A 102-m-long ice core retrieved from the Malan Ice Cap on the Tibetan Plateau provides us with a historical record of the microorganisms trapped in the ice. The microorganisms in the Malan ice core are identified as α, β, and γ-Proteobacteia, and the LGC, HGC, and CFB group by means of the results of 16S rRNA sequence analysis and physiological characteristics, while the eukaryotes in the ice core are mainly composed of Chlamydomonas sp. and Pseudochlorella sp. based on the phylogenetic examination of the 18S rRNA gene. The microbial populations show observable differences at different depths in the ice core, reflecting the effects of climatic and environmental changes on the distribution of the microorganisms in the glacier. Examination of the Malan ice core shows four general periods of microbial concentration, which correspond to four phases of temperature revealed by δ18O values in the core. Observations also indicate that microorganism concentrations tend to be negatively correlated with the temperature at a relatively long timescale and, to some extent, positively correlated with mineral concentrations. The present study demonstrates that more microorganisms are associated with colder periods while fewer microorganisms are associated with warm periods, which provides us with a new proxy for the reconstruction of past climatic and environmental changes by means of ice core analysis.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Investigations of Antarctic ice cores have revealed that large atmospheric dust particles can transport attached microorganisms and become trapped in ice [Abyzov et al., 1998; Priscu et al., 1999; Christner et al., 2000]. These microorganisms are embedded into the fractures and gas bubbles of the deep glacial ice by the winter freezing and summer melting processes [Cameron et al., 1972; Wing and Priscu, 1993; Gordon et al., 2000; Adams et al., 1998; Squyres et al., 1991]. Thus the biomass and microbial populations at the different layers of the ice core most likely reflect the prevalent climate, wind directions, and individual climatic events or human activities that occurred at the time of deposition or when the organisms grew on the surface of the glacier. Previous studies on glacial algae have shown that the physical and chemical conditions of the glacier determine the characteristics of the organism community, such as alga biomass and species. There are apparent differences in the glacial alga biomass in different geographic regions because of the distinct ecological environments. The glacial alga communities have unique structures and distributions on the glaciers in Patagonia in southern South America, Alaskan and other parts of North America, the Windmill Island region of Antarctica, New Zealand, and the Himalayas [Ling and Seppelt, 1990, 1996; Mataloni and Tesolín, 1997; Thomas and Duval, 1995; Takeuchi, 2001; Takeuchi and Koshima, 2004]. The studies by Yoshimura et al.  showed clear annual changes of algal biomass, which were consistent with the variations of the physical and chemical parameters recorded in a 7-m-long ice core from a Himalaya glacier. All these glacial biological clues suggest that sediment layers in ice cores may contain valuable bio-information records of climate and environment, which help reveal effect of the past climatic and environmental changes on the distribution of microorganisms and terrestrial sediment particles in glacier.
 Although many papers have been published relating to climatic and environmental changes on the Tibetan Plateau since the successfully drilling of the Dunde ice core in 1986 [Thompson et al., 1989], few deal with the microbioorganisms in ice. Only recently have the microbiological characteristics in this extreme environment been investigated in order to understand life in the ice of this region [Christner et al., 2000]. However, full understanding of the biogeochemistry and microbial life in these habitats and their relation to climate and environment awaits further investigations.
 Here, we try to extend our current understandings regarding microorganisms in perennial Tibetan Plateau ice. In particular, we bring together our recent microbiological studies and the chemical analysis from the Malan ice core in an attempt to relate the microorganisms in ice to the climatic and environmental changes. In this study, we also examined the differences in the bacterial population distribution in the different layers of ice core, which bacteria were recovered by means of aerobic incubation at low temperature, and the direct polymerase chain reactions (PCR) of 16S rRNA gene.
2. Study Sites and Ice-Core Analysis
 The Malan Ice Cap (35°48.40′N, 90°35.34′E), with an elevation more than 6050 m and total area of almost 200 km2, is located in the north-central Tibetan Plateau (Figure 1). The Malan ice core (10 cm diameter, 102 m long) was drilled at an elevation of 5680 m on the Ice Cap. The ice core was transported back to the cold room (air temperature between −18° and −24°C) at the Key Laboratory of the Ice Core and Cold Regions Environment of the Chinese Academy of Sciences. To keep the ice core frozen and in good quality during transportation, a freezer with temperature range from −15° to 18°C powered by a generator was put on a big truck. The ice core was cut into pieces of about 1 m long to put in the freezer. The visible stratigraphic features were described in the field immediately after the ice core was recovered. The sections were split lengthwise into four portions for different purposes and analyses within walk-in freezers (−18° to −24°C). One part of the four portions of ice core was used for microbial investigation.
 The Malan Ice Cap is a cold glacier. The average air temperature at the drill site can be represented by the 10-m ice temperature which was measured as −6.5°C. The temperature condition of the glacier implies a good condition to record original environmental and biological information brought by snowfall. However, the direct radiation on the Tibetan Plateau is very strong all-the-year-round because of high elevation. The strong direct radiation may not produce much melt water which can disturb environmental and biological record in ice but makes the densification process develop year round. On the other hand, the annual snowfall in the region is quite low (about 150 mm). The combination of the two facts makes the ice formation process on the glacier very fast: Snow transformed into ice within 1 year. The in situ observations confirm that the depth of the surface snow on the glacier is only 35 cm and there is immediately bubble ice under the surface snow. Owing to the ice boundary between annual layers, there is no melt water penetration between different annual layers even when there is some melt water in summer. Therefore the original environmental and biological information, including microorganism, would be recorded without disturbance after snow deposition. The biological records in the 102-m-long ice core is, therefore, reliable in the ice core.
 The ice cores sections used for δ18O measurement were cut into 2250 samples using a band saw within walk-in freezers (−18° to −24°C). The surface ice from each sample was scraped away with a stainless steel microtome. The δ18O value of each sample was determined by a Finnegan MAT-252 mass-spectrometer with a precision within ±0.5‰.
 Six samples were selected from the “dirty” layers of the inner core for mineral analysis and were melted. Twenty to fifty milliliters of melt water was precipitated to concentrate the particles and was dried in oven at 105°C for 24 hours, and the remained particles were weighed. The size of particles in the melt water was measured at ∼1000 times magnification using a microscope. The composition of particles was examined by a Dmax-2400 X-ray diffraction system.
 For microbial analysis (length about 15 mm, diameter 5 mm), the ice sample (containing sow and ice) at 0.17 m ice depth were decontaminated by cutting away the 10-mm annulus with a sterilized sawtooth knife. The other 22 samples were decontaminated by the above method, and then the remaining inner core was rinsed with iodine (2.5%) and bromo-geramine (1%), or with cold ethanol (95%), and finally with cold autoclaved water. The decontaminated ice samples were placed in the autoclaved containers and melted at 4°C. These handling procedures were undertaken at temperatures below 20°C within a sterile, positive-pressure laminar flow hood and described in detail by Zhang et al.  and Xiang et al. [2004a].
 Twenty-three ice samples were used for aerobic incubation with a diluted medium (5 g of NaCl, 2.5 g of glucose, 1 g of beef extract and 1 g of peptone per liter) [Zhang et al., 2002] at 5°C. In order to successfully isolate the microorganisms from ice layers with the anticipated low biomass, 10 mL melt water was added to 10 mL liquid media and incubated at 5°C. Both 200 μL of melted-water sample and pre-enriched samples were spread on the surface of agar-solidified diluted medium and incubated in duplicate at 5°C [Zhang et al., 2002]. All the distinct individual clones were obtained by streak culture and identified at genus level by the standard methods according to their morphological, cultural and physiological characteristics [Holt, 1993].
 The number of the culturable bacteria in the glacial ice was estimated by counting the average CFU (clone formation unit) on each agar plate. The culturable and total bacterial changes displayed similar tendencies, supporting our results on the amount of bacteria isolated from the Malan ice core samples [Zhang et al., 2002].
 Genomic DNA of isolates was extracted using the chloroform–isoamyl alcohol extraction procedure [Johnson, 1981] from the cells grown in 1.5 mL L-broth cultures of each isolate. The 16S rRNA gene was PCR-amplified from 20 ng aliquots of this genomic DNA, using the protocols described by Voytek and Ward . PCR products were passed through PCR purification columns (TakaRa) and partially sequenced directly with an ABI PRISM 377-96 sequencer, using primer 8f (5'-AGAGTTTGATCATGGCTCAG) which corresponds to regions 8 to 27 of the Escherichia coli 16S rRNA molecule [Brosius et al., 1978]. Then the isolates were identified by means of the phylogenetic analysis of 16S rRNA sequences [Ma, 2004].
 The SSU rRNA (16S and 18S rRNA) genes were also directly extracted from the ice samples by using the independent cultivation methods. One ice sample at 70 m from the Malan ice core was used for the genome DNA extraction and the establishment of the SSU rRNA gene library [Zhang, 2002; Zhang et al., 2003], and three samples at 35 m (Malan A), 64 m (Malan B), and 82 m (Malan D) were used for the direct PCR of the 16S rRNA gene from the melted water [Xiang et al., 2004a]. The 16S rRNA gene was PCR-amplified from 20 ng aliquots of this genomic DNA from the melt water of Malan C using the above method [Zhang et al., 2003]. The 18S rRNA gene was amplified from the same genomic DNA of Malan C ice sample using the similar procedure above, but using annealing temperature at 50°C and the forward primer (5'-TCYGGTTGATCCTGCCRG) which corresponds to regions 8 through 25 of the Escherichia coli SSU rRNA molecule [Brosius et al., 1978]. The 16S rRNA gene was also amplified directly from the melt water from the ice samples Malan A, B and D. Before use, all reaction tubes, tips and solution, except for the Taq DNA polymerase, were decontaminated described by Xiang et al. [2004a]. All reagent transfers were undertaken within a sterilized laminar flow hood. Melt water (1, 30 or 50 μL aliquots) was added directly to PCR mixtures (final volume of 100 μL) that contained 5 pmol of the primers 8f and 1492r (reverse, 5'-CGGTTACCTTGTTACGACTT-3'). Reaction mixtures were placed at 95°C for 9 min, and then subjected to 43 cycles of PCR amplification by incubation for 1 min at 94°C, 1 min at 52°C and 2 min at 72°C with a final extension at 72°C for 5 min. Four 16S rRNA and one 18S rRNA gene libraries were established by DNA ligation with the pMD 18-T vector, purification and transformation into competent E. coli JM109 cells. Recombinant transformants were selected by blue and white screening. The SSU rRNA clones with insert were obtained by PCR amplification as procedure described by Voytek and Ward . The unique individual clone was examined by RFLP (Hae III), and was partially sequenced by the above Sequencer with universal rRNA-specific primers. The sequences obtained have been deposited and are available on line in the GenBank nucleotide sequence database (http://www.ncbi.nlm.nih.gov/blast.cgi) under accession numbers AY378209-AY378228 for the isolates, AY322483-AY322483, and AY121823-AY121832 for the environmental clones.
 Three to ten milliliters of melt water of 23 ice samples was used for the total bacterial counting. The cells in the melt water were fixed with buffered glutaraldehyde (2% final concentration) and stored in 60-mL Nalgene bottles at 4°C in darkness before analysis. Acridine orange, allowing easy separation between the non-autofluorescing cocci and similar size mineral particles, was added (2% final concentration) [Takacs and Priscu, 1998, and Priscu et al., 1999] and the cells in the aliquots were collected by filtration onto Nuclepore polycarbonate membrane filters (2-μm pore size; Whatman). At least 100 cells were counted from each melt water sample using epifluorescence microscopy with both ultraviolet and blue-light excitation (Olympus BH-2, Tokyo).
3. Methodological Considerations
 Various methods have been used for the decontamination of the outer surfaces of ice cores in order to recover only organisms and nucleic acids from the interior of the ice sample [Priscu et al., 1999; Christner et al., 2000; Zhang et al., 2002; Sheridan et al., 2003; Rogers et al., 2004]. Rogers et al.  reported a systematic comparative study of several decontamination protocols with sodium hypochlorite (5.25%), ethanol (95%), NaOH/HCl (1N or 10N), and H2O2 (6% and 35%), heated probes, drilling, sterile water ablation, mechanical ablation, and UV irradiation. The results showed the 5% sodium hypochlorite to be the most effective oxidizer. All the treatments appear to be effective to some extent, however UV irradiation appears least effective. For the Malan ice core, we took several precautions and aseptically removed the surface ice, and then several decontaminants were used for the decontamination. For example, the mechanical ablation was for the surface firn sample, and iodine, bromo-geramine, and cold ethanol for the ice column. Because all the procedures were performed below 20°C, and inoculation and cultivation of the melt-water was maintained at 5°C, the inner core samples we used are unlikely to have been contaminated, or the cold-adapted bacteria impaired.
 With the application of molecular biological techniques, clone libraries of the 16S rRNA gene have been used to investigate the microbial communities in glacial habitats [Christner et al., 2003; Brinkmeyer et al., 2003; Zhang et al., 2003; Xiang et al., 2004a]. These investigations help us interpret the common characteristics within the microbial communities of the different geographical glacial environments. However, this approach suffers from certain limitations because of variable nucleic acid extraction efficiencies or biased amplification in the PCR. The limitations may result in our underestimate the diversity of bacteria in ice. Despite the limitations, the relatively high overall diversity (due to the very low biomass material from the ice melt-water) presented here suggests that our bacterial analyses were based on a substantial fraction of the bacterial community. The isolate Flavobacterium sp. ML-E9 AY378227 [Ma, 2004] was also detected using both DNA extraction (e.g., clone AY121829) and direct PCR of the 16S rRNA gene (e.g., AY322485) from the same ice layer, with both the clones and isolate being identical to Flavobacterium xijianggens AF433173 from Chinese Number 1 Glacier.
4. Microorganisms in the Malan Ice Core
 The physiological and chemical examination of the glacial isolates from the Malan ice core showed the diverse bacteria (Table 1). The aerobic culturable bacteria have been identified as Proteobacteria, CFB (Cytophaga-Flavobacterium-Bacteroides), LGC (low-G+C gram-positive bacteria) and HGC (high-G+C gram-positive bacteria) phylum, including Acinetobacteium sp., Alcaligenes sp., Achromobacter sp., Arthrobacter sp., Bacillus sp., Brevibacter sp., Flavobacterium sp., Microbacterium sp., Micrococcus sp., Nocardia sp., Paenibacillus sp., and Pseudomonas sp. [Zhang et al., 2002]. Working with the same glacial isolates by means of 16S rRNA gene analysis, Ma  also reported various similar bacteria recovered from the Malan ice core (not listed in Table 1) although the phylogenetic analysis of 16S rRNA gene of the isolates was not completely consistent with the results of physiological and chemical analysis. The 16S rRNA sequence analysis revealed the bacterium Phyllobacterim sp. (AY378225), Sphingomonas sp. (AY378228), Cryobacterum psychrophilum (AY378220), Duganella zoogloeoides (AY378226) and Sporosarcina sp. (AY378222), besides the above bacterial genera. Some strains in the ice core, such as Alcaligenes sp., Pseudomonas sp., and Brevibacter sp., identified by physiological and chemical analysis fell into Pseudomonas sp. (AY378224), Sphingomonas sp., and Brachybacterium sp. (AY378215), respectively, based on the phylogenetic analysis of 16S rRNA gene sequences. Some bacteria have similar phenotypic properties, but belong to different phylogenetic groups [Holt, 1993], which may partially explained our discrepancy between the analyses of 16S rRNA gene sequence and physiological and chemical characteristics. Moreover, the universal primer sets based solely on 16S rRNA gene sequences can bias diversity analyses [Koike et al., 2003; Tajima et al., 2001], and our knowledge of the biodiversity adherent to glacial microorganisms is still limited. In this context, ribosomal intergenic spacer analysis (RISA), which employs 16S and 23S rRNA gene-specific primers, might be important for the future study of microbial community structure and speciation in glacial environment.
Table 1. Microorganisms Recovered From the Malan Ice Corea
TC, total bacterial concentrations (cells mL−1); CN, culturable bacterial numbers (cfu mL−1); ND, no detection; similarity, sequence similarity of the environmental clone to the closest reported bacterium.
 Our analyses show that Bacillus sp., which belongs to LGC group, is the most common species occurring throughout the Malan ice core (Table 1), and that Brachybacterium sp. (HGC), is the second most common species present in the ice core. Several other bacteria such as Arthrobacter sp. and Microbacterium sp. of HGC group, Flavobacterium sp. of CFB (Cytophaga-Flavobacterium-Bacteroides) group, and Acinetobacteium sp. of Proteobacteria group also commonly occur in the core. This diversity of culturable bacteria is consistent with the 16S rRNA analysis of the direct extraction of DNA from the same four ice samples [Xiang et al., 2004a]. The phylogenetic analysis of the 16S rRNA gene libraries shows that the bacterial community structures in the Malan ice core includes α, β and γ-Proteobacteria, and the CFB and other eubacteria groups [Xiang et al., 2004a]. Further analysis reveals that the dominant bacteria closely clustered with those from samples from cold environments, such as sea ice and other glaciers. The growth properties of these isolates further in the ice core shows that most of these isolates appear to grow well from 2°C up to 30°C or 37°C [Ma, 2004], consistent with psychrotolerant species. This provides us evidence that most bacteria in this glacier are cold-adapted species.
 Phylogenetic analyses of the 18S rRNA sequences from the ice layer at 82 m in the Malan ice core show the diverse eukaryotes community composition in the ice core (Table 1), including Chlamydomonas sp. and Pseudochlorella sp., of which Chlamydomonas sp. is the dominant species [Zhang, 2002].
 The present study shows distinct differences between the bacterial populations at different depths, which were characterized by the repeated occurrence of some bacterial population groups at different ice layers (Table 1). For example, Brachybacterium sp. has been repeatedly isolated at different ice core depth such as 0.2, 12.4 and 28.0 m, and other deeper layers. Arthrobacter sp. has been detected at 28.0, 34.8, 54.3 and 63.8 m. Xiang et al. [2004b] have studied the vertical distribution of 16S rRNA in the ice core. The results indicate the predominance of β- and γ-Proteobacteria in ice. As shown in Figure 2, some bacteria such as γ-Proteobacteria appear at 82 m (e.g., Acinetobacteium sp., clone Malan D-10 AY322492, and D-11 AY322493) and 35 m (e.g., clone Malan A-86 AY322489) and 70 m (e.g., clone Malan C-33 AY121825). That some bacterial groups repeatedly occur at different depths indicates their ubiquitous and dominant property in the glacial environments. This is consistent with previous reports on the predominancy of some glacial alga species during specific environmental condition represented by deposition time [Yoshimura et al., 1997; Takeuchi et al., 1998]. This dominant distribution of some bacterial species suggests the preferable ecological environments on glacier at time of their deposition.
 Since the ice layers at different depths correspond to different periods [Thompson et al., 1989, 2000; Yao and Thompson, 1992; Wang et al., 2003] with different climatic conditions, the distribution of microorganisms at different depths might reflect the effect of the climatic and environmental changes at the time of their deposition on the microbial distribution in ice, or the response of microorganisms to the variations of past climate and environment.
5. Mineral Particles in the Malan Ice Core
 Our initial study shows microbial differences between clean ice and dust layers. In order to distinguish the possible source of the microorganisms, we analyzed the mineral particles in the Malan ice core. These analyses show differences in composition, size, and concentration of the insoluble mineral particles at different depths (Table 2). The concentration of the insoluble mineral particles (dust) changes with ice core depth, exceeds 22 mg mL−1 in all six measured “dusty” layers, and reaches 120 mg mL−1 at the 94 m ice depth. Quartz and feldspar are the primary minerals, and the concentrations of chlorite and calcite are below 5%. As shown in Table 2, the concentration and diameter of minerals particles differ considerably at different depths. Generally, at the most concentrated dust layers, the concentrations of the mineral particles range from 50 to 120 mg mL−1 and the particle diameters range from 60 to 100 μm. In the layers characterized by lower concentrations of mineral particles (about 40 mg mL−1), particle diameters are also smaller (below 20 μm, except 40–100 μm at about 12 m depth). The least dusty layer (e.g., at about 78 m) only contains 22 mg mL−1 mineral particles and has even smaller particle diameters (below 20 μm). The dust layers in the ice on the Tibetan Plateau form by dust deposition each spring when dust storms prevail, or from co-precipitation in snowfall, which may be the assemblage of minerals associated with microorganisms. The more concentrated dusty layers with larger particle diameter indicate stronger or more frequent dust storms and vice versa. Table 2 also indicates that the ice layers at 12 m and 35 m containing larger particles are characterized by more microorganisms, while the least dusty layers with smaller particles contain fewer microorganisms. This combination of dust layers and microorganisms indicates the possibility that most microorganisms are brought to the glacier by dust storms. Previous studies [Yao et al., 2004; Wu et al., 2004] point out that in the Tibetan Plateau ice cores, the higher dust concentrations correlate with colder periods and lower dust concentrations correlate with warmer periods. More frequent and stronger winds appear to have brought more and larger dust particles to the glaciers during the colder periods. Warmer periods, however, seem to be characterized by less frequent and weaker dust storms, which brought fewer and smaller dust particles to the glacier [Patterson and Gillette, 1977]. On the basis of this logic, more microorganisms are most likely brought to the glacier in colder periods than in warmer periods.
Table 2. Distribution of Composition, Main Size, and Concentration of Insoluble Mineral Particles in Malan Ice Core Section
H, heavy dusty layer; M, medium dusty layer; L, least dusty layer.
quartz, feldspar, mica, and chlorite (<5%)
quartz, feldspar, calcite (<5%), dolomite (<5%) and chlorite (<2%)
quartz, feldspar, calcite (<5%), and chlorite (<2%).
quartz, feldspar, calcite (<5%), and mica (<2%)
quartz, feldspar, calcite (<5%), and mica (<2%) dolomite (<2%) and chlorite (<2%)
quartz, feldspar, and calcite (<5%)
 In few cases, the dust layers with larger and higher amount of particles may contain less microorganisms, for example at 94 m depth, there is less microorganisms (Table 2). The discrepancy between biomass and particle concentration was also reported by Yoshimura et al. . This might indicate the change of sources of dust storm where microorganisms are different from the original source. Further studies are needed to clarify this.
6. Climatic and Environmental Significance of Microbial Populations
 The results from the deep oceans [Bowman and McCuaig, 2003; Nasreen and Hollibaugh, 2002; Gosink et al., 1993; Staley and Gosink, 1999] show that different light, oxygen, and nutrient conditions determine the microbial community structure. Our studies on the Tibetan Plateau suggest that the microbial patterns in ice over the Tibetan Plateau respond to climatic and environmental changes. Therefore the vertical distribution of these microorganisms, including the cultured and uncultured prokaryotes and eukaryotes reflect past climatic and environmental conditions.
Figure 3 shows the changes of the bacterial concentrations and of δ18O along the 102-m-long Malan ice core. The preliminary results from examination of this ice core [Wang et al., 2003] provide a detailed time series of the core as far back as 1800 A.D. The present paper also gives a roughly estimated longer time series of the ice core back to 1130 A.D. by using an ice flow model based on Wang et al.'s  time series. In Figure 3, the average δ18O for the whole core is −14.2‰. Using δ18O as a proxy for temperature [Dansagaard, 1964; Lorius et al., 1979; Yao et al., 1996, 1999], periods when δ18O was less negative than −14.2‰ are interpreted as warmer periods, and ones when δ18O was more negative than −14.2‰ as colder (C2) periods. The record then indicates two warmer periods (W1 and W2) and two colder periods (C1 and C2) and from 0 to 22 m corresponding to 1999 to 1893, and from 54 to 72 m (1700–1540), 22 to 54 m (1893–1700), and 72 to 102 m (1540–1130 AD), respectively.
 Considering the limited microbial sampling in the Malan ice core (at most about 6-m intervals above 57 m and about 4-m intervals below 57 m), we used an 11-point moving average of δ18O values to evaluate with the microbial record, thereby ignoring small events. The results demonstrate an inverse relationship between temperature and bacterial concentrations (Figure 3).
 The comparison of microbial concentration with the δ18O value recorded in the ice core (from past to present) displays the following remarkable features (Figure 3). (1) The temperature appears to have fluctuated considerably, but generally risen while the total concentrations of the bacterial community declined, with fluctuations, from 410 to 50 cells per milliliter from the past to present (from bottom to top of the ice core). (2) The δ18O record displays four phases as shown in Figures 3a and 3b, marked as C1 (first colder period), W1 (first warmer period), C2 (second colder period), and W2 (second warmer period). Responding to these four periods, the bacterial concentration demonstrates four corresponding stages as shown in Figure 2d, marked as H1 (first high concentration), L1 (first low), H2 (second high), and L2 (second low). (3) There are sudden rises and falls in temperature at stage C1 and W1 indicated by δ18O, and accordingly there appear high and low bacterial concentrations within stages H1 and L1. The cultured bacteria display a tendency similar to the total bacterial concentration, changes ranging from 0 to 85 CFU mL−1 with depth (Figure 3c).
 This examination result of negative correlation between bacterial concentration and temperature represented by the variations δ18O value of the Malan ice core seems paralogistic with the recognition that the microbial reproduction is always supported by high temperature. However, this may be partially explained, if we consider the factor of biological transportation of atmospheric circulation into glacier. As shown in Figure 3d, more microorganisms appear to be generally associated with heavier dust layers. For example, microorganisms seem to have “bloomed” where the dust layers appear, and are sparser where the ice core contains less dust. These results agree with the above discussion and previous studies that revealed a general correlation between microorganism patterns and mineral concentrations [Abyzov, 1993; Yao et al., 2003]. The observation by Hu et al.  revealed that the high input of microorganisms resulted from a serious sandstorm occurred in north China, and the “dust-snow” samples contained bacteria up to 8.9 × 104 to 1 × 105 cfu mL−1. Abyzov et al.  also reported bacterial cell densities between 1.2 × 103 and 8.3 × 103 in Vostok ice core at 1665 m and 2750 m ice depth, respectively, and the highest cell density was correlated with atmospheric microparticle concentration presented during a glacial period. These studies suggest an important role of paleoclimate in the distribution of microorganisms found in ice core. Thus, the high microorganism concentrations in dust layers in the Malan ice core could be interpreted as being deposited along with mineral particles (terrestrial sediments) by more and stronger storm events during colder periods, and the low influx in clean ice layers as being deposited by less and weak storm events during warmer periods. However, as discussed above, the source of microorganisms in ice is relatively perplexing, and difficult to be answered specifically at the moment. This study just presents our initial understanding of glacial microorganisms and their relation to climatic and environmental changes. Further detailed investigations of microorganisms on a longer time scale from ice core help improve our understanding.
 It seems that psychrotolerants can survive and grow under the cold environment [Ma, 2004]. To survive and grow, the bacteria have to overcome the extreme glacial environments with cold, desiccation, and minimal nutrient availability [Kohshima, 1984, 1994; Hoham et al., 1989; Helmke and Weyland, 1995; Bowman et al., 1997]. Temperature in winter might be more important. In the winter of 2000, we measured the lowest temperature of −60°C on the Puruogangri Ice field, which is not far (about 150 km) from the Malan Ice Cap. There exists similar temperature condition on the Malan Ice Cap. To investigate the adaptation of psychrotolerants to low temperature of −60°C, more experiments are needed in the future. Although the effect of temperature on the reproduction of some microorganisms is currently not clear, the initial total amount of microorganisms transported from outside of glacier is the most important source of total microorganisms on glacier. Therefore, on a longer time scale (1999 to 1130 AD) indicated by the analysis of the 102-m-long Malan ice core, the small event such as the seasonal and annual variations of microbial mass can be ignored from the general change of microbial distribution. The tendency of microbial mass distribution with depth could substantially exhibit the effect of the variations of the past global atmospheric circulation on the distribution of microorganisms on glacier.
7. Concluding Remarks
 The Malan ice core is characterized by various microorganisms, and dominated by single cell bacteria with the total concentrations ranging from 50 to 410 cells mL−1, and concentrations of culturable bacteria isolated from the ice ranging from 0 to 85 CFU mL−1. The most common culturable bacteria in the Malan ice core are the HGC group including Arthrobacter sp., and Brachybacterium sp., the LGC group including Bacillus sp., and Sporosarcina sp., and the CFB group such as Flavobacterium sp. The common members of the uncultured community are Acinetobacteium sp., of the Proteobacteia group, and the CFB group. Chlamydomonas sp. and Pseudochlorella sp. are the eukaryotes members in the Malan ice core. The abundance of these microorganisms indicates their ability to survive in a cold ice environment.
 The microbial populations and their quantitative distribution in the Malan ice core indicate a negative correlation between bacterial concentration and temperature represented by δ18O, and a general positive correlation between microbial abundance and dust storms revealed by mineral particle concentration. In other words, more microorganisms reflect a colder climate with stronger and more frequent dust events, and fewer microorganisms reflect a warmer climate with weaker and less frequent dust storms. From this, we conclude that the detailed microbiological records in deep ice cores can be used to reconstruct the history of climatic and environmental changes, and atmospheric circulation patterns. Further investigation of these ancient microorganisms as bio-indicators provides a potentially significant direction for future research into global change.
 We are very much obligated to John N. Reeve from the Department of Microbiology of the Ohio State University and to Nozomu Takeuchi from Research Institute for Humanity and Nature, Japan, for their helpful comments on this paper. We also thank S. O. Rogers and two anonymous reviewers for their helpful comments. This study was supported by the National Natural Science Foundation of China (40121101), Key Project of Chinese Academy of Sciences (KZCX3-339, KZCX2-SW-118) and project from MOST (2005CB422004) and NSFC grant (40471025).