High diversity and distinctive community structure of bacteria on glaciers in China revealed by 454 pyrosequencing

Systematic and Applied Microbiology 38 (2015) 578–585

Contents lists available at ScienceDirect

Systematic and Applied Microbiology journal homepage: www.elsevier.de/syapm

High diversity and distinctive community structure of bacteria on glaciers in China revealed by 454 pyrosequencing Qing Liu, Yu-Guang Zhou, Yu-Hua Xin ∗ China General Microbiological Culture Collection Center, Institute of Microbiology, Chinese Academy of Sciences, Beijing 100101, PR China

a r t i c l e

i n f o

Article history: Received 21 June 2015 Received in revised form 15 September 2015 Accepted 21 September 2015 Keywords: Bacterial community Glaciers Diversity Next-generation sequencing Climate Geographical distance

a b s t r a c t The bacterial diversity, community structure and preliminary microbial biogeographic pattern were assessed on glacier surfaces, including three northern glaciers (cold glaciers) and three southern glaciers (temperate glaciers) in China that experienced distinct climatic conditions. Pyrosequencing revealed that bacterial diversities were surprisingly high. With respect to operational taxonomic units (OTUs), Proteobacteria was the most dominant phylum on the glacier surfaces, especially Betaproteobacteria. Significant differences of the bacterial communities were observed between northern and southern glacier surfaces. The rare and abundant populations showed similar clustering patterns to the whole community. The analysis of the culturable bacterial compositions from four glaciers supported this conclusion. Redundancy analysis (RDA) and partial Mantel tests indicated that annual mean temperature, as well as geographical distance, was significantly correlated with the bacterial communities on the glaciers. It was inferred that bacterial communities on northern and southern glacier surfaces experienced different climate, water and nutrient patterns, and consequently evolved different lifestyles. © 2015 Elsevier GmbH. All rights reserved.

Introduction Glaciers represent extreme low temperature environments in the world, which cover more than 15 million square kilometres and occupy approximately 10% of the Earth’s total land area. Most glaciers are located in the polar regions, and a few are located in mountainous areas above the snow line (http://nsidc. org/cryosphere/glaciers). The characteristics and types of glaciers are determined by the climate [31]. According to thermal characteristics, glaciers can be classified into three types: cold, temperate (warm) and polythermal. The temperate glaciers are at melting point throughout the year from the surface to the base [31]. However, cold glaciers are below their melting point from top to bottom, except in certain cases where the temperature of the surface layer can reach the melting point seasonally. This type of glacier only exists in regions with high latitude and high altitude where the mean annual temperature is extremely negative [27]. The cold and low nutrient concentrations make microbial communities on glaciers special and complex. The study of glacier microorganisms would yield valuable information concerning

∗ Corresponding author at: No. 1 West Beichen, Chaoyang District, Beijing 100101, PR China, Tel.: +86 10 64807586; fax: +86 10 64807585. E-mail addresses: [email protected] (Y.-G. Zhou), [email protected] (Y.-H. Xin). http://dx.doi.org/10.1016/j.syapm.2015.09.005 0723-2020/© 2015 Elsevier GmbH. All rights reserved.

microbial community composition, the mechanisms for coldadaptation of microbial cells and the origin, evolution and limit of life on Earth. As the biodiversity of microorganisms in cold environments has been receiving more attention, plenty of knowledge of the bacterial communities on glaciers have been obtained based on both the use of culture-dependent and culture-independent approaches [1,29,32,33,37,38,43]. Amato et al. [1] summarised that bacteria from cold environments belonged to the Proteobacteria group (mainly Alphaproteobacteria, Betaproteobacteria and Gammaproteobacteria), Cytophaga–Flavobacterium–Bacteroides (CFB) and low G + C (LGC) and high G + C (HGC) Gram-positive bacteria. Several investigations of the microbial community have revealed some abundant genera, such as Arthrobacter, Hymenobacter, Deinococcus, Cryobacterium, Flavobacterium, Polaromonas and Sphingomonas [26,32,36]. However, because of methodological limitations, previous studies have reported mainly on community composition rather than structure. Few microbial investigations have employed 454 pyrosequencing of the 16S rRNA gene in order to obtain a snapshot of the microbial community structure on the surfaces of mountain glaciers. Currently, little biogeographic information concerning the biosphere on glacier surfaces is available [2,28], although Xiang et al. [44] found different bacterial community patterns between two glaciers on the Tibetan plateau. Therefore, further studies should be performed to understand better the biodiversity and bacterial biogeography on glacier surfaces.

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

579

Fig. 1. The sampling sites in this study: NO1, TM and QY are located in northwest China; MD, HLG and MY are located in southwest China.

There are 48,571 glaciers [25] from southwest to northwest China, all of which are located in mountainous areas. Northwest glaciers are cold glaciers under the influence of a continental climate, and their constant temperature stratums achieve negative temperatures. However, southwest glaciers are temperate glaciers under the influence of a maritime climate, and their constant temperature stratums are at the pressure melting point [18]. The glaciers in these two geographic regions enabled us to characterise the bacteria compositions and compare their biogeographic distributions under these different climates. In this context, 454 pyrosequencing of the 16S rRNA gene was employed to gain an in-depth insight at a greater resolution into the biodiversity of bacterial communities on the surfaces of both northwest and southwest glaciers. From the biodiversity surveys and microbial biogeographic comparisons, we tried to evaluate the influences of environmental (physicochemical and climatic) variables and geographic characteristics on the biodiversity, as well as, finally, to discover a possible preliminary microbial biogeographic pattern on the glacier surfaces. Materials and methods Sample collection and description of sampling sites Six glaciers were selected for sample collection: Xinjiang No. 1 Glacier (NO1), Qiyi Glacier (QY), Toumingmengke Glacier (TM), Hailuogou Glacier (HLG), Mingyong Glacier (MY) and Midui Glacier (MD) (Fig. 1, Table 1). NO1, QY and TM are located in northwest China, whereas HLG, MY and MD are located in southwest China. Surface ice from all six glaciers was collected from the glacier tongues in September 2011 using three replicates that were 100 m apart. The samples that harboured surface dusts were chipped to approximately 0–20 cm thick by a sterile hatchet. Sterile sample bottles were used to transport the ice samples in an insulated container with dry ice and they were then stored at –80 ◦ C in

the laboratory. Ice samples were melted at ambient temperature before filtration. Three replicates of each sample were mixed and filtered together onto sterile cellulose acetate membranes (pore size 0.22 ␮m; Millipore) using the pump and filtration device for DNA extraction and the determination of physicochemical compositions. For avoiding cross-contamination, the filtration unit was rinsed three times with sterile Veolia water before each sample filtration. HLG is situated on Gongga Mountain, which is in the eastern Hengduan Mountain Range in Sichuan Province. MY is situated on Meri Snow Mountain, which is in the western Hengduan Mountain Range in Yunnan Province. MD is located at the juncture of the Nyenchen Tanglha Mountain Range with the Boshula Mountain chain (the far western chain of the Hengduan Mountain Range) in Tibet. The Hengduan Mountain Range occupies the southeastern part of the Qinghai–Tibet Plateau. As the only mountain range in China under the influence of a high-altitude westerly circulation and the monsoon circulation of the Indian and Pacific oceans, Hengduan Mountain Range has a favourable maritime climate, with very obvious wet and dry seasons. NO1 sits on the Tianshan Mountain Range, which is a large mountain range in Central Asia, running through central Xinjiang, China, with the west end stretching into Kazakhstan. In the area of the Tianshan Mountain Range, there is an obvious continental climate, with distinct cold and warm seasons during the year. TM and QY are in the territory of Gansu province Table 1 Geographic characteristics of the glaciers from northwest and southwest China. Region

Glacier

Latitude

Longitude

Altitude (m; a.s.l.)

Northwest China

NO1 QY TM HLG MD MY

43.12 N 39.25 N 39.50 N 29.56 N 29.45 N 28.45 N

86.81 E 97.75 E 96.52 E 101.97 E 96.50 E 98.76 E

3,838 4,041 4,278 3,457 3,901 2,970

Southwest China

580

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

and are not far from each other. Both glaciers are located in the hinterland of the Qilian Mountain Range. The Qilian Mountain Range is composed of multiple northwest–southeast parallel mountains and wide valleys, and forms a natural barrier in northwest China. It has an obvious continental climate and is covered by thick snow throughout the year.

Analyses of physicochemical and climatic parameters Physicochemical parameters, including pH, total nitrogen, total phosphorus, total carbon and ionic concentration (K+ , Ca2+ , Na+ , Mg2+ , Cl− , NO3 − , HCO3 − ) were determined by the Pony Testing International Group. DIVA–GIS version 7.0 (www.diva–gis.org) was used to extract the 19 bioclimate variables for each sampling site with a spatial resolution of approximately 1 km2 [17]. The Student’s t-test was performed to test whether there were any significant differences between northern and southern glaciers for the bioclimate variables and the physicochemical parameters measured.

Pyrosequencing of barcoded 16S rRNA gene amplicons Genomic DNA was extracted from each sample using the Omega Bio–Tek E.Z.N.A.® Water DNA Kit. The universal primer sets 27f (5 –AGAGTTTGATCCTGGCTCAG–3 ) and 533r (5 –TTACCGCGGCTGCTGGCAC–3 ) were used to amplify the V1–V3 hypervariable regions [42]. PCR products with barcodes were submitted to pyrosequencing on a Roche 454 GS FLX+ platform with titanium chemistry. The raw reads were sorted into different samples according to the barcodes. The data preprocessing and operational taxonomic unit (OTU)-based diversity analysis were performed using modules implemented in the mothur software platform [35]. Sequencing denoising was implemented using pre-cluster to remove sequences likely to be sequencing errors [19]. Chimeric sequences were detected and filtered using ChimeraSlayer [10]. Raw 454 sequences are available from the NCBI Sequence Read Archive under accession number SRP042028.

Classification and statistical analyses The 454 sequences were classified into bacterial taxa using the SILVA and Greengenes databases by a Bayesian approach at bootstrap values of >50%. The effective reads for each sample were clustered into OTUs using the furthest neighbour method, with a complete linkage algorithm. Good’s coverage, rarefaction curves, abundance-based richness estimators (Chao1 and ACE) and diversity estimators (Shannon and 1–Simpson) of the microbial communities were calculated in order to describe the biodiversity of the bacteria. The Student’s t-test was performed to test whether there was a significant difference in ␣-diversity between northern and southern glaciers. The OTUs from all samples together were separated into abundant and rare populations at a cutoff of 10 sequences. The “pheatmap” package in R version 3.2.2 [20,39] was used to draw a heatmap based on the abundant genera (> 0.1%) according to the complete linkage agglomerative clustering method. To analyse the beta diversity, a hierarchical cluster based on Hellinger-transformed Bray–Curtis dissimilarities was used to estimate the community similarities between the samples. A dendrogram inferred from the unweighted pair-group average algorithm was constructed and evaluated using bootstrap analyses based on 1,000 resamplings in PAST software (ver. 2.17b) [14]. Jackknife sample cluster analyses based on unweighted UniFrac pairwise distances were performed with the Fast UniFrac online application [13] and the input maximum-likelihood phylogenetic tree was constructed with FastTree [34]. Unweighted principal

coordinate analysis (PCoA) in Fast Unifrac was also used to infer the bacterial community similarities. Multivariate analyses The relationships of the microbial community with environmental (physicochemical and climatic) factors and latitude were analysed using canonical multivariate analysis with the Vegan package in R version 3.2.2 [30,39]. Detrended correspondence analysis (DCA) was used to determine the appropriate statistical model (linear or unimodal). The length of the longest estimated gradient (<3) implied that a linear model was more suitable for our data. Therefore, redundancy analysis (RDA) was used to analyse the correlations of community composition with environmental variables and latitude. The percentage abundance data of abundant genera (≥0.1%) in each library were used as the species input, and then the environmental variables and latitude were normalised and they served as the environmental input. In the preliminary analysis, environmental variables and latitude with an inflation factor larger than 5 were removed. The inflation factor of 5 and higher has been identified as an indicator of collinearity in multivariate analysis [9]. The most significant variables were selected by RDA using stepwise selection. Monte Carlo permutation tests were used to build the optimal model for the microbe–environment relationship [21]. A partial Mantel test was used to investigate the correlation between the Bray–Curtis distances of bacterial communities and the geographic distance using the Vegan package [30,39]. Results Environmental parameters of the samples on glacier surfaces Surface samples were collected from northern and southern glaciers in China (Fig. 1, Table 1). Table 2 summarises the physiochemical parameters measured and the bioclimate variables extracted. No physicochemical parameters showed any significant differences (Student’s t-test; P > 0.05), whereas most of the climatic variables showed significant differences (Student’s t-test; P < 0.05) between the northern and southern glaciers. The results showed that northern and southern glaciers differed greatly in climate. The temperatures on southern glaciers were higher than those on northern glaciers and, in addition, the southern glaciers received more precipitation than the northern glaciers. Bacterial diversity and community structure analyses of pyrosequencing sequences As shown in Table 3, a total of 72,436 effective sequence reads were obtained from six glacier samples after quality filtering. A total of 2,643 OTUs at the 97% similarity level were defined. The highest number of OTUs (941) was found in MD, while the lowest number of OTUs (356) was found in TM. The Good’s coverage estimators for all six samples were more than 96.0%, which meant the sequence reads represented the majority of bacteria existing in the glacier samples. The rarefaction curves at the 97% sequence similarity level did not reach a plateau, which suggested under-sampling even with the average of 12,073 effective reads. The curves at the 90% similarity level approached saturation, which indicated almost sufficient sampling at the phylum level (Supplementary Fig. S1). The diversity estimates (Chao1, ACE, Shannon index and 1–Simpson index) at a 3% cut-off are summarised in Table 3. The richness of bacteria (Chao1 and ACE) between northern and southern glaciers had no statistically significant difference (Student’s t-test; P > 0.05); however, the diversity of bacteria (Shannon and 1–Simpson) between northern and southern glaciers did show a statistically significant difference (Student’s t-test; P < 0.05). The bacterial diversity on

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

581

Table 2 Physicochemical properties and 19 bioclimatic variables for each sample Bio1, annual mean temperature. Bio2, mean diurnal range (mean of monthly (max temp - min temp)). Bio3, isothermality (bio2/bio7) (×100). Bio4, temperature seasonality (standard deviation × 100). Bio5, maximum temperature of warmest month. Bio6, minimum temperature of coldest month. Bio7, annual range of temperature (bio5bio6). Bio8, mean temperature of wettest quarter. Bio9, mean temperature of driest quarter. Bio10, mean temperature of warmest quarter. Bio11, mean temperature of coldest quarter. Bio12, annual precipitation. Bio13, precipitation of wettest month. Bio14, precipitation of driest month. Bio15, precipitation seasonality (coefficient of variation). Bio16, precipitation of wettest quarter. Bio17, precipitation of driest quarter. Bio18, precipitation of warmest quarter. Bio19, precipitation of coldest quarter.

pH TN (mg L−1 ) TP (mg L−1 ) TC (mg L−1 ) K+ (mg L−1 ) Ca2+ (mg L−1 ) Na+ (mg L−1 ) Mg2+ (mg L−1 ) HCO3 − (mg L−1 ) Cl− (mg L−1 ) NO3 − (mg L−1 ) bio1 (◦ C) bio2 (◦ C) bio3 bio4 bio5 (◦ C) bio6 (◦ C) bio7 (◦ C) bio8 (◦ C) bio9 (◦ C) bio10 (◦ C) bio11 (◦ C) bio12 (mm) bio13 (mm) bio14 (mm) bio15(mm) bio16 (mm) bio17 (mm) bio18 (mm) bio19 (mm)

HLG

MD

MY

NO1

QY

TM

5.63 0.16 1.16 0.50 2.36 1.32 0.53 0.91 28.5 0.15 0.05 4.7 11.9 42 5550 16.7 −11.3 28 11.4 −2.9 11.4 −2.9 1011 201 5 89 576 21 576 21

7.32 0.37 0.08 0.60 0.80 0.76 0.80 1.15 4.8 0.28 0 4.5 12.7 44 5604 17.2 −11.2 28.4 11.4 −1.9 11.4 −2.9 671 145 3 90 393 15 393 17

6.98 0.38 0.22 0.00 1.50 4.68 0.00 2.07 15.8 0.29 0 5.2 10.9 41 5189 20.8 −4 24.8 15.8 3.7 15.8 2.7 797 173 7 81 441 28 441 32

6.45 26.00 0.18 10.20 0.80 1.89 2.80 0.26 12.9 3.63 0.79 −7.6 10.8 28 9752 11 −27 38 4.3 −20.6 4.3 −20.6 316 75 2 98 201 6 201 6

6.92 1.30 0.17 0.00 2.60 25.20 3.30 11.70 20.6 2.42 0.97 −7.7 12.9 33 9171 10.9 −28.2 38.9 4.1 −18.3 3.8 −19.5 285 76 1 107 192 5 192 6

6.25 0.56 1.14 0.90 5.36 11.80 1.04 5.70 115 0.37 0.11 −8.1 13.3 34 9197 10.3 −28.4 38.7 3.5 −18.5 3.5 −19.7 228 60 1 101 153 5 153 6

Fig. 2. Relative abundances of unculturable groups in six glacial samples classified using the Greengenes database. Groups with relative abundance below 2% were summarised in the artificial group “others”. Relative abundances are based on the proportional frequencies of the OTUs that could be classified at the class level.

(8.66%), Deltaproteobacteria (6.13%), Saprospirae (6.05%), Cytophagia (5.68%), Alphaproteobacteria (5.68%) and Gammaproteobacteria (5.03%), followed by Actinobacteria (3.44%) and Sphingobacteria (2.12%). Although these classes were dominant, many differences were present among the six samples. Betaproteobacteria dominated in NO1, HLG, MY and QY, with proportions of 51.0%, 50.48%, 48.6% and 39.95%, respectively, whereas its proportion dropped to 23.6% and 22.21% in TM and MD. The highest unclassified portion was found in MD (12.86%) at the class level, while only 0.72% and 1.47% unclassified OTUs were found in HLG and MY, respectively. All the effective sequences were classified into 167 genera, among which there were 89 genera in HLG, 82 genera in MD, 71 genera in MY, 57 genera in NO1, 40 genera in QY and 53 genera in TM. More genera were found in the samples from southern glaciers than in the samples from northern glaciers. A total of 17 genera were shared by all samples: Polaromonas, Flavobacterium, Herminiimonas, Rhodoferax, Flectobacillus, Terrimonas, Janthinobacterium, Lysobacter, Ferruginibacter, Rhodobacter, Methylibium, Pseudomonas, Methylotenera, Pedobacter, Gemmatimonas, Mycobacterium and Mucilaginibacter (Supplementary Table S1). The genera Polaromonas (21.35%) and Flavobacterium (18.51%) were the most abundant in six samples (Fig. 3a). Flavobacterium accounted for a higher proportion in TM, HLG and MY (14.89–63.30%), whereas it accounted for a lower proportion in QY, NO1 and MD (0.54–4.63%). Noticeably, more than half the effective reads in TM were classified into Flavobacterium but there were less than 1% in MD. Herminiimonas accounted for a higher proportion in NO1 (52.01%) and a much lower proportion in other samples (0.01–4.82%). A total of 34.25% sequences were assigned to Rhodoferax in MY, whereas only 0.14–6.69% sequences belonged to

southern glaciers (HLG, MD and MY) was higher than that on the northern glaciers (TM, QY and NO1). The highest bacterial diversity was found in MD, followed by HLG and MY. The Greengenes and SILVA databases were used to classify the sequences, and when the results were compared most of them were in agreement. However, more sequences could be classified at the genus level using the Greengenes database (85.1%) than the SILVA database (70.38%). Therefore, the final classification employed the Greengenes database. The dominating phyla in the six samples were Proteobacteria (54.33% OTUs in total), especially Betaproteobacteria, followed by Bacteroidetes (23.87% OTUs) and Actinobacteria (4.65% OTUs). A total of 61 groups, including candidate divisions, were identified at the class level (Fig. 2). The most dominant classes on the six glaciers were Betaproteobacteria (35.19%), Flavobacteria

Table 3 Number of sequences and OTUs after filtering, coverage, richness estimates (Chao1, ACE) and diversity indicators (Shannon index, 1-Simpson) in the six samples obtained with mothur at a genetic distance of 3%. Sampling site

No. of sequences after quality filtering

No. of OTUs

Coverage (%)

ACE

Chao1

Shannon

1-Simpson

HLG MD MY NO1 QY TM

10918 12138 11351 12309 12433 13287

830 941 475 653 378 356

96.3 96.2 98.3 97.3 98.6 98.8

1765 2181 677 1507 830 642

1476 1761 692 1327 647 605

4.67 4.89 3.61 3.00 2.95 2.37

0.98 0.97 0.92 0.74 0.86 0.71

582

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

Fig. 3. Hierarchical clustering based on a distance matrix computed with Bray–Curtis dissimilarity. The dendrogram was constructed with the unweighted pair-group average algorithm and evaluated using bootstrap analyses based on 1,000 resamplings. (a) Clustering including all OTUs. (b) Clustering obtained for abundant OTUs only (>10 reads). (c) Clustering obtained for rare OTUs only (≤10 reads).

Rhodoferax in other samples. In summary, there were many differences of genera abundance between the six glaciers. Additionally, 14.86% of sequences could not be classified into any genus, which indicated that a large number of unknown bacteria existed on the glaciers. Bacterial community similarities To analyse the community similarities of unculturable bacteria, hierarchical clustering based on the Bray–Curtis dissimilarity

was employed (Fig. 4a). The results revealed that TM and QY in the Qilian Mountain Range grouped together. They also grouped together with NO1, indicating that the bacterial communities from the northern glaciers were similar. In addition, HLG and MY clustered together and were separated from TM/QY/NO1, indicating that the bacterial communities on HLG and MY from the Hengduan Mountain Range were most similar but were different from the northern glacier bacterial communities. In particular, MD from southern China formed an independent branch, which showed that the bacterial community on this glacier was clearly

Fig. 4. (a) Heatmap displaying the relative abundances of the most dominant genera (>0.1%). The dendrogram represents complete-linkage agglomerative clustering based on Euclidean dissimilarities. (b) RDA ordination plot showing the statistically significant relationship between annual mean temperature and bacterial compositions. The genera corresponding to the numbers are listed in the middle. Bio1, annual mean temperature.

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

differentiated from the others. To reinforce the results from hierarchical clustering, Jackknife cluster analysis and PCoA were performed using the online software UniFrac based on molecular evolutionary distances of the partial 16S rDNA sequences. The results of Jackknife cluster analysis and PCoA indicated uniform patterns of community similarity and validated the conclusion of the cluster analysis (Figs. S2a and 2b). Clustering of the top 37 genera (≥0.01%) based on Euclidean dissimilarities also showed a similar pattern to that of the community (Fig. 4a). In six samples, there were 419 abundant OTUs that represented 93.0% of the total sequences. On the other hand, 2,224 rare OTUs comprised 7.0% of the sequences. The clustering patterns of both the rare and abundant OTUs (Fig. 4b and c) were similar to that of the entire community (Fig. 4a). Community correlation with environmental variables and geographic characteristics RDA was performed in order to examine the relationships of bacterial community composition with environmental variables and latitude. Collinearity (strong correlation) between the variables related to temperature, precipitation and latitude was evident (inflation factor >5), which meant the collinear covariates could be disregarded [9]. Surface temperature has been shown to be the key parameter for determining the thermodynamic state of the glacier surfaces [4], since it determines the water and heat conditions that are important for glacier development [18]. Therefore, the strategy for resolving the collinearity problem of our analysis was to disregard two variables: precipitation and latitude. Stepwise forward selection of the physicochemical and temperature variables further reduced the data set to the single variable annual mean temperature. Therefore, the RDA was restricted to the annual mean temperature parameter, and it showed a significant correlation with the community compositions (Monte Carlo permutation test, p = 0.043, Fig. 3b). Axis 1, representing the annual mean temperature gradient, explained 36.7% of the variation in diversity. The result of a partial Mantel test showed a significant correlation between geographical distance and community compositions when the effects of the physicochemical variables and annual mean temperature were removed (r = 0.57, p = 0.025). Discussion Diversity and composition of microbial communities On a glacier surface, microorganisms are often exposed to solar radiation, oligotrophication and low temperatures. In this study, the bacterial communities on glacier surfaces were characterised from three northern and three southern glaciers in China. This represented a comprehensive investigation of bacterial composition and diversity on glacier surfaces using 454 pyrosequencing data, which provided much more information than previous studies [2,26,44,46]. High levels of bacterial diversity were found on glacier surfaces using the 454 pyrosequencing methods, and the taxonomic classification revealed that there were a lot of unknown taxa (14.86% unclassified sequences at the genus level), implying that many novel bacterial taxa existed on the glaciers. In our previous work, 999 bacterial isolates were collected from TM, NO1, MD and HLG sampling sites using a culture method. In order to compare these with the 454 pyrosequencing datasets, the diversity of the culturable bacteria was also analysed (Supplementary Tables S2, S3 and S4). Among the 999 bacterial isolates, numerous novel species were found, some of which have already been validly described [5,22–24,47,48]. Some differences in the bacterial

583

community compositions were also found between the 454 pyrosequencing data and the culturable data. The classes Actinobacteria and Alphaproteobacteria were the most dominant groups according to the culture-based method, which was similar to the conclusions of Peeters et al. [32] and Liu et al. [26]. However, in the pyrosequencing datasets, the most dominant group was Betaproteobacteria (averaging 35.19%), whereas Actinobacteria accounted for only 3.44%. Additionally, only 0.02% of the pyrosequencing sequences were associated with the genus Cryobacterium, which was the major cold-adapted genus in the culturable groups. It was therefore concluded that some taxa might not have been discovered by 454 pyrosequencing in our study, although the coverage exceeded 96.0%. Two possible explanations to explain the fact that only a few Cryobacterium were found in the pyrosequencing datasets can be suggested. First, there may have been low abundance of Cryobacterium in the samples. Second, the biases of genomic DNA extraction from the samples and subsequent PCR amplification may have led to low-level detection. Betaproteobacteria, the most dominant class by pyrosequencing, was also found on the surface of Bench Glacier, John Evans Glacier [38], Kuytun 51 Glacier [44], Northern Schneeferner [37] and a High Arctic polythermal valley glacier [16], which indicated that Betaproteobacteria was prevalent in the superficial environment of glaciers. Perhaps tolerance to oligotrophic environments led to these bacteria being dominant on the cold and barren glaciers [44]. Hell et al. [16] inferred that Betaproteobacteria might be correlated with nitrogen cycling on glacier surfaces and be capable of a rapid response to a dynamic physicochemical environment. Polaromonas was one of the most dominant genera revealed by pyrosequencing in this study, which was consistent with the conclusion that this genus was a dominant taxon in glacier ice and sediments worldwide [8]. Similarity and biogeography of bacterial communities In this study, the significant difference of bacterial compositions on glacier surfaces between northern and southern glaciers in China was discovered by 454 pyrosequencing analysis (Figs. 3 and 4). The rare and abundant populations also showed similar clustering patterns, which indicated that they had a similar geographic distribution to that of the community as a whole on the glacier surface (Fig. 4). For the culturable bacteria, pairwise comparison of communities on the four glaciers using sharedchao and abundance-based approaches (Sørensen’s similarity coefficient and Jaccard similarity coefficient) revealed that the communities on two northern glaciers were most similar, as were the communities on two southern glaciers (Supplementary Table S5). Therefore, the cultivation-based evidence gave support to the result of the 454 pyrosequencing analysis. The influences of environmental variables and geographic characteristics were also investigated on bacterial community compositions, and the results provided some clues to possible interactions of the climatic variables with microbial community structures. By exploring the glacier bacterial communities in China, Xiang et al. [44] concluded that the bacterial communities from glacier surfaces were different and suggested biogeography could strongly influence community structure. Xiang et al. [45] postulated that both aeolian deposition (aerosol, dust and precipitation events) and post-depositional selection processes simultaneously determined the microbial composition in the ecosystems on glacier surfaces. Our study further attested that the different annual mean temperature between northern and southern glaciers may exert a strong selective pressure for the microbial community to adapt and differentiate. In this study, HLG, MY and MD were temperate glaciers that had developed in the southern warm region (the Hengduan Mountains and the east section of the Nyainqentanglha Range) where there is

584

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585

high precipitation. On the other hand, TM, QY and NO1 were cold glaciers that had developed in the northern mountain areas, where the climate is very cold and dry [18]. With different annual mean temperatures, the surface layers of southern and northern glaciers are subjected to different intensities and durations of seasons. The surface layers of southern glaciers are at melting point throughout the year and those of northern glaciers are at subfreezing temperatures except in the summer. Thus, the microbes living on the surface layers of glaciers endure different seasonal temperatures, and their variations lead to different activity and specialised ecophysiological traits. The growth characteristics of some pure isolates provided some evidence for this conjecture. For example, the maximum tolerated growth temperature of Flavobacterium strains isolated from southern glaciers was higher than those from northern glaciers (unpublished). Water and nutrients are two key criteria for the growth of microbes on glacier surfaces. The hydraulic configuration of a glacier surface governs the distribution of water and nutrients throughout the glacier ecosystem and, furthermore, governs the efficacy of a supraglacial ecosystem [6,12,40]. Water can usually be obtained when ice is at the pressure melting point, but melting only occurs under certain thermodynamic conditions that vary both spatially and temporally. Moreover, nutrients are acquired most likely through snowmelt that is influenced by seasonal snow cover or seasonal icemelt, especially in cold glaciers [7,11]. With different annual mean temperatures, the acquisitions of water and nutrients between northern and southern glaciers were different, and they further influenced the structures of microbe communities. Additionally, although annual mean temperature and annual precipitation are collinear, precipitation also drives the ecosystem well [49]. Water and nutrients on the glacier surface can also be supplied by precipitation. This means viable microbial nutrients and water from atmospheric or supraglacial sources are very different between northern and southern glaciers, which will further result in variations of the bacterial communities. In addition, the different climatic zones may receive different bacterial inocula in discrete air masses by precipitation, aerosol and dust, which may also drive the microbial communities on the surfaces of southern and northern glaciers. Altogether, the results of this study showed that a climatically based energy hypothesis may be applicable to the diversity and distribution of microorganisms on glaciers. This hypothesis was also concluded from the diversity and distribution of macroorganisms (animal and plant) and the claims that climate constrains terrestrial taxonomic richness over broad geographic areas [15,41]. The distributions of microorganisms, as well as macroorganisms, also reflect the influence of dispersal limitation. It was obvious that the bacterial communities on the same mountain, such as QY/TM in the Qilian Mountains and HLG/MY in the Hengduan Mountains, were most similar. This provided a clue that mountains could also play a role in limiting dispersal, since natural physical barriers [3] and mountains could affect the bacterial composition and limit species distribution. Hence, dispersal limitation may also be an important factor that contributed to bacterial distribution among the glaciers. Additionally, since the cluster pattern of MD was distinct, it was also speculated that some undetected environmental conditions may affect the structure of bacterial communities on glacier surfaces. Taken together, our data demonstrated the high diversity of bacterial communities on glacier surfaces and significant community differences between northern and southern glaciers. Temperature and precipitation were found to be strong drivers for differentiating the bacterial communities on glaciers, implying the important role of the climate. Additionally, geographic distance may also play a role in shaping their bacterial communities. However, since 16S rRNA gene sequences have limited resolution, the

study of biogeography at the species and strain level could not be conducted in this study. Therefore, in the future, more information concerning the biogeography and ecological adaptation of bacteria on glacier surfaces may be generated using metagenomics at both the taxonomic and the functional levels. Conflict of interest The authors declare no conflict of interest. Acknowledgements This work was supported by grants from the National Natural Science Foundation of China [NSFC, no. 31070001] and the Knowledge Innovation Project of the Chinese Academy of Sciences [grant no. KSCXZ–YW–Z–0937]. The authors gratefully acknowledge Dr. Hong-Can Liu, Bing-Da Sun and Man-Man Wang for help with sample collection on glaciers. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.syapm.2015.09. 005. References [1] Amato, P., Hennebelle, R., Magand, O., Sancelme, M., Delort, A.M., Barbante, C., Boutron, C., Ferrari, C. (2007) Bacterial characterization of the snow cover at Spitzberg, Svalbard. FEMS Microbiol. Ecol. 59, 255–264. [2] An, L.Z., Chen, Y., Xiang, S.R., Shang, T.C., Tian, L.D. (2010) Differences in community composition of bacteria in four glaciers in western China. Biogeosciences. 7, 1937–1952. [3] Croteau, E.K. (2010) Causes and consequences of dispersal in plants and animals. Nature Education Knowledge. 1, 12. [4] Comiso, J.C., Hall, D.K. (2014) Climate trends in the Arctic as observed from space. WIREs Clim. Change. 5, 389–409. [5] Dong, K., Liu, H.C., Zhang, J.L., Zhou, Y.G., Xin, Y.H. (2012) Flavobacterium xueshanense sp. nov. and Flavobacterium urumqiense sp. nov., two psychrophilic bacteria isolated from glacier ice. Int. J. Syst. Evol. Microbiol. 62, 1151–1157. [6] Edwards, A., Anesio, A.M., Rassner, S.M., Sattler, B., Hubbard, B., Perkins, W.T., Young, M., Griffith, G.W. (2011) Possible interactions between bacterial diversity, microbial activity and supraglacial hydrology of cryoconite holes in Svalbard. ISME J. 5, 150–160. [7] Fortner, S.K., Tranter, M., Fountain, A., Lyons, W.B., Welch, K.A. (2005) The geochemistry of supraglacial streams of Canada Glacier, Taylor Valley (Antarctica), and their evolution into proglacial waters. Aquat. Geochem. 11, 391–412. [8] Franzetti, A., Tatangelo, V., Gandolfi, I., Bertolini, V., Bestetti, G., Diolaiuti, G., D’Agata, C., Mihalcea, C., Smiraglia, C., Ambrosini, R. (2013) Bacterial community structure on two alpine debris-covered glaciers and biogeography of Polaromonas phylotypes. ISME J. 7, 1483–1492. [9] Griffith, M.B., Kaufmann, P.R., Herlihy, A.T., Hill, B.H. (2001) Analysis of macroinvertebrate assemblages in relation to environmental gradients in Rocky Mountain streams. Ecol. Appl. 11, 489–505. [10] Haas, B.J., Gevers, D., Earl, A.M., Feldgarden, M., Ward, D.V., Giannoukos, G., Ciulla, D., Tabbaa, D., Highlander, S.K., Sodergren, E. (2011) Chimeric 16S rRNA sequence formation and detection in Sanger and 454-pyrosequenced PCR amplicons. Genome Res. 21, 494–504. [11] Hodson, A. (2006) Biogeochemistry of snowmelt in an Antarctic glacial ecosystem. Water Resour. Res. 42, 11. [12] Hodson, A., Anesio, A.M., Tranter, M., Fountain, A., Osborn, M., Priscu, J., Parry, J.L., Sattler, B. (2008) Glacial ecosystems. Ecol. Monogr. 78, 41–67. [13] Hamady, M., Lozupone, C., Knight, R. (2010) Fast UniFrac: facilitating high–throughput phylogenetic analyses of microbial communities including analysis of pyrosequencing and PhyloChip data. ISME J. 4, 17–27. [14] Hammer, Ø., Harper, D., Ryan, P. (2001) Past: paleontological statistics software package for education and data analysis. Palaeontol. Electron. 4, 1–9. [15] Hawkins, B.A., Field, R., Cornell, H.V., Currie, D.J., Guégan, J.F., Kaufman, D.M., Kerr, J.T., Mittelbach, G.G., Oberdorff, T., O’Brien, E.M. (2003) Energy, water, and broad-scale geographic patterns of species richness. Ecology. 84, 3105–3117. [16] Hell, K., Edwards, A., Zarsky, J., Podmirseg, S.M., Girdwood, S., Pachebat, J.A., Insam, H., Sattler, B. (2013) The dynamic bacterial communities of a melting High Arctic glacier snowpack. ISME J. 7, 1814–1826. [17] Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A. (2005) Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978.

Q. Liu et al. / Systematic and Applied Microbiology 38 (2015) 578–585 [18] Huang, M. (1999) Forty year’s study of glacier temperature distribution in China: review and suggestions. J. Glaciol. Geocryol. 21, 310–316. [19] Huse, S.M., Welch, D.M., Morrison, H.G., Sogin, M.L. (2010) Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environ. Microbiol. 12, 1889–1898. [20] Kolde, R. (2015) pheatmap: Pretty Heatmaps. Version 1.0. ˇ P. (2003) Multivariate analysis of ecological data using [21] Lepˇs, J., Smilauer, CANOCO. Cambridge University Press. [22] Liu, Q., Liu, H.C., Wen, Y., Zhou, Y.G., Xin, Y.H. (2012) Cryobacterium flavum sp. nov. and Cryobacterium luteum sp. nov., isolated from glacier ice. Int. J. Syst. Evol. Microbiol. 62, 1296–1299. [23] Liu, Q., Liu, H.C., Zhan, J.L., Zhou, Y.G., Xin, Y.H. (2013) Cryobacterium levicorallinum sp. nov., a psychrophilic bacterium isolated from glacier ice. Int. J. Syst. Evol. Microbiol. 63, 2819–2822. [24] Liu, Q., Xin, Y.H., Liu, H.C., Zhou, Y.G., Wen, Y. (2013) Nocardioides szechwanensis sp. nov. and Nocardioides psychrotolerans sp. nov., isolated from a glacier. Int. J. Syst. Evol. Microbiol. 63, 129–133. [25] Liu, S.Y., Yao, X.J., Guo, W.Q., Xu, J.L., Shangguan, D.H., Wei, J.F., Bao, W.J., Wu, L.Z. (2015) The contemporary glaciers in China based on the second Chinese glacier inventory. Acta Geographica Sinica 70, 3–16. [26] Liu, Y., Yao, T., Jiao, N., Kang, S., Xu, B., Zeng, Y., Huang, S., Liu, X. (2009) Bacterial diversity in the snow over Tibetan Plateau Glaciers. Extremophiles. 13, 411–423. [27] Lorrain, R.D., Fitzsimons, S.J. (2011) Cold-based glaciers. In: Singh, V.P., Singh, P., Haritashya, U.K. (Eds.), Encyclopedia of Snow, Ice and Glaciers, Springer, Netherlands, pp. 157–161. [28] Margesin, R., Miteva, V. (2011) Diversity and ecology of psychrophilic microorganisms. Res. Microbiol. 162, 346–361. [29] Miteva, V.I., Sheridan, P., Brenchley, J. (2004) Phylogenetic and physiological diversity of microorganisms isolated from a deep Greenland glacier ice core. Appl. Environ. Microbiol. 70, 202–213. [30] Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., O’Hara, R.B., Simpson, G.L., Solymos, P., Stevens, M.H.H., Wagner, H. (2011) Vegan: community ecology package. R Package Version 1, 17-11. [31] Paterson, W.S.B. 1994 The Physics of Glaciers, 3rd edn, Burtterworth Heinemann, Oxford. [32] Peeters, K., Ertz, D., Willems, A. (2011) Culturable bacterial diversity at the Princess Elisabeth Station (Utsteinen, Sør Rondane Mountains, East Antarctica) harbours many new taxa. Syst. Appl. Microbiol. 34, 360–367. [33] Philippot, L., Tscherko, D., Bru, D., Kandeler, E. (2011) Distribution of high bacterial taxa across the chronosequence of two alpine glacier forelands. Microb. Ecol. 61, 303–312. [34] Price, M.N., Dehal, P.S., Arkin, A.P. (2009) FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol. Biol. Evol. 26, 1641–1650.

585

[35] Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J. (2009) Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75, 7537–7541. [36] Schütte, U.M., Abdo, Z., Foster, J., Ravel, J., Bunge, J., Solheim, B., Forney, L.J. (2010) Bacterial diversity in a glacier foreland of the High Arctic. Mol. Ecol. 19, 54–66. [37] Simon, C., Wiezer, A., Strittmatter, A.W., Daniel, R. (2009) Phylogenetic diversity and metabolic potential revealed in a glacier ice metagenome. Appl. Environ. Microbiol. 75, 7519–7526. [38] Skidmore, M., Anderson, S.P., Sharp, M., Foght, J., Lanoil, B.D. (2005) Comparison of microbial community compositions of two subglacial environments reveals a possible role for microbes in chemical weathering processes. Appl. Environ. Microbiol. 71, 6986–6997. [39] Team, R.C. (2015) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [40] Tranter, M. (2005) Geochemical weathering in glacial and proglacial environments. In: Holland H.D. and Turekian, K.K. (Eds). Treatise on geochemistry. Volume 5. Elsevier, London, UK, pp. 189–205. [41] Humboldt, Av. (1808) Ansichten der Natur mitwissenschaftlichen ¨ J.G. Cotta: Tubingen, Germany. Erlauterungen. [42] Watanabe, K., Kodama, Y., Harayama, S. (2001) Design and evaluation of PCR primers to amplify bacterial 16S ribosomal DNA fragments used for community fingerprinting. J. Microbiol. Methods. 44, 253–262. [43] Wilhelm, L., Singer, G.A., Fasching, C., Battin, T.J., Besemer, K. (2013) Microbial biodiversity in glacier–fed streams. ISME J. 7, 1651–1660. [44] Xiang, S.R., Shang, T.C., Chen, Y., Jing, Z.F., Yao, T.D. (2009) Dominant bacteria and biomass in the Kuytun 51 Glacier. Appl. Environ. Microbiol. 75, 7287–7290. [45] Xiang, S.R., Shang, T.C., Chen, Y., Yao, T.D. (2009) Deposition and postdeposition mechanisms as possible drivers of microbial population variability in glacier ice. FEMS Microbiol. Ecol. 70, 165–176. [46] Zhang, S., Hou, S., Ma, X., Qin, D., Chen, T. (2007) Culturable bacteria in Himalayan glacial ice in response to atmospheric circulation. Biogeosciences. 4, 1–9. [47] Zhu, L., Liu, Q., Liu, H.C., Zhang, J.L., Dong, X.Z., Zhou, Y.G., Xin, Y.H. (2013) Flavobacterium noncentrifugens sp. nov., a psychrotolerant bacterium isolated from glacier meltwater. Int. J. Syst. Evol. Microbiol. 63, 2032–2037. [48] Zhu, L., Liu, Q., Liu, H.C., Zhou, Y.G., Xin, Y.H., Dong, X.Z. (2013) Mycetocola miduiensis sp. nov., a psychrotolerant bacterium isolated from Midui glacier. Int. J. Syst. Evol. Microbiol. 63, 2661–2665. [49] Zuur, A.F., Ieno, E.N., Elphick, C.S. (2010) A protocol for data exploration to avoid common statistical problems. Methods Ecol. Evol. 1, 3–14.