SEARCH

SEARCH BY CITATION

Keywords:

  • bacterial communities;
  • alpine grassland soil;
  • altitudinal transect;
  • semi-arid area;
  • Tibetan Plateau

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

The Tibetan Plateau, ‘the third pole’, is a region that is very sensitive to climate change. A better understanding of response of soil microorganisms to climate warming is important to predict soil organic matter preservation in future scenario. We selected a typically altitudinal gradient (4400 m–5200 m a.s.l) along south-facing slope of Nyainqentanglha Mountains on central Tibetan Plateau. Bacterial communities were investigated using terminal restriction fragment length polymorphism analysis (T-RFLP) combined with sequencing methods. Acidobacteria and Proteobacteria were dominant bacteria in this alpine soil. Redundancy analysis revealed that soil bacterial communities were significantly different along the large altitudinal gradient, although the dominant environmental driving factors varied at different soil depth. Specifically, our results showed that precipitation and soil inline image were dominant environmental factors that influence bacterial communities at 0–5 cm depth along the altitudinal gradients, whereas pH was a major influential factor at 5–20 cm soil. In this semi-arid region, precipitation rather than temperature was a main driving force on soil bacterial communities as well as on plant communities. We speculate that an increase in temperature might not significantly change soil bacterial community structures along the large altitudinal gradient, whereas precipitation change would play a more important role in affecting soil bacterial communities.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

Altitudinal gradients are often characterized by dramatic changes in climate coupled with biotic turnover over short geographical distances. Climatic factors such as temperature, precipitation, and radiation concurrently change along the altitudinal gradients under field conditions (Bryant et al., 2008; Wang et al., 2013). Mountain ecosystems and associated species are very sensitive and vulnerable to climate change (Diaz et al., 2003; Schröter et al., 2005). Previous studies on biogeography along altitudinal gradients have focused almost exclusively on macroorganisms such as plant and animal taxa (Herzog et al., 2005; Kreft & Jetz, 2007; Meier et al., 2010), yet the recent microbial distribution patterns have suggested that the ecological rules of macroorganisms that are following may not necessarily apply to microorganisms (Fierer & Jackson, 2006; Green & Bohannan, 2006; Bryant et al., 2008; Wang et al., 2011). Researchers have pointed out that the investigation of microbial diversity along altitudinal gradients could help reveal the response of soil microorganisms to the climatic change (Collins & Cavigelli, 2003). Therefore, investigating the change in bacterial communities along the altitudinal gradient may shed light on the prediction of future climatic changes scenario.

A few studies have shown that the composition of bacterial community varies with elevation. The amount of total bacteria and Gram-negative bacteria all increases with increasing elevation in the Austrian Northern and Central Alps (Margesin et al., 2009; Djukic et al., 2010). Controversially, total bacteria, methanotrophic bacteria, and ammonia-oxidizing archaea are all negatively correlated with increasing elevation (Ma et al., 2004; Giri et al., 2007; Zhang et al., 2009). On the other hand, the structure of microbial community could shift with increasing altitude (Margesin et al., 2009). However, in Arctic fjeld of Finnish Lapland, bacterial communities were relatively stable at various levels of altitude provided as long as the soil pH was similar (Männistö et al., 2007). These controversial results yield little information regarding the distribution of soil microbial communities and the impact of environmental factors on the composition and functioning of such soil bacteria along altitudinal gradients.

Climate warming is expected to cause an upward migration of vegetation zones (Dullinger et al., 2003), which probably entail changes in the composition and functioning of soil microbial communities in these vegetation zones. Although altitudinal patterns of macroorganisms and the determinative effect of environmental factors on their altitudinal patterns are well established (Bryant et al., 2008; Wang et al., 2013), we know little about how microbial diversity varies across altitudinal gradients. Furthermore, it is still unknown whether similar environmental factors shape altitudinal patterns of bacterial communities in higher altitudinal regions. So far, only few studies have examined bacterial diversity across altitudinal gradients in higher altitudes (Bryant et al., 2008; Singh et al., 2012) with different findings. Therefore, it is crucial to know more about bacterial community patterns from different mountain ecosystems and to further explore their biographical patterns.

The Tibetan Plateau, known as ‘the third pole’, is the earth's largest (2.5 × 106 km2) and highest (mean altitude 4500 m a.s.l) plateau with high snow cover, high UV exposure, and low concentrations of oxygen and nutrient. Studies have shown that this region is very sensitive to climate change (Yao et al., 2000). For most parts of Tibetan Plateau, the temperature in summer and winter season has increased 0.16 and 0.32 °C, respectively, during the past decades (Liu & Chen, 2000). Therefore, an investigating of the changes in bacterial community composition along large altitudinal gradients within this region is helpful for better understanding the response of soil microorganisms to future climate change. We selected a typically altitudinal gradient across a south-facing slope to study changes in bacterial community composition from 4400 m to 5200 m above sea level (a.s.l) in Tibetan Plateau and aimed to understand the altitudinal distribution of bacterial community structure and to explore their environmental driving factors along the altitudinal gradients.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

Field sites and soil collection

The annual precipitation was 479 mm according to meteorological observations from 1963 to 2010 at Damxung station (4288 m, c. 4 km away from our study site), which was much less than annual pan evaporation of 1726 mm (see details at Wang et al., 2013). The study sites are located along the south-facing slope of Nyainqentanglha Mountains (4400–5600 m, 30°30′–30°32′N, 91°03′E), near the grassland ecology station of Damxung County, Xizang Autonomous Region, China. Average precipitation increased from 227 mm at lower elevation (4400 m) to 420 mm at higher elevation (5100 m). Conversely, the mean annual maximum temperature decreased from 10.1 °C at the 4400 m elevation site to 4.3 °C at the top (5200 m) of the transect. The upper growth limit of alpine meadow was at 5210 m. Along the slope, Stipa capillacea and S. purpurea are dominated at the altitude of 4400–4500 m (alpine steppe-meadow ecotone), and K. pygmaea is dominated at higher level of altitude, ranging from 4600 m to 5200 m (Table 1) (Wang et al., 2013).

Table 1. Climatic and vegetation characteristics in the sampling sites along the altitudinal gradient
SitesMATa(°C)MMTa (°C)T5AnnaT5AugaT20AnnaT20AugaMAPa(mm)Dominant species bHPCb(cm)CPCb (%)PSRb
  1. The name with underline is the predominant species, and the left are companion species.

  2. MAT, mean annual air temperature; MMT, mean monthly air temperature; T5Ann, mean annual soil temperature at 5 cm soil depth; T5Aug, mean soil temperature at 5 cm soil depth for August; T20Ann, mean annual soil temperature at 20 cm soil depth; T20Aug, mean soil temperature at 20 cm soil depth for August; MAP, mean annual precipitation; HPC, height of plant community; CPC, coverage of plant community; PSR, plant species richness. S.c, Stipa capillacea; S.p, Stipa purpurea; A.w, Artemisia wellbyi; K.p., Kobresia pygmaea; K.r., Kobresia robusta; S.spp, Saussurea spp.

  3. a

    Data from Li et al. (2013).

  4. b

    Data from Wang et al. (2013).

4400 m3.710.17.813.47.613.2227S.c., S.p., A.w.3.39 ± 0.2016.23 ± 2.4014 ± 1
4500 m3.29.57.212.87.112.5237S.c., S.p., A.w.2.59 ± 0.0922.85 ± 3.3816 ± 1
4650 m2.48.56.311.56.411.4245K.p., K.r., S.spp.2.45 ± 0.1742.97 ± 5.4916 ± 1
4800 m1.67.76.910.56.110.2338K.p., K.r., S.spp.2.10 ± 0.0463.98 ± 2.9520 ± 1
4950 m0.26.04.68.74.68.7434K.p., K.r., S.spp.1.93 ± 0.0683.54 ± 3.1819 ± 1
5100 m−1.25.03.68.73.67.9420K.p., K.r., S.spp.1.99 ± 0.0681.37 ± 4.1620 ± 1
5200 m−1.64.33.87.13.47.6362K.p., K.r., S.spp.1.96 ± 0.1335.85 ± 5.1613 ± 1

In August 2010, seven sites were sampled along an altitudinal transect on the south-facing slope of Nyainqentanglha Mountains (4400, 4500, 4650, 4800, 4950, 5100, and 5200 m a.s.l; Fig. 1). At each site, three replicated sampling plots (1× 1 m) were set up previously, and three soil samples were collected from three soil depths (0–5, 5–10, and 10–20 cm) by mixing five to eight soil cores within each sampling plot. After field samples maintained using ice bags were transported to laboratory, soil samples were passed through a 2.0-mm soil sieve, stored at 4 °C for general analysis, and with subsamples were stored at −80 °C for following DNA extraction.

image

Figure 1. A view of the altitudinal transect along south-facing slope on Nyainqentanglha Mountain from which all samples were collected. The photograph was taken from an elevation of 4300 m, and the view is straight up the slope transect

(Courtesy: Zhong Wang).

Download figure to PowerPoint

Soil physicochemical analyses

Soil moisture was gravimetrically determined with drying at 105 °C for 12 h. pH was determined with a soil to water ratio of 1 : 2.5. Soil (clay/sand/silt percentage) was digested with H2O2, and then, soil particle distribution was determined with Microtrac S3500 analyzer (Microtrac). Soil inline image and inline image were extracted from fresh soil with 2M KCl solution (including absorbed nitrogen) and determined by a continuous flow analyzer (SAN++; Skalar, Breda, the Netherlands). Total nitrogen (TN) was determined using a modified Kjeldahl method. Total organic carbon (TOC) was measured with air-dried solid soil using TOC analyzer (TOC-VCPH, Shimadzu, Japan).

Soil DNA extraction, amplification of bacterial 16S rRNA gene, and terminal restriction fragment length polymorphism (T-RFLP) analysis

DNA was extracted from triplicate soil samples, each consisting of 0.5 g soil (wet weight), using a protocol described earlier (Bürgmann et al., 2003). Polymerase chain reaction (PCR) amplification was performed with the primer pair 27F/907R (Lane, 1991). The forward primer was labeled with the fluorescent dye 6-FAM (6-carboxyfluorescein; Invitrogen, Shanghai, China) at the 5′ end. Each PCR (50 μL) contained 5 μL 10 × PCR buffer (Takara), 2.5 mM MgCl2, 0.25 mM of each dNTP (Takara), 0.5 mM of each primer, 2.5 units Taq DNA polymerase, and 2 μL genomic template DNA. The PCRs were carried out as follows: after an initial 3-min denaturation at 94 °C, 30 cycles were run at 94 °C for 60 s, 54 °C for 60 s, and 72 °C for 60 s, followed by a final extension at 72 °C for 5 min. PCR products were purified with purification kits (TIANGEN, Beijing, China) and digested with the restriction enzyme MspI (Takara). The enzyme digestion was performed under reaction conditions of 40 μL containing 4 μL digestion buffer (Takara), 40 units of restriction enzyme, and c. 200 ng of the purified PCR product. Digestion product was precipitated overnight at −20 °C after adding 120 μL ethanol (100%) and 4 μL of 3M sodium acetate (pH 5.2). The pellets were washed with ethanol (70%, v/v) and resolved in sterilized double-distilled water. A portion of the purified digests was mixed with deionized formamide and internal standard of GeneScan-Rox 1000 (Applied Biosystems, Foster City, CA). The mixtures were denatured at 95 °C for 3 min, and DNA fragments were size-separated using 3130xl Genetic Analyzer (Applied Biosystems). T-RFLP patterns were produced using the genemapper software (version 3.7; Applied Biosystems), and peaks at positions between 50 and 850 bp were selected because all T-RFs fell in this range.

Cloning and sequencing

T-RFLP profiles showed that bacterial 16S rRNA gene T-RF patterns were quite similar when samples were collected from seven sites with three different levels of soil depth, and samples collected from 10 cm to 20 cm depth contained the most abundant T-RF types. Therefore, PCR products of bacterial 16S rRNA gene retrieved from 10 cm to 20 cm depth at elevation of 4950 m were used for clone library construction. PCR amplification used the same primer pairs (without FAM) as for T-RFLP. The PCR products were purified and ligated into the pTEM-T Easy Vector according to the manufacturer's instructions (Promega). Ligation mix was then transformed into Escherichia coli JM109 competent cells following the manufacturer's instructions. Afterward, 150 clones were randomly selected and sequenced with an ABI 3730xl (Applied Biosystems). After chimera check, we obtained a total of 125 available sequences for phylogenetic analysis and T-RF assignment. The coverage of the clone library was checked by rarefaction analysis using the dotur software program (Schloss & Handelsman, 2005).

Sequences were aligned with clustalw multiple alignment tool, and distance matrix analysis was performed with dnadist DNA distance matrix tool of bioedit software (Hall, 1999). Sequences sharing a 97% 16S rRNA gene nucleotide sequence similarity were identified as one operational taxonomic unit (OTU), and only one representative sequence of the same OTU was subjected to in silico digestion analysis using software mapdraw, version 5.0, one of the seven programs in the Lasergene suite (DNASTAR Inc.). The obtained results were a series of sequence fragments divided at the first restriction enzyme site (C^CGG) of MspI from the end of primer 27F.

Data manipulation and statistical analysis

Raw T-RFLP data were standardized and normalized based on t-rex program (http://trex.biohpc.org/). The relative abundance (Ap) of terminal restriction fragments (T-RFs) was calculated as the percentage of an individual T-RF in the sum of all concerned peak heights in a T-RFLP profile given by t-rex program. The ordination analyses of T-RFLP profiles were performed using r and canoco, version 4.5 software (Microcomputer Power, NY) (ter Braak & Šmilauer, 1998). We firstly ran the detrended correspondence analysis to estimate the gradient length of variables and found out that in most cases, the longest gradient was shorter than 2.0. Therefore, linear model approach, redundancy analysis (RDA) was conducted, combing the T-RFLP profiles with the log (x + 1)-normalized physicochemical, biotic, and climatic data obtained except soil pH. The Euclidean distance between samples roughly corresponds to their dissimilarity (Leps & Smilauer, 2003). canonical correspondence analysis (CCA) analysis was conducted to analyze the relationship of T-RFLP profiles with environmental factors along the soil depth profiles and vegetation types. Correlation analysis was performed with spss 16.0 (SPSS Inc.).

Nucleotide sequence accession numbers

All sequence data generated from this study have been deposited in the EMBL nucleotide sequence under accession numbers JX967581JX967705.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

Soil physicochemical properties along the altitudinal transect

Along the altitudinal transect, inline image increased with altitude until it reached the highest value (206 μg g−1 dry weight soil) in topsoil (0–5 cm) at the elevation of 5100 m (Table 2). Soil pH decreased with altitude at all soil depths, whereas soil pH generally showed higher pH values (6.56–7.02) at elevations of 4400 and 4500 m, followed by other five soil profiles (5.38–6.32; Table 2). Along the soil depth profiles, TN and C/N ratio at all sites tended to decrease along the soil depth profiles, whereas pH remained relatively constant along soil depth profiles (Table 2).

Table 2. Characteristics of the sites and major physicochemical parameters of the sampled soils
SitesSoil depthpH (H2O)Moisturea (%)TOCa (%)TNa (%)C/N ratioinline image (μg g−1)inline image (μg g−1)Texture (%)
ClaySiltSand
  1. TOC, total organic carbon; TN, total nitrogen.

  2. a

    Moisture: soil moisture.

4400 m0–5 cm6.67.51.10.1861.041.50.848.750.5
5–10 cm6.89.71.40.1793.132.91.654.444.1
10–20 cm7.010.10.50.1551.724.81.057.341.8
4500 m0–5 cm6.77.73.60.32100.341.90.463.136.5
5–10 cm6.814.42.60.22110.133.81.055.443.6
10–20 cm6.811.51.30.09100.230.51.453.844.8
4650 m0–5 cm6.014.06.50.45140.164.90.853.845.4
5–10 cm6.112.52.90.27100.142.71.459.938.7
10–20 cm6.114.62.50.15111.238.31.056.142.9
4800 m0–5 cm6.014.56.80.62141.790.30.866.432.8
5–10 cm5.917.53.60.34100.740.20.968.131.0
10–20 cm5.918.13.10.23110.933.30.968.330.8
4950 m0–5 cm5.420.68.70.68141.258.60.967.032.1
5–10 cm5.923.04.60.33171.134.41.470.428.2
10–20 cm6.018.91.60.2570.432.50.972.826.2
5100 m0–5 cm5.945.617.51.04248.7206.61.071.227.7
5–10 cm5.636.55.90.62129.773.41.077.121.9
10–20 cm5.828.54.30.321111.449.31.672.725.6
5200 m0–5 cm5.940.010.20.841513.8120.30.865.433.8
5–10 cm6.030.94.40.721012.142.31.071.827.2
10–20 cm6.332.13.10.321111.835.41.360.538.2

Bacterial community distribution

A T-RFLP fingerprinting approach was used to determine the distribution of bacterial communities along the altitudinal transect from 4400 m to 5200 m a.s.l. In total, 63 T-RFLP profiles (7 elevations × 3 soil depths × 3 replicates) were analyzed, and fifty-one T-RFs were selected as consistent profiles across all soil samples. To differentiate the differences in the overall distribution patterns of bacterial communities in each sample, the rank and the relative abundance of the T-RFs were calculated and plotted (Fig. 2). A Kolmogorov–Smirnov test was performed on all possible sample pairs to test whether they shared the same distribution pattern (reverse ‘J’ pattern), and the test results showed that, accounting for 80.5% of all cases, any two random samples were drawn from the same distribution at a 0.05 significance level.

image

Figure 2. Rank–abundance plots of the T-RFLP profiles of bacteria. The y-axis shows the relative abundance of each T-RF, whereas the x-axis is the ordinal rank of the T-RFs from most abundant (1) to least abundant (n).

Download figure to PowerPoint

Bacterial community composition

Most detected T-RFs in T-RFLP analysis could be identified from the constructed clone library (Table 3). The rarefaction curve was far from saturation with a homologous coverage of c. 54% according to calculation formula in a previous study (Kemp & Aller, 2004). Phylogenetic analyses of 125 sequences showed that the bacterial communities in this alpine grassland soil consisted of Acidobacteria, Proteobacteria, Gemmatimonadetes, Actinobacteria, Nitrospira, Planctomycetes, Verrucomicrobia, and Bacteroidetes (Fig. 3). Acidobacteria were the most abundant organisms (47 clones, 38%), followed by Proteobacteria (38 clones, 30%), Gemmatimonadetes (9 clones, 7%), Planctomycetes (5 clones, 4%), Actinobacteria (3 clones, 3%), Nitrospira (3 clones, 2%), Bacteroidetes (3 clones, 2%), and Verrucomicrobia (2 clones, 2%). The rest were of unidentified lineages (15 clones, 12%).

Table 3. Assignment of selected T-RFs to defined bacterial lineages
T-RF (bp)Affiliation
69Actinobacteria/Gemmatimonadetes
71, 121, 123, 138, 432, 489, 491 Betaproteobacteria
126, 668 Nitrospira
127Acidobacteria GP8
129Acidobacteria GP7
135Actinobacteria/Deltaproteobacteria
137 Gemmatimonadetes
145Acidobacteria GP2/Gammaproteobacteria
147Acidobacteria GP4
148, 437 Alphaproteobacteria
157 Proteobacteria
199, 287, 289Acidobacteria GP6
213 Verrucomicrobia
263Acidobacteria GP1
294Acidobacteria_GP4/GP6
473Acidobacteria GP11
492Betaproteobacteria/Gammaproteobacteria
503 Actinobacteria
508 Deltaproteobacteria
517Acidobacteria GP16
image

Figure 3. Phylogenetic affiliation and percentage of clone numbers of bacterial 16S RNA genes retrieved from alpine meadow soil at elevation of 4950 m profile at 10–20 cm soil.

Download figure to PowerPoint

Changes in bacterial communities across sites along the altitudinal gradient

Although fifty-one T-RFs in total were selected as consistent profiles across the samples, only thirty T-RFs with an Ap higher than 5% in at least 1 of 63 profiles were used for the following analyses. This cutoff was successfully adopted in previous studies (Noll et al., 2005; Rui et al., 2009). This screening process neglected the contribution of low-abundance T-RFs, which consisted of no more than 20% of total signals in the present study. Overall, bacterial community diversities showed no significant difference among seven sites (Supporting Information, Fig. S1A). Moreover, nonmetric multidimensional scaling analysis (NMDS) showed that bacteria community compositions showed no apparent difference among seven sites along the altitudinal gradients (Fig. S1B).

Relationship between bacterial communities and environmental factors along the altitudinal transect

Using the forward selection RDA of T-RFLP profiles, the bacterial community distribution in different soil samples and their relationships with environmental factors [i.e. mean annual precipitation (MAP), mean annual temperature (MAT), height of plant community (HPC), plant species richness (PSR), soil pH, inline image, etc.] along the altitudinal transect were analyzed (Fig. 4). The total percentage of variances explained by the first two axes reached up to 29.1%, 33.8%, and 28.7% of the total variances of bacterial community at each soil depth (0–5, 5–10 and 10–20 cm), respectively, passing the Monte Carlo test with 999 permutations (< 0.01; Fig. 4).

image

Figure 4. RDA of the T-RFLP fingerprints for the effect of environmental variables on the composition of the bacterial community along altitudinal gradients at different soil depths from 4400 m to 5200 m a.s.l. (a) 0–5 cm soil; (b) 5–10 cm soil; (c) 10–20 cm soil.

Download figure to PowerPoint

At 0–5 cm soil depth, five environmental factors (including MAP, inline image, pH, HPC, and PSR) were chosen for RDA analysis as their VIFs were all lower than 20 among all measured environmental factors (Fig. 4a). The first axis accounted for 17.3% of the total variation, and both inline image and MAP significantly correlated with the first axis (r = 0.78 and r = 0.60, < 0.01). Samples from the elevation of 5100 and 5200 m were obviously located on the right side of the first axis, while the sampling sites below 5100 m were clearly located on the left side of the first axis. The second axis accounted for 11.8% of the total variation. HPC significantly correlated with the second axis (r = 0.40, < 0.05). Among the above five environmental factors, pH, HPC, inline image, and PSR all had significant correlation with MAP (r = −0.832, −0.824, 0.638, and 0.569, respectively, < 0.01; Table S1A). Thus, both MAP and inline image might be the primary environmental factors that would have affected altitudinal distribution of soil bacterial communities at 0–5 cm soil depth.

At 5–10 cm soil depth, six environmental factors (pH, inline image, inline image, C/N ratio, HPC, and PSR) were selected for RDA analysis based on the forward selection method (Fig. 4b). The first axis accounted for 19.6% of the total variation, and C/N ratio significantly correlated with the first axis (r = 0.71, < 0.01). However, soil samples were mainly dispersed along the second axis accounting for 14.2% of the total variation. Specifically, samples from the elevations of 4400 and 4500 m were obviously located at the upper side of the second axis, while the sampling sites above 4650 m were clearly located at the lower side of the second axis. Furthermore, pH showed the highest correlation with the second axis (r = 0.821, < 0.01), followed by inline image and HPC (r = −0.725 and 0.616, respectively). Among the six environmental factors, both HPC and inline image showed strong positive correlation with pH (r = 0.839 and −0.794, < 0.001; Table S1B). Therefore, pH might be the most important factor that drives the altitudinal distribution patterns of soil bacterial communities at 5–10 cm soil depth.

At 10–20 cm soil depth, five environmental factors (pH, inline image, inline image, C/N ratio, and PSR) were chosen for RDA analysis as their VIFs were lower than 20. Soil samples were mainly dispersed along the first axis, which accounted for 20% of the total variation. Specifically, samples from the elevations of 4400 and 4500 m were obviously located at the right side of the first axis, whereas the sampling sites above 4650 m were clearly located at the left side of the first axis (Fig. 4c). Moreover, pH showed the highest correlation with the first axis (r = 0.813, < 0.01), followed by PSR (r = −0.662, < 0.01) and inline image (r = −0.61, < 0.01; Table S1C). The second axis accounted for 8.7%, and inline image strongly correlated with the second axis (r = 0.627, < 0.05; Table S1C). Among the five environmental factors, both inline image and PSR showed strong correlation with pH (r = −0.8 and −0.704, respectively, < 0.001; Table S1C). It, thus, appears that pH probably drives altitudinal distribution of soil bacterial communities at 10–20 cm soil depth.

Relationships between bacterial communities and environmental factors along soil depth profiles

All soil samples clustered as a function of depth (0–5, 5–10, and 10–20 cm) along the primary ordination axis, which explained 34% variation in the CCA biplot (Fig. 5a). The significance of the clustering was independently confirmed by NMDS combined with multiple response permutation procedure significance testing (P < 0.01; Fig. S2A and Table S2). Pearson correlation analysis showed that TN had the highest correlation with the first axis of CCA biplot (r = −0.57, < 0.001; Fig. 5c), followed by ammonium concentration (Fig. S2B). For instance, samples at the 0–5 cm soil depth were associated with high TN compared with samples from deeper layers. Therefore, our results suggest that TN may be a main soil factor that directly determines the observed depth-related changes in soil bacterial community compositions.

image

Figure 5. Relationships between bacterial community composition and environmental and spatial factors. (a) and (b): the CCA. (c) and (d): correlation between TN (b), C/N ratio (z-scores) (d), and CCA axes, respectively. Percentage values on axes represent cumulative percentage variation of species–environment relation explained by consecutive axes.

Download figure to PowerPoint

Furthermore, we also found an obvious clustering of soil bacterial communities along the secondary axis of CCA biplot according to vegetation types (adonis < 0.001; Fig. 5b). Specifically, bacterial communities from the alpine steppe vegetation region (altitude< 4650 m) where Stipa spp. was generally dominant clustered together, as do bacterial communities from the alpine meadow vegetation region (altitude > 4650 m) in which Kobresia spp. was dominant. Pearson correlation analysis also showed that C/N ratio had significant correlation with the second axis of CCA biplot (r = −0.57, < 0.001; Fig. 5d).

Discussions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

Based on T-RFLP analysis, 51 bacterial T-RFs were screened from collected soil samples. However, only a few phylotypes were abundant, whereas most of the phylotypes were relatively rare (Fig. 2), exemplifying the classic ‘long tail’ phenomenon (Fuhrman, 2009). Clone library showed that the division Acidobacteria and Proteobacteria dominated the Nyainqentanglha Mountains soil (Fig. 3). This is in agreement with some previous studies, such as studies carried out in Northern Fennoscandia tundra soils (Männistö et al., 2007), Arctic tundra soil (Chu et al., 2010, 2011), and Brazilian Atlantic forest soil (Bruce et al., 2010). Although Acidobacteria are commonly detected in soils, they are rarely isolated under laboratory conditions. The 47 cloned acidobacterial sequences represented at least 11 subdivisions of Acidobacteria with subdivision 1, 4, and 6 as dominant phylotypes, which is in line with some previous findings (Janssen, 2006). Furthermore, both clone library and T-RFLP fingerprinting patterns indicated that Alpha- and Betaproteobacteria were also abundant on Nyainqentanglha Mountains, and sequences related to Beijerinckiaceae and Burkholderiales matched with dominating T-RFs. Beijerinckiaceae are a family of Rhizobiales, which are free-living nitrogen-fixing bacteria. Burkholderiales are extremely diverse, and so are the metabolic capacities known for certain members (Kautz et al., 2013) and appear to be copiotrophs that require adequate moisture (Lennon et al., 2012).

Soil bacterial diversity exhibited no apparent altitudinal patterns along the south-facing slope of Nyainqentanglha Mountain and did not display unimodal pattern commonly found in plants or monotonous pattern commonly found in invertebrates either (Wang et al., 2011, 2013). This suggests that bacterial diversity may not follow the patterns of macroorganisms. Another possibility is that the low resolution of T-RFLP technique obscured the inherent distribution patterns of bacterial community along the altitudinal gradient. Nevertheless, using high-throughput pyrosequencing technique, Fierer et al. (2011) found out that bacteria living in three distinct habitats (organic soil, mineral soil, and leaf surfaces) exhibited no significant altitudinal gradient in diversity, which was in contrast to the significant diversity changes observed for plant and animal taxa across the same montane gradient. Therefore, it is likely that community T-RFLP method might not be the major cause of altitudinal distribution patterns of bacterial diversity on Nyainqentanglha Mountains. However, higher resolution techniques targeting bacterial community diversity are still needed to help exclude the possible effect of fingerprinting limitations.

Primary productivity in more than half of the world's ecosystems has been substantially limited by water availability (Heimann & Reichstein, 2008). Previous studies have shown that an optimum combination of temperature and precipitation would determine the altitudinal distribution center of cushion species and other dominant plant species at the same sites (Li et al., 2013; Wang et al., 2013). In this study, numerous environmental factors were measured (Tables 1 and 2), including climatic factors (e.g. MAP, MAT, etc.), soil pH and nutrients (e.g. nitrogen, carbon, inline image, inline image), and plant variables [HPC, coverage of plant community (CPC), PSR]. The distribution pattern of bacterial community at the top soil along the altitudinal transect mainly correlated with MAP and inline image (Fig. 4a), but did not significantly correlate with temperature, which is in line with Fierer & Jackson (2006). Previous studies have shown that soil moisture at the pore scale determined the access of microbial communities to soil organic matter (Schimel & Schaeffer, 2012). As a result, microorganisms are quite sensitive to soil moisture and precipitation variability (Blankinship et al., 2011; Hawkes et al., 2011). In our study, soil moisture showed significantly positive correlation with precipitation (r = 0.739, < 0.001) and also significantly correlated with inline image (r = 0.668, < 0.01) when precipitation effect was controlled. In other arid or semi-arid or Mediterranean regions, such as in the Chihuahuan Desert or across Israel, the diversity of soil microbial communities all followed the precipitation patterns with mean annual temperatures varied (Clark et al., 2009; Angel et al., 2010; Bachar et al., 2010). We speculate that the structure of bacterial communities may be comparable at top soils among different altitudes because of a combination of selection pressure exerted by precipitation and the availability of inline image nutrient released through the mineralization of soil organic matter. Hence, our results suggest that precipitation and inline image mainly influence the altitudinal distribution changes in bacterial communities in this semi-arid region. Precipitation, rather than temperature, could be the major driving force on the composition of soil bacterial communities as well as on that of plant communities in this semi-arid region.

Compared with top soil, pH was found to be mainly correlated with the altitudinal distribution of bacterial community compositions at 5–10 cm and 10–20 cm soil (Fig. 4b and c). This is in line with previous studies that soil pH was a very important environmental factor that determined spatial distribution of bacterial community structure in arable soil (Rousk et al., 2010), grassland soil (Kuramae et al., 2011), alkaline lake sediments (Xiong et al., 2012), mountain surface soil (Shen et al., 2013), and marine environment (Meron et al., 2012). However, all these extant studies were typically drawn from studies performed on horizontal soils or surface soils. In this study, soil pH was found to be the primary factor controlling bacterial composition at subsurface soils across the large altitudinal gradient. Taken together, all these findings might suggest that pH is a universal predictor of bacterial distribution.

Furthermore, soil pH is a complex parameter that relates to the function of parent material, time of weathering, vegetation, and climate. Because parent material and time of weathering are similar for all sites along the altitudinal transect, the trend of decreasing soil pH with altitude might be due to the difference in vegetation cover and an increase in leaching of basic cations from greater precipitation at higher elevations. Indeed, correlation analysis showed that pH had a close relationship with vegetation cover (on behalf of HPC and PSR) and inline image at subsurface soils (5–10 cm and 10–20 cm; Table S2).

The soil depth profiles represent strong environmental gradient with multiple edaphic factors changing simultaneously (Fierer et al., 2003; Cao et al., 2012), and some previous studies have shown that soil depth was an important spatial factor in determining microbial community assembly (Kemnitz et al., 2007; Hansel et al., 2008; Cao et al., 2012; Eilers et al., 2012). Although soil layers were thin (about 25 cm in thickness) at our study sites, when we compared soil bacterial community compositions from three soil depths (0–5, 5–10, and 10–20 cm), bacterial community structures were clustered depending on the level of soil depth. Based on CCA analysis, the depth-related clustering resulted from vertical changes in soil nutrient availability, such as TN and inline image content. Depth-related decreases in nitrogen-related nutrient availability indicate vertical shifts in the nutritional quality of soil organic matter and difference in decomposition rates of soil organic matter in the grassland soil, which have been shown earlier to influence bacterial community structure (Huang et al., 2011). However, the majority of previous studies have shown that depth-related soil organic C availability mainly contributed to the depth-change-related changes (Fierer et al., 2003; Goberna et al., 2005; Kemnitz et al., 2007; Ganzert et al., 2011). Hence, comparing with other regions, we speculate that nitrogen availability is the primary limiting factor that determines bacterial community structure along soil depth profiles in this alpine region, which is in line with findings by Chu et al. (2011) that there was a close association between bacterial and N availability across low Arctic tundra.

Vegetation type controls the quantity and quality of carbon and nutrients available for soil microorganisms through litter and exudates (Wardle et al., 2004; Bragazza et al., 2007; Eskelinen et al., 2009). Many studies have shown that soil bacterial communities differed dramatically among vegetation types (Angel et al., 2010; Shen et al., 2013), whereas at a landscape scale, Nielsen et al. (2010) found that soil bacterial community composition was not directly associated with plants, but determined by soil pH and C/N ratio. In our study, soil bacterial community compositions from two vegetation types are mainly distributed along soil C/N ratio gradient (Fig. 5c), and soil C/N ratio has significant correlation with the second axis of CCA ordination plot (< 0.001; Fig. 5d), which is in agreement with the results observed in Arctic tundra soils (Chu et al., 2011) and Changbai Mountains (Shen et al., 2013). The clustering of soil samples according to vegetation belts (alpine steppe VS alpine meadow) suggests that vegetation type may influence bacterial community composition through altering soil C and N availability along the large altitudinal gradient. Therefore, combining with previous studies, it seems possibly that soil C/N ratio is an important indicator that predicts bacterial communities from different vegetation belts in the alpine biomes across the globe.

Climate warming would generally cause higher temperature and varied humidity change in the future, and the most immediate effect of climate change would be the shifts in species geographical range (Thuiller, 2007). Some studies have pointed out that climate warming would increase the mortality of cushion species and move the optimum distribution center of alpine meadow upwards (Li et al., 2013; Wang et al., 2013). However, few studies have examined the sensitivity and response of Tibetan bacterial community to recent and future climatic warming. Considering the impact of environmental factors on the change in bacterial communities at surface soil along the large altitudinal gradients, we speculate that an increase in temperature might not significantly change the structure of soil bacterial community along the large altitudinal gradient, whereas changes in precipitation patterns would do so in this semi-arid area.

In conclusion, our study revealed that soil bacterial community was significantly different along the large altitudinal gradient, although the dominant environmental driving factors varied between topsoil and subsurface soil. Specifically, precipitation and soil inline image mainly determined the differences of bacterial community composition at the topsoil depth along the altitudinal gradients, whereas pH mainly determined the differences of bacterial community composition at subsurface soil. Furthermore, soil depth and vegetation also have significant influence on structuring soil bacterial community.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was funded by 973 Project from Science & Technology Department of China (2010CB951304-3), the Knowledge Innovation Program of the Chinese Academy of Sciences (KZCX2-EW-112), and the 100-Talent Program of the Chinese Academy of Sciences.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussions
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
fem12197-sup-0001-TableS1-S2_FigS1-S2.docxWord document102K

Table S1. Correlation matrix among RDA axes, environmental axes and environmental variables whose VIF lower than 20.

Table S2. Dissimilarity analysis among three soil depth (0–5 cm, 5–10 cm, 10–20 cm) based on IEG websites (http://ieg.ou.edu/).

Fig. S1. (A) Changes in bacterial diversity across the altitudinal gradient are shown in box plot; (B) Two-dimensional plots of non-metric multidimensional scaling analysis (NMDS) for the entire T-RFLP profiles according to altitude category.

Fig. S2. (A) Two-dimensional plots of non-metric multidimensional scaling analysis (NMDS) for the entire T-RFLP profiles according to soil depth category. (B) First canonical axis of the CCA analysis of bacteria T-RFLP profiles versus ammonium concentration for each sample. The axis explains 34% of the variability in the data for bacteria.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.