Diversity and Distribution of Marine Microbial Eukaryotes in the Arctic Ocean and Adjacent Seas^,

Abstract

We analyzed microbial eukaryote diversity in perennially cold arctic marine waters by using 18S rRNA gene clone libraries. Samples were collected during concurrent oceanographic missions to opposite sides of the Arctic Ocean Basin and encompassed five distinct water masses. Two deep water Arctic Ocean sites and the convergence of the Greenland, Norwegian, and Barents Seas were sampled from 28 August to 2 September 2002. An additional sample was obtained from the Beaufort Sea (Canada) in early October 2002. The ribotypes were diverse, with different communities among sites and between the upper mixed layer and just below the halocline. Eukaryotes from the remote Canada Basin contained new phylotypes belonging to the radiolarian orders Acantharea, Polycystinea, and Taxopodida. A novel group within the photosynthetic stramenopiles was also identified. One sample closest to the interior of the Canada Basin yielded only four major taxa, and all but two of the sequences recovered belonged to the polar diatom Fragilariopsis and a radiolarian. Overall, 42% of the sequences were <98% similar to any sequences in GenBank. Moreover, 15% of these were <95% similar to previously recovered sequences, which is indicative of endemic or undersampled taxa in the North Polar environment. The cold, stable Arctic Ocean is a threatened environment, and climate change could result in significant loss of global microbial biodiversity.

The Arctic Ocean (AO) and surrounding seas have traditionally been thought of as being dominated by large phytoplankton of >20 μm (67); however, recent studies show that these waters have active microbial food webs that are often dominated by cells of <3 μm (37, 57) and that cells of <5 μm are responsible for much of the carbon fixation over wide regions of the Arctic Basin (23, 31). The Arctic Ocean is an enclosed sea with a cold, moderately fresh (<30 practical salinity units) upper mixed layer of 30 to 60 m. These upper photic zone waters are separated from deeper waters by a strong halocline that is maintained by large riverine inputs and the annual formation and melting of sea ice (1). The physical isolation, perennially cold water temperatures (<0°C), and extreme annual light cycle provide a distinct marine habitat for microorganisms (15). 16S rRNA gene surveys have uncovered novel archaeal and eubacterial sequences from remote polar regions (6, 7, 12), confirming that these ambient conditions select for particular microorganisms, but there are no equivalent studies of microbial eukaryotes. North polar latitudes are predicted to warm rapidly as a result of global climate change and have already experienced significant impacts (2, 48). An assessment of current microbial diversity is therefore paramount for this region at this early stage in climate modification.

Isolated and extreme environments have been important sources of novel phylotypes (33, 47) that have contributed to the recent major reassessment of eukaryotic evolution (5). rRNA gene sequences from uncultured marine eukaryotes have also led to major revisions of eukaryotic phylogeny at multiple taxonomic levels (25, 40, 51). This is the first major study of the molecular diversity of small eukaryotes in arctic marine waters. We collected samples for DNA analysis from the AO and the Canadian and European Arctic Seas to investigate the diversity of picoeukaryotes (<3-μm-diameter cells) by analysis of 18S rRNA gene clone libraries. Our results show the dominance and diversity of radiolarians and the presence of novel lineages in diverse protist groups.

MATERIALS AND METHODS

Oceanographic sampling and environmental data.

Samples were collected as part of three nearly simultaneous oceanographic expeditions to different regions of the Arctic (Fig. 1). Canada Basin Arctic Ocean stations AO-NW01 (75°59′13"N, 156°52"9′W; maximum depth [Z_max], 801 m) and AO-NW08 (76°46′62"N, 148°57"55′W; Z_max, 3,474 m) were sampled from the Canadian Coast Guard ship Louis St. Laurent. Temperature, pressure, and conductivity measurements were done using a Sea-Bird Electronics SBE-911 conductivity-temperature-depth (CTD) profiler mounted on a General Oceanics rosette carousel equipped with 24 12-liter Niskin bottles (42). Chlorophyll a (Chl a) was analyzed on board (31), as were the major nutrients nitrate (NO₃), soluble reactive phosphorus, and silica, using standard techniques (8).

FIG. 1.

Map showing sites sampled. European Arctic sites GNB-M09 and GNB-Z59 were south of Svalbard in the GNB. North American sites were in (AO-NW08) and on the border of (AO-NW01) the Canada Basin and from the Beaufort Sea-Mackenzie Delta region (BS-MD65).

The Canadian Beaufort Sea Station BS-MD65 (133°31′19"W, 70°08′40"N; Z_max, 33 m) was sampled from the Canadian Coast Guard ship Radisson on 3 October 2002 using a Seabird CTD rosette as described above. Nutrient (NO₃, soluble reactive phosphorus, and silica) samples were analyzed on board using an ALPKEM autoanalyzer and routine colorimetric methods (24). Samples for Chl a were filtered through GF/F filters and then stored frozen (−80°C) until pigment extraction in ethanol (53).

The convergence of the Greenland, Norwegian and Barents Seas (GNB) was sampled on 26 to 28 August 2002. Stations M09 (76°19′06"N, 23°44′42"E; Z_max, 67 m) and Z59 (76°19′54"N, 3°59′12"E; Z_max, 3,231 m) were sampled from the F/F Johan Hjort (Norwegian Institute of Marine Research). Samples were collected from a Seabird 10 CTD rosette system mounted with 10 5-liter Niskin bottles. Nutrients were analyzed at the Norwegian Institute of Marine Research using standard techniques, and Chl a concentrations were determined on board (60).

These differences in methodologies among the cruises mean that absolute comparisons should be made with some caution. However, for nutrients all three methods were substantially the same colorimetric techniques optimized for the autoanalyzer systems used by the different laboratories, and all met established standards and calibration requirements (http://www.pangaea.de/Projects/JGOFS/Methods/chap8.html). Similarly, there were minor differences in Chl a methodologies; despite this, Chl a values can be practically compared for upper mixed-layer marine waters at the sites sampled (29). At all stations, water samples for DNA analysis were collected directly from the Niskin bottles into clean bottles that had been rinsed with acid (10% HCl) and then with MilliQ water, followed by three rinses of sample water prior to filtering. Canadian microbial samples were collected by filtering 1 to 2 liters of seawater under <5 mm Hg pressure. The particles were successively trapped onto 47-mm-diameter, 3-μm-pore-size polycarbonate prefilters and then onto 0.22-μm-pore size, 47-mm Durapore filters. For Norwegian samples, 4 to 5 liters of water was prefiltered as described above and microbial biomass collected in 0.22-μm Sterivex filter units with a peristaltic pumping system. Filters were frozen at −70°C in lysis buffer (40 mM EDTA, 50 mM Tris-HCl, 0.75 M sucrose) until nucleic acid was extracted.

DNA extractions.

Sample filters were thawed on ice and then digested using lysozyme (final concentration, 1 mg ml⁻¹) and proteinase K (0.21 mg ml⁻¹). Lysates were recovered and nucleic acids extracted with phenol-chloroform-isoamyl alcohol (25:24:1), followed by chloroform-isoamyl alcohol (24:1), and concentrated using Centricon-100 concentrators (Millipore) (21).

DNA amplification, cloning, and sequencing.

A total of eight clone libraries were constructed (Table 1). Eukaryotic 18S rRNA genes were amplified by PCR with eukaryote-specific primers EukA and EukB (44). Amplified rRNA gene products from several individual PCRs were pooled, cleaned using a QIAGEN purification kit, and then cloned with the TA cloning kit (Invitrogen) following the manufacturer's directions. Positive colonies were screened for restriction fragment length polymorphisms (RFLP) with HaeIII (Gibco BRL). Clones with the same RFLP pattern were grouped and considered members of the same phylotype (21). Phylotypes were sequenced using the Euk 528F primer and Big Dye (3.1) Terminator ready-reaction mix to obtain a 750- to 800-bp segment covering conserved and rapidly evolving regions of the small-subunit rRNA gene (50, 68). Additional universal eukaryotic primers (Euk 336f, 516r, and 1055f) were used to obtain a nearly complete 18S rRNA gene sequence from selected clones. Sequencing was done by the Serveis Científico-Tècnics, Universitat Pompeu Fabra (Barcelona, Spain), with an ABI3100 automated sequencer.

TABLE 1.

Stations, sample depths, and physical properties of and nutrient concentrations in water sampled for the clone libraries

Station	Depth (m)	Library	PSUa	Temp (°C)	Nutrients (μM)b			Chl a (μg liter−1)
					Nitrate	SRP	Silicate
GNB-M09	5	NOR26	33.17	4.53	0.7	0.20	1.29	2.81
GNB-Z59	5	NOR46	33.49	6.10	1.3-2.4	0.2-0.3	0.7-0.9	0.61
GNB-Z59	60	NOR50	34.86	2.82	11.9	0.8	5.4-5.7	0.09
BS-MD65	5	MD65	26.49	−0.44	0.4	0.6	3.8	0.21
AO-NW01	5	NW617	27.05	−1.36	0.1	0.6	3.8	0.19
AO-NW01	50	NW614	31.87	−0.82	5.5	1.2	12.7	0.25
AO-NW08	5	NW415	27.02	−1.4	0.1	0.6	3.2	0.07
AO-NW08	50	NW414	31.77	−1.0	4.8	1.1	10.1	0.53

^aPSU, practical salinity units.

^bSRP, soluble reactive phosphorus. Ranges indicate missing data, estimated from adjacent stations and depths.

Phylogenetic analysis.

The closest match to each sequence was obtained from NCBI BLAST (4). Poor-quality sequences and suspected chimeras were checked by using BLAST with sequence segments separately and then using the Chimera check program at Ribosomal Data Project II (Michigan State University; http://35.8.164.52/cgis/chimera.cgi?su = SSU). The sequences that passed chimeric screening were phylogenetically grouped and aligned using Clustal X v.1.83 (64); alignments were manually checked using Bioedit v.5.0.9 (27). Tree construction was done with PAUP v.4.0b10 (Sinauer Associates, Inc., Sunderland, Massachusetts), using neighbor-joining (NJ) and maximum-likelihood (ML) methods (25). Difficult or poorly aligned positions and divergent regions were eliminated using Gblocks (17) with a minimum block of five and allowed gap positions equal to half. Clade credibility was checked with a heuristic search using MrBayes v.3_0b4 (66). Accession numbers for sequences used in phylogenies are given in the supplemental material.

Nucleotide sequence accession numbers.

Sequences reported in this paper have been deposited in GenBank under accession numbers DQ055149 to DQ055172, DQ062463 to DQ062515, DQ119893 to DQ120009, DQ314809 to DQ314838, and DQ344786 to DQ344806.

RESULTS

Environmental.

The two sides of the Arctic covered a range of temperatures and salinities (Table 1). Nutrient concentrations were low in surface waters at all stations and were greater in deeper waters below the halocline. Chl a levels were low at all sites except GNB-M09 (2.81 μg Chl a liter⁻¹). For station GNB-Z59 at the western edge of the GNB transect, Chl a levels were low in the surface and extremely low in the 60-m sample (0.03 μg Chl a liter⁻¹). Canada Basin AO sites had a sharp halocline at ca. 30 m, and Chl a levels were greater in the deeper water than at the surface. This was especially marked at AO-NW08, where Chl a levels were 7 times greater at 50 m than at 5 m. Temperatures were below 0°C, except in the GNB, which had summer surface warming and was influenced by North Atlantic boundary currents (52).

Clone libraries.

Each library yielded between 96 and 288 positive clones, with a minimum of 48 and a maximum of 195 clones that were RFLP screened for individual libraries. Overall, 85% of our partial sequences (700 to 800 bp) were easily aligned and taxonomically assigned to known groups (Fig. 2 and 3). One or two archaeal 16S ribosomal sequences were amplified from nearly all stations; BLAST matches were poor for these sequences, and they are not considered further in this analysis. Metazoans were recovered from three sites. The metazoans were diverse and included hydrozoans, a polychaete, and copepods (Table 2). Among the target protist sequences, we found novel clades and groups (Fig. 2) that were unreported or rare in other environmental surveys (21, 22, 35, 46).

FIG. 2.

Novelty or discovery histogram of sequences from this study, binned by 0.5% identity to sequences in GenBank (x axis). The y axis (counts) shows the number of sequences in each bin.

FIG. 3.

NJ tree of partial 18S rRNA gene sequences from all taxonomic clades reported here (boldface). Sequences with >99% similarity were pruned from the tree for clarity. Palmata palmata is the red algal outgroup. Choanoflagellates are shown separately, as these were difficult to align within the larger tree. For Fig. 3 to 6 the scale bars represent 0.1 nucleotide substitution per site; the actual value depends on the branch lengths in the tree.

TABLE 2.

Metazoan clones from this study

Group and clonea	BLAST score	Query/subjectb	Closest match (accession no.)c	Orderd
Copepoda
NOR26.06	1,269	682/688	Calanus pacificus (L81939)	Calanoida
NOR50.09	1,140	602/603	ENI47296.00268 (AY938369.1)
NOR50.12	1,061	604/611
NOR50.44	1,088	600/607
MD65.33	1,380	777/796	Neocalanus cristatus (AF514344.1)
NOR26.08	1,243	668/672	ENI40076.00641 (AY937790.1)	Cyclopoida
NOR50.21	1,261	669/672
NOR26.24	1,235	664/667	ENI40076.00836 (AY937933.1)
NOR50.39	1,245	669/672
NOR50.50	1,189	661/668
NOR50.61	1,241	663/666
MD65.36	1,269	781/816	SCM27C52 (AY62027.1)
NOR50.01	1,144	664/684	SCM28C3 (AY665130.1)
NOR50.23	1,239	708/733
NOR50.51	1,114	687/719
NOR50.34	1,126	711/751	SCM28C58 (AY665128.1)
Hydrozoa
NOR50.10	959	512/520	SCM37C13 (AY665135.1)	Trachylina
NOR50.29	1,465	742/743	Apolemia sp. (AY93733.1)	Siphonophora
NOR50.32	1,396	725/732	Hippopodius hippopus (AF358069.1)
Polychaeta
MD65.35	1,463	779/790	Scalibregma inflatum (AF448163.1)	Capitellida

^aClone designations are from Table 1.

^bRatio of sequence match length to length of the nearest match.

^cThe designations beginning with ENI are those of unpublished environmental sequences from the Western Atlantic Ocean, and ones beginning with SCM are those of unpublished environmental sequences from the Sargasso Sea.

^dTaxonomic order of the nearest nonenvironmental sequence in GenBank as of 26 December 2005.

Stramenopiles.

Out of 236 protist sequences, 45 were >98% similar to uncultivated marine stramenopiles (MAST) (Fig. 1; Table 3). Our most frequent MAST phylotypes were within MAST clusters 1, 3, and 7. Within the phototrophic stramenopiles were sequences closest to the dictyochophyte Pseudopedinella, a pelagophyte, and the colorless ochromonad Spumella (NOR50.37, 95%) (Fig. 4a). We also recovered diatoms and bolidophytes; our diatom sequences were mostly related to polarcentric and araphid species, including Fragilariopsis cylindrus and a Gonioceros sp. originally isolated from the Arctic (Fig. 4a to d). A cluster of four sequences (novel phototrophic stramenopiles) had BLAST scores that were 95% similar to those for bolidophytes, diatoms, and the environmental clone C2_018 from the Guaymas basin (22). Analysis of the nearly complete sequences (1,710 bp) placed this cluster as a sister to the bolidophytes (Fig. 4a and b).

TABLE 3.

MAST clones from this study

Clone	BLAST score	Query/ subjecta	Closest match (accession no.)	MAST cladeb
MD65.30	1,548	808/813	DH144-EKD10 (AF290063.1)	1a
NOR46.03	1,316	691/696
NOR46.04	1,358	729/728
NOR46.12	922	482/485
NOR46.13	1,358	712/717
NOR46.23	1,396	717/732
NOR46.32	1,443	751/756
NOR50.45	1,437	759/765
NW414.26	1,380	734/740
NW414.44	1,513	787/791
NW614.38	1,300	745/772
NW614.43	1,465	798/810
NW617.04	1,340	693/696
NW617.06	1,487	788/797
NW617.08	1,298	691/703	ME1-21 (AF363190.1)
NW617.01	1,275	656/659	RA001219.2 (AY295537.1)
NW414.07	1,566	837/846	RA000412.91 (AY205431.2)	1b
MD65.09	1,612	836/841	ANT12-11 (A363197.2)	1c
NOR26.02	1,364	710/716
NOR26.19	1,358	707/713
NOR46.17	1,483	763/768
NOR46.21	1,437	737/741
NOR46.25	1,465	752/755
NW414.05	1,495	790/798
NW414.08	1,568	821/827
MD65.08	1,530	791/796	BAQA232 (AF372760.1)
MD65.42	1,516	789/797	ME1-22 (AF363191.1)
MD65.13	1,439	735/738	DH148-5-EKD53 (AF290083.2)	2
NOR46.18	1,485	778/785
MD65.32	1,413	764/781	HE000427.21 (AY381157.1)	3
MD65.45	1,392	753/770
NW614.11	1,179	699/731	UEPACRp5 (AY129069.1)
NW414.28	1,413	753/761	ANT12-10 (AF363196.2)	7
NW414.39	1,637	838/842
NW614.54	1,560	799/803
NW617.17	1,455	753/758
MD65.10	1,505	837/851	BL010320.6 (AY381207.1)
MD65.29	1,592	806/807
NOR26.17	1,411	715/716
NOR26.40	1,495	767/770
NOR46.19	1,493	753/753

^aRatio of sequence match length to length of the nearest match.

^bMAST clade classifications are from reference 40.

FIG. 4.

Classification of photosynthetic stramenopiles (PS) from this study (boldface). Panel a shows the overall ML tree structure from 1,710-bp alignments of the 18S rRNA gene, showing the position of a novel PS cluster (represented by clone NW414.14) within the photosynthetic heterokonts. Panels b to d show the remainder of all PS environmental sequences from this study (boldface) over 750 bp. (b) The novel phototrophic stramenopile cluster and bolidophytes from this study; (c) centric diatoms; (d) mostly araphid diatoms. ML trees are shown; NJ and Bayesian trees had similar topologies. Bayesian clade support was 100% at major branches.

Alveolates.

Nearly 40% of our phylotypes were alveolates. ML and NJ methods clearly separated all but one of these, NOR50.43 (<94% similarity with dinoflagellates), into four major groups: dinoflagellates, novel alveolate groups I and II (35), and ciliates (Fig. 1 and 5a to c). Nearly all of the dinoflagellate sequences were 97 to 99% similar to known dinoflagellates, with the exception of one cluster from GNB-M09 that was 96% similar to Cochlodinium polykrikoides (Fig. 5c). The 18S rRNA gene is generally poor at resolving phylogenetic relationships within the dinoflagellates, and many morphologically distinct genera are often >98% identical (63). Our ML analysis was still useful for defining the closest relatives of our environmental sequences and their relationships to each other. For example, one Beaufort Sea sequence was identical over 750 bp to the common polar dinoflagellate Heterocapsa rotundata (37) (Fig. 5a).

FIG.5.

ML analysis of 750-bp fragments of our alveolate sequences (boldface) and closest GenBank matches. The three trees were rooted using alveolates from outside the group treated (the outgroup has been removed from the tree for clarity). (a) Dinoflagellates; (b) group I and group II alveolates; (c) ciliates (Strom A and B are two distinct Strombidium clusters). NJ and Bayesian trees were topologically similar, and nodes were supported by 97 to 100% Bayesian posterior probabilities (not shown). Distance measures are for ML.

Among alveolate groups I and II, we found several phylotypes from both sides of the Arctic that were 99% similar to deep Guaymas Ocean Basin sequences (22) (Fig. 5b). Alveolate group II was diverse, with several sequences from the AO and Beaufort Sea stations being closest to Amoebophrya spp. and the remainder being most similar to other environmental sequences (Fig. 5b). Our ciliate phylotypes were mostly distantly related to known species, with one exception: several GNB sequences were >99% similar to the tintinnid genus Tintinnopsis (one is shown in Fig. 5c). The closest BLAST matches for a majority of ciliate sequences were to one Strombidium strain (SNB99-2), and a closer analysis showed that these made up two separate lineages (Strom A and B in Fig. 5c). Most of the other ciliates were Choreotricha related to Strobolids and Tintinnidae. Two sequences fell on a long branch within the Nassophorea Furgasonia blochmanni (NOR50.36) and Cryptocaryon (MD65.14), a fish pathogen.

Other algae.

We recovered sequences from four other algal classes. Cryptophytes were recovered in the Beaufort Sea, a haptophyte was recovered from the GNB, and a novel algal class (F. Not, K. Valentin, K. Romari, C. Lovejoy, and R. Massana, unpublished data) was discovered on both sides of the Arctic (Fig. 1). Prasinophyte sequences were present in all but one of the libraries and were 98 to 99% similar to three genera: Bathycoccus, Micromonas, and Mantoniella (Fig. 1). These were all most similar to cultured isolates (C. Lovejoy, unpublished data).

Other heterotrophic flagellates.

We recovered three choanoflagellate phylotypes, two from the Beaufort Sea and the other from the GNB (Fig. 3). Another six sequences were closest to an environmental sequence from Blanes Bay (Spain), BL010625.25 (97 to 98% BLAST matches). This sequence has been linked to the predatory flagellate Telonema (39).

Rhizaria.

Cercozoans and radiolarians (51) were well represented in our libraries (Fig. 6). These generally had poor BLAST matches to known or environmental sequences, except for three sequences from GNB at 5 m and one AO site, which were >98% similar to the algal predator Cryothecomonas. Two other AO sequences were more distantly related to Cryothecomonas, forming a separate cluster (Fig. 6a). Finally, several GNB phylotypes formed an additional distinct cluster (Fig. 6a) with closest BLAST matches (ca. 94%) to chlorarachniophytes, Cercozoa with green algal endosymbionts. These environmental sequences were 99% similar to each other over 1,780 bp but were <86% similar to any complete sequence in GenBank.

FIG. 6.

(a) ML tree derived from 1,750-bp sequences indicating the overall position of the rhizarian sequences from this study (boldface), with a choanoflagellate root. Topology was essentially the same for NJ and Bayesian trees, with 100% clade support at all main branches. (b to d) Expansions to include our partial environmental sequences (750 bp) from the three radiolarian groups: Acantharea (b), Spumellarida (c), and Taxopodida/Spongodiscidae (d) (see text). Distance measures are for ML.

Among the Radiolaria, we recovered 19 sequences that were most similar to a few environmental sequences reported elsewhere and <92% similar to identified organisms. The Radiolaria split into two main groups. One branch consisted of the Spumellarida (classed within the Polycystinea). The second branch split further into two major groups: Acantharea and Spongodiscidae (also currently in the Polycystinea), united with the Taxopodida (Fig. 6b to d).

DISCUSSION

Diversity.

Charting the true dimensions of eukaryotic diversity is essential to fully understand evolution and, by extension, the ecological complexity of microbial food webs. Molecular surveys provide a primary route towards this understanding, and each new environment studied has yielded new insights into particular aspects of eukaryotic diversity and evolution (5). To date these studies have revealed new lineages and unexpected diversity within previously known lineages in open oceans (21, 35, 46), coastal areas (39, 54), anaerobic sediments (19), acid rivers (71), and deep sea vents (22, 33). The Arctic proved to be a rich source of novel sequences, and this study extends the geographical record of recently discovered lineages known only from environmental sequences.

Among the stramenopiles, the majority of our heterotrophic lineages belonged to MAST clusters 1, 3, and 7, which have previously been reported from open pelagic systems, and Massana et al. (40) argued that these are planktonic and cosmopolitan and graze on bacteria. The phototrophic stramenopiles from the AO were mostly araphid diatoms (Fig. 4d), while centric diatoms and bolidophytes were recovered from the GNB (Fig. 4b and c). The difference is likely due to the histories of the water masses. The GNB cuts across southward-flowing Arctic water and northward-flowing Atlantic water, which is relatively low in silicic acid required for diatom growth (52). In contrast, Pacific water, which is the source of the upper mixed layer of the Western Arctic, is high in silicic acid (43, 61). Even on small scales, water masses can have an influence on community structure (36, 37).

We recovered one novel cluster from the AO and Beaufort Sea samples that was a sister to the bolidophytes (Fig. 4a and b). This cluster may be a candidate for the bipolar order Parmales, which has distinct siliceous plates. These organisms have never been brought into culture or sequenced and are uniquely described from environmental electron microscopy studies (11).

The members of alveolate group II were diverse, with several sequences from the AO and Beaufort Sea stations being closest to Amoebophrya spp. and the remainder being most similar to other environmental sequences (Fig. 5b). Amoebophrya is an alveolate that is parasitic on dinoflagellates and currently classed within the dinophycean order Syndiniales; the type taxon Syndinium turbo is a zooplankton parasite (63). It seems likely that all group II alveolates are parasitic with picoplanktonic life stages (69). Recently Ellobiopsids, which are also parasites on zooplankton, have been found to be phylogentically affiliated with group I alveolates (58), suggesting that this group may also be parasitic. The ubiquitous distribution of group I and II alveolates in the sea suggests that these organisms are a fundamental component of marine microbial ecosystems and that the ecological impact of parasitism in open marine waters is underestimated.

Rhizaria.

Nikolaev et al. (51) have suggested a monophyletic origin of bikont amoeboid eukaryotes. This supergroup (18) includes marine Cercozoa, Foraminifera, and Haeckel's Radiolaria (26). Among the GNB cercozoa, we found sequences that were 98 to 99% similar to those of the algal predator Cryothecomonas and another more distant clade. This genus was originally described from sea ice (65) and is easily identified microscopically and commonly recorded elsewhere, including the Canadian Arctic (28, 37). Cryothecomonas was not found among the AO sequences. Other GNB phylotypes formed an additional distinct cluster on the same branch as the chlorarachniophytes. Chlorarachniophytes are a primary endosymbiotic group containing chlorophyll b.

The AO libraries were particularly rich in radiolarians. The phylogenetic position of Radiolaria proposed by Haeckel, especially the skeletal Polycystinea and Acantharea, has generated considerable debate, with a few environmental sequences provoking recent phylogenetic reassessments (34, 70). Nikolaev et al. (51) describe three major lineages: Acantharea, Polycystinea, and Taxopodida. The Polycystinea sequences used in that analysis all belonged to the Spumellarida, and the order Taxopodida was suggested on the strength of one freshwater protist, Sticholonche ankle, and two marine environmental sequences, DH145-KW16 and CS_E043. The environmental sequence AT4-94 from mid-Atlantic ridge sediment (33) was outside the Taxopodida. Subsequently, Takahashi et al. (62) found that polycystinean Spongodiscidae grouped with DH145-KW16 and were more closely related to acantharians than to the colonial and nonskeletonal Polycystinea in the Spumellarida; they did not include the Sticholonche sequence in their analysis. Addition of our sequences resulted in tree topologies that place DH145-KW16 and AT4-94 into a monophyletic clade within Taxopodida that includes solitary shell-bearing Spongodiscidae (Fig. 6a), confirming that Polycystinea are paraphyletic and in need of taxonomic revision (62).

Several of our sequences were closest to the Antarctic DH145-HA2, and our ML analysis suggests that, rather than being an independent lineage, these are at the base of the Spumellarida (Fig. 6c). Another five of our sequences were acantharians (Fig. 6b); these were 98% similar to the hydrothermal vent sequence C3_E029 taken from sediment cores in the deep Guaymas Basin (22), suggesting wide adaptation or long-distance transport of these organisms. The abundance of skeleton-bearing radiolarians in the Canada Basin suggests they are present and active in surface cold arctic waters. Paleoceanographic studies have previously documented the widespread distribution of silica spicules, and tests of the Polycystinea in arctic sediments (3, 10) and knowledge of the life stages of this group will be valuable as a tool to compare sediment records with current conditions.

Size fractionation.

Despite 3-μm prefiltration, we recovered 18S rRNA gene sequences from larger organisms, notably dinoflagellates and ciliates (Fig. 3) and metazoa (Table 2). This phenomenon has been reported elsewhere (39, 54) and may be the result of flexible cells that can be forced through the 3-μm filter pores, cell breakage during sample collection, or sloppy feeding by zooplankton. The diversity of metazoan sequences (Table 2) suggests retention of either dissolved free DNA adhering to small particles or DNA-containing particles (13, 20), rather than contamination by one errant zooplankton. DNA readily binds to silica particles (49), and broken diatom frustules may provide such a source of silica. Dinoflagellates and ciliates are also usually >3 μm; however, those without rigid cell walls may be able to deform sufficiently to pass through the filter. Ciliates, dinoflagellates, and other naked protists also produce small, slow-sinking minipellets (14, 59) that may retain DNA from either the predator or its prey. These alveolates and other phagotrophic protists would have been the main grazers at this time of year and actively graze on each other (32). The retention of small particulates in the upper water column may explain the presence of the 18S rRNA genes from these organisms. At least some of the diversity reported from environmental surveys of unseen picoeukaryotes may be an artifact of DNA preservation in cold saline-buffered waters, in addition to incomplete sequence data for described organisms (9, 56). Future research comparing the diversity of the larger size fraction and the application of specific probes combined with microscopy (for example, fluorescence in situ hybridization) may help to resolve the origin of seemingly large-celled organisms in the smallest size fractions.

Our prefiltration technique yielded many sequences from picoplanktonic organisms, for example, marine stramenopiles (39), the prasinophyte Bathycoccus, and species with picosize life stages, such as the parasitic Amoebophrya (55). The radiolarian sequences may have come from either small zoospore stages or preserved DNA. Molecular surveys might conservatively be thought of as evidence of the phylotype's presence in the recent past (a footprint) in combination with community diversity at the moment of collection.

Diversity.

In contrast to other studies (35, 46) and all of our other libraries, one library (NW415) was surprisingly lacking in diversity. Except for one ciliate and one MAST, all sequences recovered were either diatoms or Taxopodida (Fig. 3 and 6 and Table 3). This region of the Arctic is historically covered in thick multiyear ice, but in 2002 warm conditions caused a retraction of the ice cap over the Western Arctic (42), exposing these waters to high surface irradiance for the first time. The low levels of nutrients, especially nitrate, which is considered the limiting nutrient over much of the Arctic (15, 16, 30, 38), suggest that the strong halocline suppressed an upward flux of nutrients needed to support microbial growth, resulting in low biomass in the gyre system. This low-diversity biological community was a rare marine example of colonizer species, equivalent to primary succession on land following glacial retreat. With the ongoing effects of climate change in this region, such conditions and depauperate microbial assemblages may be increasingly common. The newly open waters would not be a substitute for the lost productive ice edge habitat over shallow shelves, which currently supports marine mammals and birds over most of the polar regions (16).

Conclusion.

Some sequences from both sides of the Arctic were >99% similar. Among heterotrophic protists, several sequences were closely related to environmental clones from the deep ocean, which is perennially cold. Phylotypes with <99% similarity to other sequences could be unique species, ecotypes adapted to cold waters, or broadly temperature-tolerant cosmopolitan species. In total, 42% of our sequences were <98% (a standard microbial benchmark of genus-level diversity) similar to publicly available sequences. Overall, we report new representatives from five of eight major marine eukaryotic lineages (5). The remote AO Canada Basin proved to be a rich source of evolutionarily informative sequences, and the importance of radiolarians in these waters was previously unknown. The stable cold temperature of these waters and nutrient supply rates are likely to be the main factors selecting for community species composition (38). Global predictions are that the Arctic could warm as much as 10°C within several decades (45). Higher temperatures, increased water column mixing due to loss of ice cover, and changing current patterns (41) mean that uniquely polar phylotypes are a vulnerable component of global genetic diversity.