Comparison of the microbial diversity at different depths of the GISP2 Greenland ice core in relationship to deposition climates


*E-mail; Tel. (+1) 814 865 3330; Fax (+1) 814 863 5304.


This study presents comparative geochemical, microbiological and molecular analyses of Greenland GISP2 ice core samples representing different depths, ages, deposition climates, in situ temperatures, and gas and ionic compositions. Our goal was to determine whether specific organisms, preserved at different depths, correlate with past climate characteristics recorded chronologically in ice layers. Three clear ice samples were selected from 2495, 2545 and 2578 m to represent distinct climatic periods with milder (−45°C), colder (−51°C) and warmer (−39°C) deposition temperatures, and two Marine Isotope Stages, MIS3 (2495 m) and MIS4 (2545 and 2578 m). Results showed higher microbial abundance in ice deposited during colder climates with higher in situ ion content. The constructed universal SSU rRNA gene clone libraries were dominated by Gram-positive sequences (55–65%), and had fewer Proteobacteria (69%) and Archaea (1%). The 2495 m library differed from the other two by being dominated by Actinobacteria (55%) rather than Firmicutes. Fungi were more prevalent in the colder climate (40%). For comparison, a library was constructed from an older silty ice sample (3044 m) possibly originating from underlying permafrost with different in situ characteristics (high temperature, high methane and higher cell numbers). It showed significantly different diversity not found in the clear ice libraries. The bacterial and fungal isolates from the clear ice samples were related to organisms originating from Asian deserts, marine aerosols and volcanic dust, suggesting these environments as sources of deposited microorganisms. The observed differences in microbial diversity patterns, especially with the 2495 m library, support the idea that local climate conditions and global atmospheric circulations at different time periods have influenced the origin and composition of the microbial populations preserved at different depths of Greenland ice. Further investigations may lead to the development of microbial ‘markers’ for identifying specific deposition climates.


Glacier ice is a unique habitat because it preserves microbial life and past climate records chronologically for hundreds of thousands of years. Microbial cells and other biological material, such as pollen gains, spores, plant debris and insects, associated with dust particles are propelled by the wind from neighbouring and distant places, deposited on the ice surface, and gradually, as snow accumulates, embedded into the deeper ice layers. The first microbial studies of deep glacier ice in Antarctica suggested that in-depth fluctuations in cell numbers were related to different dust concentrations and climate changes (Abyzov et al., 1998). The hypothesis that microbial populations in ice layers may reflect the prevalent climate events during deposition was recently explored using cores from the Tibetan plateau. Xiang and colleagues (2005) observed variable quantitative distributions of bacterial isolates obtained from a 15.3 m deep Muztag Ata glacier ice core. Zhang and colleagues (2006) found a positive correlation between concentrations of Ca2+, which serves as an indicator of dust, and higher bacterial diversity in the Puruogangri ice core (83.45 m) and concluded that more species are deposited during colder periods and fewer during warmer climates. A similar suggestion for an important role of deposition climate in the distribution of microorganisms was made for a 102 m long ice core retrieved from the Malan Ice Cap (Yao et al., 2006).

The Greenland ice sheet is the second largest on Earth, covering 1.7 million km2 and providing the longest ice core paleoclimatic records in the Northern hemisphere. The 3053 m deep ice core drilled during Greenland Ice Sheet Project 2 (GISP2) showed a sharp transition at 3040.34 m between the air borne clear ice and the silty ice, which is intermixed with the underlying permafrost. Detailed elemental and isotopic analyses of CH4, O2 and N2O in GISP2 ice have been used to reconstruct past atmospheric compositions and climatic changes for the last 110 000 years covering the last glacial period including the last glacial maximum (LGM) at 18–23 ka (Brook et al., 1996; Sowers, 2000). The GISP2 climate records during this period are characterized by repeated millennial scale rapid fluctuations between warmer and colder conditions (> 10°C difference) known as Dansgaard-Oeschger (D-O) events (Johnsen et al., 1992; Dansgaard et al., 1993; Grousset et al., 2000). The colder periods (denoted stadials) tend to have higher dust concentrations compared with the warmer (interstadial) periods (Mayewski et al., 1997). For example, during the LGM Greenland received ∼30 times more dust than it does today. The aerial dispersal of dust across the globe is estimated to be 2.2 × 109 tons per year (Goudie and Middleton, 2001) with 1 g of soil containing approximately 1 × 109 cells (Whitman et al., 1998).

Past atmospheric studies used dust in ice cores (which is presumably the major transporter of microbial cells) as a tracer of global atmospheric circulations (Kang et al., 2003; Grousset and Biscaye, 2005). On the basis of clay mineralogy and Sr, Nd and Pb isotopic ratios it was suggested that the eastern Asian deserts are the main source of dust deposited over Greenland (Biscaye et al., 1997; Svensson et al., 2000; Bory et al., 2003). Moreover, oceanic air masses deposit marine aerosols (Fischer et al., 2007; Hutterli et al., 2007) along with the ubiquitous transport of volcanically derived aerosols (Zielinski et al., 1996; Zdanowicz et al., 1999) and some extraterrestrial dust accretions (Karner et al., 2003). If certain types of microorganisms found at different depths can be linked with specific source regions, then microbial markers could be added as a new tracer in helping to identify the dust origin during specific time periods.

Another important aspect of trace gas composition studies is that the sample-to-sample variability detected in ice cores from Greenland and Antarctica may be caused by in situ microbial activity (Sowers, 2001). The possibility that microorganisms may alter gases trapped in glacial ice intensifies the need to determine the microbial distribution, diversity and activity throughout the glacier in order to relate certain microbial organisms and processes with the extensive geochemical information.

Our long-term goal is to create a comprehensive picture of the microbial populations in the GISP2 ice by studying core samples representing different depths, ages, deposition climates, in situ temperatures, and gas and ionic compositions. A major aim is to determine whether specific organisms correlate with deposition climate characteristics, global origin or specific ice core gas anomalies. Unfortunately, addressing this aim is difficult due to the complexity of the factors contributing to the apparent microbial diversity, the limited ice core samples and the inherent biases of the methods for studying microbial diversity. To help circumvent these problems and gain comparative insight, we applied exactly the same methodological procedures to different ice core samples.

Here we examined the microbial distribution in over 30 samples throughout the GISP2 core and present geochemical, microbiological and molecular data from three Greenland ice core (GIC) samples from 2495, 2545 and 2578 m representing milder (−45°C), colder (−51°C) and warmer (−39°C) deposition temperatures respectively. The microbial diversity in these clear ice samples was also compared with results from a 3044 m deep silty ice sample that may have originated from the underlying permafrost millions of years ago. We applied both culture-independent and culture-dependent approaches to these samples to obtain a comprehensive picture of microbial diversity rather than to compare these approaches and their well-known biases and discrepancies. We found that the clear ice deposited during the colder climate had a greater microbial abundance, which corresponded to higher in situ ion content. The phylogenetic diversity found in the three clear ice samples differed from each other and from the diversity in the silty ice sample. These results indicate that the varying climate conditions have influenced microbial diversity and underscore the need for further investigations that may lead to the development of microbial ‘markers’ for specific deposition climates.

Results and discussion

Gas analyses and microbial distribution in GISP2 ice samples from various depths

Trace gas records from ice cores provide a unique means of reconstructing the composition of past atmospheres that characterize past climates. Coupling gas and microbial abundance measurements was important to assess whether the anomalous trace gas concentrations might result from in situ microbial activity (Flückiger et al., 2002; 2004; Sowers et al., 2003; Tung et al., 2006). Analyses were performed on the gases (N2O, CH4, O2) trapped in 31 GISP2 ice core samples, including clear ice above 3040 m and silty ice below 3040 m. The elemental and isotopic trace gas measurements of the clear ice samples were consistent with previous data from the GISP2 and Taylor Dome ice cores (Bender et al., 1994; 1995; Brook et al., 1996; Sowers et al., 2003). Analyses of five silty ice samples showed extremely high CH4 concentrations (1–14%) with highly depleted 13C/12C ratios (−74‰ to −79‰), suggesting that these values likely resulted from in situ microbial CH4 production. Interestingly, the concentrations of N2O in the basal ice were generally only slightly (< 100%) higher than atmospheric values indicating minimal in situ N2O production.

The vertical distribution of microbial abundance was measured in these same 31 ice core samples by our previously established flow cytometry protocols (Miteva et al., 2004). Results showed varying cell concentrations throughout the 3053 m deep GISP2 (Fig. 1). Cell numbers in most clear ice samples were lower (103−105 ml−1) compared with the silty ice (> 107 ml−1). The available continuous major ion composition data (Mayewski et al., 1993; 1997) provided detailed information for comparisons. Interestingly, the higher peaks in cell numbers, exceeding 106 per ml, detected in samples from 2021, 2545, 2954 and 3039 m, corresponded to higher Ca2+ concentrations. The cell enumeration of five melted silty ice samples (3042, 3044, 3046, 3048 and 3052 m) showed even greater cell abundance (107 to over 108 cells ml−1), which complements our previous results with a sample taken between 3042.67 and 3042.80 m (Sheridan et al., 2003).

Figure 1.

In-depth microbial abundance in 31 GISP2 ice core samples in relation to Ca2+ concentrations (retrieved from (Mayewski et al., 1997), and in situ and deposition temperatures (Cuffey and Clow, 1997). Symbols: black dots (cell numbers), opened squares (Ca2+ concentrations (ppb, parts per billion), vertical double lines (MIS, marine isotope stages). The grey area marks the three clear ice samples (2495, 2545 and 2578 m).
m.b.s., meters below surface.

The fluctuations of microbial cell counts (103−105 ml−1) found in the present study for the shallow (111–150 m) and deeper samples (1000–3039 m) of the clear ice were consistent with the limited published Greenland ice measurements. Yung and colleagues (2007) found 7.0 × 103 total cells per ml at 94 m depth in GISP2, while Tung and colleagues (2005) detected between 104 and 105 cells per ml in several deeper clear ice samples. Our observation of peaks of cell numbers in several clear ice samples that correlated with Ca2+ concentration data (Ca2+ is a proxy for dust) can be explained by more cells attached to a higher amount of airborne dust particles during deposition. Thus, our results are consistent with those found for other glaciers where higher bacterial input was related to more dust particles carried via more intensive atmospheric circulation (Abyzov et al., 1998; Xiang et al., 2005; Zhang et al., 2006).

Geochemical rationale for sample selection

Following the analysis of the quantitative microbial distribution throughout the GISP2 core, we addressed the question of whether any differences in the microbial abundance and diversity at different depths are related to specific past climate characteristics. Based on the available data on deposition temperatures and chemical composition of the ice, we selected three ice core samples from 2495, 2545 and 2578 m that were deposited under different climate conditions 57 000, 63 200 and 68 000 years ago during the early part of the last glacial period (Fig. 1, Table 1). The samples were in close proximity (within 84 m) and had similar in situ temperature (−21°C) so microorganisms endured similar conditions and times in the ice matrix. The major features distinguishing these samples were the fluctuating depositional climate conditions (milder, −45°C, colder, −51°C, and warmer, −39°C respectively), and the chemical composition of the precipitation. The deposition temperatures of the three clear ice samples reflect the established millennial scale climate variations that may have influenced the intensity of aerosol fluxes over Greenland. The GIC 2495 m sample was deposited during the start of an abrupt warming period, while GIC 2545 m and GIC 2578 m samples represent the preceding cycle of gradual cooling ending with a typical stadial stage of very low temperature for GIC 2545 m ice deposited 63 000 years ago (Fig. 1). Sample GIC 2545 m also contained the highest Ca2+ and generally the highest salt concentrations among the clear ice samples, indicating a higher overall dust concentration.

Table 1.  Temperature and chemical composition (μg l−1) characteristics of GISP2 ice samples.
Core depth
Top-bottom (m.b.s.)
Deposition T
in situ
T (°C)
  1. Data for the chemical composition from Mayewski and colleagues (1997).

  2. The chemical composition of sample 3044 m was measured at the Water Quality Laboratory, Penn State Institute of the Environment.

  3. NA, not applicable; m.b.s., metres below surface.

2495.4–2495.62−45 (milder)−215744.92.23.919.6234.078.298.1196.6
2545.0–2545.25−51 (colder)−2163.
2578.5–2578.75−39 (warmer)−216814.43.851.57.543.930.556.8117
3044.24–3044.4NA−9> 110300079040046408600348056.07900

Thus, we hypothesized that differences in local climate conditions during deposition plus the intensity and dynamics of atmospheric circulations during transport from the major dust source in Asia would affect the microbial abundance, diversity and survival of specific trapped organisms. In order to test this hypothesis, we compared these three clear ice samples by applying similar protocols for ice core sampling, gas measurements, cell enumeration and molecular analyses. For an additional comparison, we also analysed a much older sample from the silty ice (3044 m) with a higher in situ temperature (−9°C) that possibly originated from the underlying permafrost millions of years ago and had different in situ characteristics. In contrast to the clear ice, this sample had significantly higher ion concentrations (Table 1).

Gas analyses of selected ice core samples

We analysed the elemental and isotopic composition of trapped gases in the four selected samples to assess whether there was any indication of in situ production of CH4 and N2O associated with the microbial populations at each depth. It is important to note that natural variations in these trace gas species have been found to correlate with different climates. In general, lower trace gas concentrations are observed during colder climate and higher concentrations are characteristic of warmer conditions. Results from the three clear ice samples deposited during different climatic periods (Table 2) were similar to other contemporaneous ice samples from both the GISP2 and Antarctic ice cores (Sowers et al., 2003; Spahni et al., 2003) and showed no anomalies.

Table 2.  Gas and isotope analyses of selected GISP2 clear and silty ice sample.
Ice core
depth (m.b.s.)
δ13C of
δ15N of
δ18O of N2O
  1. m.b.s., metres below surface; ppm, parts per million; ppb, parts per billion.

3044.2811 600−76.364127.037.6

In stark contrast to the clear ice samples, the gas composition of the silty ice sample clearly illustrated in situ CH4 (and possibly N2O) production with methane concentration more than four orders of magnitude higher than the clear ice samples (Table 2). Additionally, the isotopic composition of the CH4 in the silty ice sample was much lower than the clear ice samples. Because biogenic CH4 is isotopically depleted relative to the atmosphere, the low δ13CH4 and δDCH4 values are consistent with our suggestion that in situ CH4 production occurred within the silty ice. However, we cannot exclude the possibility that some (or all) of the elevated CH4 in the silty ice originated in the permafrost soils in central Greenland before the present ice sheet began accumulating. The N2O concentration in the 3044 m silty ice was ∼50% higher than the clear ice samples and ∼30% higher than present day values (∼321 ppb) (Table 2). The N2O isotope data from this single silty ice sample were not, however, measurably different from preanthropogenic or glacial/interglacial values (Sowers et al., 2003).

Microbiological characterization of selected ice core samples

Following decontamination the three clear ice samples and the silty ice sample were subjected to detailed microbiological and molecular analysis. The flow cytometric analysis of the ice samples stained with SYTO 13 or the LIVE/DEAD kit allowed estimation of total, live, and ultrasmall cells including bacteria, fungi and spores. Results (Fig. 2) showed that: (i) GIC 2545 m, deposited during the coldest climate, had the highest number of cells, 7.1 × 105 ml−1, but relatively fewer live cells (2.5%); (ii) GIC 2578 m, deposited during the warmer conditions had a lower cell number (3.6 × 105 ml−1) but the highest percentage of live (6.3%) and 68% small cells (< 1 μm); (iii) the enumeration values for GIC 2495 m, deposited during the warming period from stadial to a milder interstadial climate, were intermediate. In contrast to the clear ice samples, the silty ice sample had nearly two orders of magnitude higher cell numbers (1.3 × 107 ml−1) with 6.1% live cells. The clear ice results for the microbial abundance, viability and cell size distribution are consistent with the trend found in other ice cores where higher cell numbers appear related to more active atmospheric circulation during stadial periods resulting in higher dust/cell deposits. The role of mineral particles in protecting microbial cells from extreme conditions, providing attachment surfaces, nutrients and ions, and influencing cell physiology and survival, has been acknowledged (Priscu et al., 1999; Christner et al., 2000; Zhang et al., 2006) but is still far from clear. In addition to the atmospheric factors before deposition (desiccation, UV irradiation), the in situ ice characteristics impact its physical structure and the liquid veins between ice crystals, which can be the major habitats for microbial cells (Price, 2000).

Figure 2.

Microbial abundance of total (white dotted bars), live (white bars) and small cells (black dotted bars) in three GISP2 clear ice samples (GIC 2495 m, GIC 2545 m and GIC 2578 m) and one silty ice sample (GIC 3044 m), estimated by flow cytometry.

Decontamination efficiency and DNA yields

We conducted a variety of tests to demonstrate that our sampling and decontamination procedures efficiently eliminated external contamination while preserving the in situ microbial diversity. One test included analysis of the total cell number and the proportion of live cells in the outside melt water of the original ice core and in each of the 50 ml melt fractions collected after decontamination. As expected, the number of total and live cells in the outer ice core layer was slightly higher (e.g. 2.3 × 106 ml−1 versus 7.1 × 105 ml−1 the inner fractions for GIC 2545 m sample). However, following the hypochlorite treatment, all melted fractions of the same sample had very similar total cell numbers and non-significant fluctuations in the quantity of live cells.

The choice of methods for our comparative molecular diversity analyses was based on the microbial abundance estimates and preliminary experiments. Our optimized DNA extraction protocol still only provided DNA equivalent to 2.5–6.2% of the total possible based-on estimates from the total number of cells and an assumed per-cell DNA content of 1.6–11.4 fg (Bakken and Olsen, 1987; Sandaa et al., 1998). The amounts were particularly low for the clear ice samples, which were of limited volume and had relatively low microbial content (3–7 × 105 cells ml−1). Because test PCR experiments produced little or no products even after two rounds of amplification, we applied whole-genome amplification (WGA) using REPLI-g to increase the amount of DNA for cloning. Preliminary extensive WGA experiments demonstrated successful and accurate representation with as low as 0.1 pg starting DNA. Evaluation of the possible biases of this method included cloning PCR amplified Archaeal 16S rDNA performed with original and WGA amplified DNA from the silty ice and showing similar diversity (data not shown). The application of the universal PCR primer set 515F-1492R had the advantage of amplifying SSU rRNA genes from all major domains, Bacteria, Archaea and Eukarya. The possible biases for either WGA and PCR of not equally amplifying representatives of the different microbial groups would be consistent for all samples. Although such biases cannot be excluded, the use of the same protocols throughout made these comparisons possible.

Comparative analyses of small subunit (SSU) rRNA gene clone libraries

We constructed four SSU rRNA gene libraries from the clear and silty ice samples designated as K (GIC 2495 m), L (GIC 2545 m), M (GIC 2578 m) and J (GIC 3044 m) respectively. A total of 420 clones (67 for K, 119 for L, 90 for M and 144 for J) of SSU rRNA genes amplified with universal 515F-1492R primers were grouped into operational taxonomic units (OTUs) by ARDRA (amplified ribosomal DNA restriction analysis). Between 6 and 7 OTUs were found per library and all clones with identical ARDRA patterns had highly similar sequences. Analyses showed the presence of bacterial, eukaryotic and a few archaeal sequences in the ice core samples. The relative distribution of phylogenetic group varied in different clone libraries (Fig. 3). The library from GIC 2495 m (−45°C at deposition) had the highest proportion of actinobacterial sequences (55%), followed by eukaryotic sequences (33%) and Proteobacteria (9%). The library from GIC 2545 m, deposited in the coldest climate, showed a very different picture with two major groups, the Firmicutes (54%) and Eukaryota (40%), and a few Proteobacteria and Actinobacteria. Interestingly, Firmicutes also dominated (65%) the third clear ice clone library from GIC 2578 m, deposited in the warmest conditions, with lower representation of Eukaryota, Actinobacteria and Proteobacteria at 19%, 11% and 6% respectively.

Figure 3.

Distribution of SSU rRNA gene clones representing major phylogenetic groups in libraries obtained from GISP2 clear ice samples and one silty ice sample designated K (GIC 2495 m), L (GIC 2545 m), M (GIC 2578 m) and J (GIC 3044 m) respectively.

In contrast to the domination by Gram-positive organisms in the clear ice libraries, the clone library obtained from the 3044 m deep silty ice sample using the same SSU rRNA universal primers showed significantly different diversity dominated by Gammaproteobacteria (47%). Firmicutes and Actinobacteria together were represented by only 19% of these clones along with a significant group of eukaryotic sequences (32%). Unfortunately, the primers used did not amplify many archaeal SSU rRNAs but the few sequences that were found were related to the psychrophilic methanogenic species Methanococcoides burtonii and M. alaskense. The same species were detected in an Archaeal clone library (manuscript in preparation), which strongly supports the idea for in situ production of methane.

The phylogenetic relationships between sequences representing the major groups were examined in more detail. The combined phylogenetic tree for the three clear ice libraries (Fig. 4) showed both similarities and differences between clones from libraries K (GIC 2495 m), L (GIC 2545 m) and M (GIC 2578 m) respectively. The low G+C Firmicutes sequences that were highly represented in two of the clear ice libraries, GIC 2545 m and GIC 2578 m, had similar sequences which grouped with two sporeforming species, Bacillus muralis and B. asahii, as well as with several Bacillus glacial isolates. At least three clones, L45, M59 and M13, may represent novel taxa showing distant relationships (< 95%) to halophilic and alkaliphilic marine isolates belonging to B. koreensis and B. psychrosaccharolyticus respectively.

Figure 4.

Phylogenetic relationships of bacterial, archaeal and eukaryotic SSU rRNA gene clones identified from three GISP2 clear ice samples designated K (GIC 2495 m), L (GIC 2545 m) and M (GIC 2578 m) respectively. The tree is based on distance analysis (neighbour-joining algorithm with Juke-Cantor model). Bootstrap values greater than 50% generated from 1000 replicates are shown above the nodes. Accession numbers for closely related environmental sequences and cultured representatives retrieved from databases are included.

Sequences related to those from high G+C Gram-positive organisms were found in all three libraries with dominant representation among clones K from GIC 2495 m. As seen from the tree (Fig. 4), the actinobacterial diversity was low, restricted mostly to the genus Propionibacterium and one clone related to Pseudonocardia petrolophila.

Proteobacterial sequences comprised a relatively low portion in the clear ice clone libraries. These included representatives of three of the six known groups of Proteobacteria: Alpha, Beta and Gamma, with high similarity to the following genera: Methylobacterium, Ralstonia, Pseudomonas, Citrobacter. Several bacteria from frozen environments were among the closest relatives. One clone, L36, was significantly different from any sequence in the database with < 94% similarity to Alphaproteobacteria from air and a closest validated relative Sphingomonas aquatilis, isolated from mineral water at 93%. Finally, a significant number of sequences related to fungi (Acremonium) were detected in all clear ice libraries (Fig. 4).

The phylogenetic tree for clones from the silty GIC 3044 m sample clearly contrasts with the results from the clear ice samples (Fig. 5). The Firmicutes sequences from the silty ice clone library (J) formed a cluster, related to the non-sporeforming carnobacterium Atopostipes suicloacalis and the alkaliphilic marine bacterium Marinilactobacillus psychrotolerans. Similarly, all actinobacterial sequences formed a separate cluster, most closely related to Brevibacterium, which was not found in the clear ice samples. Surprisingly, Gammaproteobacteria were dominant in this clone library and formed a large tight cluster related to the genus Stenotrophomonas. The eukaryotic sequences also differed from those in the clear ice libraries. Phylogenetically they were mostly related to plants, unlike the fungal sequences found in the clear ice samples.

Figure 5.

Phylogenetic relationships of bacterial, archaeal and eukaryotic SSU rRNA gene clones identified from a silty ice GISP2 sample J (GIC 3044 m). The tree is based on distance analysis (neighbour-joining algorithm with Juke-Cantor model). Bootstrap values greater than 50% generated from 1000 replicates are shown above the nodes. Accession numbers for closely related environmental sequences and cultured representatives retrieved from databases are included.

Statistical analysis of the clone libraries

The biodiversity indices (Table 3) were calculated for the four clone libraries. The Shannon indices of diversity were relatively low compared with other glacial environments (Mikucki and Priscu, 2007). Diversity was higher in the silty ice sample and lower in the clear ice clone libraries that could be due to biases introduced by DNA extraction from low biomass samples, PCR procedures, or the WGA. However, because we obtained these libraries by consistently applying the same procedures, the observed phylogenetic differences are likely related to environmental factors such as different deposition climates and in situ parameters. The frequency of ARDRA patterns was used to calculate the coverage of the library. Results showed that the major part of the microbial diversity originating from the extracted ice core DNA was detected with higher coverage for the three clear ice libraries, 94%, 96% and 95%, respectively, and relatively lower representation for the silty ice sample of 87%. This broader coverage may be also due to the use of universal primers that allow amplification of SSU rRNA genes from all three major domains.

Table 3.  Phylotype richness, diversity and evenness values for Greenland ice core microbial populations at different depths.
IndexaSSU rDNA clone libraries
GIC 2495 mGIC 2545 mGIC 2578 mGIC 3044 m
  • a. 

    Indices were calculated according to Hill and colleagues (2003).

  • b. 

    Phylotype richness, S is the number of OTUs (distinct ARDRA patterns).

  • c. 

    Shannon index, H is calculated from the proportion of clones in i-th OTU to the total number of clones (pi): H = −Σpi ln pi.

  • d. 

    Evenness, E was calculated as E = H ln−1 S.

  • e. 

    Simpson's index, D = Σpi.


Recovery of isolates

Results from direct cultivation of microorganisms from the three clear ice samples on three agar media after aerobic and anaerobic incubation of plates at 18°C and 5°C showed low isolate recovery. Despite the indication for live cells, microbial culturability at 18°C estimated as the total number of colonies relative to live cell numbers was in the range of 2–5.6%. The majority of colonies recovered (> 103 ml−1) were related to fungi. Bacterial colonies grew aerobically on R2A and TSA at 18°C after 70 days. Microbiological characteristics of the bacterial isolates showed pigment formation by all isolates except the Bacillus strains, wide range of growth temperatures and tolerance to high salt concentrations. Only two isolates, KI-2 and KI-3, did not grow in media with 5% and 7% NaCl, but they grew with 2% NaCl in the medium. Four isolates (KI-3i, KI-4y, LI-1y, KI-6) grew in the presence of 12% and 20% NaCl usually used for selection of true halophiles.

Analysis of the 16S rDNA sequences showed that bacterial isolates were related to Alpha- and Gammaproteobacteria (Methylobacterium, Paracoccus, Pseudomonas), Actinobacteria (Sanguibacter, Micrococcus, Brevibacterium, Microbacterium, Brachybacterium) and Firmicutes (Bacillus) (Fig. 6). Interestingly, the highest number of isolates, representing all above-mentioned groups, was obtained from the GIC 2495 m sample deposited in the intermediate milder climate. Only Actinobacteria were cultivated from the GIC 2545 m sample and they were related to Brachybacterium and Brevibacterium. Finally, the two bacterial isolates obtained from the GIC 2578 m sample with lowest microbial content were related to a sporeformer from the Bacillus cereus group and a Microbacterium organism from Culex mosquito gut respectively. It is worth noting that microbial species previously reported from glacial ice and permafrost were among the closest relatives of our isolates. Other close relatives originated from aerosols, deep-sea sediments, volcanic soils or cold Asian deserts.

Figure 6.

Phylogenetic relationships of 16S rRNA genes of bacterial isolates obtained from three GISP2 clear ice samples designated KI (GIC 2495 m), LI (GIC 2545 m) and MI (GIC 2578 m) respectively (bolded). The tree is based on distance analysis (neighbour-joining algorithm with Juke-Cantor model). Bootstrap values greater than 50% generated from 1000 replicates are shown above the nodes. Accession numbers and origins of closely related environmental cultured representatives retrieved from databases are included.

A surprisingly high number of fungi were cultivated from the clear ice samples both at 18°C and 5°C on heterotrophic media not specifically designed for isolating fungi. More fungal colonies grew on 1× and 0.1× R2A than on TSA mostly after incubation at 18°C. The highest number of fungi (1 × 104 ml−1) was recovered from GIC 2545 m sample consistent with our results showing 40% fungi-related clone sequences. Interestingly, the recovery of fungi from sample GIC 2495 m was very low (50 colonies per ml) at 18°C while a much higher number of fungal colonies grew on the same media at 5°C. Distinguished morphological types were further studied by SSU rRNA gene amplification with the universal primers 515F-1492R from extracted DNA of 16 pure cultures. Sequence analyses of PCR products representing 10 ARDRA groups (Fig. 7) identified isolates that were related to Ascomycota fungi Phaeosphaeriopsis obtusispora, Geomyces pannorum, Cryptospotiopsis radicicola and several Penicillium species, including P. expansum, previously found in Antarctic and Greenland glacial ice (Abyzov, 1993; Ma et al., 2000; Gunde-Cimerman et al., 2003; Ma et al., 2005; Frisvad, 2008).

Figure 7.

Phylogenetic relationships of SSU rRNA genes of fungal isolates obtained from three GISP2 clear ice samples designated Kf (GIC 2495 m), Lf (GIC 2545 m) and Mf (GIC 2578 m) respectively (bolded). The tree is based on distance analysis (neighbour-joining algorithm with Juke-Cantor model). Bootstrap values greater than 50% generated from 1000 replicates are shown above the nodes. Accession numbers and origins of closely related environmental cultured representatives retrieved from databases are included.

Diversity and long-term survival of microbial populations

We included the culture-dependent approach in our diversity study to verify cell culturability and to see if we could obtain isolates related to those predicted with the clone libraries. Several similar 16S rRNA gene sequences were found both among clones and isolates such as Methylobacterium-related clone K20 and isolate KI-2, Pseudomomas-related clone K27 and isolate KI-3i, and Brevibacterium-related clones J54, J86 and isolate LI-1y. In general, the detected bacterial groups of high and low G+C Gram positives and Proteobacteria in the studied GISP2 samples were similar to those found in our previous studies of deep Greenland ice and by other authors in Arctic and Antarctic glacial ice (Christner, 2002; Christner et al., 2003; 2005; Miteva et al., 2004; Xiang et al., 2004; 2005). These included the genera Methylobacterium, Sphingomonas, Citrobacter, Stenotrophomonas, Pseudomonas, Bacillus, Marinilactibacillus, Brevibacterium, Micrococcus, Sanguibacter and Brachybacterium. The consistency of identifying isolates and clone sequences belonging to similar genera has been noted and explained mostly by the ubiquitous distribution of certain species in geographically distant icy environments and possessing similar adaptation strategies allowing them to survive for long times under extreme conditions (Priscu and Christner, 2004; Christner et al., 2008; Miteva, 2008).

An important finding of this work was the abundance of Bacillus sequences in two of the clone libraries L (54%) and M (65%) from samples deposited in colder and warmer climates, respectively, and their total absence in the third library K, deposited at an intermediate climate that was dominated by Actinobacteria (although two Bacillus isolates, KI-6 and MI-1, were obtained). Several studies found few sporeformers and significantly more Actinobacteria among isolates (Christner et al., 2003; Miteva et al., 2004; Mosier et al., 2007). Johnson and colleagues (2007) suggested that high G+C Actinobacteria possess mechanisms allowing them to outlast other microbial organisms. Bacilli have been reported as the most numerous isolates from the Malan glacier, Tibet (Yao et al., 2003), and we also recovered more than 400 Bacillus isolates from the deep Greenland ice after applying a selective filtration–incubation procedure that may have enhanced spore recovery (Miteva and Brenchley, 2005).

The significant portion of fungal sequences (19–40%) in our clone libraries, mostly related to Acremonium, shows that their spores and/or DNA can be preserved. In addition, the large number of fungi cultivated directly from the three clear ice samples provides new evidence that these lower eukaryotes or their spores remain viable in glacial ice for long times. Many fungi, including species related to our isolates, have been previously isolated from Antarctic and Greenland ice (Abyzov, 1993; Ma et al., 2000; 2005; Gunde-Cimerman et al., 2003; Butinar et al., 2007; Turchetti et al., 2008).

Linking microbial diversity and climate

Most studies, including ours, suggest that the quantitative distribution of microbial cells and other aeolian material in glacial ice is influenced by local and global climate events with more dust particles and cells dispersed over greater distances during colder climatic periods when winds are generally more intense (Abyzov, 1993; Priscu and Christner, 2004). Here, we also explored the intriguing possibility that the prevalent climate events during transport and deposition may have altered the microbial composition and diversity. The varying phylogenetic diversity of microbial sequences and isolates revealed in the three GISP2 clear ice samples suggests that different microorganisms or spores were deposited during the corresponding time period and some survived buried deeper in the ice for more than 60 000 years. Because these samples were deposited under different climate conditions, we addressed the question: Do specific microorganisms or microbial groups correlate with a particular local deposition climate?

First, the deposition times of the three clear ice samples GIC 2495 m, GIC 2545 m and GIC 2578 m differ by 5000 years and represent different stages of the millennial scale local climate fluctuations, milder, −45°C, colder, −51°C, and warmer, −39°C respectively (Fig. 1). Specifically, the aerosol fluxes in Greenland during the colder periods (GIC 2545 m) had higher atmospheric aridity and/or enhanced wind intensity causing elevated aerosol depositional fluxes, whereas the abrupt shifts to warmer and moist periods (GIC 2495 m) resulted in less intense aerosol fluxes with lower dust content. In addition, the ice from the GIC 2495 m represents the start of Marine Isotope Stage 3 (MIS 3) lasting between 30 and 60 ka, while the other two deeper samples have been deposited during MIS 4, which lasted between 60 and 80 ka. The climate during the transition between MIS 4 and MIS 3 (about 60 000 years ago) was characterized by gradual warming with several rapid climate shifts and gradually rising sea levels (Chappell, 2002; Landais et al., 2004; Huber et al., 2006). Altogether, these events caused diverse local and global atmospheric circulation patterns leading to variable content of sea-salt and mineral dust aerosols that likely influenced the snow chemistry and microbial composition. Thus, we would expect significant differences between GIC 2495 m and the other two samples. Indeed, the clear distinction between the three clone libraries indicated the presence of specific microbial populations dominated by different groups in the corresponding ice core samples. For example, library K from GIC 2495 m was dominated by Actinobacteria whereas libraries L and M were dominated by Firmicutes, mostly Bacilli, which were not found in the GIC 2495 m sample. Also, there were more proteobacterial sequences in library K than in the other two libraries. It is noteworthy that this same sample yielded the highest number and most diverse cultivated isolates, suggesting that the 2495 m sample was distinct in terms of the microbial composition and diversity presumably related to the very different climate conditions.

Another interesting observation was the correlation of the fungal sequences to deposition climates, which were highest (40%) in GIC 2545, deposited during colder climate and correspondingly lower in the milder (33%) and warmer (19%), which provides another indication of the dependence of their abundance and dispersal mode on climate.

Source and origin of deposited microorganisms

The strong recent interest in atmospheric microbiology has yielded a growing database of SSU sequences related to airborne microorganisms. Although these are derived mostly from present day sampling, they provide information about common microbial species found in different aerosols and evidence for temporal and spatial shifts in microbial compositions driven by meteorological conditions (Fierer et al., 2008; Morris et al., 2008). The possible active role of the microbial cells as radiative agents, cloud condensation nuclei and ice nucleators has been studied in relationship to atmospheric processes (Ariya and Amyot, 2004; Christner et al., 2008).

Here we use the valuable ice core record in an attempt to link diversity results to biogeography and particularly to the Asian deserts from where most of the Greenland dust originated. Past atmospheric records preserved in Greenland ice cores have been used to infer local climate conditions (temperature, snow accumulation), regional climate (wind-blown sea salts), and global climate events blowing dust from geographically distant sources. Mineralogical ion compositions of ice cores provide important clues about the origin of deposited aerosols. Thus, increased Ca2+ indicates mineral dust (which in Greenland originates from Taklamakan, Tenger and Mu-Us Asian deserts), presence of Na+ and Cl- indicate sea salt aerosol deposition, and volcanically derived aerosols are characterized by high SO42− and Cl- concentrations. The presence of all these ions in the three clear ice samples (Table 1) with significantly higher concentrations in sample GIC 2545 m suggests that dust and aerosols of different origins had been deposited, likely dominated by terrestrial dust, based on the very high Ca2+ concentration.

An important result of the phylogenetic analyses of our bacterial and fungal isolates was that their closest relatives originated from mineral dust or marine, or atmospheric environments suggested as sources of Greenland ice deposited aerosols. Additional links come from the geographic location of some of those environments in South-east China, cold Indian desert, Yellow sea where the corresponding microorganisms have been isolated. Interestingly, most of the latter are fungi or sporeforming bacteria, suggesting that the spores of such organisms might be transported over long distances. Particularly relevant to our results is a recent detailed identification of bacterial isolates from Gobi desert sand and from dust carried by Asian dust storms to Japan revealing that most Asian desert-originating bacteria were related and appeared to be halotolerant Bacilli (Echigo et al., 2005; Hua et al., 2007).

During intense atmospheric circulations, sea-spraying may transport diverse microbial cells in sea-salt aerosols which may, in turn, be distributed over larger distances (Lopez-Garcia et al., 2001; Aller et al., 2005; Griffin et al., 2006). Many of our isolates and clones sequences were related to organisms of marine origin suggesting a substantial input of marine aerosols over Greenland. This is in congruence with the high Na+ and Cl- ion concentrations in the samples.

Although not quantitatively significant the identification of isolates related to organisms found in volcanic soils or submarine caldera was meaningful. It should be noted that periods of increased volcanisms have been recorded in Greenland ice and one of them is between 56 000 and 61 000 years before present coinciding with the deposition time of GIC 2495 m ice (Zielinski et al., 1996). Bay and colleagues (2005) used an especially constructed dust logger to detect very thin layers of volcanic ash throughout GISP2 core and found correlations to millennial scale climate changes. It would be of great interest to analyse the microbial diversity in specifically selected samples of volcanic origin.


Our study addresses the question of whether the different climate conditions recorded in glacial ice have influenced the origin and diversity of the microorganisms transported by atmospheric processes and preserved in the Greenland ice sheet. The possibility of a relationship between microbial composition or physiological types in ice layers and climate is intriguing. Glacial ice provides the unique opportunity to search for both spatial and temporal patterns of microbial diversity and to address questions central to the immerging field of microbial biogeography. Although current knowledge in this field is still limited, often controversial, and posing more questions than answers, its importance for understanding the factors shaping microbial diversity patterns over space and time is widely recognized (Green and Bohannan, 2006; Martini et al., 2006; Fierer et al., 2008; Green et al., 2008). The differences in the microbial distribution and diversity at different depths of GISP2 found in this study support the idea that the composition of the microbial populations existing in Greenland ice is related to deposition climates and might be indicative of their origin. This conclusion is based on the following major results:

  • • Microbial abundance at different GISP2 depths correlated positively with increased Ca2+ (dust) due to more intensive atmospheric circulations during colder periods, which is in congruence with studies of other glaciers.
  • • The phylogenetic comparisons by culture-independent and culture-dependent methods of three ice core samples deposited during different climate conditions showed distinct microbial representation, which contrasted with the silty ice. The clear differences between the three clone libraries indicate that different microbial populations, dominated by different groups, exist in the corresponding ice core samples.
  • • The observed microbial diversity patterns could be linked to specific climate characteristics in Greenland such as millennial scale D-O temperature fluctuations and the transition between MIS4 (60–80 ka) and MIS3 (30–60 ka).
  • • The phylogenetic relationships of our bacterial and fungal isolates to organisms originating from Asian mineral dust, marine, or volcanic aerosols suggested that these were the sources of microorganisms entrapped in Greenland ice.

These results underscore the need for further investigations that could lead to microbial ‘markers’ as new tracers of specific past deposition climates. The availability of extensive chronological climate records and variable microbial life preserved in ice cores for hundreds of thousands of years provides an unique opportunity for future interdisciplinary studies finding new correlations between microbial diversity, biogeography and global climates.

Experimental procedures

Ice core sampling for gas measurements

The method used to extract and analyse the trapped gases in ice has been previously discussed (Sowers, 2000; Sowers and Jubenville, 2000). The procedure involves gas liberation using a cheese grater extraction technique starting with ∼1 kg ice samples. The liberated gas was collected in a pre-evacuated 35 cm3 sample tube suspended in a liquid helium Dewar. The sample tube was then removed from the Dewar and attached to a separate vacuum manifold servicing a Shimadzu (Kyoto, Japan) model GC-8a gas chromatograph equipped with an electron capture detector for N2O and O2 concentration measurements. A Hewlett Packard 5890 Series II gas chromatograph equipped with a flame ionization detector was used for CH4 measurements. Concentrations were determined by comparison with NOAA/CMDL-calibrated air mixtures (Butler et al., 1989). Measurements of δ13CH4 (13C/12C ratio) and δDCH4 (D/H ratio) were subsequently made using continuous flow isotope ratio mass spectrometry (CF-IRMS). Isotope data was reduced and reported on VPDB and VSMOW standard scales using previously documented techniques (Sowers et al., 2005; Sowers, 2006).

Ice core sampling for microbiological analyses

The ice core samples from GISP2 corresponding to depths 2495, 2545, 2578 and 3044 m below the surface were aseptically sampled for microbiological analyses in a UV and ethanol-sterilized flow laminar hood using the procedure of Rogers and colleagues (2005) applied as follows: before sampling, ice cores that had been stored at −34°C were transferred to −20°C for 18 h. They were immersed in 5% sodium hypochlorite for 10 s, followed by three consecutive washes with autoclaved MilliQ water and gradual melting in sterile funnels. Fractions of 40–50 ml were collected in individual sterile bottles and kept frozen at −80°C.

Cell enumeration by flow cytometry

Cell enumeration was performed on melted ice core samples either after gas analyses or after decontamination for microbiological analyses. Immediately after gas analyses the melted ice core samples (∼800 ml) were transferred in sterile bottles and subjected to cell enumeration. Aliquots of decontaminated melt fractions were analysed before freezing. Samples (0.5 ml) were stained with 2 μl 0.5 mM green fluorescent SYTO13 (Molecular Probes, Eugene, OR) for 30 min in the dark and analysed on a FC500 Beckman flow cytometer (Miami Lakes, FL) using red fluorescent pre-counted beads (Polysciences) as an internal standard. We also enumerated the ultra-small cells and applied the LIVE/DEAD kit (Molecular Probes) to quantify the proportion of viable cells (Miteva and Brenchley, 2005).

Genomic DNA extraction from ice core samples and whole genome amplification

The low biomass and the difficult to lyse cells made extracting sufficient DNA for cloning and sequence analysis challenging. To overcome these problems, we used two approaches: first, we combined a DNA extraction protocol with consecutive bead beating steps with increasing times. Second, we performed WGA on the DNA that was extracted using the REPLI-g™ kit (Qiagen, Germantown, MD). Genomic DNA was extracted from 60 to 80 ml from the innermost melt fractions using the Ultra Clean Microbial DNA kit (MoBio Laboratories, Solana Beach, CA) and applying 0.5–3 min bead-beating in a MiniBeadbeater-8 Cell Disrupter (Biospec Products). DNA yield was quantified using the PicoGreen® dsDNA kit assay (Molecular Probes).

Cloning of SSU rRNA genes and sequence analyses

The WGA-amplified genomic DNA was used to obtain small subunit rRNA genes by PCR with universal primers 515F-1492R and Ready-to-go PCR Beads (GE Healthcare, Biosciences, NJ). The PCR profile was: initial denaturation at 95°C for 5 min, 35 cycles, consisting of 95°C for 1 min, 55°C for 1 min and 72°C for 1.5 min with a final elongation step for 7 min at 72°C. Purified products were ligated into the PCR-Script™ Amp vector (Stratagene, La Jolla, CA). Ligation products were transformed into Z-Competent Escherichia coli DH5α cells (Zymo Research, Orange, CA). Transformants containing plasmids with inserts were grown in LB broth with ampicillin (100 μg ml−1) at 37°C and plasmid DNA was extracted by the boiling procedure and analysed by electrophoresis. Candidate plasmids with inserts were then used for re-amplification with T7-T3 primers. ARDRA of the PCR products with either RsaI or MspI was used to group the clones in OTUs. Individual PCR-amplified SSU rDNA products were purified and sequenced on an ABI Hitachi 3730XL DNA Analyzer at Penn State Nucleic Acid Facility. Sequences were compared with those from the GenBank database and aligned with reference sequences using ClustalX (Thompson et al., 1997). paup 4.0 Beta 8 (Swofford, 2002) was used to generate the phylogenetic trees. Chimeric sequences were identified using chimera-check ( and the Mallard program (Ashelford et al., 2006). Statistical analysis of the clone libraries was performed according to Hill and colleagues (2003). Coverage of the clone libraries was calculated using the Good's method (Good, 1953) where C = [1 − (n/N)] × 100, and n is the number of OTUs represented only once and N is the total number of clones.

Isolate recovery and 16S rRNA gene sequence analyses

Aliquots of melted ice were plated on 1× and 1/10× R2A or TSA (Tryptic Soy Agar) and incubated both aerobically and anaerobically in Gas Pouches at 5°C and 18°C. Growth of isolates was further tested on the same media at 2°C, 10°C, 18°C, 25°C, 30°C and 37°C.

Genomic DNA was extracted from cells using the PureGene kit (Gentra Systems, Minneapolis, MN). The SSU rRNA genes were amplified with 63F-1387R bacterial primers or 515F-1492R universal primers using Ready-to-go PCR Beads (GE Healthcare Biosciences, NJ). ARDRA with RsaI or MspI (Promega, Madison, WI) was used to group the isolates. The SSU rDNA products representing each distinct pattern were purified and sequenced.

Nucleotide sequences accession numbers

The SSU rRNA gene sequences for the clones and isolates used in the phylogenetic trees have been submitted to the GenBank database under accession numbers EU889135EU889252.


We thank Thomas Roberts and George Wood for their technical help, Elaine Kunze for the help with the flow cytometry and Jennifer Loveland-Curtze for the helpful discussions. This research was supported by NSF Grant MO 0347475, DOE Grant DE-FG02–93ER20117 and Penn State Astrobiology Center – NASA-Ames Cooperative Agreement NNA04CC06A.