonok et al. 2001; Worden et al. 2004; Durand et al. 2001), closely followed by Synechococcus with a slightly larger size average of 0.87 mm (Worden et al. 2004) and ranging from 0.5 to 2 mm (Murphy and Haugen 1985). When converted in terms of carbon, the variations found over time in Synechococcus and Prochlorococcus total biomass are mainly determined by changes in cell abundance (Durand et al. 2001), consequence of this average size stability. In comparison, Picoeukaryotes fluctuations in biomass appear as a function of both, changes in cell abundance, and changes in mean size, which more likely reflects changes in species composition of the picoeukaryotic fraction (Worden et al. 2004). The determination of carbon conversion factors based on the knowledge of the taxonomical biogeography of ecologically relevant Picoeukaryotes was still a necessary but unaccomplished goal. Microbial phylogeny The observed distribution patterns of the micro organisms are strongly dependent on the method by which organisms are classified. If clear boundaries have been identified between macroscopic species such as in mammals or plants, the categorization of bulk bacteria into distinct phylogenetic bacterial groups has been established only recently. The first approaches for understanding bacterial phylogeny arose at the end of the 19th century and were based on the study of the metabolic and morphological similarities between bacteria isolated on agar plates. However, this method soon was seen as biased by the unrealistic concentration levels of organic matter and nutrient used in culture media, falsely stimulating particular traits of bacterial metabolism. From the 70´s to the 90´s, the developments of culture independent techniques emerged with improvements made in nucleic acid extraction and sequencing methods. In parallel, the identification of genetic markers universally shared by organisms allowed to conceptualize a new representation of life, not any more based on morphological and physiological criteria but on genetic comparison of the conserved small subunit ribosomal RNA sequences which organized all live beings into three-domains composed of Archaea, Bacteria and Eukarya (Woese et al. 1977, 1987). Developments in the late 1980’s of the Fluorescence in situ hybridization (FISH) method combined to epifluorescence microscopy, based on the targeting Initial approaches 26 of rRNA by fluorescently labeled oligonucleotide probes, allowed the in situ identification and quantification of different phylogenetic bacterial groups, with a specificity of identification spanning from the species level to the level of phyla and domain (see review by Amann and Fuchs 2008). A Biogeography of microbial populations at some phylogenetic level was then possible (Alfreider et al. 1996; Llobet-Brossa et al. 1998; Murray et al. 1998; Simon et al. 1999; Kirchman et al. 2005). Alphaproteobacteria was shown to dominate in marine coastal waters of Delaware Bay (Kirchman et al. 2005), or in the northwestern Mediterranean Sea (Alonso-Sáez et al. 2007), contrasting with Betaproteobacteria found more abundantly in freshwaters (Glöckner et al. 1999). The SAR11 cluster, a distinct branch within the Alphaproteobacteria phylum and probably the most abundant bacterial group in the surface ocean, dominates particularly in nutrient-depleted areas such as oligotrophic waters of the Sargasso Sea (Morris et al. 2002) and in coastal Mediterranean waters (Alonso-Sáez et al. 2007). The Rhodobacteraceae group of marine Alphaproteobacteria has been identified in most marine environments (Buchan et al. 2005), and is generally more abundant in bacterial communities associated with marine algae (Buchan et al. 2005). Bacteroidetes (previously known as Cytophaga-Flavobacteria-Bacteroidetes) constitute one of the major groups of picoplankton (Glockner et al. 1999; Kirchman 2002), abundantly represented in a variety of ecosystems such as cold waters (Simon et al. 1999; Abell and Bowman 2005), coastal waters (Eilers et al. 2001; O’Sullivan et al. 2004; Sáez-Alonso et al. 2007), accounting for as much as half all bacterial cells counted by FISH in California coastal seawater samples (Cottrell and Kirchman 2000), in offshore conditions (Simon et al. 1999; Abell and Bowman 2005; Schattenhofer et al, 2009) and generally associated to phytoplankton blooms (Simon et al. 1999). Figure 5. A phylogenetic tree of bacteria showing the major groups identified in Blanes Bay and contribution to total cells as determined by 454 analysis of 16SrRNA (Data of C. Pedrós-Alió and T. Pommier, drawing by J.M. Gasol). Introduction 27 Biome-related patterns versus continuum hypothesis We have seen above how we can divide the picoplankton community either in functional groups (determined by flow cytometry), in activity groups (determined by fluorescent activity probes) or in phylogenetic groups using certain oligonucleotide probes. The patterns in microbial group distribution (flow cytometrically, activity-based, or phylogenetically determined) can be predicted by taking two different approaches, from either the characterization of different marine ecosystems with specific biogeochemical properties (Longhurst 1995; 1998) or assuming that the relative contributions of the different groups vary along a continuum of physical parameters such as sea surface temperature or chlorophyll a (Gasol et al. 1997; Li 1998). If the first strategy assumes the existence of distinct boundaries by the division of oceans into different marine provinces (Longhurst 1995), in contrast, “the continuum hypothesis” assumes the study of microbial community structures over a large range of parameters, smoothing out most of the variability found at smaller scales. For example, heterotrophic bacteria are known to increase following chlorophyll a concentration and temperature at large scales (e.g. Li et al. 2004). Scales of variability: coastal vs open-ocean ecosystems The ecological function of the different picoplankton groups can be inferred from the study of their distributions at different spatial and temporal scales. Seasonality in picophytoplankton groups has been often observed (Campbell et al. 1997; Jacquet et al. 1998; Li 1998; Grégori et al. 2001; Li and Dickie 2001). Large spatial scale studies have shown that the relative contribution to picoplankton community structure varies not only with ecosystem trophic level (Zhang et al. 2008), but also with temperature and stratification of the water column (Bouman et al. 2011), suggesting that microbial community structure does not vary at random but might represent ecological indicators of water mass properties. Similarly, general distribution patterns of heterotrophic bacterial abundance and bacterial activity have been identified across a range of trophic levels (as estimated from chlorophyll a concentration) (Cole et al. 1988; Billen et al. 1990; Ducklow and Carlson 1992; Bird and Kalff 1984). Similarly, the biomass ratio of heterotrophic bacteria to autotrophic phytoplankton (which can be traced back to Odum 1971) was shown to decrease over a large continuum of chlorophyll a concentration, reaching or even exceeding unity in waters of low chlorophyll a levels (Fuhrman et al. 1989; Cho and Azam 1990; Li et al. 1993; Buck et al. 1996), indicating dominance of heterotrophy in the more oligotrophic environments. Different relationships were later described among different types of ecosystem, the slope varying from freshwaters, coastal to open-ocean waters (Simon et al. 1992; del Giorgio and Gasol 1995; Gasol et al. 1997) (Figure 6). Direct determination of carbon conversion factors for ecologically-relevant photosynthetic picoeukaryotes Thomas Lefort, Fabrice Not, Ian Probert, Dominique Marie and Josep M. Gasol 01 Conversion factors for photosynthetic Picoeukaryotes 36 chapter I 37 ABSTRACT Discrepancies in conversion factor (CF) values used to translate abundance to biomass limit determination of the ecological importance of photosynthetic Picoeukaryotes (pPeuk, < 3 mm). In order to constrain these conversion factors, we determined the cell size and the C and N content of 16 different monospecific pPeuk cultures. Since the cultures were not axenic, two different protocols were used to correct for the presence of bacteria: 1) estimation of bacterial C and N content in each culture by flow cytometry, image analysis and standard bacterial conversion factors; 2) flow cytometric sorting of cells to remove bacteria prior to analysis. Cellular C and N contents varied from 230 fgC cell-1 (±1.21%) and 38.8 fgN cell-1 (±2.73%) for Ostreococcus to 21800 fgC cell-1 (±23.61%) and 4920 fgN cell-1 (±14.11%) for Pycnococcus. Correcting for bacterial carbon resulted in decreases of pPeuk cellular C content values by 7 to 33%. The efficiency of bacterial removal by cell sorting was always superior to 74%. We describe new relationships between cell volume and C and N content for the range of cell sizes considered (1.38-5.06 µm), and an average cellular carbon per unit volume (C/V) ratio for global unspecific pPeuk communities of 467 fgC mm-3 (±4%). An average CF of 1540 fgC cell-1 (±12.01%) for a cell volume of 2.14 mm3 was estimated from a mixture of pPeuk cultures. We also suggest that more specific CFs can be chosen for certain ecosystem types based on the known composition of the pPeuk communities. Conversion factors for photosynthetic Picoeukaryotes 38 INTRODUCTION In many oceanic regions tiny unicellular photosynthetic organisms (i.e. picophytoplankton, cell diameter ≤ 2-3 µm) contribute significantly to carbon fluxes (Agawin et al. 2000; Bell and Kalff 2001). Despite relatively low abundance compared to marine cyanobacteria (Synechococcus and Prochlorococcus), photosynthetic Picoeukaryotes (pPeuks) have been shown to dominate in various marine settings in terms of contribution to biomass (Partensky et al. 1996; Blanchot et al. 2001; Worden et al. 2004) and bulk primary production (Li 1994). The biomass of a phytoplankton population can be estimated by converting cell abundance to overall quantity of carbon by means of a C-per-cell conversion factor (CF). In this context, different values have been used for the relationship between cell volume and cellular C content (C-per-unit-volume CF). Initially, diatoms were used to empirically determine such CFs (Mullin et al. 1966), but because of the presence of vacuoles, large phytoplankton contain less C and N per unit volume than smaller phytoplankton. More appropriate C-per-unit-volume CFs for smaller organisms have been obtained by applying non-linear regression models to data obtained from measurement of unialgal cultures, mostly from the nanoplankton size range (3-20 µm), by Verity et al. (1992) and others. Variations among phytoplankton taxa in chemical composition, and consequently in cellular C and N content, have long been highlighted (e.g. Strathmann 1967; Moal et al. 1987). Patterns of pPeuk biomass have been described via large-scale surveys of pPeuk cell abundances from of a wide range of ecosystems (Li et al. 1992; 1995). However, discrepancies in the CF used limit inter-comparisons between such studies. Estimation of pPeuk biomass in samples from an Atlantic Meridional Transect cruise was performed using a CF of 1.5 pgC per algal cell (Zubkov et al. 1998, 2000) obtained from the C-to-volume ratio of 0.22 pgC mm-3 described for organisms <4 mm (Mullin et al, 1966; Strathmann 1967, Booth 1988) applied to an average cell volume estimated by microscopy and image analysis to be 6.8 ± 6.0 mm3 (Zubkov et al. 1998). Analyses of Peuk community structure and biomass distribution in the central north Pacific Ocean (Campbell et al. 1994) and in subarctic to subtropical oceans (Zhang et al. 2008) used 2.1 pgC cell1 as the abundance-to-biomass CF for Peuks derived from the carbon-to-volume ratio of 0.36 pgC mm-3 from Verity et al. (1992) for an average cell volume of 6.22 mm3 evaluated by microscope analysis. Other studies have used a carbon-to-volume value of 0.24 pgC mm-3 directly measured from Peuk culture isolates (Worden et al. 2004). Peuk community composition has been shown to vary according to oceanic region and nutrient characteristics of the water masses (coastal or open-ocean, eutrophic or oligotrophic, see review by Worden and Not 2008). Prasinophyceae (Archaeplastida) typically dominate Peuk communities in coastal waters (Not et al. 2004), while Prymnesiophyceae (Haptophyta), and chapter I 39 to a lesser extent Chrysophyceae and Pelagophyceae (both Heterokontophyta), appear to make up a large fraction of pPeuk communities in more open ocean ecosystems (Mackey et al. 2002; Fuller et al. 2006; Liu et al. 2009), being present but less abundant in oligotrophic Pacific regions (Vaillancourt et al. 2003) and in the Sargasso Sea (Goericke 1998). PPeuk community composition should be taken into account when determining biomass. Algal cultures can be used to determine CFs, one of the conditions being that the cultures should be taxonomically and physiologically representative of the species that dominate in the ocean. Cultures of the prasinophyte Micromonas pusilla have been used for CF calibration (Durand et al. 2002; Grob et al. 2007), this species being one of the most abundant and cosmopolitan of all pPeuks described to date (Thomsen and Buck 1998), dominating pPeuk communities all year long in coastal systems such as the English Channel (Not et al. 2004) as well as in the Norwegian and Barents seas and near the polar front (Not et al. 2005). Ostreococcus, another prasinophyte, has been shown to be abundant in coastal Pacific surface waters with a maximum at the DCM (Countway and Caron 2006) using Q-PCR. Considered as the smallest eukaryote described to date, Ostreococcus isolates have also been used to determine C-to-biovolume CFs with reported values ranging between 0.24 pgC mm-3 (Worden et al. 2004) and 0.422 pgC mm-3 (Grob et al. 2007). For other ecologically important pPeuk groups such as the Chrysophyceae and Prymnesiophyceae, cultures of the picoplanktonic size range have never been used to constrain CFs. In this study, unialgal cultures were used to investigate whether a range of pPeuks exhibit C-to-biovolume relationships similar to those previously described for larger phytoplankton organisms. We used 16 pPeuk culture strains, as representative as possible of the three ecologically important divisions (Archaeplastida, Haptophyta, Heterokontophyta) that mainly compose natural pPeuk communities. Cultures were harvested in exponential growth to determine cell size and cellular C and N content. Since axenic cultures are difficult to obtain, we used a flow cytometric cell sorting methodology to minimize biases in the determination of pPeuk C content due to the presence of bacteria in the non-axenic cultures. We describe specific empirical relationships between cellular C content and cell biovolume for different members of the Picoeukaryote phytoplankton community. Conversion factors for photosynthetic Picoeukaryotes 40 MATERIALS AND METHODS Selection of Picoeukaryotes and culture conditionsSixteen unialgal, non-axenic pPeuk culture strains were selected from the Roscoff Culture Collection (www.sb-roscoff.fr/ Phyto/RCC) based on their representativeness in terms of diversity and abundance in natural marine ecosystems (Table 1). Cells were grown under different light conditions (Table 1) in 75 cm2 tissue culture flasks in either K medium (Keller and Guillard 1985; Keller et al. 1987) or f/2 medium (Guillard and Ryther 1962). Two successive growth cycles were monitored daily using a FACSCanto II flow cytometer (Becton Dickinson, San Jose, CA) to enumerate cell abundance, following the protocol described in Marie et al. (1999). The first cycle was used to determine the timing of the mid-exponential growth phase for each of the 16 cultures. The second growth cycle was performed in triplicate for each of the 16 cultures. Average growth rates for each culture strain are presented in Table 1. Cell size and volume determination. The triplicate cultures of the 16 different strains were harvested midway through the exponential phase of growth. Mean cell diameter was determined with a Cell-Lab-Quanta SC Flow Cytometer (Beckman Coulter) calibrated before each size determination with 3 mm beads (Polysciences) diluted in MilliQ water. Cell counts and diameter analyses were performed after plotting FL3 (red fluorescence) against EV (electronic volume) parameters. Cell size was assumed to be normally distributed and the peak of distribution was taken as representative of the arithmetic mean cell diameter. Biovolumes were subsequently calculated assuming cells of all strains to be spherical. C and N determination. Duplicate samples for C and N measurement from each of the triplicate cultures of the 16 strains were collected by gentle filtration onto 25 mm glass fiber filters (Whatman GFF, previously ashed for 4 hours at 450ºC and kept in the dark under axenic conditions). The volumes filtered ranged from 10 to 100 ml according to the cell density in each culture. The filters were placed into sterile cryovials and frozen at −20°C until analysis with an Elmer 2400 CHN analyzer. Methods for correcting C estimations (1) Flow cytometric cell sorting of Peuk cultures. Flow cytometric cell sorting was used to correct for errors in biomass estimation due to presence of bacteria in 11 of the cultures (RCC245, RCC287, RCC480, RCC239, RCC927, RCC 361, RCC 299, RCC 419, RCC 422, RCC 497, RCC 504). In the other 5 strains either cell density was not high enough or the bacteria/ chapter I 41 Table 1. Phototrophic picoeukaryote cultures used in this study and geographical origins of each isolate. All are from the Roscoff Culture Collection. Also shown are the parameters of culture media, temperature, and light conditions used to obtain the exponential growth. N.A: information not available Ecosystem assignment: C for Coastal and OO for Open-Ocean were chosen based on the current knowledge (see text). More detailed information about the Peuk cultures on www.sb-roscoff.fr/Phyto/RCC STRAIN DIVISION CLASS Genus species Growth Medium Temperature ºC Light regime mEm-2.s-1 Growth rate d-1 Area of isolation Ecosystem assignment RCC 245 Archeaplastida Prasinophyceae Pycnococcus sp. K20 ºC 4 0.89 Mediterranean Sea C RCC 287 Clade VIIA K20 °C 100 0.86 Pacific ocean O-O RCC 419 Bathycoccus prasinos K15 °C 150 0.93 English channel C RCC 422 Ostreococcus sp. K15 °C 150 0.83 English channel C RCC 299 Micromonas pusilla Clade A K20 °C 100 0.85 Pacific ocean C RCC 497 Micromonas pusilla Clade C F/2 20 °C 100 0.94 Mediterranean Sea C RCC 927 Prasinoderma singularis K20 °C N.A 1.00 Pacific ocean C RCC 656 Haptophyta Prymnesiophyceae Chrysochromulina sp. K20 °C 100 0.95 Atlantic ocean O-O RCC 361 Imantonia rotunda K15 °C 150 1.00 English channel C RCC 703 Heterokontophyta Bacillariophyceae Minutocellus sp. K20 ºC 4 0.93 Indian Ocean C RCC 101 Pelagophyceae Pelagomonas calceolata K20 ºC 4 0.86 Atlantic ocean O-O RCC 480 Chrysophyseae Ochromonas sp. K25 ºC 100 0.83 Indian Ocean O-O RCC 446 Dictyophyceae Florenciella parvula K15 °C 150 0.85 English channel C RCC 239 Bolidophyceae Bolidomonas mediterranea K20 °C 100 0.88 Mediterranean Sea O-O RCC 503 Pinguiophyceae Phaeomonas sp. F/2 20 °C 100 0.90 Mediterranean Sea C RCC 504 Eustigmatophyceae Nannochloropsis gatidana F/2 20 °C 100 0.94 Mediterranean Sea C Conversion factors for photosynthetic Picoeukaryotes 42 pPeuk ratio was very high, hence a mathematical correction for the contaminating bacteria was applied (see below). Using a FACSAria flow cytometer (Becton Dickinson) with freshly prepared 0.2 mm filter-sterilized seawater as sheath fluid, algal cells were discriminated from bacterial cells in the SSC (side scatter) versus FL1 (green fluorescence) plot after DNA staining with SYTO13 (Molecular Probes, Eugene, USA; 5 mM) and then sorted in purity mode. Analysis and cell sorting were made using a 70 µm nozzle, with a sheath pressure of 70 psi and the sample flow was adapted to maintain the particle rate below 1,000. The 11 sorted strains were subsequently filtered for CHN analysis as described above. A 1 ml aliquot from each of the sorted samples was preserved with glutaraldehyde (0.25% final concentration) and stored in liquid nitrogen for subsequent determination of algal and bacterial abundance by flow cytometry. (2) Estimation of bacterial C and N by image analysis. 1 ml samples from culture triplicates were preserved with glutaraldehyde (0.25% final concentration), flash frozen in liquid nitrogen, and then stored at -80ºC until analysis. In order to enumerate bacteria, 0.2 ml of the preserved samples was stained for 10 min in the dark with SYBRGreen I (Molecular Probes, Eugene, USA) (dilution x10,000). Bacterial abundance was then measured with a FACSCalibur flow cytometer (Becton Dickinson) equipped with a laser emitting at 488 nm and the standard filters setup. Data were acquired in log mode and analysis was performed using the Cell-Quest software (Becton Dickinson) using the SSC versus FL1 plots (Marie et al. 1997; Gasol and del Giorgio 2000). 0.5 ml of the samples stored for bacterial abundance were stained with DAPI (final concentration 5 μg ml−1) for 5 min and filtered through 0.2 μm pore-size black polycarbonate filters (Porter and Feig 1980). Filters were mounted on microscope slides with non-fluorescent oil (R. P. Cargille Lab., Inc.) and stored frozen. Bacteria were counted by epifluorescence microscopy with a Nikon Labophot microscope. About 200 to 400 bacteria per sample were counted and bacterial cell size was determined by image analysis following Massana et al. (1997). To convert bacterial biovolumes to cellular C content, the allometric relationship CCC (pg cell-1) = 0.12 x BB0.72 proposed by Norland (1993) was used. The calculated bacterial volumes ranged from 0.11 mm3 to 0.31 mm3. Blank corrections. Since only a small volume was filtered for the 11 sorted cultures and the values of C and N content per filter were very low, a blank correction was applied to account for C and N in filters. C and N were measured for triplicate pre-combusted dry GFF filters. To correct for the filter effect, 0.019 µmol N and 0.517 µmol C were respectively subtracted from all PON and POC results. In order to account for contributions of C and N from the growth media, C and N were measured for triplicate pre-combusted GFF filters through which 30 ml of sterile K/2 medium had been filtered. An average value of 1.55 mmol C (± 0.12 SD) per "wet" GFF filter was chapter I 43 determined. Measurement of cellular C and N content in a mixture of pPeuk cultures. In order to test whether the average cellular C and N value of a mixture of pPeuk cultures corresponded with the average cellular C and N values determined by the different estimation methods, 5 ml samples from 15 different cultures in exponential growth were mixed in 75 cm2 tissue culture flasks in triplicate. Samples were then taken for mean size determination and PON and POC measurements as described above. Statistical analyses. Least squares regression analysis was used to determine the relationship between cellular C content and biovolume. The relationships were fitted using natural log transformations. Equations of the regressions are presented as log (Y) = a + b log (X) with Y= fgC per cell; a=intercept; b=slope; X =biovolume (mm3). In order to test whether the slopes and intercepts of the relationships were significantly different, Student’s t-tests were conducted after applying model I regression analyses. All conducted in JMP 7 software (SAS institute Inc). Conversion factors for photosynthetic Picoeukaryotes 50 Equations N observations R square Root Mean square error Intercept (±SE) Lower 95% Upper 95% Slope (±SE) Lower 95% Upper 95% (1) Cellular C non-axenic 16 0.69 0.38 2.68 (±0.2)** 2.25 3.11 1.08 (±0.19)** 0.67 1.49 (2) Cellular C corrected by bacteria 16 0.72 0.35 2.65 (±0.19)** 2.25 3.06 1.08 (±0.18)** 0.7 1.47 (3) Cellular C after bacterial sorting 11 0.8 0.22 2.78 (±0.13)** 2.49 3.08 0.88 (±0.15)** 0.56 1.21 (4) Cellular N non-axenic relationship 16 0.54 0.45 2 (±0.24)** 1.48 2.52 0.94 (±0.23)** 0.44 1.43 (5) Cellular N corrected by bacteria 16 0.56 0.44 1.97 (±0.24)** 1.47 2.48 0.94 (±0.22)** 0.46 1.42 (6) Cellular N after bacterial sorting 10 0.43 0.50 2.03 (±0.31)* 1.31 2.76 0.83 (±0.34)* 0.05 1.61 (7) Cellular C per volume unit (bacterial calculation) 16 0.005 0.35 2.66 (±0.19) 2.26 3.06 0.05 (±0.18) -0.33 0.42 (8) Cellular C per volume unit (after flow cell sorting) 11 0.007 0.22 2.79 (±0.13)** 2.50 3.07 -0.12 (±0.14) -0.44 0.20 Table 4. Parameter estimates of log-log bivariate fits of relationships between mean cellular carbon and cellular nitrogen content as related to cell biovolume. Relationships 1-6 follow the equation: Log C (fgC per cell)= Intercept (±SE) + slope (±SE) x Log (Biovolume mm3) Equation 7 and 8 follow the equation: Log C/V (fgC per mm3)= Intercept (±SE) + slope (±SE) x Log (Biovolume mm3) **p<0.0001 *p<0.0008 chapter I 51 alter the relationship between pPeuk C content and biovolume. The relationships between N content per cell (measured in fgN per cell) and biovolume (measured in mm3) are represented also by linear fits as: Non-axenic relationship: LOG (fgN cell-1) = 2.00 ± 0.24 + 0.94 ± 0.23 LOG Size (mm3) (eq. 4) Corrected by calculation: LOG (fgN cell-1) = 1.97 ± 0.24 + 0.94 ± 0.22 LOG Size (mm3) (eq. 5) After flow cytometric sorting: LOG (fgN cell-1) = 2.03 ± 0.31 + 0.83 ± 0.34 LOG Size (mm3) (eq. 6) As for cellular C content, bacterial N in the cultures apparently does not affect the relationship between cellular N content and biovolume (eq. 4 and 5). No significant differences were observed (Student’s t-test) between intercepts and slopes of the different relationships 4, 5 and 6. The intercept of relationship 5 appeared below that of eq. 6 as a consequence of subtraction of bacterial N. The slope of relationship 6 was also lower than slopes 4 and 5. However, smaller R2 values and high root mean square errors indicated a worst fit of this relationship (Table 4). Determination of average cellular C content and C/V in a mixture of pPeuk culturesThe direct determination of cellular C content of a mixture of 15 different cultures and calculation of the average cell diameter was compared to the application of the general relationships previously described. The average C content per cell of the mixture was 1540 fgC cell-1 (±12.01%) and fell inside the confidence limits of the different relationships 1, 2 and 3. This C content per cell corresponded to a weighted average diameter of 1.60 mm, equivalent to a biovolume of 2.14 mm3 (Table 2). The ratio between cellular C content and cell biovolume (C/V) represents an additional factor needed to convert pPeuk cell abundance (of known size) into biomass. We tested the significance of the relationships between this ratio (C/V) and volume and found slopes not significantly different from 0 (Table 4), indicating that the cellular C density (expressed as carbon per unit volume) did not decrease with cell volume for the tested cultures, i.e. that small cells did not contain significantly more C per unit volume than larger cells (within the range assayed). Therefore, for each estimation method, we calculated the average and median C/V (considered as constant within the range of biovolume in this study) and compared this with the empirically determined C/V ratio of the mixture (Figure 2A). Dividing the empirically determined average Conversion factors for photosynthetic Picoeukaryotes 52 cellular C content of 1540 fgC cell-1 (obtained after correction for bacterial C by calculation) by the weighted average biovolume of 2.14 mm3 (Table 2, Figure 2A), the average C/V ratio of the mixture was 717.5 fgC mm-3 (±12%). The average C/V ratio values determined in non-axenic conditions, after correction for bacterial C by calculation, and after flow cytometric cell sorting were 822.0 fgC mm-3 (±103%), 715.6 fgC mm-3 (±99%), and 555.0 fgC mm-3 (±52%) respectively. No significant differences (Student’s t-test) were found between the 3 different average C/V ratio values and the average C/V ratio value measured for the mixture. However, since the data were not normally distributed and the extreme C/V ratio values greatly distorted the averages, we also determined the median C/V ratio values in non-axenic conditions, after correction for bacterial C by calculation, and after flow cytometric cell sorting of 484.0 fgC mm-3, 444.5 fgC mm-3, and 472.9 fgC mm-3, respectively. The overall median C/V ratio was thus determined as 467 fgC mm-3 (±4%) (Figure 2A). A B Figure 2. Comparison of Picoeukaryote carbon content per unit of volume (fgC mm-3). (A) Depending on the estimation method used in this study. The green horizontal lines correspond to the average value per method. The grey horizontal line corresponds to the overall average. (B) Depending on the ecosystem origin of the Picoeukaryotes isolates: coastal or open-ocean. The green horizontal lines correspond with the average value per method. Figure 2A chapter I 53 A B Figure 2B The average C/V ratios for groups of species designated as representative of coastal and open-ocean ecosystems (Table 1, Figure 2B) were 584 fgC mm-3 (±36.50%) and 1005 fgC mm-3 (±31.5%), respectively, but did not appear to be statistically different (Student’s t-test). The 25th and 75th quartiles for the coastal C/V ratio were measured at 308 fgC mm-3 and 615 fgC mm-3, respectively. For open-ocean species, the 25th and 75th quartiles were more distant, 309 fgC mm-3 and 1985 fgC mm-3, respectively. The median values for groups of species from each ecosystem were 454 fgC mm-3 for coastal and 435 fgC mm-3 for open-ocean conditions. Conversion factors for photosynthetic Picoeukaryotes 54 DISCUSSION A selection of unialgal pPeuk cultures representative of naturally abundant and ecologically important species was used in this study. In our culture conditions, the cell size (diameter) of cultures ranged from 1.38 mm for Bathycoccus to 5 mm for Chrysochromulina sp., with an average for the 16 strains of 2.73 mm. While the canonical cell size for picoplankton is 0.2-2 µm (Sieburth et al. 1978), most Picoeukaryotes are in fact within the size range 2-3 µm (e.g. Le Gall et al. 2008). The CV of measured cell diameters varied from 0.33% for Minutocellus sp. to 30% for Florenciella parvula RCC446 (Table 2). The CV was low for most cultures, indicating that the coulter counter was well adapted for determining small cell sizes. The high CV value recorded for Florenciella parvula may have been due to cellular dimorphism, as cytograms of red fluorescence versus volume discriminated two differently sized populations (average diameters 2.61 µm and 4.53 mm), confirmed also by light microscopy. For this reason, RCC446 was excluded from the mixture of pPeuk strains used to determine the average cell size and C and N content of a representative Picoeukaryotes fraction. The average cell diameter of 15 different cultures (1.60 mm ± 66.54%), calculated from the quantity of added cells and their respective sizes, was lower than the arithmetic average of measures of the 15 cultures, determined at 2.73 mm. This discrepancy was mainly due to the fact that cell sizes were not normally distributed, smaller cells contributing more to the mixture than larger cells. The weighted average value is coherent with the average size of natural pPeuk communities, measured to be 1.74 ± 0.13 mm in the eastern South Pacific (Grob et al. 2007). The cellular C:N ratio varied from 3.06 (±1%) for Pelagomonas calceolata to 18.0 (±123%) for Ochromonas sp.. The average C:N value was 7.0 (±45%), close to the value of 6.6 often considered representative of nutrient replete cells (Goldman et al. 1979; Sakshaug et al. 1984). The C:N ratio was not correlated with cell size across the size range studied and was not impacted by growth rates. Apart from Ochromonas sp., no remarkable differences were observed between taxonomic types of algae. The very high CV of cell size (123%) measured for Ochromonas sp. means the average C:N value measured for this strain must be taken with caution. Nevertheless, the high C:N ratio for Ochromonas sp. could be the result of nutrient stress (Flynn et al. 1994) or may correspond to the fact that carbohydrates have been reported to be a large fraction (up to 70%) of total C content in Chrysophyceae (Moal et al. 1987). Starch and polyunsaturated fatty acids have been reported to be abundant in the eustigmatophyte Nannochloropsis (Lepère et al. 2009) and the pinguiophyte Phaeomonas (Kawachi et al. 2002), respectively, but C:N ratio values were not particularly high in these organisms (Table 2). Staining with SybrGreen I allowed discrimination of algae and bacteria in plots of side scatter vs. green (DNA) fluorescence and green vs. red (chlorophyll a) fluorescence. Flow chapter I 55 cytometric cell sorting resulted in decreases in the proportion of bacterial cells with respect to algal cells by 74% for RCC497, 83% for RCC299, and more than 95% for the 9 other cultures. The ratio of bacteria/pPeuk concentration was the principal limitation for rendering the Picoeukaryotes cultures axenic. The highest bacteria/pPeuk value was observed for Pelagomonas calceolata RCC101, but this was still lower than average values in natural communities where there is a large variability, that we have measure to have a median value of 330 bacteria per pPeuk cell (unpublished data). This method could not be applied for cultures with very low cell density because a minimum number of cells were required per filter due to the detection limit of the CHN analyzer. Utilization of tangential flow filtration of samples could possibly have reduced this limit. After sorting, bacterial C always represented less than 1% of algal C. The 11 sorted pPeuk cultures were therefore considered axenic. While C content per pPeuk cell was expected to be lower after elimination of bacteria, this was not always the case. The measured C content after purification by sorting was sometimes higher than before sorting, particularly for smaller cells (Table 2). In contrast, C content of larger cells such as Prasinoderma singularis was under-estimated after cell sorting in comparison with other estimation methods. This may have been the result of an underestimation of algal carbon if the cell sorting procedure damaged some cells. There were no significant differences between the pPeuk C/V or N/V relationships following correction for bacteria by calculation or after cell sorting, indicating that both methods were suitable for establishing the relationship. The first attempts to describe the relationship between cellular C and biovolume were made by applying linear regression models to data from large diatoms, with cell volumes that ranged from 101 to 106 mm3 (Mullin et al. 1966; Strathmann 1967). Use of nonlinear relationships to describe cellular C and N content per unit volume has been reported as more appropriate (Verity et al. 1992; Menden-Deuer and Lessard 2000). In these studies, regression exponents of the power curve used to fit data of cellular C and N content with biovolume were significantly <1, indicating that cellular C and N densities (expressed as C or N per unit volume) decreased with cell volume, i.e. that small cells contained more C and N per unit volume than large cells. In this study, we used linear regressions to describe the relationship between cellular C and N content with biovolume. The slopes of the regressions described (equations 1 to 6) were not significantly different from 1. This can be explained by the low range of biovolumes in this study, varying only by one order of magnitude from 1.37 mm3 to 68 mm3 for RCC419 and RCC656, respectively, compared with other studies with much larger biovolume ranges, for example Verity et al. (1992) included cells from 101 to 103 mm3. Rocha and Duncan (1985), Montagnes et al. (1994) and Pelegri et al. (1999) also found slopes not significantly different from 1 and considered cellular C to be constant with biovolume. The average median C/V value established in this study at 467 fgC mm-3 (±4%) is clearly Conversion factors for photosynthetic Picoeukaryotes 56 higher than those reported in previous studies. In samples from the central Atlantic, Zubkov et al. (1998) used a CF of 1500 fgC cell-1 calculated from an average cell size of 6.8 mm3 and a C/V value of 220 fgC mm-3. With our overall median value of 467 fgC µm-3, the estimation of pPeuk biomass would have been more than two-fold higher at 3176 fgC cell-1, meaning that pPeuks would be considered to make a greater contribution to total biomass. Note that it has been suggested that the contribution of small phytoplankton to carbon export from surface layers might be much higher than previously expected (Richardson and Jackson 2007), as recently confirmed using a molecular approach in the eastern subtropical North Atlantic (Amacher et al. 2009). The C/V values recorded for the different taxa in this study were sometimes different from those obtained for the same taxa by other authors. Durand et al. (2002) reported a value of 238 fgC mm-3 for Micromonas pusilla, similar to the value we measured for Micromonas pusillaClade A (308 fgC mm-3), but lower than that for Micromonas pusilla Clade C (470 fgC mm-3). Grob et al. (2007) determined a C/V ratio for Pelagomonas of 692 fgC mm-3, a value significantly lower than that we measured for Pelagomonas RCC101 (2660 fgC mm-3). Such a discrepancy was also observed for Pycnococcus, with estimates of 2116 fgC cell-1 (Grob et al. 2007) compared to 7420 fgC cell-1 in this study. The C/V ratio of Ostreococcus was estimated to range between 233 and 247 fgC mm-3 (Worden et al. 2004), similar to our values measured between 146 fgC mm-3 and 324 fgC mm-3 depending on the correction method used. Grob et al. (2007), in contrast, found a higher value of 423 fgC mm-3. Our new data add to the variability observed among pPeuk types, even between organisms that are phylogenetically closely related. Since C/V values were constant over the range of volume measured (1-100 mm3) but volume and hence C content varied drastically among the different pPeuk taxa selected, knowledge of the composition of the pPeuk community is clearly critical for determining the appropriate abundanceto-biomass CF. For instance, a community dominated by Micromonas pusilla Clade C would have a CF of ca. 800 fgC cell-1 (Table 2), whereas a community dominated by Ostreococcuswould have a CF of 230 fgC cell-1. In reality, pPeuk communities are composed of complex assemblages of diverse species with specific contributions that are largely unknown. We found no significant differences between the C/V in organisms considered representative of coastal or open-ocean environments, although the C/V of open-ocean species was highly variable (with a relatively large distance between the 25th and 75th quartiles, Figure 2B). High C/V values suggest adaptation to more oligotrophic conditions in contrast with communities in coastal waters where nutrients are more concentrated and the need for accumulating energy is lesser. Accumulation of C in cells has been assigned to an uncoupling between photosynthetic C assimilation and the C requirements for biomass production (Dubinsky and Berman-Frank 2001), forming reserves possibly mobilized when nutrients become available again. chapter I 57 In this study we used cultures representative of a relatively wide range of genera and classes of pPeuk and our results suggest that this strategy might be appropriate for determining CFs for pPeuk biomass estimation in field studies. It is still critical to better describe pPeuk community structure at a large scale, and more detailed information about the distribution of pPeuk groups is still needed, since the choice of pPeuk representative can drastically alter CF estimation. Acknowledgements This work was supported by the EU-funded projects METAOCEANS (MESTCT-2005-019678) and ASSEMBLE (2009-227799). We particularly thank Daniel Vaulot for the use of facilities, Peter von Dassow and Daniela Mela for flow cytometry assistance, Irene Forn for help with the image analysis system and X.A. Álvarez-Salgado (IIM-Vigo) for the CHN analyses. Conversion factors for photosynthetic Picoeukaryotes 58 REFERENCES Agawin, N. S. R. Duarte, C. M. and Agustí, S. (2000) Nutrient and temperature control of the contribution of picoplankton to phytoplankton biomass and production. Limnol. Oceanogr., 45, 591--600. Amacher, J. Neuer, S. Anderson, I. and Massana, R. (2009) Molecular approach to determine contributions of the protist community to particle flux. Deep-Sea Res. 56, 2206--2215. Bell, T. and Kalff, J. (2001) The contribution of picophytoplankton in marine and freshwater systems of different trophic status and depth. Limnol. Oceanogr. 46, 1243–1248. Blanchot, J. Andre, J. M. Navarette, C. Neveux, J. and Radenac, M. H. (2001) Picophytoplankton in the equatorial Pacific: Vertical distributions in the warm pool and in high nutrient low chlorophyll conditions. Deep-Sea Res. I, 48, 297-314. Booth, B. C. (1988) Size classes and major taxonomic groups of phytoplankton at two locations in the subarctic Pacific Ocean in May and August, 1984. Mar. Biol. 97, 275--286. Campbell, L. Nolla, H. A. and Vaulot, D. (1994) The importance of Prochlorococcus to community structure in the central North Pacific Ocean. Limnol. Oceanogr. 39, 954–961. Countway, P. D. and Caron, D. A. (2006) Abundance and Distribution of Ostreococcus sp. in the San Pedro Channel, California, as Revealed by Quantitative PCR. Appl. Environ. Microbiol., 72, 2496-2506. Dubinsky, Z. and Berman-Frank, I. (2001) Uncoupling primary production from population growth in photosynthetizing organisms in aquatic systems. Aquat. Sci., 63, 4-17. Durand, M. Green, R. Sosik, H. and Olson, R. (2002) Diel variations in optical properties of Micromonas pusilla (prasinophyceae). J. Phycol., 38, 1132--1142. Flynn, K. J. Davidson, K. and Leftley, J. W. (1994) Carbon nitrogen relations at whole-cell and free amino acid levels during batch growth of Isochrysis galbana (Prymnesiophyceae) under conditions of alternating light and dark. Mar. Biol., 118, 229--237. Fuller, N. J. Tarran, G. A. Cummings, D. G. Malcom, E. Woodward, S. Orcutt, K. M. Yallop, M. Le Gall, F. and Scanlan, D. J. (2006). Molecular analysis of photosynthetic picoeukaryote community structure along and Arabian Sea transect. Limnol. Oceanogr., 51, 2515--2526. Gasol, J. M. and del Giorgio, P. A. (2000) Using flow cytometry for counting natural planktonic bacteria and understanding the structure of planktonic bacterial communities. Sci. Mar., 64, 197- -224. Goericke, R. (1998) Response of phytoplankton community structure and taxon-specific growth rates to seasonally varying physical in the Sargasso Sea off Bermuda. Limnol. Oceanogr., 43, 921--935. chapter I 59 Goldman, J. C. McCarthy, J. J. and Peavey, D. G. (1979) Growth rate influence on the chemical composition of phytoplankton in oceanic waters. Nature, 279, 210--215. Grob, C. Ulloa, O. Claustre, H. Huot, Y. Alarcón, G. and Marie, D. (2007) Contribution of picoplankton to the total particulate organic carbon concentration in the eastern South Pacific. Biogeosciences, 4, 837--852. Guillard, R. R. L. and Ryther, J. H. (1962) Studies of marine planktonic diatoms. I. Cyclotella nana Hustedt and Detonula confervacea (Cleve) Gran. Can. J. Microbiol., 8, 229–39. Guillard, R. L. Keller, M. D. O’Kelly, C. J. and Floyd, G. L. (1991) Pycnococcus provasolii gen. et sp., a coccoid prasinoxanthin-containing phytoplankter from the western north Atlantic and Gulf of Mexico. J. Phycol., 27, 39--47. Kawachi, M. Inouye, I. Honda, D. O’Kelly, C. J. Craig Bailey, J. Bidigare, R. R. and Andersen, R. A. (2002) The Pinguiophyceae classis nova, a new class of photosynthetic stramenopiles whose members produce large amounts of omega-3 fatty acids. Phycol. Res., 50, 31--47. Keller, M.D. and Guillard, R. R. L. (1985) Factors significant to marine diatom culture. In Anderson, D.M. White, A.W. and Baden, D.G. (eds.), Toxic Dinoflagellates. Elsevier, New York, pp. 113--6. Keller, M. D. Selvin, R. C. Claus, W. and Guillard, R. R. L. (1987) Media for the culture of oceanic ultraphytoplankton. J. Phycol. 23, 633–8. Le Gall, F. Rigaut-Jalabert, F. Marie, D. Garczarek, L. Viprey, M. Godet, A. and Vaulot, D. (2008) Picoplankton diversity in the South-East Pacific Ocean from cultures. Biogeosciences, 5, 203–214. Lepère, C. Vaulot, D. and Scanlan, D. J. (2009) Photosynthetic picoeukaryote community structure in the South East Pacific Ocean encompassing the most oligotrophic waters on Earth. Environ. Microbiol., 11, 3105--3117. Li, K. K. W. Irwin, B. D. and Wood, A. M. (1992) Biomass of bacteria, cyanobacteria, prochlorophytes and photosynthetic eukaryotes in the Sargasso Sea. Deep-Sea Res., 39, 501--519. Li, W. K. W. (1994) Primary production of prochlorophytes, cyanobacteria, and eukaryotic ultraplankton: measurements from flow cytometric sorting. Limnol. Oceanogr., 39, 169--175. Li, W. K. W. (1995) Composition of Ultraphytoplankton in the central North Atlantic. Mar. Ecol. Prog. Ser., 122, 1--8. Liu, H. Probert, I. Uitz, J. Claustre, H. Aris-Brosou, S. Frada, M. Not, F. and De Vargas, C. (2009) Extreme diversity in noncalcifying haptophytes explains a major pigment paradox in open oceans. P. Natl. Acad. Sci. U S A, 106, 12803--12808. Mackey, D. Blanchot, J. Higgins, H. and Neveux, J. (2002) Phytoplankton abundances and community structure in the equatorial Pacific. Deep-Sea Res. I , 49, 2561--2582. Marie, D. Partensky, F. Jacquet, S. and Vaulot, D. (1997) Enumeration and cell cycle analysis of Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 66 INTRODUCTION Understanding the distribution of plankton, i.e. picoplankton, is one of the main goals of marine microbial ecology. In addition to the Picoeukaryotes (Johnson and Sieburth 1982), two phylogenetically closely related types of photosynthetic prokaryotes: Synechococcus (Syn) (Waterbury et al. 1979) and Prochlorococcus (Pro) (Chisholm et al. 1988; Chisholm et al. 1992) compose the picophytoplankton fraction (<3 mm). The concentrations of the three groups have often been shown to peak at different periods of the year, suggesting distinct environmental controls for each of these organism types (e.g. Partensky et al. 1999a). While mesotrophic regions are generally dominated by Picoeukaryotes, low productive oligotrophic waters are generally occupied by large numbers of Prochlorococcus and to a lesser extent by Synechococcus (Jacquet et al. 2002). These are ubiquitous in oligotrophic and mesotrophic regions (Olson et al. 1990; Campbell and Vaulot 1993; Partensky et al. 1996), but are generally more abundant in nutrient rich areas (Partensky et al. 1999a). A preference of Prochlorococcus for stratified over mixed waters has also been observed (Lindell and Post 1995; Vaulot and Partensky 1992). Although the individual geographic distributions of Prochlorococcus and Synechococcus is now well documented (Partensky et al. 1999a; Partensky et al. 1999b), less is known about the photosynthetic Picoeukaryotes (pPeuk), their low numerical contribution contrasting with their dominance in the picophytoplankton biomass of many marine ecosystems (Ishizaka et al. 1997; Li et al. 1992; Li et al. 1994; Worden et al. 2004). Yet, the comprehension of the factors driving the picoplankton group distribution and their relative contribution to total picoplankton biomass is essential for understanding the dynamics of the ecosystem. Using the comparative-analysis approach, Gasol et al. (1997) showed that the ratio of heterotrophic to autotrophic biomass (first introduced by Odum in 1971) tended to decrease with increasing levels of primary productivity and, additionally, Li et al. (2001) concluded that the heterotrophic and photoautotrophic components of the picoplankton tend to complement each other so that their total biomass is more conservative than either component alone. The distribution of the different picoplankton groups has mostly been studied on relatively large time scales with sampling frequencies ranging from once per day to 1 per month, and only a few times at a higher frequency (i.e. several samples per day). Given that events of major ecological relevance often result from transient environmental perturbation (i.e. wind stress, turbulence, high irradiance…), and that the microbial life history more likely operates at short time frames, it is necessary to determine the significance of the short time scale to structuring the large scale patterns in microbial communities (Seymour et al. 2005). Episodic physical forcing at short timescales is known to induce shifts in both phytoplankton and picoplankton community structures (Guadayol et al. 2009; Pannard et al. 2008; Thomas et al. chapter II 67 2010), and light has often been also identified as the most important driver of the diel variability. Most phytoplankton species divide at specific times of the day (Gouch 1905), and even large phytoplankton such as diatoms and dinoflagellates follow diel cycles (Swift and Durbin 1972; Smayda 1975). Jacquet et al. (1998) showed that the Synechococcus cell cycle was phased with the daily light cycle, possibly enforced by a “clock” controlled by genetic factors (Johnson et al. 1996). Synchronization and phasing of cell growth for both Synechococcus and pPeuk was measured from dawn to dusk during the winter in the northwestern Mediterranean Sea and in the Alboran Sea (Jacquet et al. 1998; Jacquet et al. 2002). However, differences were reported in the equatorial Pacific where the division of Synechococcus, Prochlorococcus and Picoeukaryotes did not proceed at the same time (Vaulot and Marie 1999). Whether or not such phase differences between groups are linked to the differential sensitivity of each group to light (Sommaruga et al. 2005) remains unclear, but it was suggested that the Prochlorococcus cell cycle is tightly coupled to the irradiance levels (Jacquet et al. 2001). Moreover, the relative stability of picoplankton group cell concentrations measured on a daily and a weekly scale suggests that mortality generated by grazing and viral lysis balances cell growth and division (Landry et al. 1995). Differential grazing on Synechococcus, Prochlorococcus and pPeuk has already been described (Worden et al. 2004) and different factors that translate into preferential grazing on some bacteria have been identified, including cell size (Gonzalez et al. 1990), motility (Matz and Jürgenz 2005), surface properties (Matz and Jürgenz 2001), phylogenetic affiliation (Jezbera et al. 2005), food quality as estimated from C:N:P ratio (Shannon et al. 2007), cell viability (Landry et al. 1991) or membrane integrity (Massana et al. 2009). Thus, a detailed knowledge of grazing is needed to understand microbial diel variability and the resulting consequences on ecosystem functioning. Tight coupling between phytoplankton and bacteria should result in bacteria also following circadian cycles. As a consequence of this link, a peak of bacterial activity at noon/afternoon should directly follow a peak of DOM originated from primary production (Fuhrman et al. 1985; Gasol et al. 1998; Herndl and Malacic 1987). Conversely, absence of daily coupling between phytoplankton and bacteria would imply that bacteria are not very much dependant of the DOM produced by phytoplankton, and instead support its growth and activity from DOM coming either from non-diel grazing pressure (Nagata et al. 2000) or from allochtonous sources. Evidences for diel patterns in bacterial abundance and activity have been reported from the coastal NW Mediterranean (Ghiglione et al. 2007, Gasol et al. 1998), but how picophytoplankton variability is coupled with bacterial single cell activities has not yet been analyzed. For that objective, we followed the diel variations of picoplankton abundance by flow Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 68 cytometry sampling with a high frequency (4 hours intervals) during two cycles of 72 hours in winter 2007 in a NW Mediterranean coastal station, period of the year that presented the higher chlorophyll a levels. Combined with flow cytometry, we used molecular probes testing for bacterial activity, bacterial membrane integrity and heterotrophic nanoflagellate abundance, to determine to what extent picophytoplankton was coupled with heterotrophic bacteria, and how this coupling could be altered at the short time scale by physical forcing, such as that provoked by shifts in wind direction and strength. chapter II 69 MATERIAL AND METHODS Sampling sitesTwo diel cycles were studied during two successive three-day periods in FebruaryMarch 2007 (from 20th to 23rd February 2007 and from 26th February to the 1st of March 2007) at the Blanes Bay Microbial Observatory, a shallow (20 m depth) oligotrophic coastal station in the NW Mediterranean Sea, located 800 m offshore of Blanes, Catalonia, Spain (41º39.90’N, 2º48.03E). The sampling of surface water was performed at 0.5 m depth with polycarbonate carboys at a frequency of 6 samplings per day (every 4 hours). The samples were kept in the dark until analyses at the laboratory (less than 20 min from sampling). The first sampling of the two cycles (CDN01 and CDN 20, respectively) began at 10:00 A.M. Only one sample (CDN 14) could not be performed due to sea conditions. The temperature and salinity of the waters were measured with a SAIV A/S 204 CTD probe. Irradiance measurements during the sampling were obtained from the nearby station of Malgrat de Mar (Catalan Meteorological Service, www.meteo.cat), located at 5 km from the sampling station and at 4 m above sea level. The station recorded arithmetically averaged hourly air temperature and relative humidity at 1.5 m above ground, vector-averaged hourly wind speed and direction and global irradiance at 2 m, and accumulated rainfall at 1 m. Wave height data were collected from a scalar buoy (DATAWELL, Waverider) placed at 41º 38.49´N, 2º 48.56’E over a depth of 74 m (XIOM Network, www.boiescat.org). Chlorophyll a concentration was determined from 150 mL of seawater filtered through GF/F filters (Whatman) extracted in acetone (90% v/v), and fluorescence was measured with a Turner Designs fluorometer. Picoplankton abundancesDetermination of picoalgal and bacterial abundance was performed by flow cytometry (Gasol and del Giorgio 2000; Marie and Partensky 2006). For picophytoplankton, the samples were analyzed without addition of fixative and run at high speed (ca. 100 µl min-1), three populations (Prochlorococcus, Synechococcus, Picoeukaryotes) were discriminated according to scatter and fluorescence signals. For non-phototrophic bacteria, we choose to estimate the abundance following the NADS Viability protocol (see below), to avoid using fixatives. However, abundances were estimated also by fixing 1.2 ml samples with a 1% paraformaldehyde + 0.05% glutaraldehyde solution, and deep-freezing in liquid N2. Afterwards the samples were unfrozen, stained with SybrGreen at a 10x dilution and run at low speed (ca 15 µl min-1). Cells were identified in plots of side scatter versus green fluorescence. Heterotrophic nanoflagellate abundances were measured following the Rose et al. (Rose et al. 2004) protocol. From a stock solution of 1 mM Lysotracker Green (Molecular Probes), 1 µl was added to 99 µl of <0.2 mm MilliQ, and 3.8 µl of this diluted Lysotracker stock were added to 0.5 ml of the sample, generating a 75 nM Lysotracker final concentration. We analyzed the samples Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 70 as in Rose et al. (2004), using a combination of side scatter and green and red fluorescence plots. Samples were run alive at high (ca. 100 µl min-1) speed. Concentrations were obtained from weight measurement of the volume analyzed. Carbon Conversion factors for biomass calculationThe Synechococcus/ Picoeukaryotes biomass ratio was obtained by transforming abundances with standard biovolume to C conversion values: 250 fgC cell-1 and 1540 fgC cell-1 for respectively Syn (Campbell et al. 1994) and Peuk (Lefort et al. Chapter I). Bacterial single-cell activityMeasurements of the different physiological status of bacteria were done in two ways: I) Highly active prokaryotes, as those able to reduce 5-cyano2,3-diotolyl tetrazolium chloride (CTC; Polysciences). CTC turns into a red fluorescent formazan that is detectable by epifluorescence and flow cytometry (Sherr et al. 1999; Sieracki et al. 1999). Sample aliquots (0.4 ml) were amended with 5 mM CTC (from a fresh stock solution at 50 mM) immediately following collection and were incubated for 90 min in the dark at room temperature. CTC-positive (CTC+) cells were enumerated by flow cytometry using the FL2-versus-FL3 dot plot (Gasol and Arístegui 2007). For these analyses, we used a high speed (ca. 100 µl m-1) and a threshold set in red fluorescence. ii) Cells with intact membranes were enumerated using the NADS viability protocol, based on the combination of the cell-permanent nucleic acid strain SybrGreen I (Molecular Probes, Eugene, OR) and the cell-impermeant propidium iodine (PI, Sigma Chemical Co.) fluorescent probe. We used a 10x SG1 and 10 µg ml-1 PI concentrations. After simultaneous addition of each stain, the samples were incubated for 20 min in the dark at room temperature and then analyzed by flow cytometry. SG1 and PI fluorescence were detected in the green (FL1) and red (FL3) cytometric channels, respectively. A dot plot of red versus green fluorescence allowed distinction of the “live” cell cluster (i.e., cells with intact membranes and DNA present) from the “dead” cell one (i.e., with compromised membranes) (Grégori et al. 2001; Falcioni et al. 2008). Fluorescence and Side scatter parameters were standardized to reference Polysciences 1 µm beads Data transformations and statistical analysesTo perform the Fisher’s Kappa statistic (Davis 1941; Fuller 1976), we completed the time series with the missing CDN14 values. For that purpose, we forecasted the CDN14 values calculating the arithmetical average between the surrounding values CDN13 and CDN15. Therefore, we tested the null hypothesis that the values in the series were drawn from a normal distribution with variance 1 against the alternative hypothesis that the series had some periodic component. Kappa is the ratio of the maximum value of the periodogram, I(fi), and its average value. The null hypothesis is rejected if this probability is less than the significance level. All conducted in JMP 7 (SAS institute Inc). chapter II 71 RESULTS Background Environmental parametersThe two successive diel cycles were sampled in the winter of 2007 at the Blanes Bay coastal station. Water temperature was 13ºC (close to the minimum of the year) and salinity close to 38.30 psu, both parameters varied little over the period of observation (Table 1). Chlorophyll a concentration was only measured at the beginning of each cycle and increased from the first cycle to the second (from 0.47 to 0.89 mg L-1, Table 1). The main wind direction was N/NW (340º) during the two cycles (Figure 1B), frequently interrupted by shifts in speed and direction from North to South/SW. During the second half of the day 25th of February (between the two cycles), a pronounced change from North to South occurred concomitantly with light rainfall (not shown) and a decrease in irradiance (*, Figure 1A), Table 1. Average values and coefficients of variation of the different environmental, picoplankton community structure, and activity parameters. SD=Standard deviation. Coefficients of variation were calculated as (standard deviation)/(Mean). * indicates significantly different values (t-tests , p<0.05). AVERAGE (±SD) COEFFICIENTS OF VARIATION Parameters FIRST CYCLE SECOND CYCLE FIRST CYCLE SECOND CYCLE CHLOROPHYLL a (µg l-1) 0.47 ± 0.02 * 0.89 ± 0.03* n.d n.d TEMPERATURE (ºC) 13.43 ± 0.04 13.36 ± 0.01 n.d n.d ENV. PARAMETERS SALINITY (PSU) 38.27 ± 0.03 38.30 ± 0.01 n.d n.d ABUNDANCE (103cells mL-1) 5.76 ± 0.81* 15.70 ± 0.30* 14% 19% FL2 (REL. UNITS) 0.96 ± 0.05* 0.87 ± 0.05* 5% 6% FL3 (REL. UNITS) 1.39 ± 0.04* 1.32 ± 0.04* 3% 3% SYNECHOCOCCUS SSC (REL. UNITS) 1.16 ± 0.07* 0.86 ± 0.10* 7% 12% ABUNDANCE (103cells mL-1) 5.40 ± 0.10* 12 ± 0.17* 19% 14% FL3 (REL. UNITS) 0.62 ± 0.07* 0.56 ± 0.05* 12% 10% PROCHLOROCOCCUS SSC (REL. UNITS) 0.24 ± 0.03* 0.21 ± 0.02* 14% 8% ABUNDANCE (104cells ML-1) 1.09 ± 0.40* 1.38 ± 0.30* 40% 20% FL3 (REL. UNITS) 1.63 ± 0.07* 1.57 ± 0.06* 4% 4% PICOEUKARYOTES SSC (REL. UNITS) 0.80 ± 0.04* 0.75 ± 0.05* 6% 6% ABUNDANCE (105cells ML-1) 7.75 ± 1.13 8.27 ± 0.51 15% 6% LIVE + DEAD CELLS (105cells ML-1) 7.20 ± 0.70 7.53 ± 0.90 9% 12% CTC+ (104cells ML-1) 4.70 ± 0.90 5.70 ± 0.17 19% 20% CTC+ (%) 6 ± 0.70* 6.50 ± 2.00* 11% 32% HNA (%) 59 ± 3.00 58 ± 2.00 5% 5% HETEROTROPHIC. BACTERIA NADS-LIVE (%) 84 ± 3.15 86 ± 5.00 10% 11% HNF (LYSOTRACKER) ABUNDANCE (103 cells mL-1) 1.14 ± 0.80 0.76 ± 0.40 68% 53% SYN/PEUK BIOMASS RATIO (10-1) 0.96 ± 0.29* 1.88 ± 0.31* 30% 17%  Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 72 Figure 1. (A) Irradiance measurements (upper panels) during the first cycle (from the 20th to the 23rd of February), during the second cycle (from the 26th of February to the 1st of March 2007; and between the two cycles (from the 24th to the 25th of February 2007) and main wind direction (lower panel) expressed in percentage of time, * indicates the episode of increased turbulence, lower irrediance and shift in wind direction before the second cycle (B). The weather data come from the station of Malgrat de Mar (Catalan Meteorological Service, http:// www.meteocat.com) (C) The smoothed average wave height measured by a scalar buoy throughout the sampling period (XIOM Network, http:// www.boiescat.org), The lines were obtained using a smooth fit in software Kaleidagraph vs 3.6.2 (Synergy Software). A B C   Firstcycle  Weekend Secondcycle   In % N NE E SE S SW W NW First cycle 26.29 0.43 3.02 4.74 12.93 12.93 23.71 15.95 Weekend 30.77 2.56 7.69 4.27 14.53 9.40 18.80 11.97 Second cycle 25.41 0.66 7.26 5.61 12.87 11.22 21.12 15.94   * chapter II 73 rapidly followed by an increase of the turbulence at the onset of the 26th of February (Figure 1C). Compared with the first cycle, turbulence stayed at higher levels during the second cycle. Picoplankton community StructureHeterotrophic bacteria constituted the major component of the Picoplankton community structure during the two cycles (Figure 2). The average bacterial concentrations during the first and second cycle were 7.75 (±1.13) 105 cells ml-1 and 8.27 (±0.51) 105 cells ml-1 (Table 1). During the first cycle, picophytoplankton community structure appeared clearly dominated by photosynthetic Picoeukaryotes (pPeuk) for which the average concentration was 1.09 (±0.40) 104 cells, followed by Synechococcus and Prochlorococcus with 5.76 (±0.81) 103 cells ml-1 and 5.40 (±0.10) 103 cells ml-1 respectively (Table 1 and Figure 2). During the second cycle (week 2), a shift in community composition followed the change in wind direction and strength. Synechococcus dominated community structure with an average concentration of 1.57 (±0.30) 104 cells ml-1 (representing a 172% of increase when compared 18H 08H 18H 08H 18H 08H 18H 08H 18H 08H 18H 08H 6 10 5 6.5 10 5 7 10 5 7.5 10 5 8 10 5 8.5 10 5 9 10 5 9.5 10 5 5 10 3 1 10 4 1.5 10 4 2 10 4 0 5 10 15 20 25 30 35 Het. Bact. Cells ml-1 Stations 2 nd Cycle 26/02/2007 to 01/03/2007 1 st Cycle 20/02/2007 to 23/02/2007 Picophytoplankton Cells ml-1 Figure 2: Diel variations of Picoplankton group abundances as measured by flow cytometry. The Y left axis corresponds to the heterotrophic bacterial concentration (m) and the Y right axis to the picophytoplankton abundances: (l) for Picoeukaryotes; (n) for Synechococcus; (r) for Prochlorococcus. The grey areas correspond to dark period (from 18h00 to 7h00) also indicated by solid bars on the top axis; the error bars correspond to the range of variation of the duplicate samples. Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 74 with the average found the first week), closely followed by pPeuk and Prochlorococcus with the average concentrations of 1.38 (±0.30) 104 cells ml-1 and 1.20 (±0.17) x 104 cells ml-1 respectively (Table 1 and Figure 2) Synechococcus diel patternsAs a general tendency, Synechococcus concentration increased during the dark period, and decreased during the light period. During the first cycle (From 20th to 23rd February 2007), Synechococcus abundance followed a clear diel cycle with a significant periodicity of 24 hours (Fisher’s Kappa, p<0.05) (Table 2, Figure 2). Their concentration increased strongly during the first part of the dark period (from 6:00 to 10:00 pm), followed by a plateau until dawn. After dawn, a pronounced decrease of Synechococcus concentration was observed until a minimum reached at dusk. Table 2. Fisher’s Kappa statistic tests. * indicate significant periodic variations (p<0.05) “Undefined” indicates that no periodicity could be measured from the periodograms. N for number for observations. KAppA peRIODICITy First CyCle (hours) N=19 seCoNd CyCle (hours) N=19 SynechococcuS AbuNdANCe 24h* uNdeFiNed Fl2 24h* uNdeFiNed Fl3 24h uNdeFiNed ssC 24h19h ProchlorococcuS AbuNdANCe uNdeFiNed uNdeFiNed Fl3 24h24h* ssC uNdeFiNed uNdeFiNed pICOeuKARyOTeS AbuNdANCe 24h* 24h Fl3 24h* 24h* ssC 24h* uNdeFiNed HeT. BACT AbuNdANCe uNdeFiNed 15h live+deAd 24h* uNdeFiNed CtC+ AbuNdANCe uNdeFiNed Nd CtC (%) uNdeFiNed Nd hdNA (%) 24h15h live (%) uNdeFiNed* uNdeFiNed HNF (LySOTRACKeR)AbuNdANCe 24h15h SyN/peuK rAtio biomAss uNdeFiNed 12.6 chapter II 75 In comparison with the first cycle, a less pronounced diel pattern in Synechococcus abundance was observed during the second diel cycle, with no significant and defined periodicity (Figure 2, Table 2). A large diel abundance variation of Synechococcus concentration was measured (14% - 19%), mostly resulting from the strong increase during the second night of observation (Figure 2). Synechococcus abundance recovered a diel pattern towards the end of the second cycle, more exactly during the third light period of the second cycle (28th of February), when its concentration decreased with the pattern observed during the first week. Cell-specific pigment content (as measured by the standardized FL2 and FL3 parameters) followed a clear diel pattern during the two cycles, with a significant periodicity of 24 hours (Fisher’s Kappa, p<0.05) during the first cycle (Figure 3A, Table 2). However, this pattern was opposite to that observed for abundance (Figure 2). It generally increased from dawn to reach a maximum at dusk, corresponding with the accumulation of pigments during the growth process occurring during the lit period of the day. The decrease began just after dusk and reached a minimum at noon with a stationary period until dawn. No significant differences between cycles were measured for the fluorescence parameters (Table 1).         A B Figure 3. (A) FL3 Red Fluorescence (relative units) of Picoeukaryotes and Prochlorococcus, FL2 fluorescence (relative units) of Synechococcus, all standardized according to the fluorescence of Polysciences 1 µm beads ; Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 82 group abundances, indicating an important phasing between these parameters (Table 3). Moreover, %HNA was also significantly positively correlated with pPeuk and bacterial abundances, but negatively correlated with Heterotrophic nanoflagellate abundance (HNF). HNF diel patternsDuring the first cycle, HNF concentration increased during the light periods (Figure 7), although the periodic variations were not significant (Table 2). Less periodicity was observed during the second cycle, but after a pronounced decrease of HNF abundance observed the first day of the second cycle, a general trend of increase was finally observed. Note that while negative correlations were calculated between HNF abundances and the different picophytoplankton group abundances (significant only for Synechococcus, Table 3), positive relationships (but not significant) were observed with %CTC and %Live cells during the first and second cycle, suggesting a possible preference of HNF grazing activity for actively growing bacterial cells with intact membranes. 18H 08H 18H 08H 18H 08H 18H 08H 18H 08H 18H 08H 0 500 1000 1500 2000 2500 0 5 10 15 20 25 30 35 HNF cells ml -1 (by lysotracker) Stations 2nd Cycle 26/02/2007 to 01/03/2007 1st Cycle 20/02/2007 to 23/02/2007 Figure 7. Concentration of heterotrophic nanoflagellates (HNF) after Lysotracker staining (solid black line) during the two cycles. The lines were obtained using a smooth fit with software Kaleidagraph vs 3.6.2 (Synergy Software). The grey bars represent dark periods. chapter II 83 DISCUSSION Improved knowledge of the diel patterns in microbial parameters and the resulting diel variability may inform us about the factors controlling the growth and loss processes of marine microbes. Photosynthesis occurs only during the light parts of the diel cycle, and the cellular division of microbes commonly occurs at specific moments of the diel cycle (e.g. Vaulot and Marie 1999), although not necessarily at the same time for all organisms, nor in all oceanographic settings (e.g. Jacquet et al. 2001). The photosynthetic release of DOM is dependent upon the light cycle (Mague et al. 1980; Furhman et al. 1985; Pausz et al. 1999), and the grazing activities of most zooplankters are also circadian (Atkinson et al. 1992; Jakobsen and Strom 2004; Wikner et al. 1990) and so, the rates of DOM supply to heterotrophs are likely to follow diel patterns. Factors that affect the single-cell physiological status and activity level of marine bacteria, such as ultraviolet radiation, bacterivory and viral lyses are also known to follow diel variations (Jeffrey et al. 1996; Christaki et al. 2002; Wikner et al. 1990; Winter et al. 2004). Not all studies of diel variability encounter the above-mentioned periodicities, and in some cases microbial abundance and activity seem to vary at random. Other than lack of sensitivity of the used methods, it is interesting to describe what environmental factors facilitate that microbial populations vary with diel periodicity in some cases and not in others. Coastal communities fon instance may not show evidences of coupled microbial diel variability if heterotrophic communities depend on landor benthos-derived materials instead of using phytoplankton-derived primary production. We were particularly interested in the diel variability of the structure of the picoplankton community (understood as the differential contributions of each organism type), and in the linkage between this structure and the heterotrophic activities of the bacteria measured at the single-cell level, which could also be expected to vary following the phytoplankton and the light regime. We chose to perform the experiment at the likely time of the phytoplankton bloom to maximize the likelihood of observing coupled variability of picophytoplankton and bacteria. In brief, our results have shown 1) consistent diel variability of all picoplankton populations, including heterotrophic bacteria and HNF, 2) differences in the time of duplication and growth of different picophytoplankton groups, and 3) coupling between picophytoplankton variability and single-cell bacterial activities. We furthermore observed how a relatively small variation in weather patterns changed considerably the structure of the microbial community and disrupted most diel cycles, which started to recover a couple of days after the disruption. Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 84 Diel patterns in the picophytoplankton growth and divisionDuring the two cycles, diel patterns were observed for the different Picophytoplankton group fluorescence and scatter parameters, with a periodicity close to 24 h for the majority of the parameters studied. Synechococcus and Picoeukaryote growth, as measured by the increase of SSC and fluorescence (FL2 and FL3), occurred during the light period, indicating that light drove the synthesis and accumulation of carbon and pigment, followed during the night by division, producing smaller cells with lower scatter (Durand and Olson 1998). These measured diel variations are not particular but are apparently common for Synechococcus, Prochlorococcus, and Picoeukaryotes communities across systems (Vaulot et al. 1996; Jacquet et al.,1998; 2001,;2002; Vaulot et Marie 1999; Seymour et al. 2005; Durand et al. 2002). However, the increase of Prochlorococcus FL3 fluorescence and cell size occurred during the night period instead of at light. A minimum of chlorophyll a fluorescence (FL3) was measured at midday for Prochlorococcus and was concomitant with the maximum irradiance measured daily. On the contrary, no particular bleaching of fluorescence for Synechococcus and Picoeukaryotes was measured during the light period of the diel cycle. Vaulot and Marie (Vaulot and Marie 1999) measured similar patterns in the equatorial Pacific and observed only for Prochlorococcus fluorescence some quenching during the light period at the surface (in samples particularly exposed to high irradiance levels) and an increase of fluorescence at depth (Vaulot and Marie 1999), suggesting that Synechococcus and Picoeukaryotes were more protected against light damage than Prochlorococcus, something that can be explained by a much thicker thykaloid layer in the former genus Synechococcus and more complex photoprotective mechanisms in eukaryotes (Sommaruga et al. 2005; Llabrés and Agustí 2006). But no other decrease of FL3 for Prochlorococcus during the night was observed, suggesting that such day light minimum was more likely related to division rather than photochemical quenching, our study suggests a specific timing for Prochlorococcus division. Moreover, this specific behavior found for Prochlorococcus fluorescence and size increasing during the night period, could also suggest that other processes were supporting cell growth, more likely promoted by heterotrophy (by incorporation of organic matter) rather than by photosynthetic activity. It has been shown for example that both Synechococcus and Prochlorococcus are capable to assimilate amino acids in surface waters of the South Atlantic Subtropical front (Zubkov and Tarran 2005) and also that Prochlorococcus followed pronounced diel patterns in 3H-leucine and 35S-methionine uptake with a minimum occurring at midday as shown in the Atlantic (Mary et al. 2008). In spite that organic matter uptake by freshwater Synechococcus has been seen to follow diel patterns (Chen et al. 1991; Vila-Costa et al. 2006), no significantly different uptake rates between night and day were measured for Synechococcus during our sampling in Blanes Bay chapter II 85 (Ruiz-González et al. 2011). Disruption of the diel patterns in Picoplankton community structureChlorophyll a concentration increased from one week to the next concomitantly with the shift in picophytoplankton community structure, dominated numerically by pPeuk during the first diel cycle and by Synechococcus during the second. Besides, these changes in community structure were preceded on the 25th of February by shifts in wind direction, rainfall and turbulence conditions. If it has been shown that Synechococcus cell cycle was relatively little impacted by strong hydrological variability when compared with other picophytoplankton groups (Jacquet et al. 2002), the supremacy of Synechococcus during the second cycle could also be promoted by the resuspension by wind and turbulence that changed nutrients availability, differentially modifying the activities of the different marine organisms as it has been shown elsewhere (Cotner 2000; Garstecki et al. 2002), dramatically increasing Synechococcus growth rates (Agawin et al. 2000; Agawin et al. 2003; Lindell and Post 1995) over those of Prochlorococcus and Picoeukaryotes, as corroborated by the higher division rates measured for Synechococcus during the second diel cycle in comparison with other picophytoplankton groups (Figure 5). In comparison with the first cycle, the diel patterns of abundances during the second cycle appeared disrupted, particularly during the first and second day of observation, with a tendency for the recovery of the diel patterns towards the third day. Comparing with abundances, the pigment fluorescence and the SSC parameter of Prochlorococcus, Synechococcus and pPeuk presented a pronounced and very stable diel pattern during the two cycles, suggesting that in spite of the shift in community structure provoked by the turbulence and wind event, the single-cell biology was still following a regular day-night pattern. The fact that only the diel patterns in abundances were altered and not the other parameters indicate more likely an imbalance between growth and loss processes. As already commented by Seymour et al. (2005), different underlying factors can explain the loss processes in picophytoplankton communities, including grazing by heterotrophic nanoflagellates (HNF) (Dolan and Simek 1999), and also viral lysis (Suttle and Chan 1994). In view of the fact that the diel variations in picophytoplankton community structure indicate that loss processes do not occur at a uniform rate (Vaulot et Marie 1999), that grazing activity by HNF could vary with the picophytoplankton cell cycle (Christoffersen 1994; Dolan and Simek 1999; Christaki et al. 2002) and that diel variability of viral infection has been also demonstrated (Weinbauer et al. 1995), thereof, disruption of the diel periodicity in HNF or viruses abundance and activity could result in an increase in prey abundance. Effectively, Ruiz-González et al. (Ruiz-González et al. 2011) showed patterns of maximal grazing activity on pPeuk populations at nights and particularly Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 86 pronounced during the first cycle, but weakened during the second cycle. Similarly, the strong variability in HNF abundance showing decreasing trends during the first day of the second cycle directly after the wind event was likely to explain such disequilibria between rates since it resulted in a general increase of all picoplankton group abundances. Whether or not the observed changes between weeks in community structure are linked with the sampling of different water masses over the period covering the diel cycles is difficult to ascertain and exclude. No major variations of temperature or salinity were measured during this study suggesting that we followed a relatively stable water mass. However, a relatively and significantly lower light scattering measured the second week for Synechococcus might also suggest that we followed two different populations, a first one with high SSC values and low division rates (as estimated from the ratio of the maximum to the minimum SSC over one day) during the first week, and a second with smaller cells, but presenting higher division rates, possibly enhanced by the changes in water mass characteristics. Coupling between heterotroph and phototroph parametersDuring the first cycle, bacterial abundance followed pronounced diel patterns, strongly phased with the relative activity measurements (number of CTC+ cells as well as %HNA) peaking around midnight and strongly correlated with Synechococcus and Picoeukaryotes concentration. During the second cycle, only bacterial abundance was correlated with all the picophytoplankton groups (Pearson’s tests, p<0.05), and no coupling with bacterial activity estimators was found. In view of the fact that the abundance of active cells has been related with bacterial production and cell growth (Lovejoy et al. 1996; del Giorgio et al. 1997; Choi et al. 1996; Sherr et al. 1999), then, it is reasonable to consider that bacteria were more active during the night period, an idea supported by the observation of bulk and group-specific bacterial production particularly enhanced during the dark period (RuizGonzález et al. 2011). The tight phasing between picophytoplankton parameters and bacterial abundance or activity found during the first cycle could indicate that the release of dissolved organic matter originating from phytoplankton growth and division processes was directly used to support heterotrophic activity (Nagata et al. 2000). Similarly, the pronounced diel patterns in bacterial production measured during the first cycle (Ruiz-González et al. 2011) and the extremely high variability associated to its diel fluctuations (calculated to be 37% during the two cycles as measured by 3H-Leucine incorporation) supports also the idea that bacteria rapidly responded to diel changes in organic matter release from phytoplankton (Hagström et al. 2001). Ruiz-González et al. (2011) observed that the weekend turbulence event affected little bacterial community structure and activity, except for Gammaproteobacteria uptake activity, which increased during chapter II 87 the second cycle and followed the same increasing patterns than those measured for the different picophytoplankton groups. Pinhassi et al. (Pinhassi et al. 2004) hypothesized that the shift of bacterial community structure occurring after a turbulence event was more likely driven by the shift suffered by phytoplankton community structure than by the physical stress alone. It is known that DOM issued from egestion by grazing activity can represent up to 65% of DOM and most of bacterial C demand (Nagata et al. 2000). However, no night peaks of CTC+ or %HNA, nor a night increase of bacterial production (Ruiz-González et al. 2011) were observed during the second cycle, only the same trend of increase of both parameters with HNF abundance (only corroborated by a remarkably high but not significant correlation between CTC+ and %HNA with HNF abundance) strengthened the idea that bacteria were more likely using the dissolved organic matter released from grazing activity to support their growth rather than the excreted primary production. Implications of the short time scale variability on patternsConsequence of the shift in community structure between weeks likely to be provoked by the turbulence event between the diel cycles, the ratio between the two major contributors to picophytoplankton biomass (Syn:pPeuk) increased by two fold from one week to another, from 0.03 to 0.06, and something that points out the importance of transient meteorological events in structuring coastal planktonic communities (Guadayol et al. 2009) and that significant ecological events often results from episodic physical forcing operating at short timescales (Seymour et al. 2005). Similarly, the variability of this ratio was also important at the daily scale since the same two-fold increase was measured during both the first and second cycle and was driven by the differences in the diel pattern amplitudes. Using microscopy to count the number of cells grazed per HNF, it appeared that the number of Picoeukaryotes grazed (mostly Micromonas at this period of the year, details not shown) increased all along the first cycle (Ruiz-González et al. 2011). A differential grazing pressure operating at the diel scale on Picoeukaryotes and Synechococcus would then likely contribute to the differential group contribution to total picoplankton biomass. Acknowledgements: We thank all scientists of the CEAB-CSIC in Blanes for the help and laboratory space provided during the study and, in particular, E. O. Casamayor for help with logistics. This work was supported by the Spanish MICINN through projects MODIVUS (CTM2005-04795/MAR) and SUMMER (CTM2008-03309/MAR). We thank M. Galí for his help with Matlab and Figure 1, C. Cardelús, V. Balagué, I. Forn, I. Lekunberri and all the people who participated in the Blanes Bay winter diel study for their assistance with sample collection and processing. Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 88 REFERENCES Agawin, N. S. R. Duarte, C. M. and Agustí, S. (2000) Nutrient and temperature control of the contribution of picoplankton to phytoplankton biomass and production. Limnol. Oceanogr., 45, 591--600. Agawin, N. S. R. Duarte, C. M. Agustí, S. and McManus, L. (2003) Abundance, biomass and growth rates of Synechococcus sp. in a tropical coastal ecosystem (Philippines, South China Sea). Estuar. Coast. Shelf Sci., 56, 493--502. Atkinson, A. Ward, P. Williams, R. and Poulet, S. A. (1992) Feeding rates and diel vertical migration of copepods near South Georgia: comparison of shelf and oceanic sites. Mar. Biol., 114, 49--56. Binder, B. J. Chisholm, S. W. Olson, R. J. Frankel, S. L. and Worden, A. Z. (1996) Dynamics of picophytoplankton, ultraphytoplankton, and bacteria in the Central Equatorial Pacific. Deep-Sea Res., 43, 907--31. Campbell, L. Nolla, H. A. and Vaulot, D. (1994) The importance of Prochlorococcus to community structure in the central North Pacific Ocean. Limnol. Oceanogr., 39, 954--961. Campbell, L., and Vaulot, D. (1993) Photosynthetic picoplankton community structure in the subtropical North Pacific Ocean near Hawaii (station ALOHA). Deep-Sea Res., 40, 2043--2060. Chen, T. H., Chen, T. L. Hung, L. M. and Huang, T. C. (1991) Circadian rhythm in amino acid uptake by Synechococcus Rf-1. Plant Physiol., 97, 55--59. Chisholm, S. W. Olson, R. J. Zettler, E. R., Waterbury, J. Goericke, R. and Welschmeyer, N. (1988) A novel free-living prochlorophyte occurs at high cell concentrations in the oceanic euphotic zone. Nature, 334, 340--343. Chisholm, S.W. (1992) Phytoplankton size. In Falkowski, P.G. Woodhead, A.D. (Ed.), Primary Productivity and Biogeochemical Cycles in the Sea. Plenum Press, New York, pp. 213--237. Choi, J. W. Sherr, E. B. and Sherr, B. F. (1996) Relation between presence-absence of a visible nucleoid and metabolic activity in bacterioplankton cells. Limnol. Oceanogr., 41, 1161--1168. Christaki, U. C. Courties, H. Karayanni, A. Giannakourou, C. Maravelias, K. A. Kormas and Lebaron, P. (2002) Dynamic characteristics of Prochlorococcus and Synechococcus consumption by bacterivorous nanoflagellates. Microb. Ecol., 43, 341--352. chapter II 89 Christoffersen, K. (1994) Variations of feeding activities of heterotrophic nanoflagellates on picoplankton. Mar. Microb. Food. Webs., 8, 111--123. Cotner, J. B. (2000) Intense winter heterotrophic production stimulated by benthic resuspension. Limnol. Oceanogr., 45, 1672--1676. Davis, H. T. (1941) The Analysis of Economic Time Series. The Principia Press, Inc. Bloomington., Indiana. Dolan, J. R. Simek, K. (1999) Diel periodicity in Synechococcus populations and grazing by heterotrophic nanoflagellates: analysis of food vacuole contents. Limnol. Oceanogr., 44, 1565-- 1570. Durand, M. D. and Olson, R. J. (1998) Diel patterns in optical properties of the chlorophyte Nannochloris sp.: Relating individual-cell to bulk measurements. Limnol. Oceanogr., 43, 1107-- 1118. Durand, M. D. Green, R. E., Sosik, H.M. and Olson, R. J. (2002) Diel variations in optical properties of Micromonas pusilla (Prasinophyceae). J. Phycol., 38, 1132--1142. Falcioni, T. Papa, S. and Gasol, J. M. (2008) Evaluating the flow cytometric Nucleic Acid Double Staining Protocol (NADS) in realistic planktonic bacterial death situations. Appl. Environ. Microbiol., 74, 1767--1779. Fuhrman, J. A. Eppley, R. W. Hagström, A., and Azam, F. (1985) Diel variations in bacterioplankton, phytoplankton, and related parameters in the Southern California bight. Mar. Ecol. Progr. Ser., 27, 9--20. Fuller, W.A. (1976) Introduction to Statistical Time Series. New York: John Wiley & Sons. Garstecki, T. Wickham, S. A., and Arndt, H. (2002) Effects of experimental sediment resuspension on a coastal planktonic microbial food web. Estuar. Coast. Shelf, 55, 751--762. Gasol, J. M. del Giorgio, P. A., and Duarte, C. M. (1997) Biomass distribution in marine planktonic communities. Limnol. Oceanogr., 42, 1353--1363. Gasol, J. M. Doval, M. D., Pinhassi, J. Calderón-Paz, J. I. Guixa-Boixareu, N., Vaqué, D. and Pedrós-Alió C. (1998) Diel variations in bacterial heterotrophic activity and growth in the northwestern Mediterranean Sea. Mar. Ecol. Progr. Ser., 164, 107-124. Diel patterns of picoplankton community structure and activity in NW Mediterranean Sea 90 Gasol, J. M. and del Giorgio, P. A. (2000) Using flow cytometry for counting natural planktonic bacteria and understanding the structure of planktonic bacterial communities. Sci. Mar., 64, 197- -224. Gasol, J. M. and Arístegui, J. (2007) Cytometric evidence reconciling the toxicity and usefulness of CTC as a marker of bacterial activity. Aquat. Microb. Ecol., 46, 71--83. Ghiglione, J. F. Mével, G. Pujo-Pay, M. Mousseau, L. Lebaron, P. and Goutx, M. (2007) Diel and seasonal variations in abundance, activity, and community structure of particle-attached and freeliving bacteria in NW Mediterranean Sea. Microb. Ecol., 54, 217--231. del Giorgio, P. A. Prairie, Y. T. and Bird, D. F. (1997) Coupling between rates of bacterial production and the abundance of metabolically active bacteria in lakes, counting using CTC reduction and flow cytometry. Microb. Ecol., 34, 144—154. González, J. M. Sherr, E. B. and Sherr, B. F. (1990). Size-selective grazing on bacteria by natural assemblages of estuarine flagellates and ciliates. Appl. Environ. Microbiol., 56, 583--589. Gough, L. H. (1905) Report of the plankton of the English Channel in 1903. Report of the North Sea Fisheries Investigatory Commission (South Area), 1, 325--377. Grégori, G. Citterio, S. Ghiani, A. Labra, M. Sgorbati, S. Brown, S. and Denis, M. (2001) Resolution of viable and membrane-compromised bacteria in freshwater and marine waters based on analytical flow cytometry and nucleic acid double staining. Appl. Environ. Microbiol., 67, 4662--4670. Guadayol, O. Peters, F. Marrasé, C. Gasol, J. M. Roldán, C. Berdalet, E. Massana, R. and Sabata, A. (2009) Episodic meteorological and nutrient-load events as drivers of coastal planktonic ecosystem dynamics: a time-series analysis. Mar. ecol. Prog. Ser., 381, 139--155. Hagström, A. Azam, F. Kuparinen, J. and Zweifel, U. L. (2001) Pelagic plankton growth and resource limitations in the Baltic Sea. In Wulff, F.V. Rahm, L.A. and Larsson, P. (Ed.), A Systems Analysis of the Baltic Sea. Springer-Verlag, pp. 177--210. Herndl, G. J. and Malacic, V. (1987) Impact of the pycnocline layer on bacterioplankton: diel and spatial variations in microbial parameters in the stratified water column of the Gulf of Trieste (Northern Adriatic Sea). Mar. Ecol. Progr. Ser., 38, 295--303. Ishizaka, J. Harada, K. Ishikawa, K. Kiyosawa, H. Furusawa, H. Watanabe, Y. Ishida, H. Suzuki, K. Handa, N. and Takahashi, M. (1997) Size and taxonomic plankton community structure and chapter II 91 carbon flow at the equator. Deep-Sea Res. II, 44, 1927--1949. Jacquet, S. Lennon, J. F. Marie, D. and Vaulot, D. (1998) Picoplankton population dynamics in coastal waters of the northwestern Mediterranean Sea. Limnol. Oceanogr., 43, 1916--1931. Jacquet, S. Partensky, F. Lennon, J. F. and Vaulot, D. (2001) Diel patterns of growth and division in marine picoplankton in cultures. J. Phycol., 37, 357--369. Jacquet, S. Prieur, L. Avois-Jacquet, C. Lennon, J. F. and Vaulot, D. (2002) Short-timescale variability of picophytoplankton abundance and cellular parameters in surface waters of the Alboran sea (western Mediterranean). J. Plankton research, 44, 635--651. Jakobsen, H. H. and Strom, S. L. (2004) Circadian cycles in growth and feeding rates of heterotrophic protist plankton. Limnol. Oceanogr., 49, 1915--1922. Jeffrey, W. H. Pledger, R. J. Aas, P. Hager, S. Coffin, R. B. VonHaven, R. and Michell, D. L. (1996) Diel and depth profiles of DNA photodamage in bacterioplankton exposed to ambient solar ultraviolet radiation. Mar. Ecol. Progr. Ser., 137, 283--291. Jezbera, J. Hornak, K. and Simek, K. (2005) Food selection by bacterivorous protists: insight from the analysis of the food vacuole content by means of fluorescence in situ hybridization. FEMS Microbiol. Ecol., 52, 351--363. Johnson, P. W. and Sieburth, J. McN. (1982) In-situ morphology and occurrence of eukaryotic phototrophs of bacterial size in the picoplankton of estuarine and oceanic waters. J. Phycol., 18, 318--327. Johnson, C. H. Golden, S. S. Ishiura, M. and Kondo, T. (1996) Circadian clocks in prokaryotes. Mol. Microbiol., 21, 5--l1. Landry, M. R. Lehner-Fournier, J. M, Sundstrom, J. A. Fagerness, V. L and Selph, K. E. (1991) Discrimination between living and heat-killed prey by a marine zooflagellate Paraphysomonas vestita (Stokes). J. Exp. Mar. Biol. Ecol., 146, 139--152. Landry, M. R. Kirshtein, J. and Constantinou, J. (1995) A refined FLB-dilution approach for measuring the community grazing impact of microzooplankton, with experimental tests in the equatorial Pacific. Mar. Ecol. Prog. Ser., 120, 53--63. Lefort, T. Not, F. Probert, I. Marie, D. and Gasol, J. M. (2011) Direct determination of carbonconversion values for ecologically relevant photosynthetic picoeukaryotes. Submitted to Limnol. Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 98 chapter III 99 ABSTRACT The temporal and spatial variability of Picoplankton Community Structure (PCS) and heterotrophic activity were studied by flow cytometry and radioactive tracers during a cruise performed in the NW Mediterranean. Variability was measured at a short time scale in diel cycles performed in a coastal and an offshore station and was compared to the large time scale variability estimated from two years of survey at the Blanes Bay microbial Observatory (the coastal station). Synechococcus dominated numerically in coastal and surface waters and was the main contributor to picophytoplankton biomass in all stations, followed by Picoeukaryotes at the coastal and slope stations. The maximum Prochlorococcus contribution was constrained within oceanic wellstratified situations. While Picoeukaryote cell numbers exhibited the highest spatio-temporal variability, the lowest was found for bacterial abundance. When we compared the different sources of variability, we found that the largest one observed was at the spatial scale, vertically promoted by water column stratification, and horizontally by the differences in trophy between stations. Coastal stations presented high bacterial abundance and activity but low spatio-temporal variability. On the contrary, offshore waters presented lower bacterial abundances and activities but higher spatio-temporal variability. Finally, opposite patterns between the Synechococcus to Picoeukaryotes biomass ratio and chlorophyll a levels were observed not only spatially, but also at both the short-term and large temporal scale, representing a possible good candidate variable to act as ecological indicator of the trophic state. Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 100 INTRODUCTION The picoplanktonic (< 2-3 mm) community in aquatic environments is formed by the heterotrophic bacteria and Archaea, and the picophytoplankton, in turn composed by Synechococcus (Waterbury et al. 1979), Prochlorococcus (Chisholm et al. 1988) and Picoeukaryotes (Johnson and Sieburth 1982). The large-scale temporal and spatial distribution patterns of these groups have a strong seasonal component (Campbell et al. 1997; Jacquet et al. 1998; Li 1998; Grégori et al. 2001; Li and Dickie 2001) and their relative contribution to picoplankton community structure varies not only with ecosystem trophic level (Zhang et al. 2008), but also with temperature and with stratification of the water column (Bouman et al. 2011). While Synechococcus have been shown to dominate in nutrient-rich and coastal environments (Partensky et al. 1999), Prochlorococcus often dominate numerically in warm, more oligotrophic and well-stratified waters and generally extend much deeper than Synechococcus (Partensky et al. 1999). In comparison with photosynthetic prokaryotes, eukaryotic picophytoplankton (pPeuk) are relatively less abundant, but can dominate in terms of biomass in a variety of ecosystem (Li et al. 1992; 1993; Worden et al. 2004). Different controlling factors are underneath these distribution patterns, and macro-ecological studies have shown that not only temperature, but also nitrate and chlorophyll a concentration contribute up to 66% of the variance in picophytoplankton abundance (Li 2007). Heterotrophic bacteria are also known to increase following chlorophyll a and temperature at large scales (e.g. (Li et al. 2004), but not necessarily in a given study area (Li 2009). The fact that consistent distribution patterns of heterotrophic bacterial abundance and bacterial activity have been identified across a range of trophic levels (as estimated from chlorophyll a concentration) (Cole et al. 1988; Billen et al. 1990 Ducklow and Carlson 1992; Bird and Kalff 1984), has given support to the idea that bacteria use mainly the dissolved organic matter produced by phytoplankton and grazers to support their heterotrophic activities (Nagata 2000; Moran et al. 2002) . In comparison, the factors contributing to the variability in PCS and heterotrophic activity at shorter spatio-temporal scales have been less studied. Picoplanktonic group abundance has been shown to fluctuate drastically over short distances in the Celtic Sea (<1 km), indicating that the magnitude of variation in the PCS patterns at the short spatial scale might be greatly underestimated in comparison to larger spatio-temporal patterns (Martin et al. 2005). Similarly the relevance of diel variability is commonly disregarded as compared to monthly or annual variability. However, several studies have shown that Synechococcus, Prochlorococcus and Picoeukaryotes abundances follow daily variations, with diel oscillations of their pigment content (e.g. Jacquet et al. 1998; Jacquet et al. 2002; Lefort et al. submitted; Vaulot et Marie, 1999) and generally associated to a synchronized pattern of cell division, but this short-term variability has still not been compared to the variability at other scales. Similarly, bacterial activity can vary over short periods, likely chapter III 101 due, for example, to variations in phytoplankton extracellular release of DOC (Gasol et al. 1998; Ruiz-González et al. 2012; Lefort et al. Chapter II). Light drives these variations resulting from the balance between growth rates (linked to light, nutrient availability and nutrient quality) and mortality rates (linked to grazer, viral activity or physical stresses such as UV radiation). Short time-scale variations in bacterial activity have been observed to be at least as large as those created by the seasonal variations (Ruiz-González et al. 2012). The links between the spatio-temporal shifts occurring in both PCS and heterotrophic activities and the shifts in structure and function of whole ecosystems are unclear. If some ecological parameters (such as abundance or activity of one or all picoplanktonic groups) can be used to estimate the changes occurring in ecosystem conditions (Karr 1991; Beaugrand 2005) then these microbial community structure or activity variables might capture the complexity of the ecosystem (Paerl et al. 2003), and be useful as indicators, one of the conditions being that they must be simple enough to be routinely and easily measured (Dale and Beyeler 2001), something that picoplankton abundances and activities are. We describe here picoplankton group distribution and heterotrophic production across different spatio-temporal scales with the aim of i) identifying patterns in picoplankton group distribution and activity at each spatio-temporal scale (including the short time scales), ii) quantify and compare variability at each scale, and iii) to determine the links between the variability and different environmental and ecological factors, such as stratification and trophic level (as estimated from chlorophyll a concentration). To address these issues, we followed the different group abundances (also including viruses) and bacterial activity using flow cytometry and incorporation of radioactive tracers, during a cruise performed in the North Western Mediterranean Sea in September 2007 (mainly) and in some additional samplings. We measured for each variable the coefficients of variation as estimators of parameter variability and we compared the values at different spatio-temporal scales: the temporal scale was divided into 2: the diel scale (with two diel cycles performed at a coastal station and one at an open-ocean station) and a larger temporal scale (seasonal year-round variations in the same parameters). The spatial scale was analyzed during the transect across 5 stations from coastal to deep ocean sites, comparing the vertical profiles with the horizontal distribution of the different picoplankton groups. Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 102 Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea MATERIAL AND METHODS Sampling - Diel cyclesTwo diel cycles were sampled in September 2007 from onboard the oceanographic vessel “García del Cid”, during the cruise “MODIVUS” in the NW Mediterranean Sea (Figure 1). The first one was performed during 56 hours from the 18th to the 20th September 2007 at a coastal station (the Blanes Bay Microbial Observatory, Station C), a shallow (20 m depth) oligotrophic coastal site, located 800 m offshore of Blanes (41º39.90’N, 2º48.03E). The second diel cycle was done at an offshore Station D (40º39´4.7”N, 2º51´1.6” E), in the deepest point of the Catalan Sea. Its duration was shortened to 44 hours from the 23rd to the 25th February 2007, because of sea conditions. The samples were collected at a frequency of 6 per day (every 4 hours) with Niskin bottles mounted on a rosette with a CTD. While 2 different depths were sampled at station C (at surface: 5 m and 15 m), 4 different depths were sampled at station D (5 m, 25 m, 48 m, 65 m). We also include the data from two other diel cycle studies performed in winter at Station C (20-22 February 2007 and 26 February to 1st of March 2007) with the same sampling frequency (published in Lefort & Gasol, submitted). Figure 1. Bathymetric Map of the different sampling sites during oceanographic cruise MODIVUS, Blanes Bay coastal station C was the first station of the transect, D was the last one. Note the Blanes submarine canyon close to station CM. The diel cycles were performed at stations C and D. chapter III 103 chapter III Spatial studyBetween the two summer diel cycles, a coast to offshore transect was sampled in September 2007, and five vertical profiles taken at stations C, CM, M, MD, D (Figure 1). All the samples were kept permanently in the dark until analysis, which was done onboard. General samplesTemperature and salinity were obtained with a SAIV A/S 204 CTD probe, except in the spatial and diel cycle studies of September 2007 in which a CTD SBE 9plus was used. Chlorophyll a concentration was determined from 150 mL of seawater filtered through GF/F filters (Whatman) extracted in acetone (90% v/v), and fluorescence measured with a Turner Design fluorometer. Sampling - Seasonal survey at the Blanes Bay microbial observatory (BBMO)- To compare the diel scale with the long time scale variability, we followed picoplankton community structure and heterotrophic activity at the Blanes station C (BBMO) from January 2007 to November 2009. The samples were taken monthly, collected with polycarbonate carboys and processed by flow cytometry at the ICM facilities, 2 h after sampling. The samples were always kept at dark until analysis. Picoplankton and virus abundancesAlgal and bacterial abundance were determined using flow cytometry (Marie and Partensky 2006; Gasol and del Giorgio 2000). Three populations from the photosynthetic fraction of the picoplankton (<3 mm) (Prochlorococcus, Synechococcus, Picoeukaryotes) were discriminated according to scatter and fluorescence signals, for which the samples were run at high speed (at ca. 100 µl min-1) without additional fixative. For nonphototrophic bacteria, abundance was estimated following two different methods, a first one with the use of fixative and the second one with the NADS protocol in unfixed samples. Bacterial abundances during the long term survey at BBMO were measured taking 1.2 mL samples which were preserved with 1% paraformaldehyde + 0.5% glutaraldehyde (final conc.), and kept frozen at –80ºC until analysis by SybrGreen I (dilution x10,000) staining and flow cytometric analysis (Becton-Dickinson FACSCalibur flow cytometer). Bacteria were detected by their signature when plotting side scatter (SSC) versus green fluorescence (FL1) and FL1 vs. red fluorescence (FL3) (Gasol and del Giorgio 2000) and converted to abundances measuring the volume of sample before and after sample passage. Bacterial abundances during the oceanographic cruise were also performed with the NADS Viability protocol, based on the combination of the cell-permanent nucleic acid strain SybrGreen I (Molecular Probes, Eugene, OR) and the cell-impermeant propidium iodine PI (Sigma Chemical Co.) fluorescent probe. We used a 1:10 SG1 and 10 µg ml-1 PI concentrations. After simultaneous addition of each stain, the samples were incubated for 20 min in the dark at room temperature and then analyzed by flow cytometry. SG1 and PI fluorescence were detected in the green (FL1) and orange-red (FL3) cytometric channels, respectively. A dot plot of red versus Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 104 green fluorescence allowed distinction of the “live” cell’s cluster (i.e., cells with intact membranes and DNA present) from the “dead” cell one (i.e., with compromised membranes) (Grégori et al. 2001, Falcioni et al. 2008). Total cell abundances were the addition of the “live” and “dead” cells. VirusesViral abundance was also determined by flow cytometry. Subsamples (2 ml) were fixed with glutaraldehyde (0.5% final concentration), quick frozen in liquid nitrogen and stored at -80°C as described by Marie et al. (1999). The samples were stained with SYBRGreen I, and run at a medium flow speed following standard protocols (Brussaard 2004). Bacterial single-cell activityWe measured the abundance of highly respiring prokaryotes, i.e. those able to reduce 5-cyano-2,3-diotolyl tetrazolium chloride (CTC; Polysciences). CTC turns into a red fluorescent formazan that is detectable by epifluorescence and flow cytometry (Sherr et al. 1999a; Sieracki et al. 1999). Sample aliquots (0.4 ml) were amended with 5 mM CTC (from a fresh stock solution at 50 mM) immediately following collection and were incubated for 90 min in the dark at room temperature. CTC-positive (CTC+) cells were enumerated by flow cytometry using the FL2-versus-FL3 dot plot (see Gasol and Arístegui 2007). For these analyses, we used a high speed (ca. 100 µl m-1) and a threshold set in red fluorescence. Bacterial heterotrophic production (BHP)- BHP was estimated every 4 hours from both, radioactive 3H-leucine and 3H-thymidine incorporation. For leucine we used the 3H-leucine incorporation method described by Kirchman et al. (1985) adapted to microfuge tubes. Briefly, 4 aliquots (1.2 mL) and 2 TCA-killed controls were incubated with radiolabeled leucine (40 nmol L-1, final conc., 160 Ci mmol-1) for about 1.5 hours in the dark at in situ temperature. The incorporation was stopped by adding 120 ml of cold TCA 50% to the samples, which were stored at –20ºC until processing by the centrifugation method of Smith and Azam (1992). Bacterial production was also measured as 3H-thymidine incorporation following Fuhrman and Azam (1980) also in microfuge tubes. Samples were incubated with 10 nmol L-1 3H-thymidine (final concentration) and processed like the 3H-leucine samples. Standard cell to carbon Conversion factors for biomass estimationTo translate cell abundance into biomass, different CF were chosen for Prochlorococcus, Synechococcus and Picoeukaryotes, with respectively 53 fgC cell-1 and 250 fgC cell-1 (Campbell et al. 1994) and 1540 fgC cell-1 for pPeuk (Lefort et al. Chapter I). Integration of the dataTo compare Synechococcus, Prochlorococcus and Picoeukaryotes abundances, we calculated the integrated average at stations C and D. To estimate the diel variability, we followed the variations of the ratio of the depth-integrated average (every 4 hours) to the diel average. For the transect, the data were integrated over the photic zone for the picophytoplankton chapter III 105 (0-125 m) as well as for bacterial and virus abundance, CTC activity, and bacterial production. No depth integration was done with the long-term survey performed at BBMO since only one depth had been sampled (5 m). Statistical analysisTo estimate the variability of each abundance or activity parameter, we calculated the respective coefficient of variations (CV) expressed for the diel scale as the standard deviation of the depth-integrated values divided by the diel averages, For the variability of the vertical profiles, CVs were expressed as the standard deviation of the depth-integrated values divided by the depth-integrated average of the station, for every station of the transect. For the horizontal variability, CVs were expressed as the standard deviation of the different depth integrated averages measured at each station, divided by the transect average. The CVs measured during the diel cycles were compared to the CVs measured during the long time-scale survey. To perform correlation analyses, we run Pearson correlations that summarized the strength of the linear relationships between each pair of response variables. All conducted in JMP 7 (SAS institute Inc). Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 106 RESULTS To describe the patterns in picoplankton group distribution, we calculated for each parameter the depth integrated average at different spatio temporal scales in 5 stations from the coastal station C to open-ocean station D (C, CM, M, MD, D, Table 1A). We estimated the variability generated by the distributions, calculating the coefficients of variation (CV) for every picoplankton group as a proxy of the amplitude of variation around the integrated average (Table 1B). Mesoscale variability of Picoplankton community structure (horizontal variabilityHeterotrophic bacteria dominated numerically the picoplankton at all the stations of the transect and at all the scales of observation (Table 1A). Despite no strong differences in the distribution were measured, as evidenced by the low variability (average of only 27%, Table 1B), bacterial concentration was relatively higher at the coastal station C than at station D (Table 1A), with a maximum at the slope station CM of 8.6 x 105 cells.ml-1 (±10%) that followed the isopycnal 27 kg m-3 to surface waters of station MD. This density dome, which possibly indicates an upwelling event or a cyclonic vortice, frequent in this region (La Violette et al. 1990), was confirmed in satellite images (details not presented) and separated a patch of lower bacterial concentration at station D with 4.3 x 105 cells.ml-1 (±71%) (Figure 2D). Compared to the low variability found in the bacterial abundance parameter, much higher variability was measured in bacterial activity, ranging from 43% to 53% (close to two fold) either when estimated for the percent of actively respiring cells (%CTC) or for 3H-Thymidine or 3H-Leucine incorporation rate values. Synechococcus followed similar spatial distribution patterns than heterotrophic bacteria as revealed by the strong correlation found between their abundances (Pearson tests, N=30, p<0.005, Suppl. Table 1C), but higher coefficients of variation were observed horizontally from coastal station C to offshore station D (Table 1B). Synechococcus dominated numerically the picophytoplankton community structure at coastal station C that was characterized by lower salinity and temperatures than the offshore station D (Figure 2A, Figure 2E). At this station, Synechococcus cell concentrations ranged from 6.8 x 104 to 8.0 x 104 cells.ml-1 at 15 and 28 meters depth respectively for a depth integrated average of 5.9 x 104 cells.ml-1 (Table 1A). Another maxima of Synechococcus abundance of 5.9 x 104 cells.ml-1 was observed at the surface in station MD, which corresponded also with the density dome (Figure 2A and 2E). Prochlorococcus dominated numerically the photosynthetic fraction of the picoplankton in more offshore conditions with a depth integrated average of 7.1 x 104 cells.ml-1, particularly at the DCM of Station D, with a cell concentration of 1.2 x 105 cells.ml-1 (Table 1A, Figure 2B). chapter III 107 Table 1. Depth averaged cell concentration (cells. ml-1) (A) and variability (Coefficient of variation) (next page) (B) of the different picoplankton groups and activity indices considered. The averages were calculated from the integrated data over the photic zone. The coefficient of variation (CV) of each parameter was calculated dividing e.g. the standard deviation of the diel depth integrated value by the diel average. For the long-term survey at station C, Picoplankton group abundances at the Blanes bay station (5 m) are averaged from January 2007 to November 2009 over the whole period. NA: not available. A STATION C STATION CM STATION M STATION MD STATION D ST C, CM, M, MD, D DEPTH AVERAGED PARAMETER DIEL CYCLE LONG TERM SURVEY VERTICAL PROFILE VERTICAL PROFILE VERTICAL PROFILE VERTICAL PROFILE VERTICAL PROFILE DIEL CYCLE TRANSECT SYNECHOCOCCUS 10 4 cells. ml -1 5.94 2.45 6.23 1.69 3.06 2.02 1.49 2.16 2.90 PROCHLOROCOCCUS 10 4 cells. ml -1 2.48 8.33 2.21 2.67 2.79 2.86 7.10 6.55 3.54 PICOEUKARYOTES 10 3 cells. ml -1 1.95 3.55 1.46 1.55 0.42 0.33 0.82 0.79 0.92 HET. BACTERIA 10 5 cells. ml -1 7.78 7.29 8.53 7.30 8.02 7.07 6.19 6.42 7.42 CTC+ CELL ABUNDANCE 10 4 cells. ml -1 5.67 --- 6.29 7.76 4.57 2.59 3.13 3.80 4.87 BACTERIAL ACTIVITY pmol Tdr L -1 h -1 10.91 NA 12.54 8.36 3.47 8.99 3.78 7.43 7.43 BACTERIAL ACTIVITY pmol Leu L -1 h -1 41.10 49 69.15 35.83 26.96 21.41 27.29 18.42 36.13 VIRUS ABUNDANCE 10 6 ml -1 8.33 N.A 6.60 6.43 5.82 4.29 6.22 5.81 5.87 SYN:PPEUK BIOMASS RATIO 6.17 2.34 6.94 1.77 11.80 9.89 2.93 4.58 8.89 CHLOROPHYLL A µg L -1 0.17 0.48 0.14 0.35 0.22 0.17 0.21 0.26 0.22  Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 114 DISCUSSION Picoplankton group variabilityPatterns in picoplankton community structure (understood as the differential contribution of each organism type to the whole community) result from the structuring effect of interconnected physical and biological processes, which generally occur simultaneously but at different scales (Ducklow 1984; Dickey 1991). These interconnected factors participate greatly in the distribution of the different picoplankton groups, and are relatively well described for Synechococcus and Prochlorococcus (Partensky et al. 1999) but less well known for pPeuk and heterotrophic bacteria. Our analysis indicates that each Picoplankton group varied differently across the different scales of analysis, and Picoeukaryote abundance showed the highest average variability and heterotrophic bacteria the lowest (Table 1B). Neither bacteria nor pPeuk communities (followed in this study by cytometry) constitute homogeneous groups but are both composed by several phylogenetic groups that likely vary differentially. While we followed the pPeuk at the bulk level by flow cytometry, without information about phylogenetic composition and relative group contribution to community structure, it is likely that the high variability measured in pPeuk bulk abundance resulted from the shifts occurring in community structure across the different stations of the transect. Several studies have shown that pPeuk community structure vary according to oceanic region, nutrient characteristics of the water masses (coastal or open-ocean, eutrophic or oligotrophic) and time of the year (see review by Worden and Not 2008). For instance, Prasinophyceae (Archaeplastida) typically dominate pPeuk communities in coastal waters (Not et al. 2004), while Prymnesiophyceae (Haptophyta), and to a lesser extent Chrysophyceae and Pelagophyceae (Heterokontophyta), contribute more to pPeuk communities in more open-ocean ecosystems(Mackey et al. 2002; Fuller et al. 2006; Liu et al. 2009). In contrast, the relatively lower bacterial abundance variability, an average of 27% (Table 1B), suggests that bacterial community structure was spatially and temporarily more stable. Comparative studies of picoplankton community structure in different ecosystems have shown that spatial variability in the bacterial heterotrophic biomass is less strong than in autotrophic picoplankton (Zhang et al. 2008), Massana et al. (2004) also noticed such discrepancy between community variability and observed a high temporal variability of pPeuk assemblages compared to the small seasonal changes that tend to occur in bacterioplankton not only in the same site that had been studied by fingerprinting by Schauer et al. (2003) but also at the SOLA station (Banyulssur-mer, France), a site relatively close from the Blanes Bay station (Ghiglione et al. 2005). However, a constant overall bacterial concentration or stability of phylogenetic group chapter III 115 contribution to community structure does not imply necessarily the stability at the OTU level. Indeed, while almost similar bacterial community structures were measured by FISH among the different stations of this same transect (Crespo et al. unpublished), dominated by SAR11 and Bacteroidetes in the coastal conditions and by SAR11 and Gammaproteobacteria in open-ocean conditions), strong differences in bacterial OTUs composition (as estimated by pyrosequencing) was found from a station to another by Pommier et al. (2010). Several studies have suggested that the limits of bulk bacterial and bacterial group concentration are set by ecological parameters such as grazing and trophic level (Gasol and Duarte 2000; Li et al. 2004, chapter IV). If we consider the possibility that the high average variability found for pPeuk abundance resulted from more pronounced grazing pressure or competition with other algae for nutrients, then the stability of the overall bacterial concentration more likely indicated resilience of the bacterial community as a whole. Ecosystem variabilityTo determine whether the variability observed was related to the type of ecosystem (Coastal station C and more Offshore influenced stations), we plotted for each picoplankton group their coefficient of variability (Table 1B) against their corresponding depth-integrated average concentrations measured at the different spatio-temporal scales (Table 1A, Figure 5). In this representation, as the Y and X scales share a common term (i.e. the CV are the standard deviation divided by the depth-integrated average), the relationships described should be taken as indicative and no statistical prediction is possible. However, it is interesting to observe that Prochlorococcus and heterotrophic bacteria followed two opposite patterns. While the variability of Prochlorococcus increased significantly with increasing cell abundance, the variability of bacterial abundance decreased with increasing bacterial cell densities (Figure 5A and 5B). Such opposite patterns indicate that the maximum Prochlorococcus concentration occurred at specific and narrower periods of the day or of the year and also at narrower spatial locations (at strongly stratified spots). The opposite pattern was observed for bacteria since a lower bacterial variability was in general measured at coastal station C and was associated with high cell densities. On the contrary, high variability was measured in offshore stations, and was associated with lower bacterial cell densities. This relative stability of the bacterial concentration at high cell densities in coastal ecosystems could indicate that shifts in bacterial community structure in coastal conditions, generated by the decrease of the abundance of a particular bacterial phylogenetic group, would be concomitantly balanced by the increase of abundance of another group. Moreover, the decrease of bacterial abundance variability from coastal station C to offshore station D was concordant with the parallel decrease of bacterial richness and evenness as estimated by pyrosequencing by Pommier Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 116   5A 5B Figure 5. Relationships between the spatio-temporal variability of the Picophytoplankton groups and bacterial activity with abundances or bacterial activity level. Prochlorococcus (A), Heterotrophic bacteria (B), 3H-thymidine incorporation rates (C), CTC cell abundance (D). chapter III 117 C, CM, M, MD, D for the different vertical averages and variability. “Transect” for the horizontal average and variability, “Cl” for the long-term average and variability at station C, “Cw” and “Cs” for the diel cycle averages and variability during winter 2007 and summer 2007 at station C, “Ds” for the diel cycle performed the summer 2007 at station D. “Cw” values were taken from (Lefort et al. Chapter II).   5C 5D Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 118 et al. (2010) from coastal to offshore during the same transect. This would suggest that the spatiotemporal variability of bacterial community structure was enhanced at low bacterial richness and evenness, and on the contrary, that the high number of different bacterial OTUs composing the heterotrophic bacterial community at station C offered the optimum conditions for ecological homeostasis, in agreement with the idea that stable ecosystems tend to have higher diversity as suggested by Pommier et al. (2010). It is important to note that bacterial cell concentration was integrated only for the photic zone, the low bacterial concentration in deeper layer was not taken into account but would have considerably increased the spatial vertical variability, resulting in the increase of the strength of the positive relationship found between variability and cell densities, (until a Pearson’s R= 0.92, details not shown). Vertical sources of variabilityThe vertical scale was the major source of variability for all the groups, likely enhanced by the stratification of the water column. While Synechococcus dominated more in coastal and surface waters, Prochlorococcus dominated numerically in offshore well-stratified waters, particularly at the DCM. Temperature and water column stability have been shown to participate greatly in the spatial and temporal patterns of the overall size structure of phytoplankton communities (Li 2002; Bouman et al. 2003). Opposite patterns have been shown between the Prochlorococcus preference for well stratified nutrient depleted waters and the Synechococcus numerical dominance during periods of vertical mixing in a variety of ecosystems such as the subtropical waters (Campbell et al. 1997; Bouman et al. 2011a); or in the Western North Atlantic (Zinser et al. 2007). The relatively high surface area to volume ratio typical of Prochlorococcus in comparison to the other picophytoplankton groups has been suggested to indicate an ecological advantage in oligotrophic conditions (Raven et al. 2005; Partensky and Garczarek 2010). However, stratification not only influences the supply of nutrient from deep waters, but also regulates the light environment of cells within the mixed layer. Thus, it is possible that the spatial variability measured in Synechococcus and Prochlorococcus distributions resulted from either different capacities to resist high irradiance exposure and rapid changes in light Conditions (Six et al. 2007) or the possible photo-inhibition of some of these groups (Vaulot and Marie 1999; Sommaruga et al. 2005; Llabrés and Agustí 2006; 2010). Several studies have highlighted the particular aptitude of Picoeukaryotes for growing in physically dynamic environments when compared with cyanobacteria (Lindell and Post 1995; Campbell et al. 1998; Steinberg et al. 2001). During the transect, Picoeukaryotes appeared to contribute more at the DCM of the slope station CM where the concentration of NO3 was particularly high (data not shown), and was coherent with the findings of Bouman et al. (2011) that showed that the abundance of picoeukaryotes increased in deeply mixed and weakly stratified waters, with moderate to high concentrations of inorganic nitrogen. chapter III 119 Importance of the short spatio-temporal scalePhysical processes, including windinduced mixing (Figure 3B) impacted greatly water column stratification and induced short spatio-temporal variability of picoplankton group distribution. At the short temporal scale, wind greatly altered the regularity of the variations at coastal station C, increasing particularly the measured variability of Picoeukaryotes and Prochlorococcus abundances, with CVs of ±72% and ±40% respectively (Table 1B) compared to only 18% at offshore station D. Taylor and Howes (1994) suggested that events of major ecological significance are likely to result from episodic environmental perturbations, more particularly if we consider that the time frames of relevance to the life history of marine organisms are relatively short (Seymour et al. 2005). We observed also (Chapter II) at station C during winter 2007 one of such disturbances at the short time scale during a wind event, with picoplankton community structure that varied after possible resuspension of bottom sediments and decrease of light availability. Wind has also been reported as a principal factor of environmental heterogeneity in a deep Polynesian atoll lagoon (Thomas et al. 2010), and is likely to affect other parameters of the water mass such as temperature, positively linked with Synechococcus abundance (Waterbury et al. 1986; Chang et al. 1996; Tsai et al. 2005). However, it is also possible that different water masses were sampled over time during the diel cycles, generating strong variability at the short spatial scale and representing a potential source of errors for the interpretation of eulerian time-series as suggested by Martin et al. (2005) and shown in their analysis of a transect in the Celtic Sea. Heterotrophic bacterial abundance and activity variabilityThe variability of the different estimates of heterotrophic activity observed at the different spatio-temporal scales (Table 1B) was 1.60 to 1.96 fold higher than the variability measured in bacterial abundance, these differences being more particularly pronounced at coastal station C than at offshore station D. Several studies have shown similar discrepancies in the variability of activity as compared to that of abundance or biomass (Cole et al. 1988; Ducklow 2000; Sherr et al. 2001), indicating that in situ bacterioplankton assemblages can undergo relatively rapid shift-up or shifts-down in metabolism depending on local environmental conditions (del Giorgio and Cole 2000). At large spatio-temporal scales, comparative analyses revealed that bacterial production increased faster than bacterial abundance (Gasol and Duarte 2000) and that the strength of this link increased in more productive environments, suggesting that bacteria use algal-derived carbon more efficiently in eutrophic waters, since the maintenance energy costs might appear to be highest in oligotrophic systems (del Giorgio and Cole 1998). The discrepancies measured in our study at the different scales between the variability of bacterial abundance versus the variability found in bacterial activity indicate that different controlling factors occurred across the different stations. Gasol et al. (2002) showed that while Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 120 bacterial abundance and growth in the most oligotrophic environments are likely regulated by top down control (i.e. protists grazing), such control in more productive waters is more likely through changes in bacterial phylogeny, size and activity community composition. In this study, the lowest spatio-temporal variability in bacterial abundance and highest spatio-temporal variability in bacterial activity were both measured in coastal station C, while the opposite pattern was measured in offshore station D. This indicated that contrarily to bacterial activity, bacterial concentration was less influenced by top down control in the coastal station than in more oligotrophic waters. Moreover, opposites trends of variability were found between bacterial production and CTC+ cell number at the diel scale at both station C and D (Figure 5C and 5D), and vertically at station CM (Table 1B). Such discrepancies in the magnitude of variations of the different bacterial activities suggest that different bacterial metabolic processes are targeted by each method. Several studies have shown that heterotrophic bacterial activity follows diel patterns (i.e. Gasol et al. 1998; Ruiz-González et al. 2011). Contrasting with Fouilland and Mostajir (2010; 2011) that argued that bacteria do not depend on primary production in oligotrophic waters but from other sources of carbon at the relatively short time scale (days and week), the relatively large variability measured at the diel scale in 3H-thymidine and 3H-leucine incorporation rates at both coastal station C and offshore station D indicates that some of the bacterial populations in a community can rapidly respond to enhanced substrate availability on a time scale of hours by increasing rates of cell-specific activity and growth (Sherr et al. 1999; del Giorgio and Cole 2000), in agreement with the comment by Morán and Alonso-Sáez (2011) stating that bacterial metabolism is strongly dependent on the local primary production and the organic matter released in situ by primary producers as previously formulated by Baines and Pace (1991). Since several studies have shown that the CTC method targets the cells with the highest respiration rates (Sherr et al. 1999b; Sieracki et al. 1999; Smith and del Giorgio 2003), the discrepancies measured at the diel scale between the low variability in CTC+ cell number and the high variability of bacterial production would indicate that production processes vary more rapidly than bacterial respiration processes, again a pattern that seems logical: maintenance processes being more stable than growth. Ecological indicators of shifts in ecosystem picoplankton structureOpposite temporal and spatial patterns were found between the ratio of Synechococcus to pPeuk biomass and chlorophyll a concentration levels, at both short and large temporal scales. While water-column stratification has been shown to be one of the main factors promoting shifts in phytoplankton community structure in coastal and temperate waters at the seasonal scale (Cushing 1989), in picoplankton community structure in subtropical waters (Bouman et al. 2011), Zwirglmaier et al. (2007) showed that Synechococcus generally presented no obvious depth preference, but did show highly specific distribution at the horizontal scale. Early considered as possible biological marker chapter III 121 of shifts in water mass properties, Synechococcus abundance indicated advection of a warm matter mass in Polar Regions (Gradinger and Lenz 1989) and high ratio values of cyanobacteria to eukaryotes abundances were used to characterize saline intrusions of subtropical origin (Jochem and Zeitzschel 1993). Despite associations between picoplankton community structure and water mass properties have been since established in large spatial scale surveys (Li 1995; Zubkov et al. 2000; Li and Harrison 2001; Tarran et al. 2001), little is still known about Picoeukaryotes distribution and their link with water mass characteristics. Looking at the contribution of each group to picophytoplankton community (considered here as composed by Synechococcus, Prochlorococcus and Picoeukaryotes) in terms of C biomass at each station (Figure 6), the major contributor to picophytoplankton biomass was Synechococcus, which ranged from 54% at the slope (station CM) to 89% at the station M. The second contributor to PCS were generally the pPeuk at both coastal and slope stations (C and CM) with 12% and 32% respectively, the second most important contributor in more oligotrophic stations (M, MD, D) was Prochlorococcus, that reached 26% of total PCS biomass at station D (Figure 6). However, this overall domination by Synechococcus expressed by calculating an average ratio Syn:pPeuk of 8.89, varied at 64% horizontally (Table 1A and 1B). The ratio Syn:pPeuk Figure 6. Ratio of chlorophyll a concentration to the transect average (depth integrated for the first 125 m) and ratio Synechococcus:Picoeukaryotes biomass with the percentage biomass contribution of each picophytoplankton group (Synechococcus, Prochlorococcus and Picoeukaryotes) in the pie charts. Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 122 decreased from 6.94 (±11%) at station C to a minimum of 1.77 (106%) at station CM (Table 1) where maximum chlorophyll a concentrations were measured and confirmed by the increase of the ratio of chlorophyll a to its transect average (Table 1A and Figure 6). These horizontal variations, driven by the differences between the ecosystem characteristics of coastal station C, slope station CM, or more offshore influenced stations highlighted opposite trends: while pPeuk contribution to biomass increased from coastal station C to the slope CM, Synechococcus contribution decreased (Table 1A). Similar to these results, Calvo-Díaz et al. (2004) showed during transects along the N and NW Iberian peninsula shelf that Synechococcus prevalence in picoplankton community structure was associated with low chlorophyll a levels whereas the total and relative abundance of pPeuk increased with phytoplankton biomass. Note that high values of the ratio Syn:pPeuk were also measured at the sea surface at station MD, a region that corresponded with the eddyassociated upwelling event (Figure 2E). Similar findings by Tarran et al. (2001) demonstrated that eddy waters contained higher contributions of pPeuk and heterotrophic bacteria. ConclusionsThe large variability measured in the different heterotrophic activities at the different spatio-temporal, particularly pronounced at the diel scale, as well as the synchrony of the variations with the picophytoplankton groups, indicate the tight coupling that can occur between bacteria and primary producers. On the contrary, the stability measured in bulk bacterial abundance was poorly indicative of the shifts that can occur at the group or the OTU level in changing environments, suggesting that further study of variability of the bacterial community structure at narrower phylogenetic level would be welcome. Similarly to chlorophyll a concentration, used in several comparative analysis as indicator of phytoplankton biomass or trophic level, the distribution of the different picoplankton groups, here indicated by the ratio of Syn:pPeuk throughout the transect, was not random but linked to the ecological characteristics of the water masses. Acknowledgements This work was supported by the Spanish MICINN through projects MODIVUS (CTM200504795/MAR) and SUMMER (CTM2008-03309/MAR). We thank C. Cardelús, V. Balagué, I. Forn and all the people who participate in the Blanes Bay summer diel study cruise for their assistance with sample collection and processing, and to the captain, crew and colleagues on board R/V G. del Cid for smooth operation. chapter III 123 REFERENCES Baines, S. B. and Pace, M. L. (1991) The production of dissolved organic matter by phytoplankton and its importance to bacteria - patterns across marine and fresh-water systems. Limnol. Oceanog., 36, 1078--1090. Beaugrand, G. (2005) Monitoring pelagic ecosystems using plankton indicators. ICES J. Mar. Sci., 62, 333--338. Bec, B. Husseini-Ratrema, J. Collos, Y. Souchu, P. and Vaquer, A. (2005) Phytoplankton seasonal dynamics in a Mediterranean coastal lagoon: emphasis on the Picoeukaryote community. J. Plankton Res., 27, 881--894. Billen, G. Servais, P. and Becquevort, S. (1990) Dynamics of bacterioplankton in oligotrophic and eutrophic aquatic environments: bottom-up or top-down control? Hydrobiologia 207, 37—42. Bird, D. F. and Kalff, J. (1984) Empirical relationships between bacterial abundance and chlorophyll concentration in fresh and marine waters. Can. J. Fish. Aquat. Sci., 41, 1015--1023. Bouman, H. A., Ulloa, O. Barlow, R. Li, W. K. W. Platt, T. Zwirglmaier, K. Scanlan, D. J. and Sathyendranath, S. (2011) Water-column stratification governs the community structure of subtropical marine picophytoplankton. Environ. Microbiol., 3, 473--482. Bouman, H. Platt, T. Sathyendranath, S. Li, W. Stuart,V. Fuentes-Yaco, C. Maass, H. Horne, E. Ulloa, O. Lutz, V. and Kyewalyanga, M. (2003) Temperature as indicator of optical properties and community structure of marine phytoplankton: implications for remote sensing. Mar. Ecol. Prog. Ser., 258, 19--30. Brussaard, C. P. D. (2004) Optimisation of procedures for counting viruses by flow cytometry. Appl. Environ. Microbiol., 70, 1506--1513. Buck, K. R. Chavez, F. P. and Campbell, L. (1996) Basin-wide distributions of living carbon components and the inverted trophic pyramid of the central gyre of the North Atlantic Ocean, summer 1993. Aquat. Microb. Ecol., 10, 283—298. Bustillos-Guzmin, J. Claustre, H. and Marty, J. C. (1995) Specific phytoplankton signatures and their relationship to hydrographic conditions in the coastal northwestern Mediterranean Sea. Mar. Ecol. Progr. Ser., 124, 247-258. Calvo-Díaz, A. Morán, X. A. G. Nogueira, E. Bode, A. and Varela, M. (2004) Picoplankton Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 130 of Prochlorococcus sp. (Prochlorophyta) strains isolated from the North Atlantic and the Mediterranean Sea. Plan. Physiol., 101, 295--296. Pommier, T. Neal, P. R. Gasol, J. M. Coll, M. Acinas, S. G. Pedrós-Alió, C. (2010) Spatial patterns of bacterial richness and eveness in the NW Mediterranean Sea explored by pyrosequencing of the 16 rRNA. Aquat. Microb. Ecol., 61, 221--233. Raven, J. A. Andrews, M. and Quigg, A. (2005) The evolution of oligotrophy: implications for the breeding of crop plants for low input agricultural systems. Annals of Applied Biology 146, 261-- 280. Ruiz-González, C. Lefort, T. Massana, R. Simó, R. and Gasol, J. M. (2012) Diel changes in bulk and single-cell bacterial heterotrophic activity in winter surface waters of the northwestern Mediterranean Sea. Limnol. Oceanogr., 57, 29—42. Schauer, M. Balagué, V. Pedrós-Alió, C. and Massana, R. (2003) Seasonal changes in the taxonomic composition of bacterioplankton in a coastal oligotrophic system. Aquat. Microb. Ecol., 31, 163- -174. Seymour, J. S. Seuront, L. and Mitchell, J. G. (2005) Microscale and small-scale temporal dynamics of a coastal planktonic microbial community. Mar. Ecol. Prog. ser, 300, 21-37. Sherr, B. F. Sherr, E. B. and Mcdaniel, J. (1992) Effect of protistan grazing on the frequency of dividing cells in bacterioplankton assemblages. Appl. Environ. Microbiol., 58, 2381--2385. Sherr, E. B. Sherr, B. F. and Cowles, T. J. (2001) Mesoscale variability in bacterial activity in the Northeast Pacific Ocean off Oregon, USA. Aquat. Microb. Ecol., 25, 21--30. Sherr, E. Sherr, B. and Sigmon, C. (1999a) Activity of marine bacteria under incubated and in situ conditions. Aquat. Microb. Ecol., 20, 213-223. Sieracki, M. E. and Sieburth, J. M. (1986) Sunlight-induced growth delay of plankton marine bacteria in filtered seawater. Mar. Ecol-Progr. Ser., 33, 19--27. Sieracki, M. E. Cucci, T. L. and Nicinski, J. (1999) Flow cytometric analysis of 5-cyano-2,3ditolyl tetrazolium chloride activity of marine bacterioplankton in dilution cultures. Appl. Environ. Microbiol., 65, 2409--2417. Six, C. Finkel, Z. V. Irwin, A. J. and Campbell, D. A. (2007) Light Variability Illuminates NichePartitioning among Marine Picocyanobacteria. PLoS ONE 2, e1341, doi:10.1371/journal. chapter III 131 pone.0001341 Smith, D. and Azam, F. (1992) A simple, economical method for measuring bacteria protein synthesis rates in seawater using 3H-leucine. Mar. Microb. Food Webs, 6, 107--114. Smith, E. and del Giorgio, P. A. (2003) Low fractions of active bacteria in natural aquatic communities? Aquat. Microb. Ecol., 31, 203--208. Sommaruga, R. Hofer, J. S. Alonso-Sáez, L. and Gasol, J. M. (2005) Differential sunlight sensitivity of picophytoplankton from surface Mediterranean coastal waters. Appl. Environ. Microbiol., 71, 2154--2157. Steinberg, D. K. Carlson, C. A. Bates, N. R. Johnson, R. J. Michaels, A. F. and Knap, A. H. (2001) Overview of the US JGOFS Bermuda Atlantic Time-series Study (BATS): a decade-scale look at ocean biology and biogeochemistry. Deep-Sea ResPart II, 48, 1405-1447. Tarran, G. A, Zubkov, M. V. Sleigh, M. A. Burkill, P. H. and Yallop, M. (2001) Microbial community structure and standing stocks in the NE Atlantic in June and July of 1996. Deep-Sea Res. Part II, 48, 963—985. Taylor, C. D. and Howes, B. L. (1994) Effect of sampling frequency on measurements of seasonal primary production and oxygen status in near shore coastal ecosystems. Mar Ecol. Prog. Ser., 108, 193—203. Tett, P. and Grenz, C. (1994) Designing a simple microbiological-physical model for a coastal embayment. Vie Milieu, 44, 39--58. Thomas, Y. Garen, P. Courties, C. and Charpy, L. (2010) Spatial and temporal variability of the picoand nanophtyoplankton and bacterioplankton in a deep Polynesian atoll lagoon. Aquat. Microb. Ecol., 59, 89--101. Tsai, A. Y. Chiang, K. P. Chang, J. and Gong, G. C. (2005) Seasonal diel variations of picoplankton and nanoplankton in a subtropical western Pacific coastal ecosystem. Limnol. Oceanogr., 50, 1221–1231. Vaulot, D. and Marie, D. (1999) Diel variability of photosynthetic picoplankton in the equatorial Pacific. J. Geophys. Res., 104, 3297–310. Waterbury, J. B. Watson, S. W. Guillard, R. R. L. and Brand, L. E. (1979) Widespread occurrence of a unicellular, marine, planktonic, cyanobacterium. Nature, 277, 293--294. Spatio-temporal variability of picoplankton community structure in NW Mediterranean Sea 132 Waterbury, J. B. Watson, S. W. Valois, F. W. and Franks, D. G. (1986) Biological and ecological characterization of the marine cyanobacterium Synechococcus. Can. Bull. Fish. Aquat. Sci., 214,71–120. Worden, A. Z. and Not, F. (2008) Ecology and diversity of Picoeukaryotes, In Kirchman, D. L. (ed.), Microbial ecology of the oceans. 2nd ed. Wiley, New York, pp. 159-205. Worden, A. Z. Nolan, J. K. and Palenik, B. (2004) Assessing the dynamics and ecology of marine picophytoplankton: The importance of the eukaryotic component, Limnol. Oceanogr., 49, 168179. Zhang, Y., Jiao, N. and Hong, N. (2008) Comparative study of picoplankton biomass and community structure in different provinces from subarctic to subtropical oceans. Deep-Sea Res. Part II, 55, 1605—1614. Zinser, E. R. Johnson, Z. I. Coe, A. Karaca, E. Veneziano, D. and Chisholm, S. W. (2007) Influence of light and temperature on Prochlorococcus ecotype distributions in the Atlantic Ocean. Time, 52, 2205--2220. Zubkov, M. Sleigh, M. and Burkill P. (2000) Assaying picoplankton distribution by flow cytometry of underway samples collected along a meridional transect across the Atlantic Ocean. Aquat. Microb. Ecol., 21, 13—20. Zwirglmaier, K. Jardillier, L. Ostrowski, M. Mazard, S. Garczarek, L. Vaulot, D. Not, F. Massana, R. Ulloa, O. and Scanlan, D. J. (2007) Global phylogeography of marine Synechococcus and Prochlorococcus reveals a distinct partitioning of lineages among oceanic biomes. Environ. Microbiol., 10, 147-61. chapter III 133 Supplementary materials 134 A C B D  Figure 1. Ratio to the diel average (depth Integrated) at Station C and D of the picophytoplankton groups (A and C) and of bacterial abundance and percentage of CTC+ bacterial cells (B and D). chapter III 135 N=13 to 15 diel CyCle st C Syn Pro PPeuK rAtio biomAss syN : PPeuK het.bACteriA CtC+ bP tdr bP leu viruses WiNd sPeed ms-1 irrAdiANCe W m-2 SynechococcuS 0.59* 0.60* -0.49 0.64* 0.57* 0.70* 0.77** -0.45 0.05 -0.32 ProchlorococcuS 0.66* -0.55* 0.71** 0.68* 0.56* 0.40 -0.41 0.51* -0.02 PiCoeuKAryotes -0.84** 0.83** 0.84** 0.60* 0.44 -0.15 0.27 -0.13 rAtio biomAss syN : PPeuK -0.56* -0.91** -0.45 -0.22 -0.16 -0.28 0.07 het. bACteriA 0.56* -0.54* 0.74* -0.20 0.26 -0.18 CtC+ 0.38 0.17 -0.05 0.26 -0.08 bP tdr 0.89** -0.37 0.21 -0.65* bP leu -0.42 -0.03 -0.61* viruses -0.26 0.23 WiNd sPeed -0.02 irrAdiANCe Table 1. Matrix of Bonferroni-corrected Pearson’s correlation coefficients between the different parameters measured during the diel cycle performed at station C (A), at station D (B), and during the transect from station C to D (C). Syn For Synechococcus cell abundance (cells ml-1), Pro for Prochlorococus abundance (cells ml-1), and pPeuk for Picoeukaryote cell abundance (cells ml-1). BP for Bacterial activity as measured by Thymidine incorporation (pmol TdR l-1 h-1) and Leucine incorporation (pmol Leu l-1 h-1). N for number of observations. One asterisk indicates significant correlations (p< 0.05) while two asterisks indicate probabilities that are below the Bonferroni-corrected probability equivalent to p=0.05 (which for the set of correlations is p=0.003 for A, p=0.004 for B and p=0.001 for C). v Table 1A Supplementary materials 136 N=12 diel CyCle st d Syn Pro PPeuK rAtio biomAss syN : PPeuK het.bACteriA CtC+ bP tdr bP leu viruses WiNd sPeed ms-1 irrAdiANCe W m-2 SynechococcuS 0.76* 0.55 0.34 0.48 0.66* 0.13 -0.18 -0.60* 0.22 0.20 ProchlorococcuS 0.21 0.51 0.67* 0.63* -0.08 0.09 -0.15 0.07 0.22 PiCoeuKAryotes -0.59* -0.09 0.22 0.52 -0.05 -0.69* 0.27 -0.03 rAtio biomAss syN : PPeuK 0.58* 0.41 -0.45 -0.01 0.19 -0.03 0.18 het. bACteriA 0.54 -0.33 -0.14 0.28 0.31 0.14 CtC+ 0.18 0.05 -0.39 0.45 -0.08 bP tdr 0.75* -0.62* 0.51 -0.50 bP leu N=12 -0.10 0.27 -0.49 viruses -0.36 0.08 WiNd sPeed -0.21 N=28 to 47 trANseCt Syn Pro PPeuK rAtio biomAss syN : PPeuK het.bACteriA CtC+ bP tdr bP leu virus SynechococcuS 0.09 0.17 0.52* 0.69** 0.18 0.52* 0.71** 0.54* ProchlorococcuS 0.06 0.05 0.37* 0 0 0.06 0.65** PiCoeuKAryotes -0.2 0.44* 0.16 0.29 0.68** 0.46 rAtio biomAss syN : PPeuK 0.52* 0 0 0.04 0.31 het. bACteriA 00.58** 0.75** 0.92** %CtC+ 0 -0.1 0 bP tdr 0.75** 0.48** bP leu 0.70** virus Table 1B Table 1C chapter III 137 stAtioN C dAy / Night n=15 stAtioN d dAy / Night n=12 SynechococcuS (cells ml-1)+ (n.s / n.s) n.s (+ / -) ProchlorococcuS (cells ml-1) n.s (n.s / n.s) n.s (+ / -) PiCoeuKAryotes (cells ml-1)+ (n.s / n.s n.s (+ / n.s) SynechococcuS : PPeuK biomAss n.s (n.s / n.s) n.s (n.s /-) het. bACteriA (cells ml-1) n.s (n.s / n.s) - (n.s / n.s) hNA n.s (n.s / n.s) n.s (n.s / n.s) live n.s (n.s / n.s) n.s (n.s / n.s) deAd n.s (+ / n.s) - (n.s / n.s) CtC+ AbuNdANCe + (n.s / n.s) n.s (n.s / n.s) CtC%+ (n.s / n.s) n.s (n.s / n.s) bP leu. iNCorP.n.s (n.s / +)(n.s / n.s) bP tdr. iNCorP.n.s (n.s / +)+ (n.s / n.s) viruses n.s (n.s / n.s) n.s (n.s / n.s) Table 2. Diel cycles at stations C and D during cruise MODIVUS. (+) and (-) indicate significantly increasing or decreasing values from day to night (t-tests, p<0.05). Signs inside parenthesis : (+/-) indicates significantly increasing values during the day, and significantly decreasing during the night. (-/+) indicates significantly decreasing values during the day, significantly increasing during the night. N.s for non significant differences. Patterns in marine bacterial group distribution, as measured by FISH, in relation to chlorophyll, temperature and salinity Thomas Lefort and Josep M. Gasol 04 Patterns in bacterial community structure 146 Table 1. Data used in this study, oligonucleotide probes considered as well as the type of Fluorescence in situ Hybridization protocol. REFERENCES TECHNIQUE TYPE SAMPLING SITES PROBES CONSIDERED Alderkamp et al. 2006 Coastal North Sea CF319, Ros537 Wietz et al. 2010 Global survey EUB338, ALF968, BET42a, GAM42a, CF319 Alonso Sáez et al. 2007 NW Mediterranean coastal waters (Blanes Bay) EUB338, ALF968, BET42a, GAM42a, SAR11-441R, ROS537 Obernosterer et al. 2007 South Pacific Ocean EUB338, ALF968, GAM42a, CF319A Piccini et al. 2006 Southwestern Atlantic EUBI-III, ALF968, GAM42a. BET42a, CF319, SAR11-441 Stoica and Herndl 2007 NW Black Sea EUB338, ALF968, BET42a, GAM42a, ROS537 AND ROS1029 Garcés et al. 2007 NW Mediterranean coastal waters EUB338, ALF968, BET42a, GAM42a, ROS538 Mary et al. 2006 English Channel EUB338 I-II-III, ALF968, SAR11-152R, SAR11-542R, RSB67, GAM42a, CF319a Garneau et al. 2006 Canada Arctic shelf EUB338, ALF968, BET42a, GAM42a, CF319a Alonso Sáez et al. 2006 NW Mediterranean coastal waters (Blanes Bay) EUB338, ALF968, BET42a, GAM42a, SAR11-441R, ROS537 Ruiz-González et al. (unpublished) NW Mediterranean coastal waters (Blanes Bay) EUB338-II-III, ALF968, BETA42a, CF319, SAR11-441R, ROS537, GAM42a Alonso Sáez et al. 2008 Western Arctic EUB338-II-III, ALF968, BETA42a, CF319, SAR11-441R, ROS537, GAM42a Baltar et al. 2007 NW Africa upwelling to Canary Coastal Transition zone EUBI-II-III, SAR11-441R, CF319a, ROS537, GAM42a Teira et al. 2008 Ría de Vigo, Atlantic EUB338, ALF968, BET42a, GAM42a, CF319a, SAR11-441R, ROS537 Schattenhofer et al. 2009 Atlantic EUB338 I-III, SAR11-441R, GAM42a, CF319a Topping et al. 2006 CARD-FISH Scotia Sea, Antarctica EUB338, ALF968, SAR11-486, SAR11-542R, ROS537, GAM42a, CF319a Lin et al. 2006 Eastern sub-basin of the Cariaco system and Black Sea EUB338, ALF968, BET42a, GAM42a, CF319 Lin et al. 2008 FISH/CARD-FISH Cariaco system EUB338, ALF968, BET42a, GAM42a, CF319 Zhang et al. 2006 South China Sea EUB338, ALF968, BET42a, GAM42a, CF319 Simon et al. 1999 Southern Ocean EUB338, ALF968, BET42a, GAM42a, CF319 Wells et al. 2003 Canadian archipelago EUB338, CF319 Longnecker et al. 2006 Oregon, Newport (West US), North Pacific EUB338, ALF968, BET42a, CF319, GAM42a Vila et al. 2004 Northern gulf of Mexico EUB338, ALF968, ROS536, GAM42a, CFB319 Kirchman et al. 2003 Delaware estuary EUB338, ALF968, BET42a, GAM42a, CF319a Agawin et al. 2006 NE Subarctic Pacific EUBI-II-III, ALF968, ALF1B, BETA42a, GAM42a, CF319a Yokokawa and Nagata 2005 North Pacific coast EUB338, ALF968, BET42a, GAM42a, CF319 Zubkov et al. 2002 Northern north Sea EUB338, ALF968, GAM42a, CF319a, RSB67 Cottrell and Kirchman. 2000 California Coast EUB338, ALF968, BET42a, GAM42a, CF319a, SAR11-A1 Lin et al. 2007 Eastern sub-basin of the Cariaco system EUB338, ALF968, BET42a Castle and Kirchman 2004 Delaware estuary and Chesapeake Bay EUB338, ALF968, BET42a, GAM42a, CF319 Yokokawa et al. 2004 Delaware estuary EUB338, ALF968, BET42a, GAM42a, CF319 Cottrell et al. 2006 Mid Atlantic bight and North Pacific gyre EUB338, ALF968, SAR11, CF319, ROS537 Kirchman et al. 2005 Delaware Estuary Eub338, Alfa968, ROS537, Beta42, Gam42a, CF319 Carlson et al. (unpublished)* FISH BATS. Sargasso Sea SAR11, CF319, ROS537 The data from BATS station by Carlson et al. were extracted from icomm.mbl.edu/oocs_summaries/OOCS_Carlson/OOCS_Carlson_final.ppt chapter IV 147 In addition, the relative abundance of the SAR11 and Rhodobacteraceae clusters was also analyzed mostly using the probes SAR11-441R (Morris et al. 2002) and Ros537 (5´-CAACGCTAACCCCTCC-3´) (Eilers et al. 2001). However, other probes were also used to target SAR11, such as SAR11/486 (5´-GGACCTTCTTATTCGGGT-3´) (Fuchs et al. 2005) and SAR11/542R (5´-TCCGAACTACGCTAGGTC-3´) (Morris et al. 2002). Probe RSB67, specific for Alphaproteobacteria subgroup Rhodobacteraceae was also used in some studies (5’-CGCTCCACCCGAAGGTAG-3’ (Zubkov et al. 2001) (see details in Table 1). Extraction of the dataMost data were from the surface ocean layer (except for the CARIACO basin data set, Table 1). Relative abundances were expressed in terms of % contribution to total DAPI counts or to total flow cytometric counts, while absolute abundances were expressed in number of cells per milliliter. The subgroup concentrations were paired with chlorophyll a concentration added to other environmental variables such as temperature (ºC), and salinity when information was available. Some data were obtained from graphs and digitized with the use of GraphClick vs. 3 (Arizona Software). Considering the break shelf as the limit between coastal and open ocean ecosystems, 20 of the studies were considered to be coastal environments (of the 20, 4 studies were clearly river-influenced or estuary habitats), 6 were open-ocean, and 6 contained both coastal and open-ocean data. Chlorophyll a measurementsChlorophyll a concentration (mg ml–1) defined operationally as the pigment amount detected from particles retained on glass-fibre membranes, by spectrophotometry or by High-performance liquid chromatography (HPLC), was used as a proxy of ecosystem trophy. All chlorophyll a concentrations corresponded to field measurements performed on the day of sampling. For the study of Castle and Kirchman (2004), chlorophyll a concentrations were not directly available but were estimated from the particulate beam attenuation coefficient (cp) results following the relationship: Chla (mg. L-1) = 2.6 cp – 0.014 as proposed by Behrenfeld and Boss (2006). Standardization of FISH and CARD-FISH resultsAs explained above, we combined the data obtained with the FISH and the MARFISH protocols. It is well known that, in most ecosystems, the CARD-FISH protocol produces higher counts, as is more sensitive than the FISH protocol (e.g. Pernthaler et al. 2002). In order to be able to compare the relative abundances (% of DAPI) and the absolute cell concentrations (calculated from the total DAPI) measured by FISH and CARDFISH, we conducted a standardization of the relative abundances, assuming that all the FISH and the CARDFISH estimates came from the same global “population”, and that both Patterns in bacterial community structure 148 methodologies had sampled enough to obtain a fair representation of the contribution of each group to the global community. For each bacterial group studied, one-way Anova and t tests were conducted on the whole data set to test whether measurements of bacterial group contribution to total bacterial community structure by FISH or CARD-FISH were significantly different. When significant differences were observed, we corrected the FISH values accordingly. E.g. for EUB+ cells, we could assume that the average 30% obtained by FISH and the average 59% obtained by CARD-FISH are both estimates of the same data. In that example case, we would bring the 30 to 59% and we would thus multiply all the EUB values by the factor 59/30. Concentrations were then computed from the percentages and the total DAPI. When no significant differences were observed between techniques, no percentage transformations were conducted. Note that this procedure assumes that the unlabelled cells by FISH that could be labeled by CARDFISH are distributed equally among all bacterial groups, and that the discrepancy between the two methodologies stands from different degrees of activity spread equally within all subgroups. This is likely not the case (large Gammaproteobacteria cells might be better detected by FISH than small SAR11 cells), but we have no other ways of accounting for these differences. We ignored specific variations in protocols (even though they are relevant, Bouvier and del Giorgio 2003). The detection of target cells by FISH is known to vary drastically among the published literature, ranging from 1% to 100% of variations for Eubacteria across 51 different published reports (Bouvier and del Giorgio 2003). Not only methodological factors such as the type of fluorochrome or the stringency conditions can significantly influence the performance of FISH, but ecological factors such as ecosystem type (coastal, open-ocean, freshwaters….) explain also a large amount of variability in target detection (Bouvier and del Giorgio 2003). We also ignored variability in probe coverage, but the known limitations of current probes should be taken into account (e.g. Amann and Fuchs 2008). Many of the group-specific probes used to target the major taxonomic groups do not have 100% group coverage and no outgroup hits, and potential for false identification exists (e.g. CF319a was designed to cover the Flavobacteria (90%) and Shingobacteria (90%) but is much less efficient at targeting the Bacteroidetes (30%) (Amann and Fuchs 2008). Multivariate analysisBoth relative (% of DAPI/total prokaryotes) and absolute abundances (cells ml-1) were used to describe the relative importance of chlorophyll a concentration, temperature and salinity. To avoid the differences generated by the scale of the independent variable (chlorophyll a, temperature, salinity) and to allow for the comparison of the relative impact of each independent variable in multivariate models, we used standardized Beta-coefficients in multiple regression models. All analyses were performed with the JMP (version 5.0.1) statistical chapter IV 149 software package (SAS Institute, Inc). The graphs of Figure 1 were done with the software Aabel 2.4 (Gigawiz Ltd. Co) with a 7*7 moving average. Linear regression analysis and analysis of covarianceTo compare slopes of relationships between log-transformed standardized absolute abundances of each bacterial group and independent variable (e.g. chlorophyll a), equations of the regressions are presented as log (Y) = a + b log (X) with Y= cells ml-1; a=intercept; b=slope; X =independent variable (e.g. chlorophyll a in mg l-1). In order to test whether the slopes and intercepts of the relationships are significantly different, Student’s t-tests were conducted after applying model I regression analyses following the methods explained by Zar (1999). To compare linear regressions and test for heterogeneity of slopes, ANCOVA tests were also performed. All conducted in JMP 5.0.1 software (SAS Institute Inc). Analysis of bacterial group distributionThe collected data of bacterial group relative abundances and environmental parameters such as chlorophyll a concentration, temperature and salinity were used to estimate the “preferred” range of environmental parameters for each bacterial group using a Quotient-Rule Analysis (QRA, Somarakis et al. 2006). Each environmental variable was divided into regular intervals for which the frequencies of occurrence were calculated and expressed in percentage. The number of intervals in every environmental variable was set to ensure that maximum occurrence per interval did not exceed 20% of all measurements. For every bacterial group, the log transformed cell concentration within each interval of environmental parameter was calculated and then was expressed as a percentage of the group abundance over the full range of environmental variable. Then, for every interval the quotient values were estimated with the equation: € Q=bacterial group abundance (%) frequency of occurence of environmental variable (%) Quotient values were smoothed using a 3-point running mean and then plotted against environmental factors, reflecting “preference” (quotient values >1) or “avoidance” (quotient values <1) for a specific variable range (interval). A non-parametric Kolmogorov-Smirnov test of goodness of fit (Zar 1999) was used to compare the cumulative frequency distribution of bacterial groups per category of environmental variable against the distribution histograms of that environmental variable. The null hypothesis (H0) considers that the observed bacterial group distribution should be at random along that environmental variable. Patterns in bacterial community structure 150 RESULTS Environmental parametersThe data set analyzed was quite representative of the world’s oceans: chlorophyll a concentrations in coastal environments had an average value of 3.97 mg l-1 and ranged from 0.05 mg. l-1 to 103.15 mg.l-1 (N=266). A shorter range and lower chlorophyll a average, 0.48 mg l-1, was measured in open-ocean conditions, 0.001 mg.l-1 - 10.01 mg. l-1 (N= 197, Tables 1 and 2). The large chlorophyll a values were measured in the Southern China Sea and in the NW African upwelling.Temperature ranged from -1.3ºC and -0.67ºC in coastal and open-ocean environments respectively, to 28.5ºC and 30.1ºC (Tables 1 and 2). Salinity of coastal waters ranged from 0.08 recorded in Delaware estuary (Kirchman et al. 2003) to a maximum at 38.7 measured at the Blanes Bay station in NW Mediterranean (Alonso-Sáez et al. 2007). In openocean waters, a minimum was measured in Canada Arctic shelf (Garneau et al. 2006) at 20.3 and a maximum at 37.43 observed in the Western Arctic (Alonso-Sáez et al. 2008) (Tables 1 and 2). Range and average bacterial group contribution to bacterial community structure (BCS) across ecosystemsAt the global scale, SAR11, Alphaproteobacteria and Bacteroidetes were the largest contributors to BCS with relative abundance averages (expressed as % of DAPI counts) of 29% (±44%), 26% (±53%) and 21% (±71%) respectively (Table 2). Almost similar contributions were found for Betaproteobacteria and Gammaproteobacteria with respectively 13% (±84%) and 11.5% (±95%). The average bacterial group relative contributions and their coefficient of variation (± CV) showed opposite patterns, low relative contributions were associated with high variability (as estimated from CVs) (Table 2). Alphaproteobacteria relative abundance in coastal ecosystem was on average 25%, for a range of 1-85% (Tables 1 and 2). In offshore conditions, there was a higher relative contribution of Alphaproteobacteria to BCS, 32% (t-tests, n Alpha=351, p<0.001) but with a lower range of variability 5-59%. SAR11 contributed 25% on average in coastal conditions, ranging from 0-59%. In offshore, an almost similar range of variations 0-67% was measured but with significantly higher average contribution, 32% (t-test, n SAR11 = 252, p<0.001). In comparison with Alphaproteobacteria and SAR11, the average relative contribution of Rhodobacteraceae was significantly higher in coastal than in open-ocean conditions with 6.4% and 4.9% respectively (but with lower significance level) (t-test, nRhodo = 281, p<0.05). The range was 1-32% in coastal areas and 0 to 18% in open-ocean environments. chapter IV 151 Global scale Coastal Open-Ocean N of observations AVERAGE % (±CV) AVERAGE CELLS ML -1 (±SD) N of observations AVERAGE % (MIN/MAX) AVERAGE CELLS ML -1 (±CV) N of observations AVERAGE % (MIN/MAX) AVERAGE CELLS ML -1 (±CV) Eubacteria 292 73.5 (±26%) 1.12 ±1.80 x10 6 239 73 (4 / 100) 1.24 ± 1.93 x10 6 53 74 (12 / 100) 5.62 ± 8.48 x10 5 Alphaproteobacteria 351 26 (±53%) 3.91 ± 10.6 x10 5 290 25 (0.21 / 85.09) 4.03 ± 11.60 x10 5 61 32 (5.13 / 58.97) 3.35 ± 3.89 x10 5 SAR11 252 29 (±44%) 1.94 ± 1.87 x10 5 106 25 (0 / 59) 1.97 ± 1.43 x10 5 146 32 (5.25 / 66.94) 1.91 ± 2.15 x10 5 Rhodobacteraceae 281 6 (±95%) 5.04 ± 8.38 x10 4 149 6 (0 / 38.32) 6.54 ± 9.84 x10 4 132 5 (0.4 / 18) 3.15 ± 5.52 x10 4 Gammaproteobacteria 289 11 (±95%) 1.92 ± 8.15x10 5 222 12 (0.7 / 95.82) 2.21 ± 9.12 x10 5 67 9 (0 / 22.55) 8.02 ± 1.07 x10 5 Bacteroidetes 465 21 (±71%) 2.02 ± 3.22 x10 5 301 25 (0.26 / 81.14) 2.77 ± 3.82 x10 5 164 14 (1 / 51.49) 8.22 ± 1.13 x10 5 Betaproteobacteria 174 13 (±84%) 2.08 ±3.18 x10 5 152 13 (0 / 51.17) 2.22 ± 3.30 x10 5 8 9 (4.40 / 16.23) 7.82 ± 8.20 x10 4 Temperature (ºC) 346 17.37 (±40%) 209 17 (-1.3 / 30.1) 137 17.5 (-0.67 / 29.7) chlorophyll a (µg.l -1 ) 491 2.57 (±270%) 294 3.97 (0.05 / 103.15) 197 0.50 (0.001 / 10.01) Salinity (psu) 175 27.37 (±46%) 129 23.50 (0.08 / 38.7) 46 35.21 (20.30 / 37.43) Table 2. Average and range of environmental parameters (Minimum and Maximum) and bacterial group relative and absolute abundances (% relative to DAPI or flow cytometry counts. Bacterial group abundance expressed in cells ml-1) as measured by the FISH/CARDFISH protocol in different types of ecosystems. Coastal sites are those placed on the continental platform. CV for Coefficient of Variation, SD for Standard Deviation Patterns in bacterial community structure 152 The Bacteroidetes relative abundance and range of variation were significantly higher in coastal than in offshore conditions (t-test, nBacteroidetes = 465, p<0.001). At the coast, an average contribution of 25% (range 1-81%) was found, while offshore, Bacteroidetes were contributing to an average of 15% (range <146%). In comparison with open-ocean waters where Gammaproteobacteria contribution to BCS averaged 9%, significantly higher relative contribution of Gammaproteobacteria was measured in coastal environments, averaging 12% of BCS (t-test, nGamma = 289, p<0.001) and with a very high range of variability <1-90% of the DAPI counts. The strong maximum contribution was found in coastal lagoon waters (>90% BCS) with high chlorophyll a concentration and high temperature in the southwestern coastal Atlantic (Piccini et al. 2006). As Bacteroidetes, Gammaproteobacteria and Rhodobacteraceae, higher relative abundance of Betaproteobacteria were measured in coastal than offshore with respectively 14% and 9% (t-test, nBeta = 174, p<0.001) averages. Similarly, a larger range of Betaproteobacteria contribution was measured at coastal sites (<148%) than in offshore environments (<4-16%) (Tables 1 and 2). Compared with coastal, the observation’s number for Betaproteobacteria abundance was very low in offshore conditions, mostly measured during coast to offshore transects in the south China Sea and off the Oregon coasts (Zhang et al. 2006; Longnecker et al. 2006; Table 1). Patterns in BCS across environmental parametersWe analyzed to what extent the variability measured in the bacterial group contribution to BCS was driven by different environmental parameters such as chlorophyll a, temperature and salinity. We represented with contour plots the relative contribution to BCS as a function of chlorophyll a concentration and temperature (Figure 1), and as a function of chlorophyll a concentration and salinity (Figure 2). No evident patterns in Alphaproteobacteria and Gammaproteobacteria relative contribution to BCS could be observed across the range of chlorophyll a concentration and temperatures or salinity, the maximum (> 50%) appearing at both low and high temperatures levels. However, strong effects of temperature were seen on Eubacterial relative contribution as revealed by a strong correlation found (Pearson t test, N=159, p<0.005), the maximum contribution observed at lower chlorophyll a and temperatures levels (Figure 1A and Table 3). Furthermore, the highest Bacteroidetes contribution (>18%) was observed at high chlorophyll a levels but within a large range of temperature from 2ºC to 17ºC (Figure 1C). Even more pronounced patterns were observed when focusing on Alphaproteobacterial groups: SAR11 and Rhodobacteraceae (Figure 1B and 1D, Figure 2B). While highest SAR11 relative contribution were observed at chapter IV 153 A Figure 1: Contour plots of bacterial group relative abundances (expressed as % of DAPI counts) as a function of temperature (ºC) and chlorophyll a concentration (log transformed, mg l-1). The bacterial groups considered are Eubacteria (A), Rhodobacteraceae (B), Bacteroidetes (C), SAR11 (D), and Gammaproteobacteria (E). B C D E Patterns in bacterial community structure 154 (1) TEMPERATURE AND LOG CHLOROPHYLL A (2) SALINITY AND LOG CHLOROPHYLL A Bacterial groups Dependant variable NOBS R2 F ratio BETA TEMPERATURE BETA CHLOROPHYLL A NOBS R2 F ratio BETA SALINITY BETA CHLOROPHYLL A Bacteria total Log Abs. abundance 254 0.43 96 (p<0.0001) 0.24* 0.61* 146 0.22 20 (p<0.0001) -0.29* 0.23* Rel. abundance (%) 137 0.20 18 (p<0.0001) -0.34* -0.22* 97 0.04 3 (n.s) -0.13 -0.31* Eubacteria Log Abs. abundance 114 0.41 41 (p<0.0001) 0.17* 0.58* 81 0.18 10 (p<0.0001) -0.11 0.37* Rel. abundance (%) 185 0.03 4 (p<0.05) 0.01 -0.20* 137 0.09 8 (p<0.0001) 0.22* -0.13 Alphaproteobacteria Log Abs. abundance 156 0.06 6 (p<0.005) 0.15 0.22* 115 0.04 3 (p<0.05) 0.01 0.22* Rel. abundance (%) 166 0.08 8 (p<0.0005) 0.20* -0.07 73 0.10 5 (p<0.05) 0.10 -0.33* SAR11 Log Abs. abundance 156 0.07 7 (p<0.005) 0.16* 0.30* 73 0.01 0.40 (n.s) 0.04 -0.10 Rel. abundance (%) 164 0.08 8 (p<0.0005) 0.01 0.30* 67 0.21 10 (p<0.0001) -0.30* 0.37* Rhodobacteraceae Log Abs. abundance 158 0.22 24 (p<0.0001) 0.04 0.50* 67 0.29 14 (p<0.0001) -0.35* 0.42* Rel. abundance (%) 173 0.01 0.07 (n.s) 0.01 0.01 135 0.01 0.17 (n.s) -0.06 -0.04 Gammaproteobacteria Log Abs. abundance 154 0.03 3 (p<0.05) 0.10 0.18* 128 0.06 5 (p<0.05) -0.21* 0.10 Rel. abundance (%) 249 0.05 8 (p<0.001) -0.15* 0.18* 143 0.15 13 (p<0.0001) 0.10 0.45* CFB Log Abs. abundance 220 0.34 57 (p<0.0001) 0.04 0.59* 123 0.22 18 (p<0.0001) 0.11 0.53* Rel. abundance (%) 79 0.08 4 (p<0.05) -0.01 0.32* 62 0.56 41 (p<0.0001) -0.78* 0.08 Betaproteobacteria Log Abs. abundance 47 0.24 8 (p<0.001) 0.21 0.37* 40 0.32 10 (p<0.001) -0.37* 0.31* *: significant values (p<0.05) Nobs: Number of observations n.s: Non significant values Table 3. Results of Multiple Linear Regression analysis. Bacterial relative abundances (expressed as % of DAPI counts and standardized according to FISH and CARDFISH, see M&M) and Log of absolute abundances (cells ml-1) were considered as a linear function of both: 1: Temperature (ºC) and Log chlorophyll a (mg l-1) , 2: Salinity (psu) and Log chlorophyll a (mg l-1). Beta coefficients (Beta) represent the contribution of each independent variable (Tº, chlorophyll a or salinity) to the prediction of the dependant variable (relative or absolute abundances). chapter IV 155 high temperatures and low chlorophyll a concentration (Figure 1D), Rhodobacteraceae relative abundance increased with increasing levels of chlorophyll a concentration (Figure 1B and 2B) and was strongly influenced by salinity (Pearson t test, N=70, p<0.05) (Table 3). Similarly, particularly pronounced effects of salinity on Betaproteobacteria relative contribution were obvious (Pearson tests, N=66, p<0.005), reaching values of >18% at salinities <6 (Figure 2A). The distribution of bulk bacteria was not related significantly to the examined environmental parameters (Table 4). However, the study at narrower phylogenetic levels showed that the contribution of the different bacterial groups to community structure did not vary uniformly across the intervals of the different environmental variables (Table 4). While Quotient rule analysis and quotient curve plots showed similar patterns in the preference of Eubacteria, Alphaproteobacteria and Gammaproteobacteria for increasing chlorophyll a levels (even more pronounced for Betaproteobacteria), SAR11 on the contrary exhibited pronounced ”avoidance” and relatively lower contributions to BCS at increasing chlorophyll a levels (Figure 3A, Table 4) with quotient curve values < 1 at intermediate chlorophyll a levels. Gammaproteobacteria, Alphaproteobacteria, and to a lesser extent Eubacteria followed similar patterns of preference and avoidance for increasing levels of temperatures (Figure 3B, Table 4). Not one but several intervals of “preference” and “avoidance” were observed for these groups along the gradient of temperature. Figure 2: Contour plots of bacterial group relative abundances (expressed as % of DAPI counts) as a function of salinity (psu) and chlorophyll a concentration (Log values, mg. l-1). Bacterial groups are Betaproteobacteria (A) and Rhodobacteraceae (B). A B References 258 kingdoms. PNAS, 74, 5088—5090. Woese, C. R. (1987) Bacterial evolution. Microbiol. Rev., 51, 221—271. Worden, A. Z. Nolan, J. K. and Palenik, B. (2004) Assessing the dynamics and ecology of marine picophytoplankton: The importance of the eukaryotic component, Limnol. Oceanogr., 49, 168179. Worden, A. Z. and Not, F. (2008) Ecology and diversity of picoeukaryotes. In Kirchman, D. L. (ed.), Microbial ecology of the oceans. 2nd ed. Wiley, pp. 159-205. Wright, R. T. H. and Hobbie, J. E. (1965) The uptake of organic solutes in lake water. Limnol. Oceanogr., 10, 22—28. Yentsch, C. S. and Yentsch, C. M. (1979) Fluorescence spectral signatures: The characterization of phytoplankton populations by the use of excitation and emission spectra. J. Mar. Res., 37, 471--483 Zhang, Y., Jiao, N. and Hong, N. (2008) Comparative study of picoplankton biomass and community structure in different provinces from subarctic to subtropical oceans. Deep-Sea Res. Part II, 55, 1605—1614. Zimmerman, R. (1977) Estimation of bacterial numbers and biomass by epifluorescence microscopy and scanning electron microscopy. In Microbial ecology of brackish water environment. SpringerVerlag, New York, pp. 103—120. Zobell, C. E. (1946) Marine Microbiology. Chronica Botanica, Waltham, Mass. Zubkov, M. V. Sleigh, M. A. Tarran, G. A. Burkill, P. H. and Leakey R. J. G. (1998) Picoplankton community structure on the Atlantic transect from 50ºN to 50ºS. Deep-Sea Res. I, 45, 1339--1355. Zubkov, M. V. Sleigh, M. A. Burkill, P. H. and Leakey, R. J. G. (2000) Picoplankton community structure on the Atlantic Meridional Transect: a comparison between seasons. Progr. Oceanogr., 45, 369—386. References 259 REMERCIEMENTS / AGRADECIMIENTOS / ACKNOWLEDGEMENTS 260 Ha llegado el momento tan esperado de los agradecimientos, que significa el fin de la travesía del desierto! Si por ahora, la alegría por haber acabado domina, también realizaré dentro de poco que esto suena como el final de una pagina de mi vida en Barcelona. Ahora sé un poco mas (o mucho mas) lo que significa hacer una tesis y todos los compromisos que supone llevarlo a cabo (ver detalles con Juliette), sé también que no hubiera sido posible hacerlo sin la ayuda de muchos. La primera persona que me gustaría agradecer es, por supuesto, a mi jefe Josep Gasol a quien debo el liderazgo de esta tesis. Me dio la oportunidad y la libertad de aprender muchos aspectos del trabajo de científico, enseñándome una parte de su conocimiento, incluyendo su exigencia. Le agradezco haber sabido desbloquear muchas situaciones científicamente y humanamente complicadas, y me quedaré con el recuerdo de una de sus mayores calidades: ser hombre de palabra y de buen consejos. También me acordaré que me llamó Groucho Marx cuando escribí frases del capitulo III totalmente oscuras Je tiens aussi à remercier amicalement Fabrice, et pour le fait qu’il m’ait donné avec Silvia, l’opportunité d’aller en Antarctique sur le Tara, m’offrant le privilège de vivre ce que peu de gens ont la chance de connaitre, et m’apportant l’oxygène indispensable pour affronter cette dernière année d’écriture. Me gustaría agradecer también a mis compañeros de despacho, empezando con Estella con quien he compartido muchas bromas absurdas. Te tengo que devolver el cronometro que me diste para ponerse las pilas a leer nuevos abstracts. No funcionó conmigo pero gracias. Gracias a Andrea y Beatriz por las ayudas de optimismo en los momentos mas obscuros de esta tesis. Mas recientemente, agradezco a Bibiana y Nuria por haber compartido los procesos finales de este doctorado y también a Ivo para las pausas y las conversaciones surrealistas sobre el peligro de no encriptar los emails, tema a primera vista extraño, pero que me ayudó a pasar las ultimas etapas estresantes de la tesis. Gracias a Montse por su ayuda con el Catalán y a Cristina por su sentido del humor y su acento madrileño. A Irene, Vanesa y a Clara por todas las veces o les he cortado en su trabajo con mis preguntas. También gracias a Juancho y a Clara por la ayuda en matar hierbas y desplazar piedras... Agradezco a Luisa por haber intentado enseñarme el Catalán, que por supuesto me hubiera gustado controlar mas. Pero que sepáis que no me iré de aquí completamente orgulloso Para mi defensa, os tendréis que acordar de mi acento horroroso en castellano o/y en ingles y que, quizás, no os habéis perdido mucho a no escucharme hablar un poco mejor vuestro idioma. Enfin je remercie les miens pour leur soutien affectueux et leur confiance, grâce à eux beaucoup de choses furent rendues possibles et la vie plus douce. Je remercie mes amis proches pour leur fidélité. Enfin, merci à Juliette, car sans elle et son intelligence a tous égards, rien de tout cela n’aurait été possible. 261 Fundings This thesis was supported by the European Union 6th Framework Program METAOCEANS through a PhD fellowship to Thomas Lefort, The METAOCEANS project (MESTCT-2005-019678) is a Marie Curie host fellowship for early stage training on meta-analysis and comparative analysis in the marine sciences. “Elucidating the structure and functioning of marine ecosystems through synthesis and comparative analyses” Other projects that contribute partially to the completion of this thesis were the Spanish MCINN projects SUMMER (CTM2008-03309/MAR funded by the MICINN, P.I.: Dr Rafel Simó), and ASSEMBLE (2009-227799), the project MODIVUS (CTM2005-04795/MAR, P.I.:Dr Josep Gasol) and the project STORM (CTM2009-09352/MAR).