Water mass-specificity of bacterial communities in the North Atlantic revealed by massively parallel sequencing

Abstract

Bacterial assemblages from subsurface (100 m depth), meso- (200-1000 m depth) and bathy-pelagic (below 1000 m depth) zones at 10 stations along a North Atlantic Ocean transect from 60°N to 5°S were characterized using massively parallel pyrotag sequencing of the V6 region of the 16S rRNA gene (V6 pyrotags). In a dataset of more than 830,000 pyrotags, we identified 10,780 OTUs of which 52% were singletons. The singletons accounted for less than 2% of the OTU abundance, whereas the 100 and 1000 most abundant OTUs represented 80% and 96% respectively of all recovered OTUs. Non-metric Multi-Dimensional Scaling and Canonical Correspondence Analysis of all the OTUs excluding the singletons revealed a clear clustering of the bacterial communities according to the water masses. More than 80% of the 1000 most abundant OTUs corresponded to Proteobacteria of which 55% were Alphaproteobacteria, mostly composed of the SAR11 cluster. Gammaproteobacteria increased with depth and included a relatively large number of OTUs belonging to Alteromonadales and Oceanospirillales. The bathypelagic zone showed higher taxonomic evenness than the overlying waters, albeit bacterial diversity was remarkably variable. Both abundant and low-abundance OTUs were responsible for the distinct bacterial communities characterizing the major deep-water masses. Taken together, our results reveal that deep-water masses act as bio-oceanographic islands for bacterioplankton leading to water mass-specific bacterial communities in the deep waters of the Atlantic.

Figure 1

Location of the stations where samples for pyrosequencing were taken during the Transat-1, Transat-2 and Archimedes-2 cruises in the North Atlantic. Characteristics of the samples are given in Table S1.

Figure 2

Depth distribution of the 5, 6 – 100 and 101 – 1,000 most abundant OTUs and the singletons. Inset: Rank-frequency distribution of the 1,000 most abundant OTUs.

Figure 3

Non-metric multidimensional analysis based on relative abundance of all OTUs except the singletons. Individual samples were affiliated to their respective water-mass. Superimposed circles represent clusters of samples at similarity values of 45 and 55% (Bray-Curtis similarity). Characteristics of the samples are indicated in Table S1. The final solution was based on 25 iterations with a final stress of 0.12.

Figure 4

Similarity in relative abundance of all OTUs except the singletons (in %, calculated through SIMPER analysis) between each station and the northernmost station of the transect (station 27, Transat-2) within bathy- and mesopelagic water layers.

Figure 5

Canonical correspondence analysis of the relative abundance of the 1,000 most abundant OTUs (a) for all samples, (b) for all samples without depth and latitude as explanatory variables, (c) for bathypelagic samples and (d) for mesopelagic samples. Monte Carlo permutation tests for the first and all axes were highly significant for all the four CCA analyses (p < 0.002). Abbreviations of the environmental and bacteria-related variables: lat, latitude; O₂, oxygen concentration; AOU, apparent oxygen utilization; DOC, dissolved organic carbon; DON, dissolved organic nitrogen; Sal, salinity; NH₄, ammonium concentration.

Figure 6

Relative abundance and affiliation of the 1000 most abundant bacterial OTUs of each zone of the Atlantic Ocean (subsurface n=6, mesopelagic n=20 and bathypelagic n=19). (a) on the phylum level, (b) the class level within Proteobacteria, (c) the order level within Gammaproteobacteria.

Figure 7

Relative abundance and affiliation of the singletons of each zone of the Atlantic Ocean (subsurface n=6, mesopelagic n=20 and bathypelagic n=19). (a) on the phylum level, and (b) the class level within Proteobacteria.

Figure 8

Deviation (in %) from the mean relative abundance of the 100 most abundant OTUs for all the samples for the different water masses.