Microbial communities in peatlands along a chronosequence on the Sanjiang Plain, China

Microbial communities play crucial roles in the global carbon cycle, particularly in peatland ecosystems under climate change. The peatlands of the Sanjiang Plain could be highly vulnerable to global warming because they are mainly located at the southern limit of northern peatlands. In this study, the alpha diversity and composition of bacterial communities in three different minerotrophic fens along a chronosequence were investigated. We captured a rich microbial community that included many rare operational taxonomic units (OTUs) but was dominated by a few bacterial classes that have frequently been detected in other peatland ecosystems. Notably, a large diversity of methanotrophs affiliated with Alpha- and Gammaproteobacteria was also detected. Bacterial alpha diversity and composition varied as a function of peat depth and its associated physical-chemical properties, such as total carbon, total nitrogen, pH and bulk density. We also found that bacterial community turnover (beta diversity) to be significantly correlated with soil age, whereas bacterial alpha diversity was not.

Despite covering only 6-8% of the world's terrestrial ecosystems, northern peatlands store approximately 30% of the global soil carbon (C) 1 . Thus, as one of the largest atmospheric carbon sinks and CH 4 sources, northern peatlands have played an important role in global C cycles and climate change throughout the Holocene (from approximately 11600 years ago to the present) 2, 3 . In peatlands, the sequestration of C arises from an imbalance between inputs via primary productivity and losses through microbial decomposition, which releases C to the atmosphere and as dissolved organic carbon (DOC) exports 4,5 . Both environmental and physico-chemical conditions impact C cycling through primary production and decomposition processes. Methane production is a consequence of the permanently waterlogged condition of the peat layers, where anaerobic decomposition prevails, and many studies have shown that microorganisms play a crucial role in C decomposition and methane production process [6][7][8][9][10] . Although microbial communities in the peatlands of Europe, America, Canada and the UK have been investigated [11][12][13] . However, there are relatively few studies on the peatlands in more temperate regions [14][15][16] . The Sanjiang Plain, located in the temperate climate region, is the largest area of freshwater marshlands in China 17 . As these peatlands are mainly located at the southern limit of northern peatlands, they may be more vulnerable to global warming than other areas 18 . Thus, it is imperative that we deepen our understanding of the microorganisms in this ecosystem.
The peatlands of the Sanjiang Plain have developed under certain topographic conditions during the Holocene or earlier 19 , and they vary in age from 500 to 10,000 years 20 . Correspondingly, these areas differ with regard to C accumulation rate. Accordingly, this chronosequence of peatlands offers an exceptional opportunity for studying links between the chronological characterization (soil age), soil properties and microbial community structure in peatland ecosystems. The influences of vegetative communities, water hydrology and soil properties in peatland microbial communities have been examined in many studies [21][22][23][24] , though the association between chronological characterization and the distribution of the microbial community structure has been neglected to date.
In the present study, high-throughput Illumina sequencing of 16S rRNA genes was applied to examine the bacterial communities of the Sanjiang Plain, the southern edge of northern peatlands. Peat cores were collected from three fens that began to develop in this area during different periods. The greater sequencing depth achieved by high-throughput sequencing allows for the capture of the less abundant and uncultured taxa and thus supplies a more thorough characterization of peatland bacterial diversity. Moreover, such chronological characterization facilitates the description of potential links between bacterial communities and soil age as well as the C accumulation rates.

Results
Physico-chemical and chronological characterization of peat cores. Nine peat cores were retrieved from fens, S (Shenjiadian), H (Honghe), and Q (Qindelie), on the Sanjiang Plain, northeastern China (Fig. 1). With a depth of 100 cm, the peatlands located in Q are shallower than those at the other sites; mud deposits occur at depths greater than 100 cm. Accelerate mass spectrometry (AMS) dating results indicate that Sanjiang Plain peatlands began to develop over a wide range of dates during the Holocene (Fig. S1). The peat was dated from 637 to 11,496 cal. yr BP (Table 1). Peat cores from Q exhibited the most ancient chronological dates, with even the 30-60-cm section being dated to 3136 ± 1325 cal. yr BP, which was older than the age of the bottom section (100-200 cm) of the youngest fen, S at 2881 ± 1338.95 cal. yr BP, Overall, the mean C accumulation rates of the peat cores decreased with depths with one exception, the maxima of the mean C accumulation rates of the peat cores from S were observed in the 60-100-cm section (Table 1).
A summary of the soil physico-chemical characteristics is presented in Table 1. Soil pH was consistently acidic and varied from 4.80 to 5.91. Soil total C (TC) and N (TN) ranged from 29.74 to 458.54 g/kg and from 2.95 to 21.62 g/kg, respectively. Overall, they were decreased with depths at all sites, and the soils from S, the youngest fen, showed the highest total C and N contents. The C/N ratio was approximately 1.5 times lower in the lowest section than in the surface section of the same site.

Sample Location Depth
community function, only highly abundant OTUs (relative abundance higher than 0.1% and detected in more than 6 samples) were used for the calculations. The bacterial communities of the surface peat of the S, H, and Q fens shared 42 common OTUs (Fig. 4a), and bacterial communities varied with depth at each site. For example, in the soil sampled from S, the unique OTU numbers of the 0-30, 30-60, 60-100, and 100-200-cm sections were 56, 28, 18, and 25, respectively (Fig. 4b).
Bacterial communities differed across depths and sites. The relative abundance of the dominant phylum, Proteobacteria, increased with increasing depth across all sites and was higher in younger peatlands, in contrast, Actinobacteria and Acidobacteria sequences decreased with increasing depth (Fig. 2). From top to the bottom, the relative abundance of Actinobacteria in the peat cores from S, H and Q fens, decreased from 23.60% to 13.16%, from 15.10% to 12.62%, and from 24.24% to 8.06%, respectively, and that of Acidobacteria decreased from 16.29% to 1.53%, from 12.21% to 4.72%, and from 16.78% to 4.50%, respectively Notably, a large diversity of methanotrophs affiliated with Alpha-and Gammaproteobacteria was detected (Fig. 5). Methylocystaceae were the dominant methanotrophs across all peat samples and were primarily observed in the deep and aged soils.
Association of bacterial diversity and composition with soil age and C accumulation rate. Using multiple alpha diversity indices, we assessed whether soil age and the C accumulation rate are significantly related to different aspects of microbial alpha diversity. However, we observed no significant relationships, positive or negative, among these parameters (all p > 0.05). Nonetheless, principal coordinate analysis  (PCoA) of unweighted UniFrac distance ordinations did show that the bacterial community structures across the 33 soil samples were affected by soil age (Mantel test, r = 0.195, p < 0.05), but not the C accumulation rate (r = −0.119, p = 0.889) (Fig. 7). In regression analysis between the PCoA scores and soil ages, a significantly linear regression with the PCoA horizontal axis was found (r = 0.701, p < 0.001), which indicating that the variation in the bacterial community could be explained by soil age along the PCoA horizontal axis. This finding is consistent with the results of CCA analysis.

Discussion
We captured a rich microbial community from the peat cores collected from the Sanjiang Plain, at the southern edge of northern peatlands. However, we found that the bacterial community to be dominated by a few bacterial classes and OTUs that have frequently been detected in other peatland ecosystems. The most abundant phyla, Proteobacteria, Actinobacteria, Acidobacteria, Bacteroidetes and Firmicutes, have frequently been detected as dominant groups in other peatland ecosystems (oligo-mesotrophic, ombrotrophic and tundra) 5,25,26 .
Due to the vertical gradients in the soil properties we measured, such as TC, TN, pH, bulk density, and C/N ratio, as well as the availability of oxygen (O 2 ) and other electron acceptors, such as SO 4 2− and, NO 3 − , mentioned by others 5, 27 , the peat soil provides vertically stratified heterogeneous habitats for bacteria. In the present study, bacterial alpha diversity and composition varied as a function of peat depth and its associated physico-chemical properties, and pH, TC, TN, and bulk density showed significant effects on bacterial community structure.
The dominant microbial taxa detected at the surface are members of Acidobacteria, Gammaproteobacteria, Alphaproteobacteria (Acetobacteraceae), and Verrucomicrobia (Chthoniobacter). Acidobacteria, which are reportedly dominant among soils rich in organic matter, are involved in degradation of lignocellulose and cellulose [28][29][30] . A substantial number of carbohydrate-utilizing Acidobacteria and Alphaproteobacteria from acidic peatlands have been cultivated under aerobic conditions 31 . Candidatus Solibacter, the most abundant taxon among our samples, has been shown to be able to break down, utilize, and biosynthesize of diverse structural land-storage polysaccharides and exhibits resilience to fluctuating temperatures and nutrient-deficient conditions 32 . According to the research of Xing et al. 17 , potential peatlands cover an area of approximately 10,520 km 2 on the Sanjiang Plain and currently store ~0.26 Pg C. In response to future climate change and CO 2 fertilization, both net primary productivity and CH 4 fluxes will increase 33 . Methanotrophic bacteria can act as a natural barrier capable of significantly reducing the flux of methane into the atmosphere 34 , and bacteria that perform aerobic and anaerobic methanotrophy were detected in our samples. Although the relative abundance of type I and type II methanotrophs detected in the present study was similar to that in other peatlands 25,35 , a broader sequence diversity (14 OTUs) of putative methanotrophs belonging to Alpha-and Gamma-proteobacteria was observed. The OTUs detected in this study, are most likely to act as key players in methane cycling in this environment. Further studies on the isolation and characterization of these microorganisms are needed and will facilitate our understanding of their physiological functions and ecological importance.
By comparing the AMS dating and soil properties of the cores of peat that developed at different ages, we found that soil properties, such as TN, TC, C accumulation rate and pH, were correlated with peat age. These peatlands initiated at different times, were affected by the climatic conditions, such as temperature and precipitation, etc, and underwent different initiation and decomposition processes, resulting in the respective soil properties. It is worth noting, that we also found that bacterial community turnover (beta diversity) to be significantly correlated with soil age, bacterial alpha diversity was not. We suggest that the effect of soil age on the bacterial community was due to its correlation with soil properties. The structure and activity of bacterial communities involved in C decomposition and release processes can directly affect the C accumulation in peatlands and other ecosystems 4, 5 , Nonetheless, when investigated the linkage of organic C accumulation and microbial community dynamics in a sandy loam soil, Zhang et al. found that C accumulation promoted the macroaggregation and reduced the effective diffusion coefficient of oxygen, causing changes in microhabitats and a shift in microbial communities 36 . We did not detect significant correlation between bacterial community turnover and the C accumulation rates. The role of bacteria in C cycling in these peatlands requires further verification via field experiments and laboratory ecophysiological studies.
In summary, we found that the peatlands on the Sanjiang Plain developed over a wide range of dates during the Holocene, with the peat cores being dated from 637 to 11,496 cal. yr BP. The relative abundance, distribution, and composition of the microbial communities in three different minerotrophic fens on the Sanjiang Plain, at the southern edge of northern peatlands, was investigated by next-generation sequencing. We captured a rich microbial community that included many rare OTUs but was dominated by a few bacterial taxa that have frequently been detected in other peatland ecosystems. Notably, a large diversity of methanotrophs affiliated with Alpha-and Gammaproteobacteria was detected. In addition, bacterial alpha diversity and composition varied as a function of peat depth and its associated physical-chemical properties, such as TC, TN, pH and bulk density. We also found that bacterial community turnover (beta diversity) to be significantly correlated with soil age, though this was not observed for bacterial alpha diversity.

Methods
Site description and sampling. We collected peat cores from three different minerotrophic fens, Shenjiadian (S), Honghe (H), and Qindelie (Q), on the Sanjiang Plain (129°11′-135°05′E, 43°49′-48°27′N), northeastern China (Fig. 1). The three fens began to develop during different periods of the Holocene 20 . This region is a large alluvial plain that is crossed by three major rivers, the Heilong River, Ussuri River and Songhua River, and has a total area of 10.9 × 10 6 ha, an elevation of <200 m and a slope grade of <1:10,000. Over 70% of this region is dominated by freshwater wetlands that have developed in ancient riverbeds and waterlogged depressions 37 , and approximately 30% (or nearly 3.3 × 10 4 ha) of this region is covered by peatlands that developed under certain topographic conditions during the Holocene or earlier 38 . The present climate of the plain is a temperate humid or subhumid continental monsoon climate. The mean annual temperature ranges from 1.4 to 4.3 °C, with an average maximum of 22 °C in July and an average minimum of −18 °C in January. The mean annual precipitation is 500-650 mm, and 80% of the rainfall occurs between May and September 39 . In addition to precipitation, the fens receive water inputs from groundwater and they are primarily covered with sedges (Carex lasiocarpa). In May 2012, triplicate cores were collected from each fen using a Russian peat corer, the cores were subsampled for chronological, microbiological and physico-chemical analyses. The storage of the soil samples for chronological analysis is described in more detail in the work of Zhang et al. 18 . For the microbiological and physico-chemical analyses, peat cores were sectioned into depth intervals of 0 to 30, 30 to 60, 60 to100 and 100 to 200 cm and; homogenized in sterile bags. After subsampling, the peat soils were stored in a sample incubator with a cooling function (2 to −10 °C) before being transferred to the laboratory. Chronological and physico-chemical characterization was conducted immediately after the samples arrived. Samples for DNA extraction were immediately frozen at −80 °C.
Chronological and physico-chemical characterization. Subsamples with a volume of 3 cm 3 were used for loss on ignition (LOI) with sequential combustion at 500 °C to estimate the organic matter content 40 . The bulk density of a 1-cm interval of each peat core was calculated according to the dry weight and volume of each subsample. Ash-free (organic matter) bulk density was calculated from the measurements of the bulk density and organic matter contents. All subsamples for AMS dating were dated with an accelerator mass spectrometry system at the Institute of Earth Environment, CAS. The AMS 14 C dates were calibrated into calendar ages using the program Calib 7.02 based on the INTCAL 13 calibration dataset 41 . Carbon accumulation rates were calculated by multiplying the organic carbon content (using 52% C in peat organic matter) by the bulk density and dividing by the age interval 42 . Soil pH was measured in a 1:5 soil/water suspension 43  Sequence data preprocessing and statistical analysis. Raw sequences were divided into sample libraries via sample-specific barcodes and truncated after cutting off the barcode and the primer sequence. Forward and reverse reads with at least 10-bp overlaps and less than 5% mismatches were merged using FLASH 27 . Quality filtering of the raw tags was performed according to the QIIME (V1.7.0, http://qiime.org/index.html) quality control process 45 , and all sequences shorter than 200 bp in length and with an average quality score lower than 25 in the raw reads were removed. The remaining sequences were subjected to chimera removal using the UCHIME algorithm (http://www.drive5.com/usearch/manual/uchime_algo.html). UPARSE 46 (version 7.0.1001, http://drive5.com/uparse/) was employed to classify the operational taxonomic units (OTUs) at the 97% similarity level. The longest sequence that had the largest number of hits to other sequences in each OTU was screened as a representative sequence. All the OTUs with sequence numbers ≤ 2 were removed in the subsequent analysis. The RDP classifier (version 2.2, http://sourceforge.net/projects/rdp-classifier/) was used to annotate taxonomic information for each representative sequence. Representative sequences were aligned using the Greengenes Database 47 (version 2011, http://greengenes.lbl.gov/cgi-bin/nph-index.cgi), with a minimum identity of 80%. To study the phylogenetic relationships among different OTUs and the differences between the dominant species in different samples (groups), multiple sequence alignments were performed using the MUSCLE software (version 3.8.1, http://www.drive5.com/muscle/) 46 . For alpha and beta diversity analyses, the sequences were rarefied to depths of 21,811 sequences per sample, to minimize the effects of a different sampling efforts. Alpha diversity indices, including Chao1, Shannon, Simpson, ACE and PieLou equitability, were calculated. Beta diversity among the microbial communities was evaluated using both weighted and unweighted UniFrac distances. Evolutionary relationships were analyzed with neighbour-joining phylogenetic trees constructed in MEGA6. Pearson correlation analyses were applied to evaluate the relationships between the soil geochemical 48 . Differences in soil properties across samples were determined by ANOVA in IBM SPSS (version 19.0, Chicago, IL, USA) 49 . Relationships between the taxonomic diversity and the soil properties were examined by linear regression analyses using SPSS. Principal coordinate analysis (PCoA) was conducted using the ape package (version 4.1) in R (version 3.3.3) 50 . Constrained correspondence analysis (CCA) was performed using the vegan package (version 2.4-1) in R 48 . A Mantel test was used to study the relationship between bacterial similarities and overall environmental factors.
SCIENtIfIC REPORTS | 7: 9567 | DOI:10.1038/s41598-017-10436-5 We also performed post hoc permutations using the function "envfit" in the vegan package to detect associations of the microbial community composition with environmental variables. A Mantel test was also used to examine the correlation between the UniFrac distances and soil age, and C accumulation rate. Heatmap plots were generated in R using the pheatmap package 51 . The Illumina sequencing data obtained in the present study have been deposited in the NCBI SRA database under the accession number SRP082472 . The OTU table used for alpha and beta diversity analyses is provided as  Supplementary Table 3.