Correspondence: Angelina Lo Giudice, Dipartimento di Biologia Animale ed Ecologia Marina (DBAEM), Università di Messina, Viale Ferdinando Stagno d'Alcontrès, 98166 Messina, Italy. Tel.: +39 090 6765533; fax: +39 090 393409; e-mail: firstname.lastname@example.org
A polyphasic approach that included PCR-dependent and PCR-independent molecular techniques was applied to analyze the prokaryotic community in surface waters of shallow Antarctic lakes. The in situ abundance of different bacterial groups was determined by the fluorescence in situ hybridization, whereas bacterial diversity was investigated by 16S rRNA gene sequencing of bacterial clones and isolates. The different approaches allowed the identification of the significant microbial components of the lake bacterioplanktonic communities, indicating a predominance of Flavobacterium, Pseudomonas, and Polaromonas (up to about 56% of total sequences). These genera also appear to be important in freshwater systems elsewhere in the world. Interestingly, closest blast matches to our sequences were predominantly from polar lakes and ponds, in addition to streams and glaciers, suggesting a bipolar distribution of freshwater lake bacterioplankton. Bacteria that are more traditionally associated with the marine environment were also detected, thus indicating an external input by atmospheric deposition and/or seabird excreta. Finally, a slightly different microbial community occurred in the lake at Inexpressible Island that was characterized by low N : P ratio and very high conductivity value, reinforcing the idea that physicochemical and trophic status may affect the structure and composition of the bacterioplankton assemblages in Antarctic lakes.
In Antarctica, the seasonally or permanently ice-free regions, which represent only the 2% of the total continent surface, constitute the largest cold desert on Earth and remain largely unexplored (Bargagli, 2005). In particular, Antarctic lakes include freshwater and saline systems that have experienced little or no anthropogenic impact and, therefore, harbor pristine biotopes. Such systems can be considered sensitive indicators of environmental changes because snow and ice cover markedly affect all ecological variables (Quayle et al., 2002). Owing to a unique combination of extreme environmental stresses (i.e. low temperature, low photosynthetically active radiation, nutrient limitation, period of ice cover, limited availability of liquid water, and short growing season), Antarctic lakes support simple truncated food webs with no fish and a limited presence of grazing and burrowing organisms. These environments are generally dominated by organisms of the microbial loop, including bacteria, protozoa, and phytoplankton (Laybourn-Parry, 2002, 2009). Nevertheless, in the case of the prokaryotic communities that inhabit Antarctic lakes, information is restricted to a few regions. With respect to the application of a culture-independent approach, there is a substantial literature on maritime Antarctica (e.g. Pearce, 2003, 2005; Pearce et al., 2003, 2005, 2007; Schiaffino et al., 2009; Villaescusa et al., 2010), in addition to the continental Dry Valleys and Vestfold Hills lakes (e.g. Bowman et al., 2000; Gordon et al., 2000; Van Trappen et al., 2002; Izaguirre et al., 2003; Laybourn-Parry et al., 2004; Mosier et al., 2007). Conversely, only limited attention has been generally devoted to the culturable fraction of the microbial community (Van Trappen et al., 2002; Peeters et al., 2011), and studies have mainly attempted to isolate psychrophilic organisms of major importance for biotechnology, that is, production of cold-adapted enzymes, presence of antifreeze proteins, and tolerance to salts and desiccation (Van Trappen et al., 2002; Stingl et al., 2008). This is because such microorganisms are subjected to a high selection pressure and very likely possess novel biochemical adaptations. Interestingly, a number of new Antarctic lake bacterial species have been described in the last decade (Reddy et al., 2003; Sheridan et al., 2003; Van Trappen et al., 2003, 2004a, b, c, 2005 ; Jung et al., 2004; Chen et al., 2005).
To the best of our knowledge, the analysis of the whole microbial community in the northern Victoria Land (Ross Sea Region, East Antarctica) has been restricted exclusively to soil systems (e.g. Barrett et al., 2006; Niederberger et al., 2008; Aislabie et al., 2009). Moreover, only a single report exists about the taxonomic description of a bacterium that was isolated from a saline lake in such region (Poli et al., 2007). As a result of the paucity of knowledge about the microbiology of lakes in the northern Victoria Land, this study was aimed at describing the microbial communities inhabiting three shallow lakes (max depth, 4 m), located at the Crater Cirque, Inexpressible Island, and Luther Vale areas (Fig. 1), which possess unique characteristics as regards origin and the nature of the surrounding areas. FISH was applied to achieve a general overview on the bacterial community composition, and bacterial diversity was more deeply analyzed by sequencing the 16S rRNA genes of bacterial clones and isolates.
Results from this study help further the understanding of the prokaryotic community in lakes of continental Antarctica, providing an insight into the ecologically significant microbial components within this extreme ecosystem.
Materials and methods
The water bodies studied are located in the northern Victoria Land (East Antarctica).
The lake at Crater Cirque (CC; coordinates: 72°36′ S–169°21′ E; distance from the coast: 48 km; a.s.l.: 85 m; estimated surface: 15 300 m2), in the vicinity of Cape Hallett where dry cirques and saddles occur, consists of a cirque on the south wall of Tucker Glacier, immediately west of its junction with Whitehall Glacier. The lake is colonized by red and green algae, and in the surrounding granitic rock walls, there are nests of Wilson's petrels (Oceanites oceanicus), Antarctic skuas (Catharacta maccormicki), and snow petrels (Pagodroma nivea), as well as running streams and relatively lush growths of mosses and lichens.
The Luther Vale occupies approximately 100 ha at 200 m a.s.l. and is underlain by metamorphic rock. The Luther Lake (LH; coordinates: 72°37′ S–169°89′ E; distance from the coast: 8 km) is a large ice-covered pond located in the center of the Luther Vale cirque, along with several smaller ponds that are fed by snow-pack water run-off from snowfields on the north-western side of the Luther Peak. Such higher elevation area, standing about 18 km southeast of Mount Peacock in the Admiralty Mountains and overlooking Edisto Inlet in northern Victoria Land, does not appear to receive significant marine inputs, indicating a terrestrial source of organic matter, that is, algae, mosses, and lichens. The Luther Lake was partially covered by ice at sampling time.
The lake at Inexpressible Island (INI; coordinates: 74°53′ S–163°43′ E; distance from the coast: 0.01 km; a.s.l.: 1 m; estimated surface: 26 700 m2) is the closest to the sea among the investigated basins. The island lies close south of the Northern Foothills at the outer edge of the Nansen Ice Sheet. The Inexpressible Island area includes 14 lakes that are located in sufficiently close proximity to one another to present similar physico-chemical characteristics. This area is almost completely free from glaciers and culminates at 390 m. It is characterized by ice-covered moraines and disintegration moraines, with isolated conical relief and circular depressions accommodating shallow lakes. The lake is located near penguin (Pygoscelis adeliae) rookeries or nesting south polar skuas (C. maccormicki).
Hereinafter, the investigated lakes will be indicated by the site acronym, that is, CC, INI, and LH for the lake at Crater Cirque, Inexpressible Island, and Luther Vale, respectively.
Surface water samples (first 10 cm; 10 L) were collected from the lakes at CC, LH, and INI in January and February 2005. Sampling was performed manually using acid-washed polycarbonate tanks. The bottles were rinsed three times with the sample prior to filling. Conductivity, pH, and temperature were measured on site using portable meters.
For the estimation of total picoplankton (TPP), samples were immediately fixed with formaldehyde (final concentration, 2%), filtered on black polycarbonate membranes (pore size, 0.22 μm; diameter, 25 mm) and kept frozen until processing. In particular for the TPP estimation, cells were stained with DAPI (4′,6-diamidino-2 phenylindole) at a final concentration of 1 μg mL−1 (Porter & Feig, 1980). More than 300 cells per sample were counted in randomly selected eye fields. Counts were determined by epifluorescence microscopy as previously described (La Ferla et al., 2004).
Samples for cultivable heterotrophic bacteria (VC) were processed within approximately 2 h after sampling. Serial dilutions were prepared (1 : 10 and 1 : 100, using filter-sterilized freshwater), and 100 μL of each dilution was spread plated in three replicates on R2A (Difco). Colony-forming units (CFU mL−1) were counted after incubation at 4 °C in the dark after 1 month.
Community composition analysis
Quantitative analysis of community composition by FISH
The abundance of different phylogenetic groups was determined by FISH as described by Glöckner et al. (1999). Cells were concentrated from water samples on white polycarbonate filters (diameter, 25 mm; pore size, 0.22 μm) and subsequently fixed for 30 min at room temperature by overlaying filters with a freshly prepared paraformaldehyde (final concentration, 4%) phosphate-buffered saline (PBS; 130 mM NaCl, 10 mM Na2HPO4, and 10 mM NaH2PO4, pH 7.4) solution (3 mL). The fixative was removed by applying vacuum, and filters were washed twice with 3 mL of PBS and, finally, with distilled water. Filters were air-dried and stored in a sterile Petri dish at −20 °C until processing.
Filter sections were hybridized with the CY3-labeled oligonucleotide probes listed in Table 1 (MWG-Biotech, Germany) and mainly targeting phylogenetic groups of the domain Bacteria. A negative control probe, nonEUB (5′-ACTCCTACGGGAGGCAGC-3′), was also used for nonspecific probe binding. Results with this negative control probe, which accounts for autofluorescence of cells and nonspecific probe binding, were subtracted from the percentages detected with probes for the bacterial groups. Filters were mounted on glass slides with Citifluor AF1 (Citifluor Ltd., Canterbury, UK) and enumerated by epifluorescence microscope (Axioplan, Zeiss) equipped with specific filter sets for CY3. For each sample and probe, minimum 20 fields and 200 cells were enumerated.
Table 1. Sequences of oligonucleotide probes used for FISH
Probes EUB338I, EUB338II, and EUB338III were equimolarly mixed together to obtain the EUB-mix; the probes LGC354a, LGC354b, and LGC354c were equimolarly mixed together to obtain the LGC-mix; Probes DELTA495a, DELTA495b, and DELTA495c were equimolarly mixed together to obtain the DELTA-mix; Probes CYA664 and CYA762 were equimolarly mixed together to obtain the CYA-mix.
Values represent percent of formamide in the hybridization buffer.
For genomic DNA extraction and cloning, subsamples (5 L) were filtered on sterile 47 mm diameter, 0.22 μm pore size membranes (Millipore), which were used for subsequent analyses. Subsamples were subsequently stored at −20 °C until processing. Genomic DNA was extracted as previously described (Michaud et al., 2009). Briefly, membranes were pretreated with 150 μL of a 5 mg mL−1 lysozyme solution for 10 min. Minced filters were then placed in a sterile 2-mL Eppendorf tube and subjected to DNA extraction using the RNA/DNA Extraction Kit (Qiagen) following the manufacturer's instructions. Finally, DNA was precipitated by adding 0.7 volumes of 100% isopropanol followed by a wash with ice-cold 70% ethanol and after air-drying resuspended in 50 μL of deionized sterile water. The quantity and quality of the DNA were checked by agarose gel electrophoresis (1%, w/v) in TAE buffer (0.04 M Tris–acetate, 0.02 M acetic acid, 0.001 M EDTA), containing 1 μg mL−1 of ethidium bromide.
Extracted DNA was used to construct clone libraries. The PCR amplification of 16S rRNA gene was performed with an ABI 9600 thermocycler (PE; Applied Biosystems) using the forward primer 8f (5′-AGAGTTTGATCCTGGCTCAG-3′) and the reverse primer 907r (5′-CCGTCAATTCCTTTRAGTTT-3′). The reaction mixtures were assembled at 0 °C and contained 1–10 ng DNA, 10× buffer, 1.5 mM MgCl2, 150 ng of each forward and reverse primers, 250 μM dNTP, 0.5 units of PolyTaq polymerase (Polymed), and sterile distilled water to a final volume of 20 μL. The PCR program was as follows: 3 min at 95 °C followed by 30 cycles for 1 min at 94 °C, 1 min at 50 °C, 2 min at 72 °C, and a final extension step of 10 min at 72 °C.
The 16S rRNA gene fragments were cloned into the pGEM Easy Vector System (Promega) according to the manufacturer's instructions. The resulting ligation products were used to transform Escherichia coli ElectroMAX DH10B cells (Invitrogen). Between 100 and 200 inserts were subsequently PCR amplified from lysed white colonies using vector-specific primers, M13f (5′-GTAAAACGACGGCCAGT-3′ and M13r (5′-CAGGAAACAGCTATGACC-3′), under the same previously described PCR conditions. The results of all the amplification reactions were analyzed by agarose gel electrophoresis as described above. The amplified 16S rRNA gene fragments were directly purified from PCR reaction mixture using the QIAquick PCR Purification Kit (Qiagen) according to the supplier's instructions.
Cultivable heterotrophic bacteria
Colonies were randomly picked from agar plates used for viable counts. To ensure the purity, each colony was subcultured at least three times under the same conditions. A single colony of each strain was lysed by heating at 95 °C for 10 min. PCR amplification of 16S rRNA gene was carried as described above for clones with the forward primer 27f (5′-AGAGTTTGATCCTGGCTCAG-3′) and reverse primer 1492r (5′-CTACGGCTACCTTGTTACGA-3′).
All isolates are part of the Italian Collection of Antarctic Bacteria (CIBAN) of the National Antarctic Museum (MNA, www.mna.it) ‘Felice Ippolito’ kept in our laboratory at the University of Messina. They are currently maintained on R2A slopes at 4 °C and routinely streaked on agar plates from tubes every 6 months to control purity and viability. Antarctic strains are also preserved by freezing cell suspensions at −80 °C in R2A Broth (which have the same composition of Difco R2A minus agar) to which 20% (v/v) glycerol is added.
Amplified rDNA restriction analysis (ARDRA) of isolate and clones
ARDRA was carried out as previously reported (Michaud et al., 2004; Lo Giudice et al., 2010). Briefly, 5 μL of each PCR mixture, containing approximately 1.5 μg of amplified 16S rRNA gene, were digested with 3 units of the restriction enzyme AluI (Fermentas) in a total volume of 20 μL at 37 °C for 3 h. The enzyme was inactivated by heating at 65 °C for 15 min, and the reaction products were analyzed by agarose (2.5%, w/v) gel electrophoresis (at 90 mV for 90 min) in TAE buffer containing 1 μg mL−1 of ethidium bromide. A GeneRuler™ 100-bp DNA Ladder (Fermentas) was applied to each gel as a band reference.
As only the presence/absence of each band was considered, a matrix was generated by the mere visual comparison of the restriction patterns obtained by ARDRA. Antarctic isolates and clones were grouped into operational taxonomic units (OTUs), assuming that a single OTU was made up of bacteria belonging to the same species, which showed the identical AluI restriction pattern. For isolates, colony morphology of strains, which showed identical ARDRA patterns, were compared to check for eventual errors.
One to three representative strains/clones (where possible), showing identical ARDRA pattern, were randomly selected for 16S rRNA gene sequencing and analysis. All singletons (strains/clones from OTUs made of a single member) were sequenced. Clones and isolates, which were named with the suffix ALC (Antarctic lake clone) and ALI (Antarctic lake isolate) followed by the OTU number, were further processed by 16S rRNA gene sequencing and analysis.
Phylogenetic affiliation of isolates and clones
Automated sequencing of 16S rRNA gene from clones or isolates was carried out by cycle sequencing using the dye-terminator method. Sequencing was carried out at the Sequencing Service of the Macrogen Laboratory (Korea). The closest relatives of clones/isolates were determined by comparison with 16S rRNA gene sequences in the NCBI GenBank and the EMBL databases using blast, and the ‘seqmatch’ and ‘classifier’ programs of the Ribosomal Database Project II (http://rdp.cme.msu.edu/). All sequences with similarity ≥ 97% were considered to represent one phylogenetic group or phylotype. Sequences were further aligned using the program clustal w (Thompson et al., 1994) to the most similar orthologous sequences retrieved from database. Each alignment was checked manually, corrected, and then analyzed using the neighbor-joining method (Saitou & Nei, 1987) according to the model of Jukes–Cantor distances. Phylogenetic tree was constructed using the mega 5 (Molecular Evolutionary Genetics Analysis) software (Kumar et al., 1993). The robustness of the inferred trees was evaluated by 500 bootstrap resamplings.
Nucleotide sequence accession numbers
Nucleotide sequences have been deposited in the GenBank database under the accession nos. JQ229546–JQ229623 (see Supporting Information, Table S1 for details).
Data analyses and diversity indices
Bray—Curtis similarity was computed among samples, based on the clones, isolates, and FISH results, expressed as relative phyla percentages. A subsequent cluster analysis (Group average) was performed. The relative distribution of OTUs in each sample (merging clone and isolate sequence) was used to calculate coverage values (Good, 1953) and the nonparametric Chao1 estimator (Chao, 1984), using the freely downloadable software spade (Chao & Shen, 2003). The Chao1 nonparametric estimator estimates of the probable total number of phylotypes present in the sample (Lee & Chao, 1994).
Moreover, more conventional Margalef richness index (d), the ln basis Shannon index (H′), the Pielou evenness index (J′), and the Simpson reciprocal index (1/D) were calculated. The Shannon index is a general diversity index that is positively correlated with species richness and evenness, and it is more sensitive to change in abundance of rare species. Margalef richness index standardizes the number of species encountered against the total number of individuals encountered. Finally, the evenness (Pielou index, J), reflects the relative importance of each taxon within the entire assemblage, while the Simpson reciprocal index is a measure of dominance and represents the chance that two randomly sampled individual will belong to the same species, and it emphasizes common species. All calculations (with the exception of Chao-1) were carried out using primer 6 software, version 6β R6 (Copyright 2004; PRIMER-E Ltd).
At sampling time, the water temperature was 0.9, 1.8, and 4.1 °C in LH, CC, and INI, respectively. The pH values measured in all lakes were generally alkaline reaching the highest value (9.7) in LH, followed by INI and CC (8.2 and 7.4, respectively). The conductivity was particularly high in INI (1995 μS cm−1) and low in CC (13 μS cm−1), whereas a value of 321 μS cm−1 was measured in LH.
Differences in the mean microbial abundances were determined among the three investigated lakes. Viable counts on R2A plates were highest in INI (3.9 ± 0.3 × 103 CFU mL−1) and similar for CC and LH (2.4 ± 0.2 and 2.3 ± 0.4 × 103 CFU mL−1, respectively). Conversely, the highest total count value (4.5 ± 1.6 × 105 cells mL−1) after DAPI-staining was determined in samples from LH, followed by CC and INI (3.9 ± 1.1 and 2.8 ± 0.9 × 105 cells mL−1, respectively). Heterotrophic bacteria on R2A plates represented 0.5–1.4% of total counts.
The relative abundance of bacterial groups was examined by performing FISH reactions with bacteria-specific probes (EUB338 mix), an Archaea-specific probe (ARCH915) and 12 different bacterial group-specific probes (listed in Table 1). The hybridization levels (percentage of DAPI-stained bacteria – that is, Eubacteria plus Archaea – that were visualized by FISH) were 70.0, 69.3, and 62.1% in samples from INI, CC, and LH, respectively. With the set of probes used to detect the major divisions within the domain Bacteria, it was possible to affiliate between 52.9% and 60.8% of the EUB338 hybridized cells with known bacterial groups. Between 8.1% and 8.5% of the EUB338 counts remained unaffiliated. The percentage of bacteria detected with a negative control probe varied from 0% to 2% of the DAPI-stained bacteria.
Results from the FISH analysis for each lake are shown in Fig. 2. When considering probes that target heterotrophic bacteria, the Firmicutes were the least represented group (between 0.5% and 2.6% of DAPI-stained cells) in all the lakes. The Cytophaga-Flavobacterium (CF) group of the Bacteroidetes (25.1%) clearly dominated over the Proteobacteria (12.4%) and Actinobacteria (9.4%) in INI. Conversely, the Proteobacteria (18.6%) were more abundant in CC, followed by the Actinobacteria and CF group of the Bacteroidetes (15.6% and 9.7%, respectively). These three main phyla were equally dominant in LH (between 10.8% and 12.2% of DAPI-stained cells).
In particular, among the Proteobacteria, the Beta- (5.4%) and Alphaproteobacteria (4.8%) were more abundant than the Delta-, Gamma-, and Epsilonproteobacteria (between 1.2% and 3.8%) in CC. Conversely, the Delta- and Betaproteobacteria appeared to occur in higher numbers in INI and LH.
Picocyanobacteria were abundant, comprising 10.2–16.3% of the lake microbial populations. Finally, the Archaea that were targeted by the probe ARCH915 represented 0.7–1.1% of the DAPI-stained cells.
Genetic fingerprinting and phylogenetic affiliation of clones and isolates
The composition of the microbial communities was investigated by the analysis of 16S rRNA gene sequences of clones and isolates. Overall, a total of 501 clones (153 from CC, 174 from INI, and 174 from LH) and 401 isolates (166 from CC, 127 from INI, and 108 from LH, respectively) were initially screened by amplified 16S rDNA restriction analysis (ARDRA) patterns generated using AluI. Representative clones and isolates were named with the suffix ALC and ALI, respectively, followed by the OTU number. All sequences with similarity ≥ 97% were considered to represent one OTU.
Results in Table S1 and Fig. 3 are based on the blast and Seqmatch/classifier analyses of the sequences and also on the phylogenetic affiliation of clones/isolates as depicted in Fig. 4a and b. Overall, the phylogenetic analysis of sequences from clones/isolates revealed the distribution of Antarctic lake bacteria within 78 separated OTUs (listed in Table S1) that fell into well-defined phyla that compose the bacterial lineage as follows: CF group of Bacteroidetes (34.2%), Beta- (24.9%) and Gammaproteobacteria (19.2%), Cyanobacteria (9.0%), Actinobacteria (8.5%), Alpha- (3.3%), Gemmatimonadetes (0.4%), and Planctomycetes (0.2%). Only the 0.3% of sequences shared the highest degree of sequence identity with unclassified bacteria.
Heterotrophic bacteria found in the 16S rRNA gene clone libraries were seldom obtained in culture, even if they were not always retrieved from the same lake. In this regard, OTU-sharing among lakes was seldom observed (Table S1). Both for clones and isolates, the lakes CC and INI shared exclusively those sequences (one and three, respectively) that were common to all lakes, as it is specified below. Regarding the clones, a single OTU that was affiliated to Polaromonas sp. ALC-CC-C1 was common to all lakes. In particular, the lake LH shared two and five OTUs with INI and CC, respectively. Among isolates, three OTUs were present in all lakes, that is, Pseudomonas spp. ALI-INI4 and ALI-CC18, and Psychrobacter sp. ALC-LH-B7. The lake LH shared two and eight OTUs with INI and CC, respectively.
As it is shown in Fig. 3, sequences affiliated to the Alpha- Beta-, Gammaproteobacteria, CF group of Bacteroidetes, and Cyanobacteria were detected in all lakes within clones and/or isolates, although to different extents. Conversely, the Actinobacteria (both clones and isolates) occurred only in CC and LH, whereas Planctomycetes and Gemmatimonadetes were obtained only from the LH clone library. In particular, affiliates to the genera Polaromonas (among the Betaproteobacteria), Pseudomonas (among the Gammaproteobacteria), and Flavobacterium (among the CF group of Bacteroidetes) appeared to be predominant in the three investigated Antarctic lakes.
On the basis of the diversity indices, which were computed on the merged clone and isolate datasets, LH appeared to be the most diverse and rich in species (Shannon 3.14 and Margalef 7.62, respectively) among the investigated lakes (Table 2). The remaining two lakes were comparable for all the computed indices although INI had significant higher species richness. Moreover, the INI lake presented the highest Simpson reciprocal index (3.78), thus suggesting that such lake is dominated by some species mostly belonging to the Pseudomonas and Flavobacterium genera. Finally, the INI lake was characterized by the highest Chao-1 index value (107.3).
Table 2. Diversity indices
N, no. of screened clones/isolates; S, OTU richness; d, Margalef index (richness); J′, Pielou index; H′, Shannon index; 1/D, Simpson reciprocal index.
This study represents the first assessment of the prokaryotic community structure and composition in the surface waters of shallow Antarctic lakes of the northern Victoria Land (East Antarctica). A polyphasic approach that included PCR-dependent (16S rRNA gene sequencing of bacterial clones and isolates) and PCR-independent (FISH) molecular techniques was adopted. Consistently with other studies performed in Antarctic freshwater lakes (Van Trappen et al., 2002; Pearce, 2003; Pearce et al., 2003; Villaescusa et al., 2010), such approach allowed the identification of significant microbial components of the lakes bacterioplankton communities, with the inclusion of same classes of bacteria known to be important also in freshwater systems elsewhere in the world, such as the CF group of Bacteroidetes and Betaproteobacteria.
Moreover, phylogenetic groups that may exist in Antarctic lakes at low levels were detected in this study thanks to the application of FISH. For example, as reported by Pearce (2003) for the Moss lake, we determined a low abundance of Archaea by FISH, suggesting that they probably do not play a prominent role in the Antarctic lakes. Additional information obtained by FISH regarded the occurrence of Firmicutes, Deltaproteobacteria, and Epsilonproteobacteria, which have been often reported as constituents of Antarctic sediments and microbial mats (Peeters et al., 2011). As shallow lakes were considered, it is plausible to assume the origin of such bacteria from the benthic compartment by sediment removal processes. Finally, the FISH data also suggested the importance of the unicellular pico-cyanobacteria that often overcame the relative abundance of the other phylogenetic groups, mainly in the lakes CC and LH. Their predominance has been previously observed in other Antarctic oligotrophic lakes, that is, the Moss Lake (Pearce, 2003) and Sombre Lake (Pearce et al., 2003), suggesting that they may represent a conspicuous fraction of the pelagic community in terms of biomass in more oligotrophic conditions (Andreoli et al., 1992; Pearce et al., 2005).
Overall, the cluster analysis that was computed on all molecular results (not shown) highlighted that the bacterial communities that inhabited CC and LH tightly grouped together (71.6%, 95.2%, and 89.5% similarity for clones, isolates, and FISH, respectively) and were distinct from the lake INI. This finding may be a consequence of several factors (e.g. the distance from the sea and the presence/absence of nesting seabirds), but it may be more probably explained by the different physicochemical and trophic status that characterized the lake INI during the sampling period, confirming the previous observations in other Antarctic lakes (Pearce, 2000, 2005; Pearce et al., 2005; Villaescusa et al., 2010). The lake was characterized by a very high conductivity value, which suggests an intrusion of marine waters and a temporal brackish feature of such lake, and appeared to be naturally eutrophyzed by marine birds (mainly Adelie penguins and south polar skuas) (Borghini et al., 2008). Such features may be responsible, for example, for the lack in INI of Actinobacteria (which are generally thought to be associated with soil habitats, but are of major importance in freshwater environments; Peeters et al., 2011), as well as for the low abundance of pico-cyanobacteria (whose presence is not favored by relatively high N : P ratio; Kahru et al., 2000), and the consistently pronounced abundance of the Gammaproteobacteria (which are typically marine), Interestingly, in this regard, sequences of bacteria that are more traditionally associated with the marine environment (e.g. Alpha- and Gammaproteobacteria) were retrieved not only from the lake INI that was closest to the sea, but also for CC and LH. Thus, it is to be not excluded an input of bacteria from the marine environment into Antarctic lakes through atmospheric deposition (aerosols) and/or seabird excreta, both possibly increasing salts and nutrient supply to inland lakes (Camacho, 2006).
The phylogenetic analysis allowed assigning a large number of sequences that were reported by blast or Seqmatch/classifier as unidentified bacteria with their nearest phylogenetic neighbor (Fig. 4). Interestingly, the latter was generally from lakes and ponds, in addition to streams and glaciers (ice and snow), located in polar areas, thus strengthening results by Pearce et al. (2007) and Villaescusa et al. (2010), who demonstrated the bipolar distribution of freshwater lake bacterioplankton. As was the case for the study by Peeters et al. (2011), a number of sequences were also found to have a cosmopolitan distribution, even if they were mainly associated with cold habitats.
Both the Proteobacteria and CF group of Bacteroidetes were well represented among clones and isolates. The 32 OTUs belonging to the Proteobacteria phylum fell into three of the five classical classes: Alpha, Beta, and Gamma (Fig. 4a). The Alphaproteobacteria (nine OTUs) formed two subgroups, with three OTUs (which represented the 50% of the sequences in such class) that clustered around Sphingomonas spp. (mainly retrieved from cold environments). The Betaproteobacteria contained 17 OTUs distributed among several genera of the phylum, for example Polaromonas, Simplicispira, Acidovorax, Hydrogenophaga, and Herminiiimonas. The former was represented by two OTUs (i.e. Polaromonas spp. ALC-INI-B3 and ALC-CC-C1) that included the 12.1% of total sequences. Polaromonas spp. are generally dominant in clone libraries and culture collections from polar and high-elevation environments, mainly in glacial ice and sediment samples (Darcy et al., 2011). Their wide distribution across the globe is hypothesized to be dependent on air-dispersion processes (Darcy et al., 2011). To date, only few well-established Polaromonas species exist, with Polaromonas vacuolata that was firstly isolated from Antarctic seawater (Irgens et al., 1996). The two Polaromonas phylotypes, that is, ALC-INI-B3 and ALC-CC-C1, were closely related to Polaromonas sp. 1020 (EF423330) and Polaromonas sp. clone IC3028 (HQ595202) that were both isolated from a glacier, thus suggesting their adaptability to low temperatures. The Gammaproteobacteria (six OTUs) clustered in two main groups. The first one was composed by a single OTU (i.e. ALI-INI31), which was strongly related to Stenotrophomonas maltofila RBE1CD-58 from a Colombian River. The larger group branched in two subclusters represented by Psychrobacter spp. (two OTUs) and Pseudomonas spp. (three OTUs), respectively (Fig. 4a). In particular, in this investigation, the latter represented the 16.8% of total sequences. Pseudomonas is a cosmopolitan genus that has been found in several environments including freshwater and terrestrial habitats from Antarctica (Pearce et al., 2003; Saul et al., 2005). Interestingly, the phylotypes ALC-LH-A3 and ALI-CC18 were closely related to Pseudomonas sp. BSs20145 and Pseudomonas sp. BF02_S14 that were previously obtained from polar sites.
The 31 OTUs within the CF group of Bacteroidetes formed two distinct clusters: the first one included exclusively Flavobacterium spp. (19 OTUs), whereas the second cluster branched in two subclusters represented by Algoriphagus spp. (six OTUs) and Pedobacter spp. (six OTUs), respectively (Fig. 4b). In particular, sequences related to the genus Flavobacterium have been often reported for Antarctica with several novel species that were described (McCammon et al., 1998; McCammon & Bowman, 2000; Humphry et al., 2001; Van Trappen et al., 2002, 2003, 2004c, 2005; Yi & Chun, 2006; Peeters & Willems, 2011). In this study, at least seven phylotypes (26.7% of total sequences) within such genus had bacteria previously isolated from Antarctic freshwater habitats as closest relatives (98–99% similarity). Among these, Flavobacterium frigoris (sequences ALI-LH2, ALI-INI1, ALI-INI18, ALC-LH-E2, and ALC-LH-A1 in this study) and Flavobacterium segetis (sequence ALI-LH1 in this study) are at present only known from Antarctica, thus suggesting their endemic feature.
Interestingly, among the Proteobacteria and the CF group of Bacteroidetes, a number of genera (i.e. Polaromonas, Simplicispira, Hydrogenophaga, Psychrobacter, Pseudomonas, Flavobacterium, and Pedobacter) were often common to more than one lake investigated. Besides, they have been frequently reported in oligotrophic and eutrophic Antarctic freshwater systems, thus suggesting their wide diffusion in Antarctica (e.g. Van Trappen et al., 2002, 2003, 2004a, b, c, 2005 ; Pearce, 2003; Pearce et al., 2003, 2005, 2007; Villaescusa et al., 2010; Peeters et al., 2011). Remarkably, in this investigation, the majority of sequences (504 sequences of 902) clustered within the genera Flavobacterium, Pseudomonas and Polaromonas (Fig. 4a and b), suggesting that they may play a key role within the bacterioplankton communities that inhabit lakes of the northern Victoria Land. Members in such genera hydrolyze a wide variety of organic compounds (e.g. several carbohydrates and biomacromolecules).
The phylogenetic tree in the Fig. 4b also includes Actinobacteria, Planctomycetes, Gemmatimonadetes, and filamentous cyanobacteria that were detected exclusively by the clone library analysis (with the exception of Actinobacteria). In particular, sequences affiliated to the Planctomycetes and Gemmatimonadetes were exclusively obtained from LH. These microorganisms may exist in Antarctic soil environments at low levels (Smith et al., 2006; Yergeau et al., 2007) and, interestingly, members of both phylogenetic groups have been recently reported in the Luther Vale soil (Niederberger et al., 2008); thus, it is plausible to assume that they might origin from the surrounding terrestrial habitat (e.g. through seeding by wind).
The well-developed microbial mats that were observed in all lakes at sampling time (Borghini et al., 2008) could explain the occurrence of sequences from the filamentous cyanobacteria Leptolyngbya and Phormidium in the clone libraries (Fig. 4 and Table S1). These organisms have been frequently found in different Antarctic biotopes. In particular, the strong abundance of the genus Phormidium in the lake INI is in line with previous microscopy observations made for microbial mats in the same area (Fumanti et al., 1997; Cavacini, 2001).
In conclusion, the synergistic application of different methods allowed obtaining complementary information on the microbial community inhabiting Antarctic lakes of the northern Victoria Land, thus enlarging our knowledge on the prokaryotic diversity in these extreme habitats. Interestingly, this study demonstrates that the investigated lakes harbor bacteria that are common to both Antarctic and Arctic freshwater systems, thus highlighting their diffusion in Antarctica (including endemic species) and their bipolar distribution. Moreover, our results suggest that the microbial community composition in Antarctic shallow lakes can be often locally and/or temporally affected by external influences that include atmospheric deposition, seabird excreta, and seawater intrusion, in addition to exchanges between the lake waters and both the benthic and soil compartments.
L.M. and A.L.G. wish to thank their colleague Chiara Agnorelli (University of Siena, Italy) for assistance during sample collection, and all of the staff at ‘Mario Zucchelli’ Station, for the logistic help and support, which made possible the expedition. We also thank the Editor Prof. Max Häggblom and three anonymous reviewers for helpful comments on the manuscript. This research was supported by grants from the National Antarctic Research Program (PNRA), Italian Ministry of Education and Research (PEA 2004, Research Project PNRA 2004/1.6), and from the National Antarctic Museum (MNA).