A molecular phylogenetic survey of sea-ice microbial communities (SIMCO)


*Corresponding author. Tel.: +61 (3) 6226 2776; Fax: +61 (3) 6226 2642; E-mail: john.bowman@utas.edu.au


16S rDNA clone library analysis was used to identify bacterial biodiversity in a variety of sea-ice microbial communities (SIMCO). DNA was extracted from seven Antarctic sea-ice samples and one Arctic sea-ice sample and 16S rDNA PCR-amplified using universal and Archaea-specific primers. Recombinant 16S rDNA clones were obtained and dereplicated using restriction fragment length polymorphism analysis (RFLP). After RFLP analysis, 100 distinct phylotypes (a unique clone or group of clones with sequence similarity of >0.98) were defined. From the clone libraries 16S rDNA sequences of bacterial and eukaryotic origin were detected, however Archaea were not detected either with universal or Archaea-specific 16S rDNA primer sets. Bacterial phylotypes grouped within the α and γ proteobacteria, the CytophagaFlavobacteriumBacteroides division, the Gram-Positive bacteria and the orders Chlamydiales and Verrucomicrobiales. The majority of bacterial phylotypes were affiliated with heterotrophic taxa and many grouped closely with cultivated genera and species. Eukaryotic clones were affiliated with a variety of autotrophic and heterotrophic nanoplankton and included a large number of chloroplast 16S rDNA genes. The findings of this investigation corroborated culture data indicating bacterial biodiversity increased in SIMCO displaying high levels of primary production, however the bacterial communities within SIMCO were highly heterogeneous at the genus/species-level between different samples. A comparison of Antarctic and Arctic SIMCO revealed certain sea-ice dwelling bacterial genera are common at both poles.

1. Introduction

Sea ice contains much of the productivity of the Southern Ocean and strongly influences global energy budgets and atmospheric–oceanic interactions in the polar zones [1]. Sea-ice microbial communities (SIMCO) have been shown to have a major role in trophic food webs and occur as surface populations, within ice floes, and concentrated near the sea-ice:seawater interface [2]. Sea ice is temporally and spatially highly variable and is for the most part seasonally transient constantly being broken up and reshaped by wind and ocean currents. Sea ice is also a relatively extreme environment with internal temperatures ranging from −1°C to as low as −50°C in winter. The salinity of sea ice brines within channels and cracks of the sea ice (formed when salt is ejected during freezing) can rise as high as 150‰ while the salinity can drop to <10‰ as the sea ice melts. The relative development and complexity of the SIMCO is thus primarily determined by physical forces [3]. In any case, sea ice is a biologically active habitat and on a per volume basis is more productive than the underlying pelagic zone [4].

The assemblages of organisms that form the SIMCO can be quite complex though it mostly consists of ice-adapted diatom species and bacteria. Well-developed communities will usually support a microbial loop and a host of allochthonous and autochthonous protists, zooplankton and small metazoa [2].

Bacterial activity and populations in sea ice are tightly coupled to and dependent upon algal primary productivity, playing an important role in secondary mineralization of dissolved and particulate organic matter [5]. At certain times, bacteria can dominate within sea-ice habitats [5,6]; in general though they make up only a small proportion of the total SIMCO biomass [7]. Most bacteria isolated from sea ice have been found to be pigmented, highly cold-adapted with both free-living and epiphytic bacteria present [9]; some are able to form gas vesicles [8].

Extensive cultivation and identification of bacteria from SIMCO indicated about half of the taxa isolated are psychrophilic while the rest are psychrotolerant [10]. The number of psychrophilic taxa and overall number of taxa successfully isolated increased when algal-rich sea-ice samples were investigated [10]. The numbers of psychrophilic bacteria (enumerated by most probable number counting) along with total bacterial direct count also increased proportionally with chlorophyll a concentration and algal population density [11]. Most taxa isolated from sea ice belong to the γ proteobacteria and the CytophagaFlavobacteriumBacteroides (CFB) division [10]. Several of these taxa have now been fully characterized and named, a list of these species has been published by Nichols et al. [12] with others also having been described by Gosink et al. [13,14] and Junge et al. [15]. Biogeographical studies of sea-ice bacterial species [16], indicated various psychrophilic genera are present at both the north and south poles, e.g. Octadecabacter spp. and Glaciecola spp., however it has not been conclusively shown whether both polar regions harbor the same psychrophilic sea-ice species.

The amount of specific data available on sea-ice dwelling bacteria is still limited, by gaining knowledge of these bacteria, insights into oceanic polar ecosystem community structure can be obtained especially in establishing the functional roles that bacteria play in the sea-ice ecosystem and to determine their relationships and physiological characteristics within the broader context of polar and marine zones. From a practical standpoint sea-ice bacteria due to high levels of cold adaptation also have strong biotechnological potential including sources of cold-adapted enzymes, ω-3 and ω-6 polyunsaturated fatty acids and novel pigments [12].

Previous studies investigating bacteria within SIMCO have been limited to cultivation techniques, in this study the variability and extent of bacterial biodiversity in sea ice obtained over a wide geographical range is established using direct cloning of 16S rDNA sequences amplified from sea-ice DNA. From this a more comprehensive analysis of bacterial species predominating in sea-ice can be obtained. Sea-ice samples, containing SIMCO of different complexities were investigated to determine whether the presence of algal assemblages lead to shifts in the bacteria community structure and complexity that so far have only been roughly suggested by cultivation data.

2. Materials and methods

2.1 Sea-ice samples

Sea-ice cores 1, 4, 7, 8, 9 and 10 were collected at various locations in the Antarctic pack ice between Casey and Davis bases during November–December 1996 on Australian National Antarctic Research Expedition voyage 2 of the Aurora Australis. Fast ice was collected by A. McMinn from McMurdo Sound in November 1997 while the ice sample from the Canadian Arctic was collected by C. Lovejoy during the Northern Open Water cruise during July 1998. All cores were obtained using a SIPRE corer with care taken not to disturb the lower platelet ice layer and algal assemblage if present. Cores were immediately transferred to sterile plastic bags in which they were melted the same day at 4°C in an equivalent volume of sterile artificial seawater (Sigma Chemical Co., St. Louis, MO, USA). Samples were then filtered onto 0.2-μm pore-sized filters (Millipore) which were subsequently stored at −80°C until processed.

2.2 DNA extraction

Extraction of DNA from filters used a modification of the procedure of Fuhrman et al. [17]. Filter fragments were suspended in 2 ml STE (0.1 M NaCl, 10 mM Tris–HCl, 1 mM disodium EDTA, pH 8.1) buffer and cell material lysed by boiling with 1% (w/v) sodium dodecyl sulfate for 2 min in 15-ml polypropylene centrifuge tubes. Cellular debris was removed by centrifugation (10 min at 10 000×g) using a model 5417c Eppendorf centrifuge. The supernatant was transferred to fresh tubes and DNA was precipitated by adding 7 ml absolute ethanol and 0.7 ml 10.5 M ammonium acetate (at −20°C for 2 h). High molecular mass DNA was then pelleted by centrifugation in a swinging bucket rotor at 3800×g (Sorvall Super T 21 centrifuge, DuPont). Following air-drying the DNA pellet was suspended in 0.5 ml TE (10 mM Tris-buffer, 1 mM disodium EDTA, pH 8.1) and then extracted with 0.5 ml TE-saturated phenol by gentle mixing. The aqueous layer was then separated by microcentrifugation and the lower organic phase removed carefully. Extraction was repeated the same as above with 0.5 ml 3:1 phenol:chloroform and then with 0.5 ml 24:1 chloroform:isoamyl alcohol. The DNA was precipitated with 1 ml absolute ethanol and 0.12 ml 10.5 M ammonium acetate and pelleted at 4°C. The supernatant was poured off, the pellet air-dried and the pellet suspended in 0.1–0.3 ml of sterile MilliQ water. The quality of DNA samples was examined by agarose electrophoresis and concentrations determined using fluorimetry (using Hoechst dye 33258 and a Model TKO 100 DNA fluorimeter, Hoefer Scientific).

2.3 PCR, clone library construction and screening of 16S rDNA genes

Primers used to for 16S rDNA amplification from the sea-ice DNA included primers S-*-Univ-0519-a-A-18 (5′-GWATTACCGCGCKGCTG-3′) and S-D-Bact-1492-a-A-21 (5′-ACGGYTACCTTGTTACGACTT-3′) [18]. This primer combination is effective for the amplification of the vast majority of bacteria and Archaea but less efficient for amplification of eukarya and will not amplify 18S rDNA genes from Bacillariophyta (diatoms). Specific primers were also used for PCR amplification of archaeal 16S rDNA including the following primer pairs: S-D-Arch-0021-a-A-20 (5′-TTCCGGTTGATCCYGCCGGA-3′)/S-D-Arch-0958-a-A-19 (5′-YCCGGCGTTGAMTCCAATT-3′) and S-D-Arch-0021-a-20/S-*-Univ-1392-a-A-15 (5′-ACGGGCGGTGTGRC-3′) [19]. PCR conditions utilized and methods used for clone library construction and clone dereplication by restriction fragment length polymorphism analysis (RFLP) analysis have been described previously by Bowman et al. [20,21].

2.4 Phylogenetic analysis

Sequence data were checked using the program Sequence Navigator (Applied Biosystems) to resolve ambiguous base positions and by the CHECK_CHIMERA program [22] to determine the presence of any PCR-amplified hybrid sequences. Sequences were compared to the GenBank nucleotide database library by GAPPED BLAST on-line searches [23] (http://www.ncbi.nlm.nih.gov/blast/blast.cgi). Sequences were then manually aligned to selected GenBank sequences as well as to sequences obtained from the Ribosomal RNA Project II database [22] (http://www.cme.msu.edu/RDP/html/index.html). Phylogenetic analysis was then performed by using the program package PHYLIP [24] as described by Bowman et al. [20,21]. The sequences determined in this study are deposited under GenBank accession numbers AF277467–AF277565 and AF333080–AF333086.

2.5 Diversity analysis

For calculation of diversity indices, the libraries were normalized to 50 clones each by rarefaction [25] using the program RAREFACT.FOR [26]. Estimates of diversity (Shannon–Weaver index), dominance concentration (Simpson Index) and evenness (Equitability index) were calculated using equations described by Krebs [27] and which are also detailed in Bowman et al. [20]. The diversity or Shannon–Weaver index measures the average degree of uncertainty (synonymous with diversity) of predicting the species of a given individual picked at random from a community. The index varies from a value of 0 for communities with only a single species to high values for communities having many species (phylotypes in the case of this study), each with a few individuals. The dominance concentration or Simpson Index is based on the probability of drawing a pair of individuals of the same species. Dominance values that approach 1 indicate that only a few species predominate in the sample. Evenness indicates the homogeneity of distribution of species present within a given population. An evenness value that approaches 1 indicates that many species occur in the sample at similar levels. Species richness was estimated using the non-parametric model of Chao [28] and is a statistical extrapolation of the total number of species present within the sample. Estimation of biodiversity coverage used the method described by Mullins et al. [29]. Biodiversity coverage is based on the comparison of unique and non-unique phylotypes within a community with a value of 100% indicating that all unique phylotypes within a sample have been obtained. Pair-wise comparisons of clone libraries were carried out determining a similarity coefficient (s) determined with the following formula [30]:


where A and B are the numbers of phylotypes in libraries A and B, respectively and C is the number of shared phylotypes.

3. Results and discussion

3.1 Samples and clone library analysis

Sea-ice sampling was limited to the polar spring/summer period (November–December) and no set spatial structure was used in the sampling due to the availability of ship-time and access to suitable sea-ice areas. Sea-ice samples were collected at random over a wide area (Table 1) and contained widely variant phytoplanktonic content. Southern Ocean pack ice cores 1 and 7 had no visible SIMCO present while samples 4, 8, 9 and 10, Arctic (Baffin Bay) and McMurdo Sound samples had distinct platelet ice algal assemblages present, indicated by a distinct brown–green discoloration present at the base of the ice core. All samples possessing an algal assemblage, except the McMurdo sample, were dominated by Entomoneis and Nitzschia diatom species. The McMurdo sample contained mostly Nitzschia spp.

Table 1.  Sea-ice samples investigated in this study
Sample numberIce typeSite sampled
1packSouthern Ocean, 64°51′S 109°37 E
4packSouthern Ocean, 62°55′S 92°37′E
7packSouthern Ocean, 64°18′S 74°07′E
8packSouthern Ocean, 66°21′S 76°54′E
9packSouthern Ocean, 66°21′S 76°58′E
10(multi-year) packPrydz Bay, 68°35′S 77°58′E
McMurdofastMcMurdo Sound, 77°30′S 165°24′E
ArcticpackBaffin Bay polynya, 73°07′N 66°48′E

Clone libraries based on universal primers were constructed from all samples with 50–60 clones per ice sample containing 16S rDNA inserts dereplicated using RFLP analysis. From this about 100 phylotypes were discerned with each phylotype consisting of either a unique clone or a group of clones that have sequence similarities of 0.98 or greater. From 6 to 18 16S and 18S rDNA phylotypes were recovered from each ice sample.

Biodiversity coverage for these libraries was relatively high ranging from 42.9 to 57.1% (Table 2). Theoretically, about half of the biodiversity present in the samples was sampled in the universal libraries. Archaeal-specific PCR primer sets were also employed, however surprisingly no PCR amplification products were obtained from any sample. Archaea, mostly of Marine groups I and II are relatively common in Antarctic seawater [31] however, survey results indicate their populations begin to decline substantially at the time when the ice samples were obtained [32]. It is also possible that during phases of sea-ice formation when there is a transient suppression of biological activity and turnover [33], archaeal cells along with other microorganisms may be selected against and fail to colonize the ice matrix. Further analyses using PCR-based procedures and fluorescent in situ hybridization (FISH) are required to further establish the presence or absence of Archaea in sea-ice samples.

Table 2.  Biodiversity characteristics of the sea-ice samples investigated
Sample numberCoverage (%)Diversity (H′)Dominance (S′)Evenness (J′)Number of phylotypesa (bacteria)Species richnessa (bacteria)16S rDNA clones (%)a
       chloro-plast/plastidCFB group
  1. a 18S rDNA and Chlamydiales clones were not considered in these calculations.


3.2 Affiliation of phylotypes

Distribution of the sea-ice sample 16S rDNA phylotypes on the basis of phylogenetic groupings is shown in Fig. 1. Most bacterial phylotypes detected, grouped within the γ and α proteobacteria (Fig. 2A) and CFB division (Fig. 2B). Phylotypes also grouped in the orders Verrucomicrobiales and Chlamydiales and the Gram-positive bacteria (Fig. 2C). The presence of eukaryotes, which normally dominates the biomass of SIMCO, was indicated by a high incidence of chloroplast-derived 16S rDNA clones (Fig. 2C) in several of the clone libraries. The incidence of which appeared to increase with increasing algal density within the samples (Fig. 1). 16S rDNA genes from chloroplasts appear to be mostly derived from diatom species, which usually make up the bulk of the SIMCO biomass [5,7]. A group of AT-rich phylotypes (SIC.4-6, SIC.8-6, SIC.9-3, ARCTIC_ICE-13, SIC.4-13) detected in most of the algal-rich ice samples was affiliated with small subunit RNA genes from mitochondria (Fig. 2C). The source of the mitochondrial genes is uncertain due to the low level of similarity these clones had with sequences available on the GenBank database. 18S rDNA phylotypes were detected sporadically in several samples. Pack ice sample 7 was dominated by a single phylotype derived from a sarcomonad flagellate (Fig. 2D) a group not known previously to exist in sea ice (A. McMinn, personal communication), which could be allochthonous grazers. Pack ice sample 4 contained 18S rDNA derived from the common pyrsemniophyte Phaeocystis antarctica and several clones derived from a dinoflagellate related to Prorocentrum and relatives. All of the algal-rich samples contained at least one phylotype derived from free-living turbellarians of the order Acoela. The flatworms are probably grazers within the algal assemblage feeding on algae, bacteria, protists and possibly zooplankton [34] suggesting an active microbial loop exists in these samples.

Figure 1.

Phylogenetic distribution of sea-ice clones derived from universal 16S rDNA clone libraries with sea-ice samples arranged in order of increasing phytoplanktonic density.

Figure 2.

Figure 2.

Phylogenetic trees (clustered by Neighbor-joining of maximum likelihood values) showing positions of sea-ice phylotypes to closest related 16S rDNA sequences from either cultivated or cloned sources. Clones are indicated in bold-type (SIC clones include those from Southern Ocean sites). Species flagged with asterisks have been isolated from polar sea ice and from other Antarctic samples including seawater and sediment. Species indicated in quotation marks are currently not nomenclaturally valid names. Some sediment and seawater Antarctic isolates come from unpublished data. Numbers at branch nodes are bootstrap values (only values of 60% and greater are shown). Phylogenetic groups depicted: (A) γ and α proteobacteria; (B) CFB division; (C) other bacterial divisions/groups including Verrucomicrobia, Chlamydiales, mitochondrial 16S rDNA, chloroplast rDNA and Gram-positives; and (D) eukaryotes (18S rDNA).

Figure 2.

Figure 2.

Phylogenetic trees (clustered by Neighbor-joining of maximum likelihood values) showing positions of sea-ice phylotypes to closest related 16S rDNA sequences from either cultivated or cloned sources. Clones are indicated in bold-type (SIC clones include those from Southern Ocean sites). Species flagged with asterisks have been isolated from polar sea ice and from other Antarctic samples including seawater and sediment. Species indicated in quotation marks are currently not nomenclaturally valid names. Some sediment and seawater Antarctic isolates come from unpublished data. Numbers at branch nodes are bootstrap values (only values of 60% and greater are shown). Phylogenetic groups depicted: (A) γ and α proteobacteria; (B) CFB division; (C) other bacterial divisions/groups including Verrucomicrobia, Chlamydiales, mitochondrial 16S rDNA, chloroplast rDNA and Gram-positives; and (D) eukaryotes (18S rDNA).

3.3 Comparisons with cultivated strains and microscopically observed morphotypes

Most seawater bacterial cells belong to the γ and α proteobacteria and CFB group as indicated by studies using FISH, including a variety of Southern Ocean seawater samples [35,36]. A similar predominance of taxa from these phylogenetic groups is also found in SIMCO as suggested by both cultivation [10] and 16S rDNA clone data. Many bacterial phylotypes detected in this study group closely (similarity>0.95) with species originally isolated from Antarctic and Arctic sea ice as well as some other Antarctic habitats as shown in Fig. 2. Based on phylogenetic inferences most of the phylotypes are derived from bacterial genera only known to be heterotrophic. This result is quite different with clone library studies of seawater in which a high proportion of phylotypes detected have no cultivated equivalents or are autotrophs, e.g. Marine group I and II Archaea, α and γ SAR clone groups, ammonia oxidizers, cyanobacteria etc. [29]. It is possible that sea-ice formation processes and the transient availability of primary production biomass are sufficiently selective factors that only fast growing, resilient pelagic species or species tightly associated with sea-ice phytoplankton are able to significantly compete and successfully colonize sea ice.

In well-developed algal assemblages it is common to find algal cells covered with many epiphytic bacteria including a frequently observed Prosthecobacter-like morphotype [37]. The common presence of this epiphyte suggests it may have a significant role in secondary mineralization within sea ice [16]. A number of phylotypes related to Prosthecobacter (similarity 0.88–0.91) were detected in one of the algal-rich ice samples (sample 8, Fig. 2C) and theoretically could derive from cells with the Prosthecobacter-like morphotype. No confirmatory cultivation data are available to completely prove this as the morphotype has been refractory to cultivation [16]. Further work with Verrucomicrobia-specific oligonucleotide probes including probes based on sequences derived from this study would be useful in determining the distribution and significance of this distinct epiphyte.

3.4 Comparisons between sea-ice samples

To assess bacterial diversity between the sea-ice samples investigated mathematical analysis of phylotype distribution was carried out using criteria used commonly to assess the diversity of metazoa within natural environments [27]. It must be pointed out the inherent biases of PCR can potentially distort clone library data and thus clone incidence in clone libraries is not necessarily a good indication of abundance of the species in the sample. Data presented in Table 2 have been used simply as a means to compare the samples investigated in this study with the caveat that the measurements are just estimates of diversity. On this basis, the data suggested increased biodiversity in the sea-ice samples appeared to be linked to increased algal density in most samples studied except for the McMurdo Sound sample. This is reflected in the number of phylotypes (11 to 18 compared to 6 to 7) found in the algal-rich samples (samples 4, 8 and 9; Arctic sample) compared to samples possessing low algal levels (samples 1 and 7). In addition, this difference was suggested by the Arctic sample and Southern Ocean samples 4, 8 and 9 possessing higher diversity (0.76–1.13 compared to 0.40–0.49) and evenness (0.75–0.92 compared to 0.48–0.59) indices and lower dominance concentration values (0.09–0.18 compared to 0.45–0.58) than samples 1 and 7. There were also distinct differences in terms of estimated species richness (Table 2).

Increasing algal density also seems to result in an increasing proportion of CFB clones (Table 2). In pack ice samples 1 and 7 the proportion of CFB clones was only 0–7% but in more algal-rich samples (e.g. 4, 8, 9 and the Arctic sample) this level increased to 20–27%. Cultivation studies and studies measuring populations of sea-ice bacteria epiphytes suggest CFB or CFB-like bacteria contribute very strongly to bacterial populations in well-established sea-ice assemblages, in particular those dominated by diatoms such as Entomoneis[9–11]. Up to 70% of the bacterial biomass associated with Southern Ocean pelagic algal blooms is comprised of taxa of the CFB division as shown by analysis by FISH [35,36] and it seems logical this group is also predominant in SIMCO.

The McMurdo Sound sample differed from the other samples in that it had a distinct algal assemblage but possessed a low diversity, strongly dominated by chloroplast 16S rDNA clones (65% of clones, Table 2). The ice core sample was collected in the late spring (November) from a sea-ice sample dominated by Nitzschia diatom species (A. McMinn personal communication) and examinations of McMurdo Sound sea ice by Grossi et al. [9] indicate that bacterial:algal population ratio in late spring Nitzschia algal assemblages are quite low. Its only until well into the summer period (January) when bacterial populations begin to bloom rapidly. This is unlike assemblages dominated by the diatom Entomoneis which have much higher epiphytic bacterial populations [9]. This difference may have led to a reduced chance of sampling bacterial 16S rDNA clones from the McMurdo Sound sample.

Though the diversity within individual samples could not be considered exceptionally high compared to other environments such as soil and sediment, only a few phylotypes were shared between each library and no one phylotype appeared in more than three separate libraries. Inter-sample similarity was measured numerically by pair-wise comparisons of clone library data between each of the libraries (Table 3). Interestingly, it was found that even for samples collected from the same region (samples 8 and 9) there was little similarity between datasets with inter-sample similarity ranging from only 0 to 15%. The limited inter-sample similarity may reflect the high level of sea-ice temporal–spatial variability but may also derive from limitations of the clone library analysis technique.

Table 3.  Pair-wise comparison of bacterial phylotypes from universal 16S rDNA primer of clone libraries constructed from different sea-ice samples
Sea-ice sample1478910McMurdoArctic
 Similarity (S)

A single Arctic sea-ice sample was investigated to determine whether taxa present in the sample were also found in Antarctic samples. Though none of the Arctic phylotypes exactly matched those of Antarctic samples, several phylotypes (e.g. ARCTIC_ICE phylotypes 1, 3, 6, 7 and 12) were highly similar (>0.97) to various Antarctic clones and strains previously isolated from Antarctic sea ice and sediment including Octadecabacter, Psychroserpens, Psychroflexus-like sea-ice isolate IC076 [10] and the TAYNAYA-9 clone group (Fig. 2A). One Arctic phylotype was very similar (0.98) to ‘Flavobacterium xylanivorum’ a strain isolated from sediment collected in the British Antarctic (Humphry et al., unpublished).

This study reports the first examination of SIMCO using molecular methods. The lack of Archaea in the samples was surprising and examination of SIMCO is required using different methods, in particular combined rRNA hybridization/FISH procedures that would provide a more quantitative analysis of the communities present. More cultivation work and molecular analyses are required for Arctic sea-ice habitats in which the microbial diversity still remains largely unstudied.


This study was supported by Grants from the Australian Research Council and from the Antarctic Science Advisory Committee (project 1012). The authors would like to thank personnel at the Australian Antarctic Division, the crew of the Aurora Australia, Andrew McMinn (University of Tasmania) and Connie Lovejoy (Laval University, Que., Canada) for advice and for sea-ice samples. We would like to thank Peter Grewe and Bronwyn Innes (CSIRO Marine Research Division) for access to automated DNA sequencing facilities.