Diversity of cultured photosynthetic flagellates in the northeast Pacific and Arctic Oceans in summer

. During the MALINA cruise (summer 2009), an extensive effort was undertaken to isolate phytoplankton strains from the northeast (NE) Paciﬁc Ocean, the Bering Strait, the Chukchi Sea, and the Beaufort Sea. In order to characterise the main photosynthetic microorganisms occurring in the Arctic during the summer season, strains were isolated by ﬂow cytometry sorting (FCS) and single cell pipetting before or after phytoplankton enrichment of seawater samples. Strains were isolated both onboard and back in the laboratory and cultured at 4 ◦ C under light/dark conditions. Overall, we isolated and characterised by light microscopy and 18 S rRNA gene sequencing 104 strains of photosynthetic ﬂagellates which grouped into 21 genotypes (deﬁned by 99.5 % 18 S rRNA gene sequence similarity), mainly afﬁliated to Chlorophyta and Heterokontophyta. The taxon most frequently isolated was an Arctic ecotype of the green algal genus Micromonas (Arctic Micromonas ), which was nearly the only phytoplankter recovered within the picoplankton ( < 2 µm) size range. Strains of Arctic Micromonas as well as other strains from the same class (Mamiellophyceae) were identiﬁed in further detail by sequencing the internal transcribed spacer (ITS) region of the rRNA operon. The MALINA Micromonas strains share identical 18 S rRNA and ITS sequences suggesting high genetic homogeneity within Arctic Micromonas . Three other Mamiellophyceae strains likely belong to a new genus. Other green algae from the genera Nephroselmis, Chlamydomonas , and Pyramimonas were also isolated, whereas Heterokontophyta included some unidentiﬁed Pelagophyceae, Dictyochophyceae (Pedinellales), and Chrysophyceae ( Dinobryon faculiferum ). Moreover, we isolated some Cryptophyceae ( Rhodomonas sp.) as well as a few Prymnesiophyceae and dinoﬂagellates. We identiﬁed the dinoﬂagellate Woloszynskia cincta by scanning electron microscopy (SEM) and 28 S rRNA gene sequencing. Our morphological analyses show that this species possess the diagnostic features of the genus Biecheleria , and the 28 S rRNA gene topology corroborates this afﬁliation. We thus propose the transfer of W. cincta to the genus Biecheleria and its recombination as Biecheleria cincta .


Introduction
Arctic phytoplankton undergoes a high seasonal variability with most of the biomass occurring during late summer (Sherr et al., 2003;Wang et al., 2005). During this period, freshwater inputs from rivers and ice melting in the Beaufort Sea lead to strong stratification of the water column. Consequently, phytoplankton depletes the surface layer of nutrients, especially inorganic nitrogen (Carmack and MacDonald, 2002).
The summer composition of photosynthetic pico-and nanoplankton has been investigated in great detail from the northeast (NE) Pacific to the Beaufort Sea during the MALINA cruise in summer 2009 (Balzano et al., 2012). Terminal restriction fragment length polymorphism (T-RFLP) and cloning/sequencing approaches have confirmed the ubiquity of Arctic Micromonas, which occurred in the NE Pacific, dominated the Bering Strait and was nearly the unique photosynthetic picoplankter found throughout the Beaufort Sea in both nitrogen-depleted surface waters and nitrogen-replete deep chlorophyll maximum (DCM) waters. It is not known whether such ubiquity and exclusivity covers intraspecific differences between populations occurring under different seawater conditions or whether populations are rather homogeneous and all adapted to variable conditions. In contrast, nanoplankton was more diverse and dominated by cultured microorganisms mainly belonging to diatoms, Chrysophyceae, and Pelagophyceae.
Despite obvious biases, culturing approaches permit a better characterisation of the strains isolated by the combination of microscopy and molecular methods (Le Gall et al., 2008). To date, existing datasets on Arctic phytoplankton are based either on light microscopy (Okolodkov and Dodge, 1996;Booth and Horner, 1997;Lovejoy et al., 2002;Sukhanova et al., 2009) or cloning/sequencing (Lovejoy et al., 2006;Luo et al., 2009;Lovejoy and Potvin, 2011), but few studies have performed large scale isolation efforts in the Arctic.
The present study aimed at the detailed characterisation of strains isolated during the MALINA cruise. One of our goals was to assess whether the main Arctic species are endemic or occur in other oceans. During the MALINA cruise, we isolated about 200 strains from the NE Pacific, the Bering Strait, the Chuckchi Sea, and the Beaufort Sea using different approaches (flow cytometry sorting, single cell pipetting). About half of the strains belonged to diatoms and will be investigated in a parallel study. Here, we characterise photosynthetic flagellates by 18 S rRNA gene sequencing. We also sequenced the internal transcribed spacer (ITS) region of the rRNA operon from our strains of Mamiellophyceae to assess whether Arctic Micromonas is genetically homogeneous or consists of several distinct genotypes, and if the other Mamiellophyceae strains isolated here correspond to a new genus. Finally, we characterised in further detail, by scanning electron microscopy (SEM) and 28 S rRNA gene sequencing, two dinoflagellate strains belonging to Woloszynskia cincta, a recently described species (Siano et al., 2009), and propose a taxonomical revision of the species.

Sampling
The MALINA cruise took place on board the Canadian research vessel CCGS Amundsen during the summer of 2009 from Victoria (British Columbia, Canada) to the Beaufort Sea (Table 1, leg 1b) and then throughout the Beaufort Sea (leg 2b). Seawater samples were collected with a bucket from the surface during leg 1b and at different depths with Niskin bottles mounted on a CTD frame during leg 2b. Water temperature, salinity, nutrient concentrations, and the phytoplankton composition were obtained from the MALINA database (http://www.obs-vlfr.fr/Malina/data. html).

Strain isolation
Phytoplankton strains were isolated both onboard and back in the laboratory. Onboard, strains were isolated on 5 ml glass tubes by flow cytometry sorting (FCS) either directly from the seawater, as well as from samples concentrated by tangential flow filtration (TFF) , or from enriched seawater samples. Samples were enriched by mixing 4.5 ml of 2 fold diluted medium with 0.5 ml of seawater in 5 ml glass tubes and by incubating the tubes under light-dark conditions for at least three days prior to isolations. Media used for the enrichments included f/2 (Guillard, 1975), K (Keller et al., 1987), Jaworski (http://www.ccap.ac.uk/media/recipes/ JM.htm), Erd-Schreiber (Kasai et al., 2009), and PCR-S11 (Rippka et al., 2000). Seventeen medium enrichments were spiked with 9.6 µM GeO 2 (Sigma-Aldrich, Saint-Quentin, France) to prevent the growth of diatoms (Supplement ,  Table S1). All strains were maintained on a 12 : 12 light-dark cycle and transferred weekly to new medium. Samples and cultures from the surface were incubated under white light (100 µmoles photons m 2 s −1 ) while samples from deeper layers were incubated under blue light (10 µmoles photons m 2 s −1 ).
One to six months after the MALINA cruise, more strains were isolated in the laboratory using single cell pipetting or FCS from TFF concentrated or enriched samples. Some strains were found to be non-unialgal or contaminated by small heterotrophs and were further purified using single-cell FCS (Supplement , Table S1). FCS was carried out using a Biogeosciences, 9, 4553-4571, 2012 www.biogeosciences.net/9/4553/2012/ FACSAria (Becton Dickinson, San Jos' e, CA, USA) either on board or back in the laboratory. For each strain between 1 and 20 000 cells were sorted either into 96-well plates or directly into 5 ml glass tubes prefilled with K/2 (Keller et al., 1987) medium. Different cell populations (picoeukaryotes, nanoeukaryotes, and microeukaryotes) were discriminated based on side scatter as well as orange and red fluorescence following excitation at 488 nm as described previously . Sorting was done in purity mode and samples were immediately transferred at 4 • C. For single cell pipette isolation, TFF concentrated or enriched seawater samples were observed using an inverted microscope Olympus IX71 (Olympus, Hamburg, Germany) and 1.5 ml from each sample were collected and transferred into a 24-well Iwaki plate (Starlab, Bagnieux, France). A sample aliquot was transferred into a new well containing sterile medium and this step was repeated 4 times for a final 100 000 fold dilution of the enriched sample. Single cells were then collected using a Nichipet EX 0.5-10 µl (Starlab, Bagnieux, France), transferred again into new plates containing sterile media and incubated at 4 • C under light-dark conditions for 1 to 2 weeks.

Molecular analyses
Genomic DNA was extracted from 104 strains of photosynthetic flagellates: a volume of 2 ml was collected from the cultures during the stationary-state growth phase, centrifuged at 11 000 rpm for 10 min, and 1.8 ml of supernatant removed. The genomic DNA was then extracted using Qiagen Blood and Tissue kit (Qiagen, Cortaboeuf, France) as described previously (Balzano et al., 2012).
The ITS region of the rRNA operon was amplified from 28 Mamiellophyceae strains, most of them (24) belonging to Arctic Micromonas, using the universal primers ITS-1 (5 -TCC-GTA-GGT-GAA-CCT-GCG-G-3 )  and ITS-4 (5 -TCC-TCC-GCT-TAT-TGA-TAT-GC-3 ) which amplify very small portions of both 18 S and 28 S rRNA genes and the whole ITS region (White et al., 1990). PCR reactions were performed with an initial incubation step at 94 • C for 2 min, 40 amplification cycles (94 • C for 35 s, 46.2 • C for 35 s, and 72 • C for 1 min), and a final elongation step at 72 • C for 10 min.
18 S rRNA, ITS, and 28 S rRNA amplicons were purified using Exosap (USB products, Santa Clara, USA) and partial sequences were determined by using Big Dye Terminator V3.1 (Applied Biosystems, Foster city, USA). A highly variable region of the 18 S rRNA gene was sequenced using the internal primer Euk528f (5 -CCG-CGG-TAA-TTC-CAG-CTC-3 , Zhu et al., 2005). The ITS region and the 28 S rRNA gene were sequenced using the primers ITS-4 and D1R, respectively. Sequencing was carried out on a ABI prism 3100 sequencer (Applied Biosystems, Foster city, USA).
Based on this preliminary analysis, the full 18 S rRNA gene was sequenced for at least one strain per genotype using primers 63f and 1818r, described above. Twenty-seven full 18 S rRNA sequences were aligned with environmental sequences from the MALINA cruise (Balzano et al., 2012) as well as with other reference sequences from Genbank (http: //www.ncbi.nlm.nih.gov/nucleotide), as described above. A total of 180 sequences were finally aligned. Highly variable regions of the alignment were manually removed. Phylogenetic relationships were analysed using maximum likelihood (ML) and neighbour joining (NJ) methods (Nei and Kumar, 2000). Different models of DNA substitutions and associated parameters were estimated on 1553 unambiguously aligned positions using MEGA5 (Tamura et al., 2011). A General Time Reversible (GTR) model with gamma distributed invariant sites (G + I) was then selected as the best model to infer the ML 18 S phylogeny. A Tamura-Nei model (Tamura and Nei, 1993) was used for the NJ phylogeny. For both methods, bootstrap values were estimated using 1000 replicates. The ML topology was used for all phylogenetic trees shown in this paper, which were constructed using MEGA5 (Tamura et al., 2011).
For some Pedinellales species, only a portion of the 18 S rRNA gene is available in literature. Therefore, we aligned only the corresponding portion of our Pedinellales sequences and inferred a partial 18 S phylogeny. The tree was constructed from an alignment of 37 sequences from Pedinellales as well as other Heterokontophyta based on 434 unambiguously aligned positions.
Since all the 24 ITS sequences obtained for Arctic Micromonas (Mamiellophyceae) were identical, only three of them were considered for the phylogenetic analysis. These sequences were aligned with sequences from other Mamiellophyceae strains from our study as well as from previous works (Slapeta et al., 2006), for a total of 18 sequences. 425 unambiguously aligned positions were used and the phylogenetic tree topology was inferred by the ML method using a Kimura 2-parameter model (Kimura, 1980), and a discrete gamma distribution (5 categories (+ G, parameter = 0.4993)) was used to model evolutionary rates. NJ method and bootstrap values were calculated as described above.
The 28 S rRNA gene sequences from the two dinoflagellate strains (RCC2013 and FT56.6 PG8) isolated from the MALINA cruise were aligned with 33 reference sequences from other dinoflagellates, and 542 unambiguously aligned positions were considered. Different models of DNA substitution were estimated and a GTR model with a discrete gamma distribution (5 categories (+ G, parameter = 0.59)) was used to infer ML phylogeny, whereas NJ phylogeny and boostrap values were calculated as described above.

Microscopy
At least one strain per genotype was observed using light microscopy. Cells were collected during the exponential growth phase and observed using an Olympus BX51 microscope (Olympus, Hamburg, Germany) with a 100X objective using differential interference contrast (DIC). Cells were imaged with a SPOT RT-slider digital camera (Diagnostics Instruments, Sterling Heights, MI, USA) either directly or after fixation with 0.25 % acidic lugol solution (0.6 M KI, 0.39 M crystalline iodine and 1.6 M CH 3 COOH, Sigma Aldrich, Saint-Quentin, France). Micrographs are available at http://www.sb-roscoff.fr/Phyto/RCC for a large set of strains.

Results
Using a range of techniques we isolated 104 strains of photosynthetic flagellates from different Arctic regions. Ninety-three strains have been deposited to the Roscoff Culture Collection (RCC), whereas the others have been lost or discarded subsequently. Complete information is available at http://www.sb-roscoff.fr/Phyto/RCC. After a preliminary phylogenetic analysis, the strains were grouped into 21 genotypes for which the full 18 S rRNA gene was subsequently sequenced.

Chlorophyta, Mamiellophyceae
Arctic Micromonas. Fourty-one strains belong to Arctic Micromonas and were isolated from the northern stations of leg 1b and from 10 stations of leg 2b (Table 2, Supplement, Table S1) at different depths.
Cells are spherical, 2 µm in diameter with a flagellum about 5 µm long ( Fig. 1.1). Consistent with a previous study (Lovejoy et al., 2007), the full 18 S rRNA gene sequences from our Micromonas strains RCC2306 and RCC2308 group with other Arctic sequences forming a sub-clade (94 % ML bootstrap support) within clade B sensu Guillou et al. (2004). This sub-clade is distinct from Micromonas sequences recovered from tropical and temperate waters (Fig. 2, Chlorophyta, Mamiellophyceae). Although our strains have been isolated from both oligotrophic and mesotrophic waters, ITS sequences were identical for all strains, as well as identical to previously published ITS sequences of Arctic Micromonas (CCMP2099, Fig. 3).
Bathycoccus prasinos. We isolated one strain representative from another picoplanktonic Mamiellophyceae, B. prasinos. Unfortunately, this strain was subsequently lost. This strain shares 99.8 % 18 S rRNA and 99.5 % ITS rRNA gene sequence identity with B. prasinos CCAP K-0417 isolated from the Gulf of Naples.
In contrast to Micromonas, the genus Bathycoccus is genetically homogeneous with very little sequence divergence (Guillou et al., 2004;Worden, 2006), and our strain was genetically identical to several strains collected from different oceans. B. prasinos has been previously shown to occur in the Beaufort Sea (Lovejoy et al., 2007), and it was recovered by T-RFLP during the MALINA cruise at only four stations (Balzano et al., 2012), suggesting a marginal contribution to summer photosynthetic picoeukaryotes.
Undescribed Mamiellaceae. From two stations in the Bering Sea, we isolated three other strains of Mamiellophyceae. Cells from these strains are hemispherical, 4 µm wide, and possess a long (15 µm) flagellum and a second very short (1 µm) one ( Fig. 1.2-1.4). A very pale reddish eyespot and a pyrenoid-like inflated body are also visible. These morphological features correspond to those typical of Mantoniella squamata, although electron microscopy is required for the identification of this species (Moestrup, 1990). The full 18 S rRNA gene sequences from RCC2285 and RCC2288 cluster with two environmental sequences, from MALINA and the Baltic Sea, respectively ( Fig. 2, Chlorophyta, Mamiellophyceae), forming a very robust (100 % bootstrap support, for both ML and NJ) clade distinct from the most closely related genera (Micromonas and Mantoniella). ITS phylogeny confirms this finding, although the branch grouping RCC2285, RCC2288, and RCC2497 is less well supported (71 % bootstrap) in ML (Fig. 3). Both 18 S rRNA and ITS phylogeny indicate that our strains fall within the family Mamiellaceae but probably belong to a new genus (Figs. 2-3). Detailed electron microscopy of the cell ultrastructure, the flagellar hair, and body scales would be necessary to confirm this.
Nephroselmis. Three strains (RCC2490, RCC2498, and RCC2499) were isolated from the Bering Strait, with cells 3 to 5 µm long ( Fig. 1.5), pear-shaped with two unequal flagella (http://www.sb-roscoff.fr/Phyto/RCC, RCC2498). Based on the 18 S rRNA gene sequence, these strains belong to the same genotype. They cluster together (100 % ML and NJ bootstrap support) with sequences from N. pyriformis recovered from different oceanic regions and separate from other Nephroselmis species (Fig. 2, Chlorophyta, Nephroselmidophyceae). Since the 18 S rRNA gene appears to be a good molecular marker for identifying Nephroselmis up to the species level (Nakayama et al., 2007), our data suggest that our strains belong to N. pyriformis, a cosmopolitan species occurring in temperate, tropical, but also western Greenland polar waters (Moestrup, 1983;Lovejoy et al., 2002;Nakayama et al., 2007).
Chlamydomonas. We found two genotypes belonging to this genus. Cells from strain RCC2488 (referred as Chlamydomonas sp. I) are approximately 10 µm long and 5 µm wide, with an ovoid shape ( Fig. 1.6). Their 18 S rRNA gene sequences is identical to that of the freshwater species C. raudensis ( Fig. 2, Chlorophyta, Chlorophyceae), which has been previously reported in an Antarctic lake (Pocock et al., 2004). Chlamydomonas sp. I clusters with C. raudensis and C. parkerae within the Moewusii clade sensu Pocock (Pocock et al., 2004).
Carteria. Strain RCC2487 belongs to the genus Carteria. Cells are almost spherical, approximately 30 µm long and 25 µm wide ( Fig. 1.8). Our strain is genetically affiliated with CCMP1189 isolated from Arctic waters, and both strains group with C. radiosa, C. obtusa, and a freshwater Carteria sp., forming a very robust (100 % ML and NJ bootstrap support) clade ( Fig. 2) which likely corresponds to the Carteria I clade (Suda et al., 2005). Members from this clade usually occur in temperate water, and to the best of our knowledge this is the first record of an Arctic strain belonging to this clade.
Pyramimonas. Eleven strains, belonging to four distinct genotypes have been isolated. Cells are spherical to pear-like shaped, 5 to 10 µm long and 3 to 6 µm wide ( Fig. 1.9-1.12). A pyrenoid in the middle or apical region of the cell, a chloroplast with three to four lobes, and a lateral reddish eyespot may be visible in light microscopy. Strains from the different genotypes are undistinguishable in light microscopy and a certain degree of morphological variability in terms of shape (spherical to pear-shaped) and presence of eyespot may occur within the same strain.
Haptolina. Strains RCC2299 and RCC2300 were isolated from the NE Pacific (Table 2). Cells are spherical, about 5 µm in diameter with two yellow-brown chloroplasts and two flagella ( Fig. 1.13). The spines and the haptonema are not visible in light microscopy. The taxonomy of Prymnesiales has been recently revised with the description of the new genus Haptolina and the transfer to this genus of a number of species previously affiliated to Chrysochromulina, including H. ericina and H. hirta (Edvardsen et al., 2011), which are the two species clustering with RCC2300 (92 % ML bootstrap support, Fig. 2, Prymnesiophyceae). These two species cannot be discriminated using the 18 S rRNA gene, but other taxonomic markers such as the 28 S rRNA gene could have helped for the identification (Edvardsen et al., 2011). This clade has a sister clade which includes H. fragaria and an environmental sequence from MALINA ( Fig. 2, Prymnesiophyceae), and these two clades are well supported and delineate the genus Haptolina as shown previously (Edvardsen et al., 2011).
Imantonia. Strains RCC2298 and RCC2504 contain cells approximately 3 µm long, spherical or pear shaped (3 µm long and 2 µm wide, Fig. 1.14). Two lateral chloroplasts and two flagella are located in the wider part of the cell. A single species, I. rotunda, has been described for this genus to date. Strain RCC2298 shares 99.8 % 18 S rRNA gene identity with I. rotunda strain ALGO HAP23 (GenBank accession number AM491014), as well as two unidentified Imantonia strains (Fig. 2, Prymnesiophyceae). Representatives of the genus Imantonia have been previously recorded in high latitude (Backe- Hansen and Throndsen, 2002) and temperate (Percopo et al., 2011) waters.

Cryptophyta, Cryptophyceae
Rhodomonas. The eleven Cryptophyceae strains isolated from one NE Pacific and five Beaufort Sea stations belong to the same genotype. Cells are ovoid, approximately 20 µm long and 10 µm wide, with two greenish-brown chloroplasts and a short furrow extending posteriorly ( Fig. 1.15). Cells possess two equal flagella inserting into a ventral furrow. The genus Rhodomonas can be distinguished from the closely related genus Storeatula because the latter lack the furrow (Deane et al., 2002).
The full 18 S rRNA gene sequence from RCC2020 clusters with R. abbreviata (81 % ML bootstrap support, Fig. 2, Cryptophyceae). Genus level phylogeny is not well resolved for Rhodomonas; the RCC2020/R. abbreviata clade branches with other Rhodomonas species but also with other genera such as Rhinomonas, Storeatula, Cryptomonas, and Pyrenomonas (Fig. 2 Fig. 2. Full 18 S rDNA phylogenetic tree including at least one sequence from each genotype found within the strains isolated during the MALINA cruise. The tree has been split into five groups (Heterokontophyta, Chlorophyta, Dinophyceae, Prymnesiophyceae and Crytophyceae); two fungal sequences (Phoma herbarum AY337712 and Sidowia polyspora AY544718) have been used as outgroups and are not shown for clarity. The tree was inferred by maximum likelihood (ML) analysis using MEGA5. 1553 unambiguously aligned positions were considered from an alignment of 180 nucleotide sequences. The strains sequenced in the present study are labelled in red, the environmental sequences recovered during the MALINA cruise (Balzano et al., 2012) are in blue, and other reference sequences from the Genbank are in black. Full circles indicate genotypes isolated from nitrogen depleted waters (surface waters from the leg 2b); full squares, genotypes isolated from mesotrophic waters; and empty circles, genotypes isolated from both conditions. The tree with the highest log likelihood (−26101.3937) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches from left (ML, 1000 replicates) to right (NJ, 1000 replicates). "−" indicates that bootstrap values < 70 % were obtained for the corresponding node. Poorly supported clades (< 50 % bootstrap support) have been removed. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+ G, parameter = 0.4722)). The rate variation model allowed for some sites to be evolutionarily invariable ((+ I), 27.2360 % sites). The tree is drawn to scale, with branch lengths estimated as the number of substitutions per site. polyphyletic genus and its key diagnostic features may represent the characters of the clade (Deane et al., 2002).

Alveolata (Dinophyceae)
We isolated and sequenced from the Beaufort Sea (Table 2) two strains of dinoflagellates (RCC2013 and FT56.6 PG8) belonging to a single genotype. Strain RCC2013 has been observed both in light and electron microscopy, whereas the second strain was lost before these microscopy analyses could be carried out. Cells are almost spherical, approximately 10 µm in diameter, with a shallow and descending cingulum, a deep sulcus, and a bright yellow eyespot ( Fig. 4.1, arrow). In electron microscopy, four series of plates in the epicone and three in the hypocone are visible (Fig. 4.2 and 4.3), as well as an elongate apical vesicle (EAV, see Moestrup et al., 2009a, for the definition of the EAV, Fig. 4.4 and 4.5).
The morphology of this strain perfectly matches with Woloszynskia cincta Siano, Montresor and Zingone, a species described from the Mediterranean Sea (Siano et al., 2009) and reported also in the Pacific Ocean (Kang et al., 2011). This identification is corroborated by genetic data. The 18 S rRNA gene sequences from the MALINA strains share 99.9 % identity with the W. cincta strain from the Pacific Ocean (Kang et al., 2011), and the 28 S rRNA gene sequences of our strains share 100 % identity with the W. cincta from both the Pacific Ocean and Mediterranean Sea. In both 18 S and 28 S rRNA gene sequence phylogenies, W. cincta form robust clusters with sequences of the genus Biecheleria (18 S: 100 % bootstrap for both ML and NJ, Fig. 3; 28 S: 96 % ML, 100 % NJ bootstrap, Fig. 5), questioning the ascription of W. cincta to the genus Woloszynskia.
In recent years, the systematics of the genus Woloszynskia have been revised on the basis of both genetic and morphological data. Many species previously classified as Woloszynskia but morphologically different from the type species of the genus, W. reticulata (Moestrup et al., 2008), have been recombined in four newly described genera: Biecheleria, Borghiella, Jadwigia, and Tovellia (Lindberg et al., 2005;Moestrup et al., 2008Moestrup et al., , 2009a. In addition, three new genera of woloszynskioid dinoflagellates have been erected: Baldinia, Biecheleriopsis, and Pelagodinium (Hansen et al., 2007;Moestrup et al., 2009b;Siano et al., 2010). Morphologically, W. cincta shares with Biecheleria pseudopalustris a posterior invagination and a spiny spherical cyst (Moestrup et al., 2009a;Siano et al., 2009). Biecheleria halophila and B. pseudopalustris have a type E eyespot sensu Moestrup and Daugbjerg (Moestrup and Daugbjerg, 2007). The presence of a type E eyespot was not reported in the original description of W. cincta based on the Mediterranean strain (Siano et al., 2009), but the ultrastructural analyses of the Pacific strain (Fig. 15 in Kang et al., 2011), genetically identical to the MALINA and the Mediterranean strains (Figs. 3 and 5), proved the existence of a type E eyespot in W. cincta (Kang et al., 2011).
On the basis of our new morphological and genetic data and previously provided evidences, we therefore propose the following new combination for W. cincta:
This dinoflagellate species has a wide distribution since it has been found in tropical (Kang et al., 2011), temperate (Siano et al., 2009) and polar waters (this work).

Heterokontophyta
We isolated a total of 25 strains belonging to the classes Chrysophyceae, Dictyochophyceae, and Pelagophyceae, which grouped into 6 distinct genotypes.

Chrysophyceae
Dinobryon. Four strains have been morphologically identified as Dinobryon faculiferum. Dinobryon species can be easily identified because cells are surrounded by a cellulose lorica. In RCC2292, RCC2293, and RCC2294 cells are solitary and surrounded by a thin and cylindrical lorica 60-90 µm long and 5-10 µm wide; this lorica terminates with a long spine (Fig. 1.16-1.17). Within the lorica, cells are ovoid, approximately 10 µm long and 5 µm wide. These features are typical of D. faculiferum (Throndsen, 1997), which has been frequently observed in Arctic waters (Booth and Horner, 1997;Lovejoy et al., 2002).
Genetically, the three strains (Supplement , Table S1) belong to the same genotype, and the strains RCC2290 and RCC2293 (full 18 S rRNA gene) are grouped together and have a sister clade which includes an environmental sequence from MALINA (Fig. 2, Heterokontophyta, Chrysophyceae). Sequences for D. faculiferum as well as for other marine Dinobryon species are not available in Genbank and, surprisingly, sequences from other freshwater species such as D. sociale, D. cylindricum, and D. sertularia form a clade distinct from that of our strains. Marine species of Dinobryon could group with our sequences and form a separate clade from freshwater Dinobryon species. However the phylogeny of the overall genus is not well resolved (Fig. 2, Heterokontophyta, Chrysophyceae). More sequences from marine species will be needed to better characterise this genus.

Dictyochophyceae
Pedinellales. We isolated 10 strains from this order belonging to two distinct genotypes (Fig. 2, Heterokonto Fig. 3. ITS rRNA based phylogeny of the Mamiellophyceae strains isolated from the Beaufort Sea. The phylogenetic tree was inferred by maximum likelihood (ML) analysis. 425 unambiguously aligned positions were considered from an alignment of 18 sequences. Sequences from MALINA strains are labelled in red. The evolutionary history was inferred by using the maximum likelihood method based on the Kimura 2-parameter model. The tree with the highest log likelihood (−2718.0303) is shown. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+ G, parameter = 0.4993)). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The tree was rooted with Monomastix minuta as an outgroup. The tree has been then edited, and ML and NJ bootstrap values have been included as described in Fig. 3. Evolutionary analyses were conducted in MEGA5.   Fig. 5. 28 S rDNA phylogenetic tree inferred by maximum likelihood (ML) analysis for the dinoflagellate strains isolated during the MALINA cruise. 543 unambiguously aligned positions were considered from an alignment of 35 nucleotide sequences. The strains sequenced in the present study are labelled in red. The tree with the highest log likelihood (−6075.65) is shown. A discrete gamma distribution was used to model evolutionary rate differences among sites (5 categories (+ G, parameter = 0.63)). The tree is drawn to scale with branch length measured in the number of substitutions per site. Evolutionary analyses were conducted in MEGA5 (Tamura et al., 2011). The tree was rooted with Ceratium fusus and Ceratium lineatum as outgroups. Bootstrap values > 70 % are shown next to the branches from left (ML, 1000 bootstrap) to right (NJ, 1000 bootstrap). "−" indicates that lower bootstrap values were obtained for the corresponding node.
Biogeosciences, 9, 4553-4571, 2012 www.biogeosciences.net/9/4553/2012/ radially symmetrical and possess six peripheral chloroplasts ( Fig. 1.18 and 1.20). When viewed from the side, a stalk and a flagellum are visible ( Fig. 1.19). We are not certain of the genus level identification of our strains because morphological features such as the stalk shape (straight or coiled) and the presence of tentacles, which allow the identification of Pedinellales (Sekiguchi et al., 2003), are not visible. Genetically, MALINA Pedinellales strains cluster to two distinct groups: the first group includes 7 strains (sp. I) whereas the second group includes two strains (sp. II, Table S1). The full 18 S rRNA gene sequence from RCC2289 (sp. I) clusters with environmental sequences from MALINA and the Baltic Sea (100 % bootstrap 25 support) and form a sister clade with Pteridomonas danica (Fig. 2, Heterokontophyta, Dictyochophyceae). Partial 18 S rRNA phylogeny indicates that our sequences group with Helicopedinella tricostata (Supplement, Fig. S1), forming a well supported (94 % and 98 % ML and NJ, respectively) clade. However, sp. I probably does not belong to the genus Helicopedinella because our strains possess six chloroplasts (Fig. 2), while genus Helicopedinella is defined as containing only three chloroplasts (Sekiguchi et al., 2003).
In contrast, full length sequences from RCC2286 and RCC2301 (sp. II) cluster with the strain CCMP2098 and Pedinella squamata, forming a well supported clade (98 % and 100 % ML and NJ bootstrap support, respectively) and suggesting that our strains might belong to the genus Pedinella. Partial 18 S rRNA phylogeny indicates however that our sequences group with P. squamata as well as Mesopedinella arctica RCC382 (Supplement, Fig. S1). The attribution of RCC2286 and RCC2301 to the genus Pedinella is thus also uncertain.
Pelagophyceae. Eleven strains affiliated to this class were isolated (Supplement , Table S1) and grouped into three genotypes (Table 2) which cannot be distinguished by light microscopy. Cells are hemispherical or bean shaped in side view, about 5-7 µm long ( Fig. 1.21-1.23), and adorned with two lateral flagella and a lateral yellowish-brown chloroplast. These features might correspond to those typical of Ankylochrysis lutea (Honda and Inouye, 1995), and the cells from our strains are similar in size and shape to those of the strain RCC286 identified as A. lutea (http://www. sb-roscoff.fr/Phyto/RCC).
The 18 S rRNA gene sequences from the three genotypes branch with A. lutea into a well supported clade (98 % ML, 92 % NJ bootstrap support) distinct from the other 25 Pelagophyceae genera such as Aureococcus, Pelagomonas, and Pelagococcus. Sp. II is closely related to an environmental sequence from Baltic Sea ice (Fig. 2, Heterokontophyta, Pelagophyceae).

Isolation and identification success
The combination of both concentration by TFF and medium enrichment with FCS and single cell pipette isolation proved to be successful for isolating eukaryotic phytoplankton and preventing their contamination by heterotrophic microorganisms. Some of our cultures proved to be non-unialgal and were further purified using single cell FCS. In these cultures, the dominant genotype was initially contaminated either by other phytoplankters (especially the centric diatom Chaetoceros sp.) or by heterotrophs such as uncultured Cercozoa or a Chrysophyceae affiliated to Paraphysomonas imperforata. The latter has a cosmopolitan distribution and is an opportunistic species which often dominates enrichment cultures (Lim et al., 1999).
Several genotypes could not be identified down to the species level. In addition, Dictyochophyceae and Pelagophyceae strains could not be identified at the genus level, and we found a new genus within Mamiellaceae. Whole mount and/or thin section electron microscopy would be required to characterise these genotypes further.

Autotrophic microbial diversity revealed using culturing techniques
Significant diversity occurred within cultured photosynthetic flagellates, and 8 genotypes found here were not detected by T-RFLP or cloning/sequencing of environmental samples sorted by flow cytometry based on their chlorophyll fluorescence, and thus containing only photosynthetic eukaryotes (Table 2). Within these genotypes, Rhodomonas sp. was not targeted during sorting because it contained orange-fluorescing phycoerythrin and therefore did not appear in the T-RFLP data. However, Rhodomonas sp. was observed by light microscopy in environmental samples ( Table 2, http://www.obs-vlfr.fr/Malina/data.html) and was previously found in the North Water Polynya off Greenland (Lovejoy et al., 2002). In contrast, the other genotypes are likely to belong to rare species which can be easily cultured.
The rarefaction curve indicates that we sampled a very large portion of the community of photosynthetic flagellates during the MALINA leg 2b that could be cultivated under the conditions we used (Supplement, Fig. S2). However, if we had used a larger diversity of media and isolation strategies, we would have probably recovered many other genotypes. The four Pyramimonas genotypes are undistinguishable by light microscopy and group into two T-RFLP ribotypes (sp. I/sp. IV and sp. II/sp. III). Similarly, the different genotypes found within Pedinelalles and Pelagophyceae www.biogeosciences.net/9/4553/2012/ Biogeosciences, 9, 4553-4571, 2012 share the same T-RFLP patterns for the restriction enzymes used by Balzano et al. (2012) and cannot be discriminated by T-RFLP. Therefore, although we isolated several genotypes within the genus Pyramimonas, the order Pedinellales, and the class Pelagophyceae we cannot determine whether all the cultured genotypes were present in the environmental samples analysed in the companion paper (Balzano et al., 2012). Surprisingly, we found few dinoflagellates among both our strains and environmental samples of nanoplankton (Balzano et al., 2012). However, microscopy counts revealed the presence of several dinoflagellate species during MALINA, although never as dominant taxa. Most of them were larger than 15 µm and belonged to the genera Gymnodinium and Gyrodinium (http://www.obs-vlfr.fr/Malina/data.html).
Dinoflagellates are an important component in the Arctic (Okolodkov and Dodge, 1996), and they occur during summer in the Chukchi Sea (Booth and Horner, 1997) and the North Water Polynya (Lovejoy et al., 2002). In the Beaufort Sea, however, they seem to occur in autumn (Brugel et al., 2009) rather than in mid summer (Okolodkov, 1999;Sukhanova et al., 2009), which was the period of the MALINA cruise.

Culturable phytoplankton in oligotrophic waters
Interestingly, 8 out of the 21 genotypes found here correspond to strains isolated during leg 2b from surface waters which were depleted in inorganic nitrogen (Table 2, Supplement, Table S1). Inorganic nitrogen, which was undetectable in the surface layer during MALINA, has been shown to limit bacterial production (Ortega-Retuerta et al., 2012) and was likely to limit primary production as well. The diversity found in surface waters contrasts with the fact that oligotrophic environments are generally considered to harbour slow growing/hard to cultivate phytoplankton. For example during a similar study in the southeast Pacific, no strain could be isolated from the two most oligotrophic sites (Le Gall et al., 2008). Similarly, cultured microbes contribute very poorly to phytoplankton diversity in other oligotrophic waters such as the eastern Mediterranean Sea Man-Aharonovich et al., 2010), the Sargasso Sea , or the northeast Atlantic Ocean (Jardillier et al., 2010). This suggests that resilient ecotypes adapted to the sub-freezing temperatures and variable salinities observed in the Arctic are more easily culturable than ecotypes from warm and relatively stable temperate or tropical oligotrophic waters.
In contrast, Arctic Prymnesiophyceae from oligotrophic environments appear hard to be brought in culture. The strains isolated in this study derive from mesotrophic environments of the NE Pacific or the Bering Strait, and we could not culture any Prymnesiophyceae from the Beaufort Sea, although they occurred in environmental samples. In particular, 4 operational taxonomic units (OTUs) affiliated to the genus Chrysochromulina were observed by T-RFLP and cloning/sequencing in both surface and DCM samples during MALINA (Balzano et al., 2012). Microscopy counts also revealed the presence of Chrysochromulina spp. throughout the Beaufort Sea (http://www.obs-vlfr.fr/Malina/data.html).

Low diversity of photosynthetic picoplankton
Arctic Micromonas and B. prasinos were the only taxa of picoplanktonic size recovered during this study. Imantonia rotunda has been previously reported to be < 2 µm , but our strains of Imantonia sp. had a larger size ( Fig. 1.14). In contrast, during a similar study carried out in another oligotrophic system, the southeast Pacific Ocean, photosynthetic picoplankton was more diverse (Shi et al., 2009) and picoplanktonic strains belonging to several different lineages were successfully isolated and cultured (Le Gall et al., 2008). A higher diversity of total photosynthetic picoeukaryotes has also been reported in other warmer oligotrophic regions such as the Sargasso Sea , the Mediterranean Sea , and the northeast Atlantic Ocean (Jardillier et al., 2010).
The photosynthetic picoplankton community in the Arctic consists almost uniquely of a single Arctic Micromonas ecotype, which occurs throughout the Beaufort Sea. Since all our strains share identical 18 S rRNA and ITS sequences, Arctic Micromonas populations are likely to be highly homogeneous despite the fact that they are present in both surface nitrate-depleted waters and deeper, colder, saltier, nitrate-replete waters. The ubiquity and dominance within picoplankton of Arctic Micromonas throughout the Beaufort Sea (Balzano et al., 2012) indicates that it can grow or at least survive throughout a wide range of salinities (14 to 32 psu) and temperatures (1 to 7 • C), as well as under both nitrate-depleted (< 3 nM) and nitrate-replete (up to 6.7 µM) conditions.
Nitrate-depleted conditions in general promote the growth of picoplankton over larger cells because of the lower surface to volume ratio, and accordingly, photosynthetic picoplankton was generally more abundant than nanoplankton in surface waters of the Beaufort Sea during the MALINA cruise (http://tinyurl.com/67wn5qc). Arctic Micromonas is able to survive cold waters and long dark winters (Sherr et al., 2003;Lovejoy et al., 2007), this makes it prevail over other photosynthetic picoplankters under Arctic conditions. In the Beaufort Sea, coastal waters may reach higher (7 • C) temperatures during summer, but they remain throughout the whole year surrounded by colder waters, and the transport and survival of phytoplankton species from temperate waters is thus highly unlikely. In contrast, the Norwegian and Barents Seas are in close contact with temperate waters from the Atlantic Ocean. The photosynthetic picoplankton is more diverse there; Arctic Micromonas occurs with other Micromonas clades (Foulon et al., 2008), as well as with other Chlorophyta and Haptophyta . Consistent with this hypothesis, the higher temperatures which are observed in the NE Pacific and the Bering Strait (Table 1) explain the presence of other picoeukaryotes such as Mamiellophyceae, Chrysophyceae, and unidentified picoeukaryotes which occur along with the Arctic Micromonas (Balzano et al., 2012).

Importance of mixotrophic nano-and microplankton strains
Strains larger than 2 µm appear much more diverse than picoplankton strains. Fourteen out of 21 genotypes (Table 2) found here include strains recovered from nitrogen-depleted surface waters and often correspond to genera reported in oligotrophic systems and sometimes shown to be mixotrophic. For example, mixotrophy has been reported for both freshwater (Bird and Kalff, 1986;Domaizon et al., 2003;Kamjunke et al., 2007) and marine (McKenzie et al., 1995) Dinobryon species including D. faculiferum (Unrein et al., 2010). Dinobryon strains were isolated from nitrogen-depleted waters (Table 2), and Dinobryon cells were also observed in surface water as indicated by microscopy counts (http://www.obs-vlfr.fr/Malina/data. html) and T-RFLP (Balzano et al., 2012). Chloroplast containing Pedinellales from the Baltic Sea have been found to ingest bacteria (Piwosz and Pernthaler, 2010). Similarly, P. gelidicola, a species which shares 100 % 18 S rRNA gene identity with our strains of Pyramimonas sp. II, was also shown to feed on bacteria (Bell and Laybourn-Parry, 2003). B. cincta comb. nov. isolated from Pacific Ocean was observed to ingest several algal preys using a peduncle located between the two flagella (Kang et al., 2011). Some of the strains isolated during this study might thus be mixotrophic, and their ability to assimilate organic carbon could allow their survival and/or growth under the nitrogen-depleted conditions occurring in surface waters of the Beaufort Sea during summer.

Arctic, polar, and cosmopolitan species
Four out of the 21 genotypes found in the present study (Arctic Micromonas, Pyramimonas sp. I, Pyramimonas sp. III and undescribed Pedinellales sp. II) have a strictly Arctic distribution and 7 genotypes have been sequenced for the first time (Carteria sp., Pyramimonas sp. IV, Rhodomonas sp., D. faculiferum and the three Pelagophyceae genotypes). In contrast, the other genotypes have also been reported in other oceans (Table 2). Similarly, environmental sequences from the MALINA cruise include 34 out 46 OTUs which cluster into new or endemic lineages (Balzano et al., 2012) and previous studies also highlight the prevalence of endemic lineages among Arctic environmental clone libraries (Lovejoy et al., 2006;Luo et al., 2009). The proportion of endemic and polar OTUs within our strains may be overestimated because part of the biogeography of most marine microbes is still unknown and many genotypes found here may occur elsewhere. On the other hand, different species may share the same 18 S rRNA sequence (e.g. within the genera Pyramimonas or Haptolina), and some of our cosmopolitan genotypes may be related to different species with more restricted geographical distribution.
Pyramimonas species occur frequently in polar waters, as they have been previously reported in Arctic environments (Daugbjerg and Moestrup, 1993;Gradinger, 1996) including the Beaufort Sea water column (Olli et al., 2007;Brugel et al., 2009) and ice (Rozanska et al., 2008) as well as the Barents (Rat'kova and Wassmann, 2002) and Laptev Seas (Tuschling et al., 2000). Other Pyramimonas species occur in the Antarctic Ocean (Moro et al., 2002) where some of them were reported to form blooms in Gerlache Strait (Varela et al., 2002) and Omega Bay (McMinn et al., 2000). Some Pyramimonas species appear to be adapted to the salinity changes typically occurring in the Beaufort Sea, as they were previously found under the ice pack (Gradinger, 1996) and shown to grow across a broad salinity range (Daugbjerg, 2000).
Some of our genotypes might be indeed adapted to salinity changes since sequences from our strains of Pyramimonas sp. II, Pedinellales sp. I, and the undescribed Mamiellaceae as well as from the Beaufort Sea environment al samples (Balzano et al., 2012) match sequences from the Baltic Sea. Although the Baltic Sea is much fresher and far less cold than the Beaufort Sea, both ecosystems undergo seasonal salinity changes and (partial) winter freezing events which may promote the growth of the same species.
The biogeography of Arctic microbes is currently highly debated; similarities between Arctic and Antarctic assemblages have been reported for ice, sediment (Lozupone and Knight, 2005), soil (Chu et al., 2010), snow, air, and freshwater bacteria (Jungblut et al., 2010;Harding et al., 2011), whereas seawater bacteria show a limited dispersal ability suggesting the occurrence of a marine microbial province in the Arctic (Galand et al., 2009(Galand et al., , 2010. Similarly, eukaryotic microbes from terrestrial environments of the Arctic may also occur in Antarctic and alpine environments (Harding et al., 2011;Schmidt et al., 2011), whereas marine eukaryotes are less likely to be globally dispersed. Arctic circumpolar isolation occurs, for example, for Arctic Micromonas (Lovejoy et al., 2007), and for the planktonic foraminiferan Neogloboquadrina pachyderma (Darling et al., 2007). However, Arctic barriers have been suggested to weaken, at least for abundant species, because of the ice retreat; increased seawater flows through the Arctic likely imply the dispersion of species from the Pacific to the Atlantic Ocean (Wassman et al., 2011). For example, the Pacific diatom Neodenticula seminae appeared in Labrador Sea for the first time in 1999 (Reid et al., 2007) and Atlantic and Pacific populations of Emiliania huxleyi were found to share similar mitochondrial DNA sequences (Hagino et al., 2011). www.biogeosciences.net/9/4553/2012/ Biogeosciences, 9, 4553-4571, 2012 Interestingly, our Chlamydomonas genotypes are cosmopolitan and have a likely freshwater origin since they match sequences from freshwater environments (Fig. 2, Chlorophyta, Chlorophyceae). Our strains have been indeed isolated from Stations 670 and 680 (Table 1), which are located near the main outlets of the Mackenzie River. A previous study already found a high similarity between the Antarctic Chlamydomonas raudensis and an Arctic Chlamydomonas sp. (De Wever et al., 2009), which are both closely related to Chlamydomonas sp. I. Similarly, the freshwater flagellate Spumella comprises three globally distributed clades, one of which has been frequently found in Antarctic waters (Nolte et al., 2010).
Arctic Micromonas, undescribed Mamiellaceae, B. prasinos, and Rhodomonas sp. were found in both the NE Pacific and the Beaufort Sea (Table 2). In contrast, Haptolina sp., Imantonia sp., and N. pyriformis only occurred in the NE Pacific and/or the Bering Strait and did not appear in the Chuckchi and Beaufort Seas. The other 14 OTUs were found only in the Beaufort Sea (Table 2). Similarly, planktonic foraminifera from the Beaufort Sea were found to be phylogenetically different from those occurring in the North Pacific and rather related to North Atlantic foraminifera (Darling et al., 2007), suggesting that the Bering Strait may act as a barrier to microbial dispersion.

Conclusions
The combination of culture-dependent (this study) and culture-independent (Balzano et al., 2012) techniques provided useful insights on phytoplankton diversity in the Beaufort Sea. Photosynthetic picoplankton was almost exclusively represented by highly homogeneous populations of Arctic Micromonas which occurred over a range of temperature, salinity, nutrient and light conditions. The high diversity found for surface nanoplankton and the known ability for some of these species to feed on bacteria suggest that their presence in oligotrophic waters could be supported by a mixotrophic carbon assimilation mode.