Bacterial succession in Antarctic soils of two glacier forefields on Larsemann Hills, East Antarctica


  • Felizitas Bajerski,

    Corresponding author
    • Alfred Wegener Institute for Polar and Marine Research, Research Department Potsdam, Potsdam, Germany
    Search for more papers by this author
  • Dirk Wagner

    1. Alfred Wegener Institute for Polar and Marine Research, Research Department Potsdam, Potsdam, Germany
    Current affiliation:
    1. Helmholtz Centre Potsdam, GFZ German Research Centre for Geosciences, Potsdam, Germany
    Search for more papers by this author

Correspondence: Felizitas Bajerski, Alfred Wegener Institute for Polar and Marine Research, Research Department Potsdam, Telegrafenberg A45, 14473 Potsdam, Germany. Tel.: +49 331 2882142; fax: +49 331 2882188; e-mail:


Antarctic glacier forefields are extreme environments and pioneer sites for ecological succession. Increasing temperatures due to global warming lead to enhanced deglaciation processes in cold-affected habitats, and new terrain is becoming exposed to soil formation and microbial colonization. However, only little is known about the impact of environmental changes on microbial communities and how they develop in connection to shifting habitat characteristics. In this study, using a combination of molecular and geochemical analysis, we determine the structure and development of bacterial communities depending on soil parameters in two different glacier forefields on Larsemann Hills, East Antarctica. Our results demonstrate that deglaciation-dependent habitat formation, resulting in a gradient in soil moisture, pH and conductivity, leads to an orderly bacterial succession for some groups, for example Cyanobacteria, Bacteroidetes and Deltaproteobacteria in a transect representing ‘classical’ glacier forefields. A variable bacterial distribution and different composed communities were revealed according to soil heterogeneity in a slightly ‘matured’ glacier forefield transect, where Gemmatimonadetes, Flavobacteria, Gamma- and Deltaproteobacteria occur depending on water availability and soil depth. Actinobacteria are dominant in both sites with dominance connected to certain trace elements in the glacier forefields.


Glacier retreat due to global warming is a phenomenon which can be observed in high mountain ranges like the Alps (Haeberli et al., 2007) as well as in Arctic and Antarctic environments (Rignot, 2001; Bamber et al., 2005; Cook et al., 2005; Oerlemans, 2005). Antarctica plays an important role in the global climate system and helps us to obtain insight into past and future climate changes. It is characterized by extreme conditions and is sensitive to environmental changes (Verleyen et al., 2004; Turner et al., 2005). West Antarctica and the Antarctic Peninsula are known to be strongly affected by global warming, but recent studies also reported significant warming in East Antarctica although with regional differences and seasonal variations (Steig et al., 2009; Verleyen et al., 2011). The continent-wide average near-surface temperature trend is positive (Steig et al., 2009).

Glacier forefields form terrestrial pioneer sites and are influenced by harsh climatic conditions and low nutrient availability (Schütte et al., 2009; Duc et al., 2011). Terrestrial Antarctic ecosystems are characterized by low temperatures, soil moisture and organic matter content and high salinity (Cannone et al., 2008; Niederberger et al., 2008). Antarctica is known as a continent dominated by microbial ecosystems (Wynn-Williams, 1996; Aislabie et al., 2006) as microorganisms play an important role in primary succession, pedogenesis and biogeochemical cycling (Lazzaro et al., 2009; Schütte et al., 2009). Microorganisms support mechanical and chemical bedrock degradation through weathering processes (Frey et al., 2010) and the establishment of more complex microbial communities or higher life like plants (Tscherko et al., 2003). Apparently unaffected glacier forefields provide a unique opportunity as a natural laboratory to study the succession of pioneering plants (Miniaci et al., 2007), animals (Kaufmann, 2001) and especially microorganisms (Sigler & Zeyer, 2002; Aislabie et al., 2006; Lazzaro et al., 2009; Duc et al., 2011). Understanding microbial communities in connection to soil formation and environmental parameters will help to predict how bacterial communities will react to changing environmental conditions.

During the last 20 years, microbial communities in extreme habitats became an important research topic and were studied in the Alps (Sigler et al., 2002; Duc et al., 2011), Arctic (Liebner et al., 2009; Barbier et al., 2012) or Antarctic environments (Ganzert et al., 2011a; Yergeau et al., 2012). Former studies have shown the importance of microbial communities in alpine or Arctic glacier forefields and their role in extreme Antarctic environments; however, Antarctic terrestrial microbial communities remain poorly understood and additional knowledge is necessary. With respect to methodology, several studies have used polyphasic approaches consisting of culture-independent methods coupled with culturing techniques and sedimentological analysis (Sigler et al., 2002; Shivaji et al., 2011). Culturing experiments showed the impact of isolated bacteria on granite weathering (Frey et al., 2010). A high abundance of microorganisms with up to 109 cells g−1 soil could be observed in alpine glacier forefields (Sigler & Zeyer, 2002). In recent years, molecular techniques have become an important tool describing microbial community structures in polar regions (Wagner, 2008). Molecular fingerprints have revealed a difference in communities from younger sites, compared to older sites (Lazzaro et al., 2009), and environmental changes can influence the community structures (Odum, 1970; Tiao et al., 2012). Bacterial communities change along a glacier forefield transect correlated with shifts in pH, soil water or age (Noll & Wellinger, 2008), or they show a patchy distribution according to heterogeneous soil characteristics (Niederberger et al., 2008).

Antarctic glacier forefields are suitable sites to study microbial community development because of its special environment, geographical isolation and little anthropogenic influence. Two different glacier forefields were chosen for analysis not only for comparison, but also to illustrate possible phases of succession. The aim of this study was to investigate the influence of glacier retreat on the development of forefield soil microbial communities in Antarctica. For this approach, a combination of cultivation experiments, molecular, geophysical and geochemical analysis was applied. We used terminal restriction length polymorphism (T-RFLP) and clone libraries to determine bacterial diversity and distribution. Quantitative real-time PCR (qPCR) was carried out to estimate the abundance of bacteria in the soils. Heterotrophic bacteria were examined as a part of the whole community using plate counts.

Materials and methods

Study site

The Larsemann Hills are located on the Ingrid Christensen Coast of Princess Elizabeth Land at Prydz Bay, East Antarctica (69°30′S, 76°20′E). The study site is characterized by an ice-free area of c. 50 km2 and a marine influenced continental climate leading to intensive physical weathering processes (Stüwe et al., 1989; Burgess et al., 1994). There is very little development in the geography of the ice sheets surrounding the Larsemann Hills, and with a movement of about 0.5–2 km over the last 20 ka (kiloannum), the present ice sheet is more or less stagnant. Broknes Peninsula, one of the two main peninsulas of the Larsemann Hills, has been continuously ice free since at least 40 ka BP (Hodgson et al., 2001). Air temperatures in the coastal regions are about −18°C to −29°C in Antarctic winter and around 0°C in summer (December–February). Precipitation usually occurs as snow and amounts up to about 250 mm a−1 (ANARE, 2000; Hodgson et al., 2001). Permafrost temperatures measured at the Russian Progress Station are about −8.5°C, and the active layer is 0.7 m deep on average (Vieira et al., 2010).

Sample collection and soil analysis

Two glacier forefields on Broknes Peninsula were chosen for analysis in this study and sampled during the expedition of the research vessel ‘Polarstern’ to Antarctica in March 2007 (Ganzert et al., 2008). The first, called ‘Glacier Transect’ (S 69°24.140; E 76°20.178 to S 69°24.135; E 76°20.296), was about 80 m long, and 13 soil samples and one surface sample (0–1 cm) were taken from five profiles (Fig. S1a). In the second glacier forefield (S 69°24.221; E 76°20.813 to S 69°24.326; E 76°20.273), named ‘Black Valley Transect’ referring to its surface colour and extending over a transect of 500 m, 11 soil samples and three surface samples (0–1 cm) were harvested out of five profiles (Fig. S1b). The Black Valley Transect was situated between a small glacial snow cap on one side and the Dalk Glacier on the other. All 28 soil samples were bulk samples and the sampling depth was chosen according to the graininess and horizonation of the weathering debris (Tables 1 and 2).

Table 1. Selected soil properties and bacterial abundance in the glacier forefield called ‘Glacier Transect’ on Larsemann Hills, East Antarctica. Sample designations refer to distance from the glacier and sample depth. Conductivity and pH values were determined in the field, with exception of the asterisks-marked samples
Site (m)Depth (cm)Sample IDSand (%)Silt (%)Clay (%)Moisture (%)Conductivity (μS cm−1)pH valueC (%)N (%)CFU g−1 soil on R2ACFU g−1 soil on BRIIgene copies g−1 soil
  1. a

    Analysis was performed in the laboratory.

  2. b

    Below the detection limit, C < 0.1, N < 0.05.

  3. CFU, colony-forming units; GT, Glacier Transect.

00–7GT0/0-790.< 0.1b0.15.6 × 104 ± 1.2 × 1047.0 × 104 ± 9.2 × 1031.2 × 107 ± 6.2 × 105
07–14GT0/7-1489.< 0.1b0.11.8 × 104 ± 6.1 × 1037.8 × 104 ± 5.6 × 1031.7 × 107 ± 1.2 × 106
014–25GT0/14-2589.< 0.1b0.16.5 × 104 ± 6.5 × 1031.4 × 105 ± 6.0 × 1033.5 × 106 ± 2.6 × 105
300–13GT30/0-1395. × 104 ± 1.5 × 1032.0 × 105 ± 1.2 × 1044.2 × 107 ± 5.5 × 106
3013–28GT30/13-2884.512.52.93.712.67.7< 0.1b0.15.9 × 104 ± 1.2 × 1042.8 × 105 ± 2.4 × 1041.3 × 107 ± 3.4 × 105
550–10GT55/0-1088.< 0.1b0.13.8 × 104 ± 4.4 × 1036.3 × 103 ± 3.4 × 1036.3 × 107 ± 8.0 × 106
5510–20GT55/10-2082.314.< 0.1b0.12.2 × 104 ± 3.1 × 1031.8 × 104 ± 6.0 × 1037.4 × 106 ± 4.0 × 105
650–10GT65/0-1090.< 0.1b0.11.4 × 104 ± 1.8 × 1032.3 × 103 ± 3.5 × 1025.5 × 107 ± 2.8 × 106
6510–20GT65/10-2086.711.< 0.1b0.15.5 × 103 ± 2.1 × 1039.5 × 103 ± 4.4 × 1034.9 × 106 ± 2.1 × 105
6520–30GT65/20-3089.< 0.1b0.12.5 × 103 ± 9.6 × 1024.1 × 103 ± 2.9 × 1035.6 × 106 ± 1.4 × 105
800–1GT80/0- × 104 ± 2.3 × 1034.2 × 104 ± 5.2 × 1037.6 × 108 ± 1.8 × 107
801–10GT80/1-1092. × 103 ± 1.8 × 1037.0 × 104 ± 4.3 × 1045.8 × 107 ± 6.4 × 106
8010–20GT80/10-2093.< 0.1b0.11.6 × 104 ± 1.2 × 1034.3 × 104 ± 3.5 × 1031.1 × 108 ± 2.2 × 106
1000–1GT100/SS89.< 0.1b< 0.05b1.7 × 104 ± 2.4 × 1034.0 × 102 ± 3.5 × 1021.5 × 108 ± 1.4 × 107
Table 2. Selected soil properties and bacterial abundance in the glacier forefield called ‘Black Valley Transect’ on Larsemann Hills, East Antarctica. Sample designations refer to distance from the glacier and sample depth. Conductivity and pH values were determined in the field, with exception of the asterisks-marked samples
Site (m)Depth (cm)Sample IDSand (%)Silt (%)Clay (%)Moisture (%)Conduc-tivity (μS cm−1)pH valueC (%)N (%)CFU g−1 soil on R2ACFU g−1 soil on BRIIGene copies g−1 soil
  1. a

    Analysis was performed in the laboratory.

  2. b

    Below the detection limit, C < 0.1, N < 0.05.

  3. BV, Black Valley; CFU, colony-forming units.

260–1BV26/0- × 105 ± 7.2 × 1042.8 × 106 ± 4.2 × 1056.6 × 108 ± 4.6 × 107
261–9BV26/1-993. × 105 ± 1.6 × 1055.9 × 105 ± 5.3 × 1041.2 × 109 ± 7.1 × 107
1620–2BV162/0-267.826. × 106 ± 7.4 × 1051.0 × 108 ± 5.4 × 1061.1 × 109 ± 9.2 × 107
1622–5BV162/2-591. × 106 ± 7.8 × 1052.0 × 107 ± 8.7 × 1053.2 × 108 ± 3.2 × 107
2030–7BV203/0-789. × 105 ± 9.2 × 1042.6 × 105 ± 4.7 × 1041.7 × 108 ± 1.1 × 107
2037–11BV203/7-1188. × 105 ± 1.5 × 1042.2 × 106 ± 3.2 × 1051.6 × 108 ± 5.3 × 106
20311–18BV203/11-1882.414. × 105 ± 2.7 × 1051.1 × 106 ± 3.8 × 1051.4 × 107 ± 4.3 × 105
4460–1BV446/0-187.511.11.40.921. × 105 ± 7.1 × 1041.1 × 105 ± 1.7 × 1041.0 × 109 ± 2.3 × 107
4461–6BV446/1-684. × 104 ± 1.8 × 1043.3 × 105 ± 1.3 × 1051.8 × 109 ± 2.6 × 107
4740–2BV474/0- × 106 ± 3.5 × 1051.2 × 107 ± 1.4 × 1061.5 × 108 ± 1.2 × 107
4742–6BV474/2-688. × 106 ± 2.7 × 1051.6 × 106 ± 3.1 × 1053.1 × 107 ± 6.9 × 106
3250–1BV325/SS91.< 0.05b5.4 × 105 ± 1.2 × 1053.5 × 105 ± 1.7 × 1042.0 × 109 ± 7.8 × 107
5000–1BV500/SS190.< 0.1b< 0.05b5.0 × 103 ± 1.8 × 1031.5 × 104 ± 1.0 × 1042.8 × 107 ± 7.7 × 105
5000–1BV500/SS287.610.91.50.580.2a6.6a1.00.19.2 × 105 ± 2.3 × 1051.0 × 105 ± 3.1 × 1043.4 × 109 ± 0.9 × 108

For molecular and microbiological analysis, the samples were collected in sterile 250-mL plastic boxes (Nalgene). Samples for geochemical and geophysical analysis were stored in plastic bags. All samples were transported at −25°C on research vessel ‘Polarstern’ from the Prydz Bay (Antarctica) to Bremerhaven (Germany). The harvested material was mainly composed of weathering debris and soil precursors. According to the definition of Bockheim and for simplicity, all materials in this study are referred to as ‘soil’ (Bockheim, 1982; Ugolini & Bockheim, 2008).

Conductivity and pH were measured in a soil extract (9 g soil in 45 mL distilled water) directly in the field laboratory. The soil slurry was filtered for pH value determination, whereas conductivity was measured directly in the extract. Moisture content was determined with weighing the moist and dry soil before and after freeze-drying of about 1 kg of soil. Total carbon and nitrogen contents were determined with an automatic element analyser (Elementar Vario EL III). Total organic carbon content was measured after HCl (10%) acid digestion on the analyser Elementar VarioMaxC. For trace element analyses, a soil slurry (5 g soil + 25 mL milli Q water) was mixed in an overhead shaker for 90 min and centrifuged at 3500 g for 20 min. Anions were measured with ion chromatography (Dionex-DX320), cations with inductively coupled plasma optical emission spectrometry (ICP-OES, Perkin Elmer Optima3000XL) and math formula with titration (Metrohm Titrino 794). Grain size distribution was determined as described by Biskaborn et al. (2012) and measured in a laser particle analyser (Coulter LS 200).

Cultivation and enumeration of culturable heterotrophs

The numbers of culturable heterotrophs were determined by plating serial soil solutions (up to 10−4) on R2A (Reasoner & Geldreich, 1985) and modified BRII medium (Bunt & Rovira, 1955; Ganzert et al., 2011c) as it was described before (Bajerski et al., 2013). Colonies were randomly chosen from all enrichment culture plates to obtain potentially new isolates. Microbial DNA extraction and the amplification of bacterial 16S rRNA genes were performed as described previously (Table S1; Bajerski et al., 2013).

Soil DNA extraction

Soil DNA was extracted in triplicates out of 0.5 g material each with PowerSoil Extraction Kit (MoBio Laboratories, Inc.) according to the manufacturer's protocol. The DNA triplicates were pooled for downstream analysis.

Bacterial 16S rRNA gene copy numbers

Bacterial 16S rRNA gene copy numbers were determined by qPCR to estimate bacterial abundance. A standard was generated out of pure culture, DNA was extracted as described above, and bacterial 16S rRNA genes were amplified using 0.5 μL of each primer Uni 331F and Uni 797R (10 μM, Table S1, Nadkarni et al., 2002), 12.5 μL Syber Green Polymerase Master Mix (2×, Qiagen, Hilden, Germany) and 1 μL DNA template adjusted to 25 μL with PCR clean water. The molarity (nM) of purified PCR products (HiYield PCR Clean-up Kit; SLG) was measured with Agilent 2100 Bioanalyzer (Agilent Biotechnologies, Böblingen, Germany) and multiplied by the Avogadro constant (6.022 × 1023 mol−1) to calculate gene copies μL−1. Quantitative PCR was performed as technical triplicates with primers 338F (Lane et al., 1985) and 518R (Table S1, Muyzer et al., 1995) in a Rotor Gene cycler (Qiagen) to quantify bacterial 16S rRNA gene copies (Fierer & Jackson, 2006). Each 25-μL reaction mixture contained 12.5 μL Syber Green (2×, Qiagen), 0.5 μL of each primer (20 μM), 9.5 μL PCR clean water and 3 μL DNA template in a 10-fold dilution. The inhibition of qPCR was tested and minimized using dilution series (Rasmussen, 2001). Standards and dilutions were measured every run.

Terminal restriction length polymorphism

A semi-nested PCR protocol (Table S1) was applied to enhance PCR product using universal bacterial primers 27F (Lane et al., 1985) and 1492R (reaction I, Dojka et al., 1998) and a 6-carboxyfluorescein-labelled primer FAM-27F and reverse primer 907R (reaction II, Muyzer et al., 1995) to amplify 16S rRNA genes. PCRs were performed as described elsewhere (Barbier et al., 2012). About 150 ng of purified PCR product was used in a 20-μL digestion reaction mixture with 10 U restriction enzyme AluI (New England Biolabs, Frankfurt a. M.) and 2.0 μL corresponding buffer. Amplicons were digested in duplicates at 37 °C for 3 h and the reaction was stopped with incubation at 65 °C for 20 min (UnoCylcer, VWR). Duplicate digestions were pooled and cleaned as described above and run on an ABI 3730xl DNA analyser (Applied Biosystems, Darmstadt, Germany) at GATC Biotech (Konstanz, Germany). GeneScan™ LIZ 500® (Applied Biosystems) was used as an internal size standard.

Data processing

Raw data were analysed with Peak Scanner Software 1.0 (Applied Biosystems), and output T-RF profiles were examined according to the five-step procedure of Dunbar et al. (2001). Within one sample, triplicate peaks within the range of 0.5 bp were aligned, but duplicate peaks were allowed and proceeded further as well. Only peaks above the fluorescence threshold of ≥ 25 fluorescence units were taken into account.

16S rRNA gene clone libraries

A 16S rRNA gene fragment was amplified according to PCR reaction I of the T-RFLP (Table S1) using Syber Green Polymerase Mix (2×, Qiagen). The purified PCR product (1 μL) was cloned using the pGEM®-T-Easy vector system (Promega, Mannheim, Germany) according to the manufacturer's protocol with ligation at 4 °C over night. Reamplification was carried out in a direct colony PCR of the positive clones with each 1.0 μL forward and reverse primers M13 (10 μM, Table S1, Messing et al., 1981) recognizing the vector region, 12.5 μL Mango Polymerase Mix (Bioline, Luckenwalde, Germany) and 1 μL template. Raw sequence data were processed with Sequencher (v4.7; Gene Codes, Ann Arbor, MI) and uploaded as FASTA files to Rdp classifier (Wang et al., 2007) to obtain an overview of the community composition. Cleaned sequences were aligned with Silva Aligner (Pruesse et al., 2007), and a distance matrix was built with Arb (Ludwig et al., 2004) as an infile for Dotur (Schloss & Handelsman, 2005) to determine unique phylotypes (≥ 97% sequence similarity) using nearest neighbour clustering algorithms. A chimera check of the sequences was made with Belerophon 3.0 tool of Greengenes (Huber et al., 2004; DeSantis et al., 2006). Dotur was used to calculate rarefaction curves, Shannon–Weaver (Shannon, 2001) and Simpson diversity (Simpson, 1949) indices and Chao1 richness estimates (Chao, 1984; Chao & Lee, 1992; Chao et al., 1993). GenBank accession numbers are JX171737JX172264 for sequences of the Glacier Transect and JX172265JX173055 for sequences of the Black Valley Transect.

Generated clone sequences were digested virtually with TRiFLe (Junier et al., 2008) to identify T-RFs using the same primer set, restriction enzyme and conditions as in the soil T-RFLP analysis.

Statistical analysis

Bacterial community composition was calculated statistically with Primer 6 (Primer-E Ltd, Luton, UK). Relative abundances of all T-RFs (prior to phylogenetic assignment) were fourth-root-transformed, and hierarchal cluster analysis was performed using Bray–Curtis similarity index. A nonmetric multidimensional scaling (NMDS) analysis was applied on the transformed data set.

A redundancy analysis (RDA) was calculated with canoco 4.5 (Ter Braak & Šmilauer, 2002). The geochemical and geophysical soil parameters (distance from the glacier, soil depth, grain size, moisture, carbon and nitrogen content, pH-value, conductivity and trace elements), colony and gene copy numbers built up the species data set. Identified T-RFs were added as a supplied environmental data set. Concentrations and T-RFs were log-transformed, and all data were centred and standardized. The analysis was performed separately for each transect.


Soil properties along two glacier forefield transects

Both glacier transects are characterized by a coarse grain size with a high contingent of sand and gravel (Tables 1 and 2). Soil texture was composed of over 80% sand and < 15% silt and 5% clay in almost all samples.

Although the overall soil moisture was very low, a gradient could be observed along the forefield (Glacier Transect) and in soil depth (both transects). Water content increased with increasing depth in the Glacier Transect, for example from 0.6% (GT0/0-7) in 0–7 cm, over 1.5% (GT0/7-14) in 7–14 cm and up to 2.4% (GT0/14-25) in 14–25 cm depth. It was driest in the upper layer of the most distant profile (0.2%, GT80/0-1) and moistest in the deeper soil closer to the glacier (3.7%, GT30/13-28). Water contents were slightly higher in the Black Valley Transect (up to 14.9%, BV162/2-5) and increased with depth, for example in profile BV203 from 0.6% to 4.5%. For the Glacier Transect, soils furthest from the glacier tended to be more acidic (pH 4.9 in sample GT80/10-20) than in the vicinity of the glacier (pH 8.3 in sample GT0/0-7), while the Black Valley Transect is characterized by slightly acidic to neutral pH values between pH 6.0 (BV474/0-2) and 6.9 (BV203/7-11). The carbon content was low, often below the detection limit (< 0.1, Glacier Transect), and reached a maximum of 0.4% (GT80/0-1) and 2.0% (BV162/0-2) in the surface samples of both transects. Trace elements were distributed heterogeneously in dependence of soil depth and location in the glacier forefield (Tables S2 and S3). Cations of trace elements of the Glacier Transect were increasingly observed from profile GT0 to GT65 and almost all elements reached a maximum in GT65/10-20. Elements of the Black Valley followed different depth trends, but did not shift along the transect.

Bacterial isolates and numbers of culturable heterotrophs

Heterotrophic bacteria were examined as part of the whole bacterial community. We identified 48 isolates out of the Glacier Transect and 100 isolates out of the Black Valley Transect. Representatives of the classes Actinobacteria, Alphaproteobacteria, Betaproteobacteria, Flavobacteria and Sphingobacteria were isolated from both transects. In addition, the Black Valley Transect included species belonging to Deinococci and Gammaproteobacteria. Sequence analysis indicated that there were potentially new microorganisms among the isolates (Table S4). The number of culturable heterotrophs ranged between 2.0 × 103 and 2.8 × 105 cfu g−1 dry soil in the Glacier Transect (Table 1) and 5.0 × 102 and 1.0 × 108 cfu g−1 dry soil in the Black Valley Transect (Table 2). In general, colony-forming units were higher on BRII than on R2A media, but isolates showed a higher visible diversity on R2A.

Abundance of bacteria

Bacterial abundance was displayed by the number of 16S rRNA gene copies per gram soil using qPCR. In comparison, bacterial numbers in the Black Valley Transect (1.4 × 07 to 3.4 × 109 gene copies g−1 soil, Table 2) were higher than in the Glacier Transect (3.5 × 106 to 7.6 × 108 gene copies g−1 soil, Table 1). Bacterial gene copy numbers in the Glacier Transect slightly increased with distance from the glacier, but decreased with depth. In the Black Valley Transect, no clear trend could be observed.

Bacterial community composition

Altogether, 79 different terminal restriction fragments(T-RFs) could be observed in the two different glacier forefields, whereby 25 T-RFs were present in both transects. In the Glacier Transect, 21 of 46 T-RFs were unique and in the Black Valley 33 of 58 T-RFs.

With in silico digestion of the sequences obtained from the clone libraries, we were able to identify 48 of 76 T-RFs, which were catalogued into 12 phylogenetic groups. Thirteen additional T-RFs could belong to several groups and are highlighted with patterns in the graph (Fig. 1).

Figure 1.

T-RF-based bacterial community composition after in silico identification. Full colours represent 48 of 76 T-RFs, which were catalogued clearly into 12 phylogenetic groups. Patterns show 13 additional T-RFs that could belong to several groups. Sequences were determined on the class level.

Seven samples were chosen for cloning based on the distribution and clustering of T-RFs (as analysed by NMDS, see results of the statistics below). The following clusters and groups refer to the similarities determined with Bray–Curtis cluster analysis. Within the cluster Glacier Transect (Fig. 2), the samples GT0/7-14, GT55/10-20 and GT80/10-20 were analysed via cloning. Altogether, 528 sequences, representing 245 operational taxonomic units (OTUs), were identified from clone libraries. Taking into account rarefaction curves, richness estimators and diversity indices (Table 3, Fig. S2), our study showed a high coverage of species richness of 90% (CChao) and 88% (CACE) in GT0/7-14, but did not completely cover estimated species richness in the clone libraries of GT55/10-20 and GT80/10-20.

Table 3. Characteristics (OTUs, species richness, diversity indexes) of individual and combined bacterial 16S rRNA gene clone libraries of selected samples of the Glacier Transect (GT0/7-14, GT55/10-20, GT80/10-20) and the Black Valley Transect (BV26/0-1, BV203/11-18, BV446/1-6, BV474/0-2)
SampleClonesOTUsa,bCoveragebRichness estimators (min–max 95% CI)bDiversity Index (min–max 95% CI)b
  1. a

    OTUs defined as sequences with 98% similarity.

  2. b

    Calculated with DOTUR according to confidence interval of 95% (CI).

GT0/7-1416768908876 (71–92)77 (71–92)4.08 (3.97–4.18)0.0132
GT55/10-20183975453180 (139–261)184 (144–260)4.30 (4.16–4.44)0.0144
GT80/10-20178806865117 (97–163)122 (101–164)4.15 (4.03–4.27)0.0140
BV26/0-12281503837390 (290–562)411 (305–57)4.78 (4.66–4.91)0.0082
BV203/11-18179747066105 (87–147)113 (93–152)4 (3.86–4.14)0.0198
BV446/1-62471166867171 (144–224)172 (148–216)4.55 (4.45–4.65)0.0087
BV474/0-213734958736 (34–44)39 (35–52)3.15 (3.00–3.31)0.0505
Figure 2.

Plot of the NMDS results, showing the similarities of the T-RFs including the T-RFs of all samples of the Glacier Transect (GT) and the Black Valley Transect (BV) prior to phylogenetic assignment.

Clone libraries of the Black Valley Transect were chosen according to the main clusters: group I (BV26/0-1), group III (203/11-18), group IV (BV 446/1-6) and group V (BV474/2-6). Here, 791 sequences generated 374 OTUs (Table 3, Fig. S2). Only 34 OTUs out of 137 clones cover 95% (CChao) or 87% (CACE) of the estimated species richness in BV474/0-2, whereas 150 OTUs out of 228 sequences of BV26/0-1 describe < 40% of the whole community. The profiles BV203/11-18 and BV446/1-6 showed a mean coverage of species richness of about 70%.

T-RFs and clone library sequences affiliated to the phyla Actinobacteria and Proteobacteria were identified in all profiles of the Glacier and Black Valley Transects (Figs 1 and 3). Unique Acidobacteria T-RFs were only detected in the profiles GT0 and BV474 and at low abundance in the surface sample BV500/SS1. Acidobacteria were present in all clone libraries of the Glacier Transect with similar relative abundances between 4% and 9% and with increasing abundances up to 22% of the sequences in the profiles BV203 and BV446 of the Black Valley Transect. Actinobacteria were increasingly observed with increasing distance from the glacier in the Glacier Transect and peak fluorescence represented over 50% in the deepest layer of the profiles GT55, GT65 and GT80. In the clone libraries, Actinobacteria were the most dominant group as well, making up to 51% of all samples. In the Black Valley Transect, Actinobacteria occurred in all samples with different relative abundances in T-RFLP profiles and clone libraries (Figs 1 and 3). At a distance of 474 m from the glacier, Actinobacteria were the most dominant group in the clone library, making up 66% of the whole community, whereas T-RFs represented only 13%. In the Glacier Transect clone libraries, Alphaproteobacteria were the dominant group among the Proteobacteria and their relative abundance increased from 2.4% (GT0/7-14) to 15% in the most distant profile (GT80/10-20). Corresponding Alphaproteobacteria-T-RFs could only be detected in GT0/14-25, GT80/0-1 and GT100/SS. Deltaproteobacteria-T-RFs were explicitly present in most profiles, but only appeared in the clone library of GT80/10-20. Proteobacteria were detected in all samples of the Black Valley Transect, except in the T-RFLP profile of BV26/0-1. Within the Proteobacteria, Gammaproteobacteria were the most abundant group making up a distinctive part in the bacterial communities of BV26/1-9, BV162/2-5, BV446 and BV474. Bacteroidetes were detected in all profiles of the Glacier Transect with mainly Sphingobacteria in the clone libraries (GT0, GT55) and Flavobacteria-related T-RFs (GT30, GT55, GT65, GT80). Clone library results show that Bacteroidetes occurred at highest abundance in the vicinity of the glaciers in both transects. They were decreasingly observed in the Glacier Transect and present at both ends of the Black Valley Transect, close to the glacial caps in each case. T-RFs affiliated to Gemmatimonadetes were widely distributed along the transects, more abundant in the deeper layers and became the dominant group in GT0/14-25. A trend implying increasing abundances along the Glacier Transect forefield was found in the clone libraries, but not with T-RFLP. T-RFLP results showed that Cyanobacteria and Chloroflexi were widely distributed all over the forefields as well, but in contrast to the Gemmatimonadetes, they were highly abundant in the surface samples. According to the clone library results, Cyanobacteria were present close to the glaciers and at high abundance in BV26/0-1.

Figure 3.

Relative abundance of phylotypes determined in 16S rRNA gene clone libraries in two glacier forefields on Larsemann Hills, East Antarctica. Number of sequences: a) Glacier Transect (GT) = 167 (GT0/7-14), 183 (GT55/10-20), 178 (GT80/10-10); b) Black Valley Transect (BV) = 228 (BV26/0-1), 179 (BV203/11-18), 247 (BV446/1-6), 137 (BV474/0-2).

T-RFLP profiling revealed a highly diverse and heterogeneous bacterial community of the glacier transects. Clone libraries along the Glacier Transect showed a very similar pattern, whereas the data for the Black Valley Transect revealed a highly variable composition, which indicated different states of development of the forefields. Overall, Actinobacteria, Acidobacteria and Proteobacteria were the dominant classes, but Bacteroidetes, Cyanobacteria and Chloroflexi took a distinct part in the community composition as well. Armatimonadetes, Planctomycetes and Verrucomicrobia could be detected in the clone libraries, but not as unique T-RFs. A trend in community development along both transects was observed for certain groups, but in general, the occurrence of bacteria followed a patchy distribution.

Statistical analysis of environmental and molecular data

T-RFs were compared in a Bray–Curtis similarity and NMDS analysis to identify bacterial community patterns and to select certain samples for cloning (Fig. 2). All T-RFs retrieved from the Glacier Transect, except fragments of GT100/SS and GT0/14-25, clustered together with at least 30% similarity. These T-RFs were clearly differentiated from those of the Black Valley Transect. Cluster analysis of the Black Valley Transect T-RFs created five distinct subgroups, connected to the location in the glacier forefield and soil depth. Groups I and II resembled surface and deeper soil samples, respectively, including the profiles BV26, BV162 and BV446. Group IV contained T-RFs of the profile BV446, which shared only 20% similarity to the fragments of the other samples. The T-RFs of BV446/1-6 were also connected to the soil depth group II. T-RFs retrieved from the profiles BV203 and BV474 were assigned location-dependently to groups III and V, respectively. The T-RFs of the surface samples GT100/SS and BV500/SS1 clustered together and were neither related to the cluster Glacier Transect nor to one of the Black Valley subgroups.

The RDA plot showed the distribution of the samples, the orientation and weights of the environmental parameters and the connection to certain microbial taxa (Fig. 4). The first two axes accounted for 38.5% (PC1) and 19.3% (PC2) of the total variance in the Glacier Transect and for 36.4% (PC1) and 21.4% (PC2) of the total variance in the Black Valley Transect. Samples resembling the Glacier Transect were distributed over all quadrants and were not characterized by a certain soil property. Only two clear reverse correlated clusters were formed: GT65 and GT0. Soils of the profile GT0 were negatively correlated with trace elements (Fig. 4a–c, Quadrant I), whereas GT65 depended on cations and anions (nitrate, phosphate and fluoride) of trace elements. A trend in certain environmental parameters could be observed along the glacier forefield. Soil moisture and pH increased with depth and decreased with distance to the glacier, whereas conductivity was inversely correlated and was best explained by sulphate, chloride and sodium concentrations. Bacterial gene copy numbers were higher in surface soils distant from the glacier.

Figure 4.

Ordination Plot of the RDA results with Canoco (Ter Braak & Šmilauer, 2002), showing the weights and orientation of samples, soil parameters and T-RFs on PC1 and PC2. Soil parameters and bacterial abundances build up the Variables Species Data Set and T-RFs, which were determined without ambiguity, the Supplied Environmental Data Set. One RDA was performed for each glacier transect. For clarity of presentation, the plot has been split into samples (a, d), Variables Species (b, e) and Supplied Environmental (c, f) Data Set. GT, Glacier Transect; BV, Black Valley; Pc, Principle component.

Samples of the Black Valley Transect were distributed heterogeneously, and the distance from the glacier played a minor role according to the RDA. Position BV474 and surface samples BV500/SS1 and BV203/0-7 were characterized by a coarse grain size (sand, Fig. 4d and e, Quadrant II). On the contrary, surface samples of BV446 and BV162 were influenced by silt, clay, trace elements and salts. Conductivity was linked to sulphate, chloride and sodium and decreased with depth, whereas soil moisture and pH were positively related to soil depth and increased, although only by a small amount. Samples in the deeper soils were characterized through water availability (Fig. 4d and e; Quadrant IV); surface samples depended on conductivity and anions (Fig. 4d and e, Quadrant I). Gemmatimonadetes increased with depth and were inversely related to Actinobacteria and unclassified bacteria, which all decreased with depth.

In both transects, Gemmatimonadetes, Gamma- and Deltaproteobacteria were increasingly observed with depth and were influenced by soil moisture and soil pH. The same trend was shown for Cyanobacteria in the Glacier Transect and Betaproteobacteria and Bacteroidetes in the Black Valley Transect. Actinobacteria were positively correlated with cations and anions of trace elements that were especially present in profile GT65 (Fig. 4d and e, Quadrant I) and in the surface samples of the Black Valley in Quadrant I (Fig. 4). Acidobacteria and Ktedonobacteria occurred together, but were not influenced by any particular soil parameter.


The two analysed glacier forefields of the Larsemann Hills, East Antarctica, represent areas of different bacterial succession. The results of this study give new insights how soil parameters impact the structure and development of glacier forefield bacterial communities.

The geochemical analysis reveals a deglaciation-dependent habitat formation with very slow primary succession that is still at its beginning, although the area has already been ice free since at least 40 ka BP (Hodgson et al., 2001). In contrast to the Antarctic Peninsula, which is characterized by cold-maritime climatic conditions (Vieira & Ramos, 2003), our study site is influenced by the harsh environmental conditions of continental Antarctica (ANARE, 2000). While initial soil formation processes such as humus accumulation, acidification or brownification were observed in maritime Antarctic terrestrial habitats (Bockheim, 2008; Ganzert et al., 2011a), none of those applied to the studied glacier forefields, which were characterized by very low soil moisture and small oligotrophic nutrient pools, comparable to the extreme Antarctic Dry Valleys (reviewed in Cary et al., 2010).

The Glacier Transect is directly influenced by the glacier tongue and deglaciation. Therefore, water content and pH decrease and conductivity increases along the forefield. Due to its characteristics, the Glacier Transect resembles a ‘classical glacier forefield’ with a temporary and areal shift as observed previously for pH, organic carbon, sulphate and water content (Noll & Wellinger, 2008).

In contrast to the Glacier Transect, the Black Valley Transect extends over a larger area and is influenced by glacier and snow caps from both sites. All parameters strongly depend on the position in the slope or outflow of the glacier forefield. A trend in soil moisture, pH and conductivity could be observed in soil depth, but not as a shift along the transect. The distribution of soil parameters is influenced by strong winds, snow melting processes or the mechanical movement and downwash of clay, silt and fine particles (Schütte et al., 2009), which lead to soil heterogeneity with no clear trend of the soil characteristics along glacier forefields (Lazzaro et al., 2009; Schütte et al., 2009). Furthermore, Antarctic habitat formation is known to be influenced by local microclimate effects and local-scale variability (Cannone et al., 2008). The Black Valley Transect is characterized by the black material in the outflow, which is suggested to be from biogenic origin because of higher carbon contents at the surface and the fact that several clones and T-RFs were affiliated to Cyanobacteria. Similar (black) algae mats have previously been reported for cold deserts (Alger et al., 1997). The higher soil moisture and abundance of microorganisms indicate further that the Black Valley Transect has gone through a longer development process, representing a slightly ‘matured’ glacier forefield.

We found that the habitat formation of the glacier forefields is influenced by deglaciation processes resulting in an areal shift of soil parameters especially in the ‘classical’ glacier forefield (Glacier Transect) and it is driven by the extreme Antarctic conditions, leading to a variable and prevalently depth-related distribution of soil parameters in the slightly ‘matured’ glacier forefield (Black Valley Transect).

Clone library analysis and T-RFLP profiling reveal a high diversity for all studied profiles. The dominant taxa in the glacier forefields are Actinobacteria, Acidobacteria, Proteobacteria, Bacteroidetes, Cyanobacteria and Chloroflexi. Furthermore, clone libraries and the NMDS analysis of T-RFs show a clear difference in the bacterial community composition of ‘classical’ and ‘matured’ glacier forefields. Communities of the ‘classical’ forefield are similarly composed. They are dominated by Actinobacteria (c. 50% of the clone libraries), supplemented by a forefield-dependent shifting community distribution of Acidobacteria, Bacteroidetes, Cyanobacteria, Chloroflexi and Proteobacteria. Communities of the ‘matured’ forefield show clearly different compositions at each location in the forefield indicating a specialized bacterial community in dependence on the ecological variations along this transect. Clone library results reveal a location-dependent Actinobacteria- or Cyanobacteria-dominated community and communities that have a balanced composition without a clear dominance of one group. Some groups show an orderly succession in connection to soil parameters, in agreement with what has been reported for alpine glacier forefields (Sigler et al., 2002; Noll & Wellinger, 2008).

Soil moisture, pH and (trace) elements impact certain microbial taxa and the glacier forefield development. The comparison of environmental and microbiological data shows that several taxa (Cyanobacteria, Gamma- and Deltaproteobacteria, Bacteroidetes, Gemmatimonadetes) depend on water availability in both transects; especially Cyanobacteria and Deltaproteobacteria of the Glacier Transect occur with increasing soil moisture, pH and depth and decrease with distance from the glacier. Nutrient and water limitation have a greater effect on the microbial community structure than changing temperature regimes as shown for dry mineral Antarctic soils (Wynn-Williams, 1996; Yergeau et al., 2012). Several of the identified clones, T-RFs or isolates were affiliated to phototrophic Cyanobacteria and Chloroflexi or chemolithotrophic/chemoorganotrophic Nitrosomonas (Betaproteobacteria), Nitrospira (Nitrospirae) and Rhizobiales (Alphaproteobacteria), which provide the basis for an efficient ecosystem because of their participation in primary production. Because the glacier forefields lack essential nutrients, primary producers fulfil the tasks of assimilating inorganic carbon or fixing atmospheric nitrogen to provide substrates for other microorganisms. Bacteroidetes were most abundant in the vicinity of the glaciers, where constant low temperatures and water availability are most likely to occur. Compared to distant locations in the forefield that are much drier and can reach considerable above-zero temperatures at the soil surface in Antarctic summer, Bacteroidetes, comprising several psychrophilic representatives, are well adapted to cold conditions at the glacier tongue (Shivaji et al., 1992; Bajerski et al., 2013). Bacteroidetes were detected in various cold-affected and poorly developed habitats (Liebner et al., 2008; Ganzert et al., 2011a) contributing to biological weathering by degrading polymers (Buckley & Schmidt, 2001) and producing extracellular enzymes such as lipases, proteases and phosphatases (Hirsch et al., 1998; Aislabie et al., 2006). In this way, they support initial soil formation, which is an important function in the development of glacier forefields. Gemmatimonadetes-affiliated T-RFs in our study were positively related to soil moisture, depth and pH, in contrast to previous studies that did not show significant correlations between Gemmatimonadetes (former candidate division BD) and soil organic matter content, inorganic N concentration or soil pH (Mummey & Stahl, 2003). Environmental sequences of the Gemmatimonadetes were found to be widespread in different habitats and occurred at different geographical locations, including Antarctic cryoconite holes (Christner et al., 2003) and various soils (Mummey & Stahl, 2003). Their phylogenetic divergence and broad dispersion hint to a diverse metabolic and functional potential, which can allow them to colonize several ecological niches in the glacier forefields. Their exact functions remain unclear as of yet only one isolate has been described (Zhang et al., 2003).

Trace elements (magnesium, calcium, potassium) and salts were found to influence the presence of Actinobacteria in both transects. So far, this heterogeneous group is not known to be dependent on a certain salt or nutrient pool, but trace elements play an important role for microbial turnover in general (e.g. electron donors or acceptors, cofactors). Actinobacteria, being common in all soils, are the dominant group of the glacier forefields (especially in the Glacier Transect), and they were isolated from cold-affected habitats before (Bajerski et al., 2011; Ganzert et al., 2011b). Their successful colonization of the glacier forefields may be caused by their ability to metabolize a wide range of substrates as sole carbon source. These organic matter turnover and accumulation is also important for the habitat development in the sense of soil formation processes and as a part in the carbon cycle in relation to other microorganisms. Furthermore, they may be adapted to the low temperatures of the study site by releasing simple carbon compounds as compatible solutes that protect the organisms against freezing (Wynn-Williams, 1996; Aislabie et al., 2006).

The oligotrophic conditions of the glacier forefields support the development of an Acidobacteria-related community, because they live under extremely low nutrient conditions with very slow metabolic growth rates and tolerate fluctuations in soil hydration (Ward et al., 2009). With this lifestyle, they are well adapted to extreme conditions of the glacier forefields and common in Antarctic terrestrial environments (Saul et al., 2005; Ganzert et al., 2011a), but their concrete function remains unknown due to the lack of isolates.

The comparison of environmental and biological data shows that soil parameters impact certain microbial phyla. There is a clear difference in the community composition of ‘classical’ and ‘matured’ glacier forefields, with an undifferentiated and a specialized bacterial community, respectively.

Comparing Antarctic sites and taking into account the poorly developed soils and severe conditions of the studied glacier forefields, a surprisingly high bacterial diversity and abundance was observed. Although the communities still remain undersampled in some cases (coverage of species richness estimators < 50%), high diversity indexes (Shannon–Weaver > 3) indicate a diverse bacterial community. Interestingly, the diversity at all sites was higher than in similar Antarctic cold desert soils (Smith et al., 2006) and comparable (although still higher) to the functional diversity of microbial communities of glacier forefields of the Antarctic Peninsula (Pessi et al., 2012). Therefore, glacier forefields are hotspots for microbial diversity among Antarctic soil communities.

In conclusion, we show that glacier forefields are promising model systems to study soil formation along microbial successions. The results of the two studied transects indicate furthermore that microbial successions such as growth/dominance of certain phyla or taxa are not necessarily coupled to soil formation if, for instance, carbon and water are limited. This finding supports the hypothesis that terrestrial ecosystems in the state of initial habitat formation are characterized by highly diverse but undifferentiated microbial communities, which preserve a broad range of genetic potentials. Low metabolic activity seems to be one important aspect to maintain these diverse communities.


The authors wish to thank the shipboard scientific party, particularly H.-W. Hubberten (Alfred Wegener Institute for Polar and Marine Research) and the crew of the expedition ANT-XXIII/9 in 2007 with RV Polarstern. Special thanks go to Lars Ganzert (Finnish Forest Research Institute) for successful field work, Oliver Burckhardt (German Research Centre for Geosciences, GFZ) for laboratory assistance and Béatrice Frank-Fahle (Helmholtz Zentrum München) and Mashal Alawi (German Research Centre for Geosciences, GFZ) for critical reading of the manuscript. This study was supported by the Deutsche Forschungsgemeinschaft (DFG) in the framework of the priority programme ‘Antarctic Research with Comparative Investigations in Arctic Ice Areas’ by a grant to D.W. (WA 1554/9).