Microbial Population Structures in the Deep Marine Biosphere

See allHide authors and affiliations

Science  05 Oct 2007:
Vol. 318, Issue 5847, pp. 97-100
DOI: 10.1126/science.1146689


The analytical power of environmental DNA sequences for modeling microbial ecosystems depends on accurate assessments of population structure, including diversity (richness) and relative abundance (evenness). We investigated both aspects of population structure for microbial communities at two neighboring hydrothermal vents by examining the sequences of more than 900,000 microbial small-subunit ribosomal RNA amplicons. The two vent communities have different population structures that reflect local geochemical regimes. Descriptions of archaeal diversity were nearly exhaustive, but despite collecting an unparalleled number of sequences, statistical analyses indicated additional bacterial diversity at every taxonomic level. We predict that hundreds of thousands of sequences will be necessary to capture the vast diversity of microbial communities, and that different patterns of evenness for both high- and low-abundance taxa may be important in defining microbial ecosystem dynamics.

The interrogation of DNA from environmental samples has revealed new dimensions in microbial diversity and community-wide metabolic potential. The first analysis of a dozen polymerase chain reaction (PCR) amplicons of ribosomal RNA (rRNA) sequence from a mixed bacterioplankton population revealed the ubiquitous SAR11 cluster (1), and a recent environmental shotgun sequence survey of microbial communities in the surface ocean has identified 6.1 million predicted proteins (2, 3). To realize the full potential of metagenomics for modeling energy and carbon flow, microbial biogeography, and the relationship between microbial diversity and ecosystem function, it is necessary to estimate both the richness and evenness of microbial population structures.

We used a tag sequencing strategy that combines the use of amplicons of the V6 hypervariable region of small-subunit (SSU) rRNA as proxies for the presence of individual phylotypes [operational taxonomic units (OTUs)] with massively parallel sequencing. Our goal was to provide assessments of microbial diversity, evenness, and community structure at a resolution two to three orders of magnitude greater than that afforded by cloning and capillary sequencing of longer SSU rRNA amplicons (4). We used this strategy to attempt an exhaustive characterization of the bacterial and archaeal diversity at two low-temperature diffuse flow vents, Marker 52 and Bag City, from Axial Seamount, an active volcano at 1520 m depth in the northeast Pacific Ocean (5, 6). These vents host archaeal and bacterial communities originating from the subseafloor, local microbial mats, symbionts of vent macrofauna, and microorganisms from the surrounding seawater (79). Although new production from hydrothermal vents may correspond to as much as 25% of the total imported carbon flow in the deep sea (10), these globally distributed habitats remain relatively unexplored, and there are few descriptions of diversity, evenness, and dispersal of their endemic microbial populations.

Marker 52 and Bag City are less than 3 km apart, but differ markedly in chemical composition and appearance. Marker 52 was sampled on bare rock; Bag City vent fluids were sampled within a clump of tube worms. Both sites had microbial mats growing on rock and tube worm surfaces (fig. S1). Relative to Bag City and most other diffuse vents at Axial, Marker 52 has a higher H2S/ΔT ratio, lower pH, and elevated alkalinity and iron levels (Table 1), all of which indicate a higher carbon dioxide content; Marker 52 fluids were effervescent at 1 atm (9).

Table 1.

Chemical and SSU rRNA tag characteristics of the two sites.

View this table:

We sequenced more than 900,000 archaeal and bacterial V6 amplicons from these two sites. Tags that differed by no more than 3% [generally considered to define microbial species (11)] were clustered (12) into OTUs to calculate rarefaction and nonparametric estimators (13). Taxonomic and statistical analyses revealed differences in community membership with very little overlap between the two sites (Table 2), which is particularly evident when comparing the fine structure of the communities (Fig. 1). For example, although ϵ-proteobacteria often dominate 10° to 80°C vent habitats, where they orchestrate the cycling of carbon, nitrogen, and sulfur (14), the richness and evenness of ϵ-proteobacterial families and genera are different at each site (Fig. 1). Nearly 6600 distinct ϵ-proteobacterial tag sequences accounted for 39% of bacterial amplicons, raising the estimate for total ϵ-proteobacterial diversity by at least one order of magnitude. Sequences identified as Arcobacter spp., a group of micro-aerophilic sulfur and hydrogen sulfide–oxidizing bacteria, dominated the ϵ-proteobacterial phylotypes at Bag City (FS312, Fig. 1), whereas sequences identified as Sulfurovum spp., a group of mesophilic microaerobes that use sulfur species as electron donors with nitrate or oxygen as electron acceptors, dominated Marker 52 (FS396, Fig. 1).

Fig. 1.

Taxonomic breakdown of bacterial V6 tags from each vent. Pie charts show the Phylum_Class_Order distribution for taxonomically assigned tags that occurred more than 1000 times; the remaining tag sequences are grouped into “Other.” The taxonomic distribution of ϵ-proteobacterial genera is shown in normalized histograms for each site, with further breakdown of the dominant ϵ-proteobacteria in additional histograms, with each color in the histograms representing a unique tag sequence. For FS312, the Arcobacter are expanded to the left with a histogram showing those tag sequences that occurred ≥10 times, followed by a histogram showing the diversity of tags that occurred 10 to 1800 times. For FS396, the Sulfurovum are expanded to the right, with a histogram showing those tag sequences that occurred ≥10 times, followed by a histogram showing the diversity of tags that occurred 10 to 8400 times. Nonparametric estimates suggested more than 900 phylotypes each of Arcobacter at FS312 and Sulfurovum at FS396.

Table 2.

Sequencing information and diversity estimates for all bacteria and archaea.

View this table:

We hypothesize that the geochemical regimes shape the ϵ-proteobacterial community structure (Table 1) (11). A few highly abundant, specific tag sequences dominated each genus at each site, but extensive sampling revealed the presence of many less common and rare variants (Fig. 1). Microdiversity within groups of bacteria and archaea has been noted previously in the marine environment (15, 16). It is clear that in some cases, these closely related organisms are ecologically distinct (15, 17).

Nearly 6000 unique sequences from a data set of more than 215,000 V6 amplicon tags identified as archaeal defined more than 1900 phylotypes. The slope of the rarefaction curve (18) for the archaea became nearly asymptotic, and nonparametric statistical analyses estimated an ultimate richness of ∼2700 archaeal phylotypes (Fig. 2 and Table 2). In contrast, despite examining nearly 690,000 tags identified as bacterial, rarefaction curves (Fig. 2) indicated that our sampling of bacterial richness was far from complete. We observed more than 30,000 unique bacterial sequences forming ∼18,500 phylotypes, and nonparametric estimates predicted the presence of ∼37,000 phylotypes (13) (Table 2), with steeply sloping rarefaction curves for many diverse classes, orders, and families (fig. S2). Even the dominant genera Arcobacter and Sulfurovum were incompletely sampled (fig. S2). The lower diversity of archaeal phylotypes agreed with other molecular surveys indicating that marine archaeal diversity is relatively limited (19); hence, our approach does not result in inflated richness estimates due to spurious data. Furthermore, extensive quality control of tag sequences ensured that the total error from PCR and pyrosequencing was less than 0.0025 per base and that sequencing error mis-assigned fewer than 1% of tags to phylotypes (20).

Fig. 2.

Rarefaction curves for total bacterial and archaeal communities at the two sampling sites FS312 and FS396 at 3% and 6% difference levels.

Comparing each unique sequence to our V6 reference database revealed well-characterized taxa, as well as many unknown microbial phylotypes. The 10 most abundant sequences occurred more than 10,000 times and were exact matches to sequences in our database, indicating that our sampling was representative. Of the 36,725 unique sequences found at the two sites, 36,180 were represented by fewer than 100 tags; of these, 13,385 were >10% different and ∼4000 were >20% different from known SSU rRNA genes. Many rare, divergent taxa account for most of the observed novel microbial diversity (4, 21).

Although this study only examined samples at two sites in the deep ocean, it has important implications for our ability to sample and identify all the ecologically relevant members of microbial communities in other high-diversity habitats, such as soils (22), microbial mats (23), and communities where low-abundance taxa may play crucial roles, such as the human microbiome. It provides a comparative population structure analysis with statistically significant descriptions of diversity and relative abundance of microbial populations. These large estimates of phylogenetic diversity at every taxonomic level present a challenge to large-scale microbial community genomic surveys. Metagenomic studies seek to inventory the full range of metabolic capabilities that define ecosystem function or to determine their context within assembled genomic scaffolds. Our results suggest that even the largest of published metagenomic investigations inadequately represent the full extent of microbial diversity, as they survey only the most highly abundant taxa (11).

In addition, the importance of microdiversity cannot be overlooked, and metagenomic community reconstructions from the two vents studied here would likely be largely chimeric assemblies of sequences from closely related phylotypes, which may mask important biological differences. Methods such as the massively parallel tag sequencing approach used here, combined with the multitude of other quantitative and descriptive tools now available to microbial ecologists, can serve as necessary accompaniments to metagenomic gene surveys as we strive to understand the impact of diversity on ecosystem function and long-term stability (24).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article