Bacterial Community Composition of South China Sea Sediments through Pyrosequencing-Based Analysis of 16S rRNA Genes

Background Subseafloor sediments accumulate large amounts of organic and inorganic materials that contain a highly diverse microbial ecosystem. The aim of this study was to survey the bacterial community of subseafloor sediments from the South China Sea. Methodology/Principal Findings Pyrosequencing of over 265,000 amplicons of the V3 hypervariable region of the 16S ribosomal RNA gene was performed on 16 sediment samples collected from multiple locations in the northern region of the South China Sea from depths ranging from 35 to 4000 m. A total of 9,726 operational taxonomic units (OTUs; between 695 and 2819 unique OTUs per sample) at 97% sequence similarity level were generated. In total, 40 bacterial phyla including 22 formally described phyla and 18 candidate phyla, with Proteobacteria, Firmicutes, Planctomycetes, Actinobacteria and Chloroflexi being most diverse, were identified. The most abundant phylotype, accounting for 42.6% of all sequences, belonged to Gammaproteobacteria, which possessed absolute predominance in the samples analyzed. Among the 18 candidate phyla, 12 were found for the first time in the South China Sea. Conclusions This study provided a novel insight into the composition of bacterial communities of the South China Sea subseafloor. Furthermore, abundances and community similarity analysis showed that the compositions of the bacterial communities are very similar at phylum level at different depths from 35-4000 m.


Introduction
Advances in high-throughput sequencing, also known as next-generation sequencing technology, including 454 pyrosequencing, Illumina sequencing etc., has significantly promoted microbial diversity and ecological studies. Deep sequencing makes it possible to precisely describe complicated microbial communities in several environments including marine, soil, animal or insects guts, which were all over 100 times more diverse than previously reported by traditional culture-dependent methods [1][2][3][4]. Marine microbial communities mediate biogeochemical ocean cycles including carbon, nitrogen and sulphur, and are probably play pivotal roles in maintaining marine ecosystem to prevent environmental changes such as warming and ocean acidification [5,6]. The ocean microbial community structure is influenced by pH, water temperature, salinity, silicate, seasonal shifts, and ocean currents as proved by surveys of the La Sal del Rey hypersaline lake located in southern Texas, USA [7], the western English Channel [8], and the western Arctic Ocean [9]. Phylotypic richness differed between summer and winter, but remarkable bacterial community structure stability was observed over time in the western Arctic Ocean [9]. Previously, the abundance of phylotypes in the oceanic microbial community was focused on abundant species because they were easily detectable. Thereafter, the 454 pyrosequencing technique revealed that most of the diversity of oceanic microbial communities is comprised of a high number of rare species and, in some cases, collectively comprise up to 75% of the abundance in their communities, named ''the rare biosphere" [3,9,10]. The rare microbial biosphere together with abundant members play important roles in the functioning of the ocean ecosystem [11].
The South China Sea is one of the largest marginal seas and lies within the West Pacific marine. The detrital fluxes of sediments of South China Sea came from three of the largest rivers in the world (Mekong River, Red and Pearl rivers), and the monsoon activity plays an important role in the cycling of organic carbon and other biogenic component of sediments, which controlled the sea surface circulation. In summer, the subtropical waters are advected into the South China Sea through the southern straits and through the Taiwan Strait to exit the South China Sea. In winter, the cold and saline waters enter the South China Sea from the north as a reversed pattern [12].
It has an average water depth of 1200 m and a maximal depth of approximately 5380 m and has long been recognized as the global center of marine tropical biodiversity [13]. There are abundant organic matters in the deposited sediments of the seafloor [14]. Bacterial community richness estimated from rRNA sequences of ocean samples revealed hundreds to thousands of phylotypes [15,16]. To date, the South China Sea bacterial community distribution patterns remain unknown. Although a few surveys based on culture-dependent, denaturing gradient gel electrophoresis and constructed PCR product clone libraries methods analyzed the bacterial diversity, the obtained numbers of operational taxonomic units (OTUs) was lower than100 sequences for each sample [17][18][19][20][21][22]. These smaller datasets result in the underestimation of species richness and generally do not describe rare populations that might represent considerable diversity.
The goal of our study was to explore questions about bacterial diversity in subseafloor sediments of the South China Sea using sequences of the V3 region of the 16S rRNA gene as determined by 454 pyrosequencing. We examined 16 sediment samples obtained from the surface of shallow and deep-sea bottoms at depths from 35 to 4000 meters. We next compared the similarity of rare and abundant phylotypes of the communities. Furthermore, we described the taxonomic composition of bacterial communities from our samples.

Results and Discussion
To investigate bacterial diversity and provide an in-depth description of relative abundance in benthic regions of the South China Sea, 454 (Life Science, Branford, CT, USA) pyrosequencing technology was used to sequence 16 samples collected from different locations of the northern region of the South China Sea, with depths ranging from 35-4000 m ( Table  1). We sequenced more than 265,000 PCR amplicons that span the V3 hypervariable region of rRNAs from our DNA preparations. Each sample produced from 13,000 to 20,000 reads. To eliminate random sequencing errors, reads were trimmed by removing the barcode and primers. Sequences with lengths less than 150bp or with ambiguous residues were also discarded. After removing potential erroneous sequences, data for the 16 samples was reduced from 21.8% to 10%, and on average data sizes were reduced by12.4%.

Taxonomic richness of benthic bacterial communities of South China Sea
Based on a BLASTN search of trimmed 454 reads in the RefHVR_V3 database to identify the closest matches, sequence tags were clustered into groups by defining the variation from unique sequences to 10% differences. These clusters were calculated for OTUs, abundance-based coverage estimator (ACE), and the Chao1 estimator. The exponential Shannon index was calculated and the Simpson index at species, genus, and family levels were defined with the sequence similarity thresholds of 97%, 95%, and 90%, respectively. Rarefaction curves were generated based on a species level. In total, 9,726 unique OTUs at the 97% threshold were obtained from the 16 samples.
A ranges of 695 (sample 16) to 2819 OTUs (sample 19) were discovered in a total of 16 benthic sediment samples ( Table 2). At the phylum level, all OTUs could be classified and belonged to 22 formally described bacterial phyla and 18 candidate phyla ( Figure 1). Therefore, the overall known diversity in the South China Sea increases to 40 different bacterial phyla and candidate phyla, which was higher than the reported number of 35 phyla from other marine habitats, including the Arctic Ocean and the Western English Channel [8,11]. In this study, we used massively parallel signature sequencing technologies to obtain more than 265,000 sequences from 16 sediment samples from depths of 35 to 4000 m in a 1 million square kilometer area surrounded by the Sanyan Bay, Luzon Island, Shantou bay, and Paracel Islands to the west, north, east, and south. The rarefaction analysis of the OTUs indicated that the bacterial community diversity of sample 19 (depth <300 m) was significantly higher than other samples (depth >300 m; Figure 2). In addition, the similar result also was obtained from other sea area samples. For example, in a study to investigate the prokaryote diversity in the subseafloor biosphere (Accession: SRP001218), 79717 OTUs (97% genetic similarity) were obtained from three shallow sea sediments (~3.5 m) but only 46836 OTUs were obtained from three deep sea samples (depth 3860 m and 1326m) ( Figure  S1A). In another study (SRP001269) for investigating the microbial biodiversity of Indian Ocean region revealed the average amount of OTUs was 25367 for the shallow sea sediment samples, and the average amount of deep sea samples OTUs only reached 17556. This suggests that the bacterial community diversity of the shallow subseafloor near the coast was richer than the deep subseafloor area. Abundance analysis showed that nine phyla account for over 95% of the total amplicons. The phyla include Proteobacteria, Firmicutes, Planctomycetes, Acidobacteria, Actinobacteria, Chloroflexi, Bacteroidetes, Gemmatimonadetes, and Nitrospirae ( Figure 3A). Proteobacteria was the most abundant phylum in all samples and accounted for 37-80% of all bacterial amplicons. As the most dominant community in marine environments, Proteobacteria has also been described in the Arctic Ocean [11], marine sponges [1], and the benthic North Sea [23].
The In Proteobacteria, Gammaproteobacteria were the most dominant class in all samples, accounting for 53.4% to 76.8%, and Deltaproteobacteria was the second most dominant class, accounting for 37.8% of tags in sample 19. However, Alphaproteobacteria and Betaproteobacteria were the second and the third most dominant classes in the deep-sea sediment samples ( Figure 3B). Gammaproteobacteria, the predominant bacterial group, prevailed over other taxa identified in several deep-sea investigations, including the Eastern Mediterranean Sea [24] and Northeastern Pacific Ocean [25]. Sequences affiliated with Desulfobacterales, Myxococcales, and Sh765B-TzT-29, dominated the Deltaproteobacteria and their common role is to regulate the sulfur cycle. The ocean represents a major reservoir of sulfur on Earth and microbial transformation of sulfur compounds has had a profound effect on the properties of the biosphere and continues to affect geochemistry [26]. The types of sulfur-metabolizing microorganisms of Deltaproteobacteria include sulfate reducers, organic sulfur utilizers, and sulfur reducers [27].
Abundance analysis of bacterial community diversity comparing the South China Sea with other marine area sediment samples was performed at the phylum level and the class level for Proteobacteria. The bacterial communities of different marine area displayed similarity in dominant groups, which including Proteobacteria (Gammaproteobacteria, Deltaproteobacteria), Planctomycetes, Firmicutes, Actinobacteria, Acidobacteria, Bacteroidetes, and Chloroflexi ( Figure S1). The phylum Proteobacteria, being most dominant group, was observed in a large proportion of shallow sea and deep-sea sediments. In the global overview, the dominant groups of bacterial communities of shallow sea sediments were similar with that of deep-sea sediments in the same sea area.
The benthic bacterial communities of the South China Sea showed similarity with other sea areas, but each area still has different characteristics in their dominant group's proportions. Such as in samples from cluster CFU1 (Atlantic Ocean near Portugal), where more than 54.7% of the sequences belong to candidate phylum OP9; in a 5000 m depth benthic sample from the Indian Ocean, the class Alphaproteobacteria reached 93.4% in total sequences. The geographical location has a strong impact on microbial community composition, and explained 22.2% of the observed differences in benthic communities [16].

Principal-component analysis of the bacterial community of the South China Sea
To determine the distribution and biogeography of the bacterial community, the 454 data were analyzed in relation to sampling locations using principal-component analysis ( Figure  4). The similarity of microbial communities among our 16 samples, collected from depths of 35-4000 m, was monitored with PCA at the phylum level and OTU0.03 levels. At the phylum level, samples collected from similar depths or locations did not contain more similar microbial communities to one another than to samples collected at other depths or locations ( Figure 4A). For example, samples 16, 18, and 20, collected from depths of 3536 m, 652 m and 431 m, respectively, fell into a cluster, while samples 19, 20, and 21 were all collected from the Shantou bay but fell into different clusters. However, a large amount of deep-sea samples clustered into a group by PCA analysis at the OTU 0.03 levels ( Figure 4B). The deep-sea samples exhibited a noticeable and regular separation from shallow sea in the first principal component (PC1), and 80.55% of the variation in the data explained in PC1. A large part of deep-sea samples corresponded to negative values and the shallow sea sample 19 had the highest value, which was more than 50. The shallow sea sample 19 had the lowest negative value in the second principal component (PC2), which represented 10.05% of the variation; a large amount of deep-sea samples fall in a range from -20 to +20.
In addition, the richness of bacterial diversity of the deep subseafloor was less than that of the shallow subseafloor ( Figure 2). Because the nutrient-limited, low energy-flux, and high press environment of deep subseafloor leads to microbial abundance, activity and turnover rates in the deep subseafloor are extremely low relative to those in other global habitats [28].

Rare biosphere in South China Sea
Deep sequencing revealed that the rare biosphere accounts for tremendous diversity of marine bacterial communities. We defined the rare phylotypes as having a frequency <0.01% and abundant phylotypes at a frequency >1% within a sample, according to previous reports [11,29] .
The rare phylotype distribution in each sample was similar and within a range of 55-64% of OTUs but comprised <5% of the sequence abundance ( Figure 5). Overall, 62 abundant phylotypes were counted in shallow sea and deep-sea sediment samples, covering 57-76% and 22% of the number of sequences, respectively. However, this only comprised <1.5% of the OTUs. In addition, 18 phylotypes of 62 abundant phylotypes were found to be abundant in some samples but rare in others. For example, the hydrocarbon-degrading Gammaproteobacteria, Cycloclasticus, a member of the rare biosphere, becomes an abundant member of the community when supplied appropriate conditions but returns to the rare biosphere by the loss of the supplied condition [30]. This suggests that the distribution of the bacterial rare biosphere is not obviously different in shallow and deep-sea sediments. The    Table 1. doi: 10.1371/journal.pone.0078501.g004 environmental changes from the Pacific coast, which results in these populations dominating the assemblage during a period of time then declining in the abundant biosphere to the point where they become undetectable [31]. The majority of sequences from the community were occupied by abundant taxonomic groups. These groups were thought to be well adapted to the environment and to contribute the most to biomass production [32,33]. Conversely, the biomass of rare groups is negligible compared to that of the abundant members of the community, and their contribution to carbon flow is relatively small. However, some members of the rare biosphere that are actively growing can significantly contribute to particular elements such as nitrogen and sulfur cycling. For example, Desulfosporosinus account for <0.01% of the total cell count but could contribute to most of the sulfate reduction in peat [34,35].
Further analyses of community structure and function are needed to investigate the interaction between rare phylotypes and marine subseafloor habitats. The 454 pyrosequencing technique still has pitfalls and may affect our results for the rare biosphere. Even the largest published metagenomic investigations inadequately represent the full extent of microbial diversity, and primer efficiency for generating 16S rRNA gene fragments is limited [10,36].

Conclusions
Overall, this study is the first metagenomic analysis using pyrosequencing to characterize a more comprehensive overview of the bacterial community of the South China Sea. The massively parallel signature sequencing of 16 samples of subseafloor sediments and data analysis allowed novel insights into the complex composition of this microbial community. Detected diversity of bacterial communities increased to 40 different bacterial phyla and 18 candidate phyla, and the majority populations of the South China Sea at different depths from 35-4000 m showed a high similarity at the phyla level. However, the shallow subseafloor showed a higher bacterial diversity compared to deep subseafloor.

Sample collection and preparation for pyrosequencing
Samples were collected from the deposited sediment of benthic regions in 16 different locations of the northern region of South China Sea (between Lat °N, 111° 23.243' and 120°0 .250' to Long °W 17° 58.927' and 22° 29.355') at water depths ranging from 35-4000 m. Samples were transferred to sterilized plastic tubes and stored at -80°C ( Figure 6). There are no specific permits required for the described sampling because collections did not involve endangered species and did not occur within a designated marine protected area, private reserve, or park.
Total genomic DNA was extracted from 1g of the sediment samples using the EZgeneTM Soil gDNA Kit (Biomiga, San Diego, CA, USA) according to the manufacturer's protocols. Bacterial 16S rRNA at the V3 hypervariable region were amplified using a set of primers designed by adding a 10nucleotide barcode to the forward primer, 8F, (5'-AGAGTTTGATCCTGGCTCAG-3') and reverse primer, 533R, (5'-TTACCGCGGCTGCTGGCAC-3').
The amplification reaction mixture contained 5 Units of Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA, USA), 1×Pfu reaction buffer, 200 µM dNTPs (TaKaRa, Dalian, China), 0.2µM barcoded primer, and 20ng genomic DNA template for a total volume of 100 µl. PCR was performed with a thermal cycler (Bio-Rad, USA) under the following condition: 5 min at 94°C, 25 cycles of 30 s at 94°C plus 45 s at 55°C plus 30 s at 72°C, and finally 5 min at 72°C. The PCR products were purified by using a PCR Purification Kit (QIAGEN, Hilden, Germany)

Sequence analysis
Raw sequence reads were filtered to eliminate the effect of random sequencing errors. The primer and barcode of each read were removed and trimmed. The sequences that (i) were shorter than 150 nucleotides, (ii) contained ambiguous bases (N), or (iii) contained homopolymer regions (>6 repetitions of the same base) were excluded.
Taxonomic identification of the reads (''tags'') were performed following the process described by Sogin et al [3]. All optimized reads were trimmed down to equal lengths (150bp), which contained the V3 hypervariable region. The tags were searched using the NCBI BLASTN tool in the reference database of hypervariable region tags (RefHVR_V3, http://vamps.mbl.edu/) based on the SILVA database, version 106 [37], and queried reads showing the minimum distance with the reference V3 tags were grouped and assigned the same phylotype. Taxonomy was assigned to each trimmed reference sequence (400bp) with Mothur 1.24.0 [38].

Diversity and statistical analysis
The sequences were grouped into OTUs sharing ≥97%, ≥95%, or ≥90% similarity by DOTUR [39]. The bacterial community richness indices (non-parametric ACE and the Chao1) and diversity indices (Shannon and Simpson estimators) were calculated using Mothur and Shannon-acetable.pl software programs (Majorbio, Shanghai, China). Rarefaction curves were calculated using Mothur and the software program Plot-rarefaction (Majorbio, Shanghai, China). Heat maps were drawn by hierarchal clustering performed in the R software environment (http://www.R-project.org) within the function "vegdist" in the Vegan Community Ecology Package. Figure S1.

Supporting Information
Bacterial community distribution in 17 ecosystems type. A, the relative abundances of different phyla in 17 ecosystems type which including 125 sediments samples; B, the relative abundances of different classes in Proteobacteria of 17 ecosystem type. The relative abundance is presented in terms of percentage in total effective bacterial sequences in an ecosystem type. NH1, sample 19 in this study, NH2, the other 15 samples except sample 19 in this study, other ecosystem type were described in Table S1. (TIF)