Picoplankton diversity in the South-East Pacific Ocean from cultures

In late

acterized morphologically and for more than 50% of them, genetically, through partial sequencing of the 18 S rRNA gene.
Among the characterized strains, the largest number are stramenopiles (Heterokontophyta) with a record of 38 strains belonging to the species Pelagomonas calceolata (Pelagophyceae).Strains from the recently described genera Bolidomonas and Florenciella have been re-isolated for the first time since their description.Two other abundant groups are the Chlorophyta, especially Prasinophyceae, and the Haptophyta, especially the genera Phaeocystis and Emiliania.A limited number of heterotrophic flagellates have also been isolated, all of them closely related to known species.Finally over a dozen of unicellular cyanobacteria strains have been obtained, some forming unusual short chains.
Overall our strategy was quite successful since it allowed us to isolate a large number of picoplankton strains but failed in two respects.First, apparently very few novel taxa have been obtained.One set of strains is related to Prasinoderma coloniale (Prasinococcales, Prasinophyceae) but their sequences are sufficiently different from the latter to probably belong to a new genus or species.The sequences of two other strains are phylogenetically affiliated to stramenopile environmental sequences, probably corresponding a new algal class.Second, very few strains have been obtained

Introduction
Thirty years ago, the existence of very small algal cells was discovered in marine waters (Johnson and Sieburth, 1982;Waterbury et al., 1979).They were termed picoplankton, defined as smaller than 2-3 µm (Sieburth et al., 1978).It was soon realized that a significant fraction of photosynthetic biomass and primary production could be attributed to these tiny cells (Li et al., 1983;Platt et al., 1983).This small size fraction was found to be more important as chlorophyll concentration decreased, i.e. as the degree of oligotrophy increased (Herbland et al., 1985).Within photosynthetic picoplankton, prokaryotes appeared early on as much less diversified than eukaryotes as they are dominated by two major cyanobacteria genera: Prochlorococcus and Synechococcus.This probably explains why we now know much more about photosynthetic picoplanktonic prokaryotes than eukaryotes.In particular, the genetic diversity of these prokaryotes has been quite well characterized (Fuller et al., 2003;Rocap et al., 2002), representatives of key genotypes have been isolated in culture, and more recently quite a few genomes have been sequenced (Palenik et al., 2003;Rocap et al., 2003).It is now possible to map the distribution of key groups of cyanobacteria in oceanic waters and to assess the existing relationships between genotypes and ecotypes (Johnson et al., 2006).
For photosynthetic picoeukaryotes, the situation is, in many respects, much less advanced, one reason being their very wide phylogenetic diversity.They belong to at least four major lineages: Chlorophyta, Haptophyta, stramenopiles (or Heterokontophyta) and Alveolata.Moreover, extensive studies of their genetic diversity from environmental samples only started less than 10 years ago (L ópez-García et al., 2001;Moon-van der Staay et al., 2001) EGU sensu stricto (i.e. for which cells are always smaller than 3 µm) have been described (Vaulot et al., 2007  1 ).Among these, knowledge about "flagship" species such as Ostreococcus and Micromonas (both belonging to the order Mamiellales, Prasinophyceae) is progressing fast since the genome of several "ecotypes" has already been (or is currently) sequenced (Derelle et al., 2006;Palenik et al., 2007).Their oceanic distribution can be mapped using techniques such as fluorescent in situ hybridization (Not et al., 2005) or quantitative PCR (Marie et al., 2006).However, this only constitutes the tip of the iceberg as molecular approaches, in particular the analysis of 18S rDNA genetic libraries from the natural environment, have pointed out to a very wide diversity at all taxonomic levels (Vaulot et al., 2007  1 ).For example, a new division of photosynthetic eukaryotes, the picobiliphytes, has been recently discovered (Not et al., 2007).
At the class, order or genus level many taxa are only known from their sequences.This is the case for example for Prasinophyceae clade VII B (Guillou et al., 2004) or for Chrysochromulina-related clades within the prymnesiophytes (Moon-van der Staay et al., 2000).For all these taxonomic groups, there is a critical need to obtain cultured representatives.This concern is especially acute in open ocean oligotrophic regions due to the difficulty to isolate and maintain organisms adapted to low nutrient conditions that are often outgrown by fast dividing "weed" species.
The BIOSOPE cruise that sailed through the center of the South East Pacific gyre, probably the most oligotrophic place on earth, offered an opportunity to obtain cultures from this unique environment.We performed sample enrichment with diluted culture medium following filtration to separate the smaller picoplankton cells from the rest of the plankton, a strategy that allowed us in the past to obtain novel taxa (Vaulot et al., 2004).We also targeted specifically photosynthetic picoeukaryotes by using flow cytometry sorting directly on board the ship.In the end, we obtained 212 cultures that have been integrated to the Roscoff Culture Collection (RCC), more than half of which were characterized genetically by sequencing partially the 18S rRNA gene.These cul-Figures Full Screen / Esc Printer-friendly Version

Sampling
Samples were taken in general at two depths (surface layer and vicinity of the chlorophyll maximum) at selected stations along the BIOSOPE cruise track (Fig. 1 and Table 1) using Niskin bottles mounted on a CTD frame.The oceanographic context of the cruise is described in Claustre et al. ( 2007)2 .

Primary cultures
We used two different strategies to obtain starter cultures.The first one was based on filtered seawater enriched with nutrients.The second one relied on single cell sorting by flow cytometry, targeting specific cell populations based on their size and pigment fluorescence.As cultures were examined several times during the cruise, many variations were attempted in an effort to increase final culture yield.

Growth conditions used on board
All cultures were incubated on board in a thermostatic cabinet set at 20 • C. Two light levels were obtained with 2 Sylvania 18 W tubes: white light around 140 µmol photons.m −2 s −1 and blue light (Moon Light Blue paper, M.E.S, Nantes, France) around 8 µmol photons m −2 s −1 .We used three types of medium: K (Keller et al., 1987)

EGU
for photosynthetic eukaryotes, Pro2 (Moore and Chisholm, 1999) for photosynthetic prokaryotes (Prochlorococcus and Synechococcus), and rice-based (Massana et al., 2004) for heterotrophic eukaryotes which were grown in the dark.Multi-well plates were wrapped with parafilm in order to avoid any evaporation during growth.

Enrichment cultures
About 500 mL of sample seawater was filtered by simple gravity through two superposed (in an effort to provide more tight size fractionation) Nuclepore filters of 47 mm diameter, with either 0.6 µm or 3 µm porosity (Whatman International Ltd, Maidstone, UK).The filtrate was partitioned into 50 mL culture flasks (Sarstedt, Orsay, France) or, at one station (HNL3), into individual wells of 24-well plates to which we added either 1/10 or 1/100 of full strength K or Pro2 medium.In order to try to promote nitrogen fixing organisms, some cultures were started by simply amending sea water with iron (as FeCl 3 ) and phosphorus (as KH 2 PO 4 at final concentrations of 3 nM and 0.4 µM, respectively.

Cultures sorted by flow cytometry
Samples were run either un-concentrated or concentrated between 5 and 100-fold by tangential flow filtration using a 100 000 MWCO (Regenerated Cellulose -RC ref VF20C4) Vivaflow 200 cassette.Concentration was usually necessary to target the rarer cells.Between 1 to 500 000 cells were sorted using a FACSAria (Becton Dickinson, San Jose CA) flow cytometer either into 24 or 48-well plates or directly into 10 mL polystyrene tubes pre-filled with medium diluted 100 times (Table 2).Different cell populations were discriminated based on side scatter as well as orange and red fluorescence following excitation at 488 nm (20 mW) and sorting was done either in purity or yield mode.On board the ship, primary cultures (either enriched or flow sorted) were checked for growth once or twice (depending on how early in the cruise they were started) using flow cytometry and inverted microscopy.Cultures that displayed growth but appeared mixed were sorted a second time.
A first set of cultures were transferred back to Roscoff on the occasion of change of crew at Easter Island at mid-cruise.At the end of the cruise, cultures from the early part of the cruise (i.e. about two months old) that showed no evidence of containing photosynthetic cells based on flow cytometry analysis were discarded.All cultures grown in multi-well plates were transferred to 10 mL polystyrene tubes and brought back to Roscoff in an ice box.
Once transferred back to Roscoff, cultures were monitored based on colour as well as with optical microscopy and flow cytometry.Cultures were purified either by serial dilution, solid medium plating, or individual cell pipetting under an inverted microscope.Strains that appeared to be pure were transferred to normal strength medium [PCR-S11 (Rippka et al., 2000), K, and rice for cyanobacteria, autotrophic and heterotrophic eukaryotes, respectively] and entered into the Roscoff Culture Collection (RCC) under new accession numbers (Table 2).

Strain characterization
Strains deposited to the RCC were characterized by optical microscopy.For each strain, pictures were taken on live cultures with an Olympus BX51 microscope with a x100 objective using differential interference contrast (DIC) with a SPOT RT-slider digital camera (Diagnostics Instruments, Sterling Heights, MI).Average cell dimension of each culture was determined from the pictures.Flagellated cells were also photographed after adding one drop of lugol to visualize flagellum shape, length and number.Cyanobacteria were identified by their colour and shape.The morphology of a few strains was confirmed by whole-mount transmission electron microscopy.Cells Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion

EGU
were fixed for 15 min with 1% glutaraldehyde.A drop of fixed cells was deposited onto formvar-coated grids.When the drop had dried, grids were rinsed with distilled water.Cells on grids were stained with a saturated solution of uranyl acetate for 20 min and rinsed with distilled water.Photomicrographs were taken with a JEOL JEM-1200EX electron microscope.
A subset of strains was characterized by their partial 18 S ribosomal RNA gene sequence.Cultures were grown in 50 mL flasks for 1-2 weeks depending on the growth of each strain and recovered by centrifugation at 11 000×g for 10 min.DNA was extracted using 3% Cethyl Trimethyl Ammonium Bromide (CTAB, Doyle and Doyle, 1990).DNA was then stored at -80 • C. The 18S rRNA gene was amplified by polymerase chain reaction (PCR) using the primer set Euk328f and Euk329r (Moon-van der Staay et al., 2000) and the HotStarTaq Master Mix (Qiagen, Courtaboeuf, France).For PCR, a 15 min initial activation step of the Polymerase at 95 • C, was followed by 40 cycles including 1 min of denaturation at 94 • C, 45 s of annealing at 57 • C and 75 s extension at 72 • C. The PCR program was finished by a final extension of 10 min at 72 • C followed by cooling at 4 • C. PCR products were purified with the Qiaquick PCR purification kit (Qiagen) and controlled by electrophoresis on a 1% agarose gel.Partial 18 S rRNA gene sequences were determined from purified PCR products by using Big Dye Terminator V3.1 (Applied Biosystems, Foster city, CA, USA) and the internal primer Euk 528f (Elwood et al., 1985) run on an ABI prism 3100 sequencer (Applied Biosystems, Courtaboeuf, France).
Sequences were compared to those available in public database with NCBI BLAST web application.Sequences were also automatically aligned using the ARB program (Ludwig et al., 2004) to a set of more than 20 000 high quality pre-aligned eukaryotic sequences available from the Silva database (database SSURef: http: //www.arb-silva.de).After manual refinement of the alignment, sequences were added to the reference tree provided with the SSURef database using the quick parsimony addition option.Sequences with high similarities were grouped together using Fast Group II (http://biome.sdsu.edu/fastgroup/fgtools.htm)with the sequence match pa- EGU rameter set at 80 % and one or two representative sequences per group were chosen along with the closest publicly available sequence.Phylogeny analysis was performed on aligned sequences with MEGA4 (http://www.megasoftware.net/,Tamura et al., 2007).A neighbour-joining tree was computed from 394 common positions based on Kimura 2-parameter model distances using 1000 bootstrap replications.Sequences have been submitted to GenBank under accession number xxx-xxx.

Isolation success
All together more than 1900 starter cultures were established during the BIOSOPE cruise (Table 1) either as enrichment cultures following filtration through either 0.6 or 3 µm or by sorting specific populations into individual wells or tubes.From one to three purification steps were in general necessary to obtain pure cultures (Table 2).For example, enrichment cultures started at the beginning of the cruise were sorted at the end of the cruise and then purified by serial dilution back in the laboratory.
In the end, we obtained 188 autotrophic and 24 heterotrophic cultures which have been deposited to the RCC (Table 2).Among these, 12 were subsequently lost and 25 remain not pure to this date.The latter are mostly autotrophic strains contaminated by heterotrophic eukaryotes.Cruise coverage was quite unequal with many strains obtained in mesotrophic regions and in the Chilean upwelling and much fewer from the central gyre (Table 2, Fig. 1).This reflects probably the difficulty to obtain cultures representative of extreme oligotrophic conditions, since nutrient additions even at relatively low concentrations are always much higher than those found in the environment.However this unbalanced coverage is not only the consequence of the environment but also of practical considerations.Cultures started early during the cruise had a chance to be screened before the end of the cruise and therefore could be re-purified on-board the ship.Conversely, cultures started late in the cruise were transported during their initial

EGU
growth phase, a period where they may be less fragile than once they have acclimated to more stable conditions.Refinements in culturing conditions that were implemented late in the cruise based upon results obtained in the first part of the cruise may also explain why our success rate was good by the end of the cruise.For example, starter cultures were sorted at the beginning of the cruise into 24 or 48-well plates.By midcruise, as we did not observe any growth under these conditions, we decided to switch to sorting into 10-mL tubes which seemed to result in a higher success rate.
Sorting was an important element since more than 65% of the final cultures had undergone a sorting step.The strategy that yielded most pure strains was first to establish an enrichment culture with either 0.6 or 3 µm filtered samples followed by sorting sometimes later.In this case, it was often not necessary to perform further purification by serial dilution, saving this labour-intensive step.Sorting directly from natural samples was rarely sufficient to produce pure cultures and in most cases a second purification step had to be undertaken.It is difficult to determine whether sorting was successful in isolating the initially targeted population.We sorted sub-populations on the base of side scatter and chlorophyll but each of these sub-populations does not appear to be uniform genetically and consists probably of a mixture of several taxa belonging to different algal classes (Shi, X. and Marie, D., unpublished).

Culture diversity
All purified cultures were examined by light microscopy, imaged digitally and their average size was determined (Table 2 and Fig. 2).No attempts were made to record measurements for a large number of cells in each cultures and these data are therefore only indicative.They confirm, however, that our efforts to target picoplankton were successful since the mode size for the culture set lies between 2.5 and 3 µm.
A large, randomly chosen, subset of cultures (115, Table 2, Table 3) was analysed phylogenetically by sequencing either partially or, in a few cases, totally the 18 S rRNA gene.A few other cultures were identified based on their phenotypic characteristics (cyanobacteria, microplanktonic species).Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion

EGU
Representatives of cyanobacteria and of three major eukaryotic divisions containing photosynthetic organisms (stramenopiles, Chlorophyta, and Haptophyta) have been obtained in culture with the former most prevalent and the latter two in almost equal proportions (Table 4).Two dinoflagellate cultures have also been isolated.
Thirteen strains of unicellular rod-shaped cyanobacteria have been obtained tentatively identified as Synechococcus.No Prochlorococcus was obtained despite the use of the Prochlorococcus specific Pro2 medium.Some of these cyanobacterial strains form short chains, exhibiting sometimes very elongated cells (Fig. 3, RCC 1027) contrasting the usual Synechococcus morphology (Fig. 3, RCC 1022).Such strains mostly originated from the HNLC station near the Marquesas islands (Table 3) and could be interesting since samples from this region displayed an unusually high fraction of chainforming and colonial picocyanobacteria (Masquelier and Vaulot, 2007).Phylogenetic analyses of the 16 S rRNA gene will be necessary to determine the exact nature of these strains.Chlorophyta, and more specifically Prasinophyceae, are important contributors to picoplankton and many strains have been isolated from marine waters in the past, some of them belonging to not yet described species (Guillou et al., 2004;Vaulot et al., 2004).Seventeen Chlorophyta strains have been isolated during BIOSOPE, mostly Prasinophyceae.Among these, 11 are related to Prasinoderma coloniale (Prasinococcales), a picoplanktonic species that can form colonies surrounded by mucus.These strains display the bilobed cup-shaped chloroplasts characteristics of P. coloniale (Hasegawa et al., 1996).However most of our strains do not seem to form colonies as P. coloniale does.Interestingly, one group of 9 sequences appear to form a separate clade (Fig. 4) with only 94.7% identity to P. coloniale (in contrast to the two other strains sharing 99.6% identity with P. coloniale) and possess large and highly similar insertions at least 330 bp long inside de 18 S rRNA gene starting at nucleotide position 862 of the P. coloniale sequence.Phenotypically, strains from these group appear slightly smaller (Table 2) than those closely related to P. coloniale.They were isolated from near-surface waters at a variety of stations, while the two strains more closely related to P. coloniale Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU originated from the Marquesas area.A culture closely related to Prasinococcus capsulatus, a species that also belongs to the order Prasinococcales, has been recovered from the chlorophyll maximum at the GYR station.Cells display a polysaccharide capsule around the cell (Fig. 3, RCC 859), typical of this species (Miyashita et al., 1993).Five Prasinophyceae closely related to the picoplanktonic species Pycnococcus provasolii (Pseudoscourfieldiales) have been isolated from two mid-depth samples in the Chilean upwelling.Eight strains belong to clade VII A of the Prasinophyceae (Guillou et al., 2004), a group which contains some cultured strains such as CCMP 1205 but for which no species has been described formally.All these strains consist of small (2 to 4 µm) coccoid cells lacking discriminating features (Fig. 3, RCC 857).Two sets of strains originated from surface waters and one set from 100 m in the HNLC zone.Two Prasinophyceae strains from the Chilean upwelling belong to clade C of the very ubiquitous species Micromonas pusilla (Guillou et al., 2004).They possess an unusually long flagellum (Fig. 3, RCC 913) that could be a diagnostic feature for that clade (Jouenne, F., personal communication).Finally, one culture (RCC 999) presents some phylogenetic affinities to the Prasinophyceae but its partial 18 S rDNA sequence does not allow us to place it in any of the existing clades defined by Guillou et al. (2004).We also isolated from one sample of the Chilean upwelling, three cultures representative of another green algal class, the Trebouxiophyceae.These strains are phylogenetically related to the recently established genus Picochlorum (Fig. 4) that regroups now salt-tolerant Nanochlorum (Henley et al., 2004).All Haptophyta cultures are part of the class Prymnesiophyceae.Sixteen strains belong to the genus Phaeocystis, three from the upwelling region being more closely related to the species P. jahnii which has been recently described from the Mediterranean Sea (Zingone et al., 1999) and forms loose colonies and 5 from the east of the gyre and upwelling regions related to P. globosa that forms spherical colonies (Fig. 3, RCC 851).We also isolated 12 strains of Emiliania huxleyi, a few calcifying (Fig. 3, RCC 867) and most naked (Fig. 3, RCC 951), corresponding probably to diploid and haploid stages, respectively (Houdan et al., 2003).Two other unidentified coccolithophorids have also Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU been obtained from the Marquesas and central gyre regions.Interestingly all Haptophyta strains were isolated from the top of euphotic zone (between 5 and 60 m).Among stramenopiles, 38 cultures are closely related to the picoplanktonic species Pelagomonas calceolata (Pelagophyceae).More than half of them are flagellated (Fig. 3, RCC 879), fitting the original description of the species (Andersen et al., 1993), and the rest, coccoid.However, the presence of a flagellum reflects probably more life cycle stages rather than taxonomical differences since both flagellated and nonflagellated strains have been isolated from the same sample (e.g. at 100 m at the HLN station).The presence of a thin theca characteristics of the species (Andersen et al., 1993) was confirmed by electron microscopy on strain RCC 879.P. calceolata has been isolated at a variety of stations (Marquesas, HLNC, center of gyre, east of gyre and upwelling) both in surface and at 100 m, demonstrating that this species is truly ubiquitous in oceanic waters.Interestingly in the center of the South East gyre, Pelagomonas strains were isolated from very deep samples down to 160 m.Two Pelagophyceae strains (RCC 986 and 1024) with 18 S rDNA sequences displaying slightly lower similarity to P. calceolata (Fig. 4) were recovered at 60 m depth from the Marquesas region.Both are picoplanktonic and coccoid, not displaying any specific morphological features.We isolated a novel strain with high similarity to Bolidomonas pacifica, a species that belongs to the recently described class of the Bolidophyceae (Guillou et al., 1999), closely related to the diatoms.Its morphology (presence of 2 unequal flagella) was confirmed by electron microscopy.This is quite interesting since to our knowledge this is the first novel isolate from this class since its initial discovery.In the same manner, we isolated from Marquesas surface waters, two strains very closely related by their 18 S rDNA sequence to the recently described Dictyochophyceae picoplanktonic species Florenciella parvula (Eikrem et al., 2004).Similarity was also confirmed by electron microscopy.Two photosynthetic stramenopile strains could not be assigned to any specific class as they shared homology with both Pinguiococcus (Pinguiophyceae) and Nannochloropsis (Eustigmatophyceae).Their 18 S sequences are almost identical to an environmental 18 S sequence (BL000921.5)recovered from

EGU
Blanes Bay in the Mediterranean Sea (Fig. 4).They could belong to a new class, although the presence of refractive intracellular granules (Fig. 3, RCC 853) is quite reminiscent of what is observed in Nannochloropsis.Unfortunately, they have been lost in early 2007 following a breakdown in the air conditioning system of our culture facility.Their loss, which was almost the only one from a quite large collection, attests of their sensitivity to change in environmental conditions and may explain why representatives of this group have not been isolated before.Three diatoms, belonging to the genera Chaetoceros (Fig. 3, RCC 1025)Thalassiosira, andMinutocellus were obtained from the upwelling region.The latter strain is quite interesting since its very small size (about 3 µm) connects it to picoplankton.Two dinoflagellates belonging to the genus Prorocentrum, P. minimum (Fig. 3, RCC 922) and P. dentatum, were isolated from surface waters, east of the gyre.
Twelve heterotrophic strains from dark cultures growing on rice medium have been identified by their 18 S rRNA sequences.Nine belong to the bicosoecid lineage of the stramenopiles.Three cultures are quite closely related to the genus Caecitellus and four more distantly related to Cafeteria.The two remaining strains were closely related to the bodonid (Euglenozoa) genera Rhynchomonas and Neobodo.All these genera are quite often recovered in cultures (Arndt et al., 2003).

Conclusions
Our large scale effort to isolate picoplanktonic strains from the Southeast Pacific Ocean allowed us to obtain of 212 novel cultures, a large number of which are of picoplanktonic size.The final number of cultures obtained is substantially higher than in previous efforts such as those linked to the PROSOPE and MINOS cruise in the Mediterranean Sea or the OLIPAC cruise in the Equatorial Pacific Ocean from which we obtained between 46 and 90 strains for each (Vaulot et al., 2004).Our initial intent was to use mostly flow cytometry sorting to establish this strains.However as we experienced technical problems with flow cytometry in the first few days of the cruise and as Introduction

Conclusions References
Tables Figures Printer-friendly Version Interactive Discussion

EGU
we observed subsequently that the yield of the initially sorted samples was quite low, we decided to combine flow cytometry sorting with more classical enrichments.This proved to be quite a good recipe, especially since sorting based on photosynthetic pigment fluorescence appears as a good way to prevent contamination of cultures by heterotrophic eukaryotes, a problem plaguing some of our previous efforts.The application of sorting either before or after enrichment did not appear to affect dramatically the type of taxa isolated (Table 2).
The final diversity achieved is quite wide since we obtained representatives of most major photosynthetic divisions (Table 4).However it is clear that we globally failed to obtain representatives of environmental sequences for which no culture is available yet.One interesting group of novel cultures was constituted by stramenopile strains RCC 853 and 862 from the central gyre which sequences were closely related to an environmental sequence from the Mediterranean Sea (Fig. 4).Although these sequences had some affinities, based on BLAST, to Eustigmatophyceae and their morphology was somewhat similar to the latter, they probably belonged to a novel class.Despite the fact that further studies are prevented since these strains have been lost, the strategy used (flow cytometry sorting followed by serial dilution) could be tried again to re-isolate them.Another interesting group is constituted by 9 cultures originating from the region east of the gyre and from the upwelling that are related to Prasinoderma but form a new clade clearly separated from the species P. coloniale (Fig. 4).They could belong to a new species within the genus Prasinoderma or form a new genus.Interestingly, they are apparently not related to any published environmental sequence.All the other cultures obtained are related to described species or at least to established cultures.In particular, we have been successful at re-isolating two genera Bolidomonas and Florenciella that our group had previously isolated and described (Eikrem et al., 2004;Guillou et al., 1999), but that had never been obtained again in culture since their initial isolation.Interestingly, B. pacifica was initially isolated in exactly the same region (between 2 and 16 • S) as the new strain (9 • S).In contrast, the only F. parvula strain available previously originated from English Channel coastal waters, a very different Introduction

Conclusions References
Tables Figures Printer-friendly Version

Interactive Discussion
EGU environment from that of the new strains.Moreover the 18S sequences of the latter differ slightly from that of F. parvula and they could belong to a novel species within this genus.Some of the cultures recovered correspond to ubiquitous species that were obtained from a wide range of environments.This is in particular the case for the two Haptophyta genera Emiliania, isolated from two of the four major regions investigated (Marquesas, east of gyre) and Phaeocystis isolated from three regions (Marquesas, east of the gyre, Chilean upwelling) mostly in surface waters.For the latter genus, our strains may correspond to at least two different species, P. globosa and P. jahnii.However, the largest number of strains obtained for a single taxon correspond to Pelagomonas isolated from a record of 13 different samples along the entire cruise track ranging from oligotrophic (St B13) to eutrophic (UPX) and from surface (5 m) to very deep (160 m) samples.Although the similarity of their 18S rRNA gene sequence is very high (average p-distance=0.0018),it is likely that these strains present quite different growth responses to factors such as nitrogen supply or light levels and belong to different ecotypes, as observed previously for example for the genus Ostreococcus (Rodríguez et al., 2005).From a biogeographic point of view, it is quite difficult to make any firm conclusion from this work.Many cultures belonging to a given taxonomic group were isolated from a variety of conditions and no specific pattern could be uncovered.Although there were some taxa unique to the central part of the gyre itself (stations 3 to 15) such as Prasinococcus and the potentially novel class mentioned earlier, one should emphasize the low number of strains isolated from this region.This is probably linked to the fact that the media we used (K, Pro2), that are quite successful in general to isolate and maintain a wide variety of picophytoplankton strains, fail to mimic the drastic oligotrophic conditions met in the gyre.Moreover future isolation effort may need to involve new culture approaches such as those successful to isolate fastidiously growing bacterial strains from the open ocean environment such as Peligibacter ubique (aka SAR11) that had escaped cultivation for quite a long time (Zengler et al., 2002)    .Phylogenetic analysis of selected strains recovered during the BIOSOPE cruise.One or two 18 S rRNA sequences from each taxonomic group was selected following clustering with Fast Group II (see Material and Methods).Neighbour-joining optimal tree with the sum of branch length=1.73998869shown.The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are displayed next to the branches.Only values larger than 60% are shown.The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree.The evolutionary distances were computed using the Kimura 2-parameter method and are in the units of the number of base substitutions per site.All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option).There were a total of 394 positions in the final dataset.Phylogenetic analyses were conducted with MEGA4 (http://www.megasoftware.net/mega4/).
Fig.4.Phylogenetic analysis of selected strains recovered during the BIOSOPE cruise.One or two 18 S rRNA sequences from each taxonomic group was selected following clustering with Fast Group II (see Material and Methods).Neighbour-joining optimal tree with the sum of branch length=1.73998869shown.The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are displayed next to the branches.Only values larger than 60% are shown.The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree.The evolutionary distances were computed using the Kimura 2-parameter method and are in the units of the number of base substitutions per site.All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option).There were a total of 394 positions in the final dataset.Phylogenetic analyses were conducted with MEGA4 (http://www.megasoftware.net/mega4/).
. To date, about 20 species that are picoplanktonic Introduction

Table 2 .
List of strains deposited to the Roscoff Culture Collection (RCC) ordered according to the sampled station.Steps 1, 2 and 3 refer to the different purification steps performed before the culture was entered to the RCC database.

Table 3 .
List of identified cultures ordered by phylogenetic group.The column rDNA cluster provides the RCC number of the reference culture for each cluster defined using Fast Group II (http://biome.sdsu.edu/fastgroup/fgtools.htm)with the parameter sequence match set at 80%.For cultures for which no sequence was obtained this column remains empty.

Table 4 .
Number of strains identified for the different phylogenetic groups.