Bacterial and fungal community structure in Arctic tundra tussock and shrub soils


  • Matthew David Wallenstein,

    1. Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, California, USA
    2. Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, Colorado, USA
    Search for more papers by this author
  • Shawna McMahon,

    1. Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, California, USA
    Search for more papers by this author
  • Joshua Schimel

    1. Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, California, USA
    Search for more papers by this author

  • Editor: Max Häggblom

Correspondence: Matthew Wallenstein, Department of Ecology, Evolution, and Marine Biology, University of California, Santa Barbara, CA 93106, USA. Tel.: +970 556 2591; fax: +970 491 1965; e-mail:


Fungal and bacterial community structure in tussock, intertussock and shrub organic and mineral soils at Toolik Lake, Alaska were evaluated. Community structure was examined by constructing clone libraries of partial 16S and 18S rRNA genes. The soil communities were sampled at the end of the growing season in August 2004 and just after the soils thawed in June 2005. The communities differed greatly between vegetation types, although tussock and intertussock soil communities were very similar at the phyla level. The communities were relatively stable between sample dates at the phyla and subphyla levels, but differed significantly at finer phylogenetic scales. Tussock and intertussock bacterial communities were dominated by Acidobacteria, while shrub soils were dominated by Proteobacteria. These results appear consistent with previous work demonstrating that shrub soils contain an active, bioavailable C fraction, while tussock soils are dominated by more recalcitrant substrates. Tussock fungi communities had higher proportions of Ascomycota than shrub soils, while Zygomycota were more abundant in shrub soils. Recent documentation of increasing shrub abundance in the Arctic suggests that soil microbial communities and their functioning are likely to be altered by climate change.


The Arctic tundra soil environment is characterized by long, cold winters followed by a few months of unfrozen conditions when the majority of microbial activity occurs. These conditions, in combination with strong nutrient limitations (Hobbie et al., 2002; Mack et al., 2004), constrain decomposition rates and result in a large accumulation of organic C Arctic tundra soils contain about 12% of the total amount of terrestrial soil C (Schlesinger, 1997). Therefore, these soils are an important part of the global C cycle, and their response to climate change will have important consequences for both ecosystem processes and global climate feedbacks. Rapid climate warming has been well documented in the Arctic (Moritz et al., 2002; ACIA, 2005), and future climate changes are expected to include continued warming, particularly in the winter, and altered precipitation patterns (IPCC, 2001). Climate warming has the potential to alter C storage in Arctic soils by increasing decomposition rates. On the other hand, warming may increase nutrient mineralization, and thereby increase net primary productivity and cause net C storage (Davidson et al., 2000; Kirschbaum, 2000). The balance of C therefore depends on the response of both plant communities and soil decomposer communities, and on their interactions.

One of the apparent results of climate change in the Arctic has been a shift in terrestrial vegetation biogeography (Chapin et al., 1995, 2005). For example, the abundance of shrubs in Arctic tundra has increased in recent decades (Sturm et al., 2001; Tape et al., 2006). This has important implications for ecosystem functioning and C storage, as they differ greatly in shrub communities compared to the tussock tundra they often displace (Weintraub & Schimel, 2005). Shrub soils appear to comprise a large pool of recalcitrant (possibly wood derived) organic matter, with a small labile pool that is N rich and drives rapid net mineralization. Tussock tundra is formed from graminoid and moss vegetation, has a very large pool of bio-available C, and is dominated by net immobilization (Weintraub & Schimel, 2005).

In association with large changes in the chemical nature of substrates being supplied to soil, there are likely parallel changes in the decomposer communities. The shift in microbial communities may play a significant role in changing the pattern of nutrient release and C flow. However, very little is known about the composition of microbial communities in Arctic soils, let alone how they change with climate or vegetation. Much of what is known predates the advent of culture-independent molecular techniques (Holding et al., 1974). Here, a clone-library-based analysis of 16S and 18S SSU rRNA genes are presented to describe the community composition of bacteria and fungi in Alaska tundra soils. The community composition in tussock, intertussock, and shrub soils were evaluated before soil freezing in August of 2004, and shortly after soil thawing in June 2005. The microbial community structure was compared between vegetation types and sampling dates. The seasonal stability of microbial communities provided insights into strategies for microbial survival in the Arctic soil environment.

Materials and methods

Soil collection and processing

Soils were collected from the Toolik Lake Long-Term Ecological Research (LTER) site on August 19 and 20, 2004, prior to seasonal soil freezing, and on June 6 and 7, 2005, shortly after the soils thawed (Fig. 1). Moist acidic tussock and shrub tundra sites were sampled. Tussock sites are dominated by Eriophorum vaginatum, which forms tussock mounds and intertussock microsites which are dominated by mosses with small shrubs (Vaccinium spp., Betula nana and others) intermixed (Bliss & Matveyeva, 1992). Shrub sites are dominated by willow (Salix spp.) and birch (Betula nana). Tussock soils are classified as Typic Aquiturbels and shrub soils are Aquic Umbrorthels. In each site, eight random locations were selected within a 50 m2 area. In the tussock tundra site, tussocks were randomly selected and the nearest adjacent intertussock area was also selected. In tussocks, soil cores were then taken (to 15 cm depth) using an aluminum tube with teeth filed into the bottom, while in intertussock and shrub soils, square soil sections were cut out with a serrated knife. Soil samples were stored intact at 4° C and shipped to a laboratory at the University of California, Santa Barbara for immediate processing. Shrub soils were visually split into organic and mineral horizons prior to processing. Soils were manually processed to remove live roots and to homogenize the samples. DNA was extracted within 24 h of processing.

Figure 1.

 Daily mean soil temperature at 5 cm depth in tussock soils (from Arctic LTER database). Sample dates are indicated by circles.

Nitrogen analyses

Following soil processing, 10 g (wet weight) of each field sample were shaken with 0.5 M K2SO4 for 1 h and were then vacuum-filtered through Whatman GMF 2-um filters. NH4+ and NO3 were analyzed using a Lachat AE flow injection autoanalyzer (Lachat Instruments, Milwaukee, WI).

DNA extraction and PCR amplification

For each soil sample, three replicates of c. 0.5 g (wet weight) were extracted using the MoBio Power Soil DNA extraction kit following the manufacturer's instructions. The three replicate extractions were pooled for further analyses. The pooled DNA extracts from each soil sample were quantified using the Quant-iT DNA assay kit (Invitrogen) and a Perkin Elmer Wallac plate reader. The extracted DNA was PCR amplified with the bacteria-specific primers 8f (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492r (5′-GGTTACCTTGTTACGACTT-3′) (Lane, 1991), and the fungal-specific primers EF4 [5′-GGAAGGG(G/A)TGTATTTATTAG-3′] and EF3 (5′-TCCTCTAAATGACCAAGTTTG-3′) (Smit et al., 1999). PCR amplification was performed using a MJ Research thermocycler in 50 μL reactions containing 50 ng of DNA, 25 μL of JumpStart REDTaq ReadyMix PCR Reaction Mix (Sigma) and 25 μg BSA, with final primer concentrations of 0.7 μM each. For the bacteria 16S rRNA gene primer set, PCR conditions were 94° C for 15 min, followed by 30 cycles of 94°C for 30 s, 52°C for 30 s and 72°C for 60 s, followed by a final extension step at 72°C for 12 min. For the fungal 18S rRNA gene primer set, conditions were identical except that the annealing temperature was 48°C. Following PCR, products were verified to be of correct length on a 1% agarose gel, and were purified using the QiaQuick PCR clean-up kit (Qiagen).

Clone-library sequencing and analyses

PCR products were cloned into One Shot® TOP10 Chemically Competent Escherichia coli using the TOPO TA Cloning® Kit (Invitrogen). Transformants were then plated on LB agar plates with Ampicillin and x-gal, and shipped to the Sym-Bio corporation (Menlo Park, CA) to complete the sequencing process. Recombinant colonies from each library were aseptically picked and grown overnight in 10% glycerol/LB broth media containing 50 ug mL−1 Ampicillin at 37°C. From each well, 1.5 μL was used in a Phi29 rolling circle amplification (Dean et al., 2001) reaction using TempliPhi (GE Healthcare) according to manufacturer recommended conditions. Sequencing was performed using DYEnamic ET Dye Terminator Cycle Sequencing Kit for MegaBACE (GE Healthcare) according to manufacturer-recommended conditions. The bacterial libraries were sequenced with primer 8f and fungal libraries were sequenced with primer EF4. Following sequencing, reactions were precipitated using EtoH, air-dried and resuspended in MegaBACE running buffer. Sequencing reaction products were electrophoretically separated on MegaBACE instruments under standard running conditions. The chromatograms were basecalled, quality scores assessed using Phred, and the sequences were trimmed for vector using Cross Match (Ewing & Green, 1998; Ewing et al., 1998). The sequences then underwent quality trimming at the 3′ and 5′ ends. Regions that did not meet a minimum quality score average (logarithmic) of 10 over an 11-base window were eliminated from the final sequence set.

Sequences were trimmed to a common 450-bp fragment and aligned using clustalx. Each sequence was then assigned a putative taxonomy based on its most closely related sequence. For bacteria, the database available at the Ribosomal database project website was utilized (Cole et al., 2005), and for fungi an 18S database from the Assembling the Fungal Tree of Life (AFTOL) project was utilized (Lutzoni et al., 2004). After aligning the sequences to these databases, a local blast search was performed and the taxonomic information of the most closely related sequence was extracted. While this approach would not be appropriate for species-level identification, it has been found to be accurate at least to the order level based on comparisons with maximum parsimony analyses (data not shown). All sequences were deposited in Genbank under accession numbers DQ508981DQ512318.

UniFrac (Lozupone & Knight, 2005) was used to test for statistical differences between sample dates and vegetation type. First, phylogenetic trees were constructed for the 16S rRNA gene sequences and another tree for the 18S rRNA gene sequences using the neighbor-joining method as implemented in PAUP 4.0b10 (Swofford, 2003). Testing was then carried out to detect differences between sample dates and vegetation using the UniFrac statistic, and conducted principal components analyses.


The total number of 16S rRNA and 18S rRNA gene sequences for each sample is summarized in Table 1.

Table 1.   Total number of sequences obtained for each 16S and 18S rRNA gene clone library for Toolik Lake soils collected in August 2004 and June 2005
Sample date/
16S rRNA gene
18S rRNA gene
August 2004
 Shrub Organic185233
 Shrub Mineral116206
June 2005
 Shrub Organic215205
 Shrub Mineral200244

Tussock tundra: tussock and intertussock soils

Extractable NH4 was not detected in the August 2004 samples, but a post-thaw pool of NH4+ was measured in June 2005 (Table 2). Extractable NO3 levels were generally low, but were higher in June than August (Table 2).

Table 2.   Inorganic N and microbial biomass C for Toolik soils collected in August 2004 and June 2005 (mean of five field samples)
VegetationSample dateNH4
(ug g−1 soil)
(ug g−1 soil)
TussockAugust 2004ND2.02
June 200521.162.91
IntertussockAugust 2004ND2.85
June 200527.586.23
Shrub organicAugust 200442.231.61
June 200579.406.80
Shrub mineralAugust 200420.561.97
June 200519.2415.59

Both bacterial and fungal communities were generally similar between tussock and intertussock soils at the phyla level (Fig. 2) based on the clone library analyses. Bacterial clone libraries were dominated by Acidobacteria, followed in relative abundance by Proteobacteria, Firmicutes and Bacteriodetes (Fig. 2a). In general, there were few major shifts in bacterial community composition at the phyla level between sample dates. Actinobacteria were much more abundant in August (prefreeze) than in June (post-thaw). In contrast to phyla-level patterns, UniFrac analyses suggest that there were significant seasonal community shifts in intertussock bacterial communities, but not in tussock soils (Table 3, Fig. 4a). Within the same sample date, bacterial communities differed between tussock and intertussock soils (Table 3).

Figure 2.

 Relative abundance of microbial taxonomic groups in Arctic tundra soils based on clone library analyses of 16S (a) and 18S (b) SSU rRNA genes for bacteria and fungi, respectively.

Table 3. P-values of sequence library comparisons for 16S rRNA genes (A) and 18S rRNA genes (B). P-values are for contrasts between each pair of clone libaries using UniFrac
Soil SampleAugust
August shrub
August shrub
June shrub
June shrub
 August intertussock 0100000
 August shrub mineral  000000
 August shrub organic   00.02000
 August tussock    0001
 June intertussock     000
 June shrub mineral      00
 June shrub organic       0
 June tussock        
 August intertussock 0000000
 August shrub mineral  000000
 August shrub organic   00000
 August tussock    0000
 June intertussock     000
 June shrub mineral      00
 June shrub organic       0
 June tussock        
Figure 4.

 Principal components analysis ordination plot for the 16S rRNA gene (a) and the 18S rRNA gene. The percent of variation explained by each principal component is indicated on the axis labels. Soils are represented by the following symbols: tussock □, intertussock ○, shrub organic Δ, shrub mineral ∇. Closed symbols represent samples collected in August 2004 and open symbols represent samples collected in June 2005.

Fungal communities were dominated by Basidomycota, especially in intertussock soils (Fig. 2b). There were also a large proportion of Ascomycota, and lesser relative amounts of Zygomycota, Chytridmycota and Glomeromycota. Ascomycota were dominated by Leotiomycetes, and Basidomycota were dominated by Agaricales. As with bacteria, the fungal communities were relatively stable between sample dates at the phyla level. Some fungal groups, such as the Cantharellales, were more abundant in June than in August (Fig. 3b). The Aphyllophorales showed opposite seasonal patterns in tussock and intertussock soils (Fig. 3b). In contrast to phyla-level patterns, UniFrac analyses suggest that there were significant seasonal shifts in both tussock and intertussock fungal communities (Table 3, Fig. 4b).

Figure 3.

 Relative abundance of fungal orders within Ascomycetes (a) and Basidiomycetes (b) based on 18S SSU rRNA gene clone libraries from Arctic tundra soils.

Shrub organic and mineral soils

Extractable NH4+ was greater in shrub soils than in tussock or intertussock soils. There was a post-thaw release of NH4+ and NO3 in shrub organic soils and of NO3 in mineral soils.

Fungal and bacterial community composition was generally similar between shrub organic and mineral soils (Fig. 2). They differed dramatically, however, from tussock tundra communities. In shrub soils, bacterial communities were dominated by Proteobacteria, with a much smaller relative abundance of Acidobacteria (Fig. 2a). Fungal communities were generally similar to tussock and intertussock soils, with notably more Acsomycota classified as Dothideomycetes, Pezizomycetes and Sordariomycetes (Fig. 2b). There were fewer Chytridmycota than in tussock and intertussock soils, and they were more abundant in shrub mineral than in organic soils. Both fungal and bacterial communities were relatively stable between sample dates at the phyla level. In both soil horizons, Betaproteobacteria were more common in June (post-thaw) soils than in August (prefreeze) soils (Fig. 2a). UniFrac analyses suggest that the species level composition of both bacterial and fungal communities changes seasonally, and differs between shrub organic and mineral soils (Table 3, Fig. 4a and b).


At the phyla and subphyla level, vegetation appears to be the primary driver of microbial community composition in Arctic tundra soils. However, UniFrac analyses revealed that seasonal shifts did occur at finer taxonomic levels. The differences between tussock and shrub soil microbial communities were much greater and occurred at a higher taxonomic level than any seasonal shifts within soils from the same vegetation type. This suggests that plants strongly regulate microbial communities through their role in substrate supply (e.g. litter, root turnover, exudates) and by modifying the physical environment in the active soil layer. Seasonal shifts at finer taxonomic levels suggest that at the species level, microorganisms may have specialized physiologies adapted to the environmental conditions associated with a particular season.

In general, the graminoid litter in tussocks is more bioavailable than shrub litter, which has a large woody component. The large, but low-quality C supply in tussock soils appears to support a large population of Acidobacteria, which generally appear to be relatively slow-growing, ‘k-selected’ bacteria (Fierer et al., in press). In contrast, Proteobacteria dominate the shrub soils which contain a small, highly labile pool of bioavailable C, and a larger very recalcitrant pool of lignocellulose (Weintraub & Schimel, 2003). It is likely that the Proteobacteria utilize the highly labile C pool in shrub soils, consistent with other evidence that they tend to grow in soils with high C mineralization rates and exhibit ‘r-selected’ characteristics (Fierer et al., in press). However, other environmental factors also differ between shrub and tussock soils and may also influence these patterns. In addition to differences in soil chemistry, the rooting architecture also differs dramatically between tussock sedges and shrubs resulting in contrasting physical conditions. Also, shrubs affect the soil microclimate by collecting drifting snow, thus insulating the soil during the winter. The importance of vegetation in driving microbial community composition suggests that increasing shrub abundance in the Arctic will alter soil microbial communities and decomposition processes. While it is now know that both the soil characteristics and the soil microbial communities differ between vegetation types, the rate at which they will change following shrub expansion is not known. There is likely to be a legacy of previous vegetation, but the consequences of these lags on soil microbial communities or decomposition processes cannot yet be predicted.

Plant competition for nutrients may also play a role in controlling microbial community composition, but in this system, plants are generally poor competitors for nutrients relative to microorganisms. For example, E. vaginatum takes up less than 2% of amino acids and NH4+ in field addition experiments (Schimel & Chapin, 1996; Weintraub & Schimel, 2005). Little is known about nutrient uptake rates for the shrub communities, but hypothesize that they may be relatively stronger competitors for organic N because most shrubs, including B. nana, form ericoid and ectomycorrhizal associations, in contrast to the nonmycorrhizal sedges. On the other hand, N availability is higher in shrub soils than in tussocks (Table 2), and it has been suggested that as available N increases, plants should depend less on organic forms (Schimel & Bennett, 2004). The effects of competition for N on microbial community composition in these systems is unknown.

Seasonal differences among communities were limited at the phyla and subphyla levels, but were evident at finer taxonomic resolution. This pattern may suggest that substrate quality controls overall microbial community composition over the long-term, but that in the Arctic, seasonal environmental changes cause community shifts among taxa with similar function but different environmental tolerances.

The pattern of seasonal community shifts occurring primarily at lower phylogenetic levels is fairly consistent with the patterns of microbial community composition that occurs in the Alpine tundra (Schadt et al., 2003) and that has been assumed to be general in tundra ecosystems (Neufeld & Mohn, 2005). This may suggest that within each phyla or subphyla, there are species or genera that are specially adapted to the winter environment. Microbes could live in these soils either by surviving cold winter conditions through active cell maintenance and/or growth (‘resistance’), or by dying back and then regrowing their populations following the spring thaw (‘resilience’) (Schimel et al, in press). Many psychrophilic fungi, bacteria and Archaea have a variety of adaptations that enable them to remain active well below freezing (D'Amico et al., 2006). These adaptations include the ability to increase the fluidity of cellular membranes (Chintalapati et al., 2004), to produce cold-active enzymes with flexible structures (Feller, 2003; Feller & Gerday, 2003; Georlette et al., 2004), and to produce antifreeze proteins and cryoprotectants (Krembs et al., 2002; Walker et al., 2006). However, inducing these tolerance mechanisms uses a large amount of resources and has high energetic costs. While it is clear that many Arctic soil microorganisms have the capacity to tolerate and survive the winter, the proportion of resistant microorganisms that actively survive the winter compared to resilient microorganisms that regrow following spring thaw is currently unknown. In this study, the majority of soil bacteria and fungi phyla did not differ greatly in abundance between samples taken prior to soil freezing and after the soils thawed. This suggests that most microorganisms survive the winter intact through resistance mechanisms. It is possible that these microorganisms are extremely resilient, but this would require that their populations had recovered to prefreeze levels only ten days after the soils had thawed, and even then soils had warmed to only 4°C (Fig. 1).

While most bacterial and fungal phyla and subphyla were relatively stable between sample dates, there were several groups that showed seasonal changes in relative abundance. For example, the Actinobacteria were consistently more abundant in the prefreeze samples than post-thaw (Fig. 2a). Speculation is that these filamentous bacteria may be physically disrupted by soil freezing and thawing. In contrast, the basidomycete Cantharelles were generally more abundant post-thaw (Fig. 3b). It is unclear whether substrate availability or physical conditions are driving seasonal changes in the relative abundance of these and bacteria and fungi.

The composition of microbial communities in these soils appear to be consistent with the limited number of other studies examining Arctic soil microbial communities. One of the first soil bacterial clone libraries was from Siberian birch-dominated shrub tundra (Zhou et al., 1997). Though their total library was limited to 43 sequences, over half of their sequences were identified as Proteobacteria, which is consistent with the shrub soils examined in this study. In another study, Kobabe et al. (2004) used fluorescent in-situ hybridization (FISH) to examine the relative abundance of major bacterial groups at different depths in Siberian tundra with sparse sedge (Carex) vegetation. They found somewhat higher ratios of Proteobacteria and Cytophaga–Flavobacterium–Bacteroides (CFB) in upper organic soils compared to Toolik Lake tussock soils, but did not measure Acidobacteria or other groups that were detected with the clone library approach. It is more difficult to compare our clone library analyses to the earlier culture-based analyses of microbial communities that were done during the International Biological Program (IBP) because of the biases of the techniques and because those studies were performed using different tundra types. The two sites with isolate data that are most similar to ours are Devon Island mesic meadow and Barrow, and these sites are similar to each other in terms of climate and soils (French, 1974), but are both meadows rather than tussock or shrub tundra. The bacterial communities from Barrow are somewhat comparable to those from our clone libraries, being dominated by Pseudomonas spp. (Gammaproteobacteria; 80–90% of isolates), Flavobacterium/Cytophaga spp. 5–10%) and Arthrobacter spp. (Actinobacteria; 10–20% of isolates; (Dunican & Rosswall, 1974). The meadow on Devon Island, on the other hand, was dominated more strongly by Arthrobacter spp. (43% of isolates), with all the gram-negative groups together accounting for less than 50% of isolates and so bacteriologically looks more like our shrub soils. It is harder to even try to compare clone library data to isolation studies of fungi, as a large proportion of isolates produce sterile hyphae and could not be classified (Dowding & Widden, 1974).

This study suggests that in the Arctic Tundra, vegetation drives microbial community structure at the phyla and subphyla level by affecting substrate quality and soil environmental conditions. However, seasonal changes in microbial community structure at finer taxonomic resolution suggest that different populations of closely related microorganisms likely differ in their tolerance to low temperatures and freezing, and may be active only under a narrow range of conditions.


We thank Patricia Holden, Allen Doyle, Keri Holland, Timothy James and the staff of the Toolik field station. This research was supported by funding from the NSF Office of Polar Programs (OPP) Arctic Natural Sciences Program and by an OPP Postdoctoral Fellowship to M. Wallenstein.