TheISMEJournal(2010)4,159–170 &2010InternationalSocietyforMicrobialEcology Allrightsreserved1751-7362/10$32.00 www.nature.com/ismej ORIGINAL ARTICLE Bacterial diversity and biogeography in deep-sea surface sediments of the South Atlantic Ocean Regina Schauer1, Christina Bienhold1, Alban Ramette2 and Jens Harder1 1Department of Microbiology, Max Planck Institute for Marine Microbiology, Bremen, Germany and 2Microbial Habitat Group, Max Planck Institute for Marine Microbiology, Bremen, Germany Microbial biogeographic patterns in the deep sea depend on the ability of microorganisms to disperse.OnepossiblelimitationtomicrobialdispersalmaybetheWalvisRidgethatseparatesthe AntarcticLowerCircumpolarDeepWaterfromtheNorthAtlanticDeepWater.Weexaminedbacterial communities in three basins of the eastern South Atlantic Ocean to determine diversity and biogeography of bacterial communities in deep-sea surface sediments. The analysis of 16S ribosomalRNA(rRNA)geneclonelibrariesineachbasinrevealedahighdiversity,representing521 phylotypes with 98% identity in 1051 sequences. Phylotypes affiliated with Gammaproteobacteria, Deltaproteobacteria and Acidobacteria were present in all three basins. The distribution of these sharedphylotypesseemedtobeinfluencedneitherbytheWalvisRidgenorbydifferentdeepwater masses, suggesting a high dispersal capability, as also indicated by low distance–decay relationships. However, the total bacterial diversity showed significant differences between the basins, based on 16S rRNA gene sequences as well as on terminal restriction fragment length polymorphismfingerprints.Noticeably,bothgeographicdistanceandenvironmentalheterogeneity influencedbacterialdiversityatintermediate(10–3000km)andlargescales(43000km),indicatinga complexinterplay oflocal contemporaryenvironmental effectsand dispersallimitation. TheISME Journal(2010)4, 159–170;doi:10.1038/ismej.2009.106;published online15October 2009 SubjectCategory: microbial population and community ecology Keywords: barrier; biogeography; deep sea; Gammaproteobacteria; spatial scale Introduction Bioirrigation by the activities of larger benthic organism as well as near-bed currents (Hughes and Biogeographic patterns in microbial communities Gage, 2004; Queric and Soltwedel, 2007) influence are traditionally explained by two factors, the sediment-water interface exchange and conse- the environmental heterogeneity and historical quently lead to the dispersal of particles and events (Martiny et al., 2006; O’Malley, 2008). On therefore of microorganism. Barriers to microbial the basis of the cosmopolitan hypothesis, ‘every- dispersal could be physical (topography) or physio- thing is everywhere, but the environment selects’ logical conditions (temperature, pH or hydrostatic (Baas-Becking, 1934), environmental conditions pressure). have long been considered to have a strong influ- In the eastern South Atlantic Ocean, the Cape ence on microbial biogeography. The effects of Basin is separated from the Angola and Guinea spatial distances (historical events) have been basinsbytheWalvisRidgethatformsabarriertothe shown to affect microbial diversity in several northward and southward flow of water below a studies (Papke et al., 2003; Whitaker et al., 2003; depth of about 3000m (Shannon and Chapman, Martiny etal.,2006; RametteandTiedje,2007). The 1991).Furthermore,theCapeBasinisdominatedby relative influences of environmental heterogeneity Lower Circumpolar Deep Water arriving from Ant- and historical events on microbial biogeography are arctica and the deepest part of the Angola and still poorly understood. In marine habitats like the Guinea Basins are filled with North Atlantic Deep deep sea, microorganisms in the surface sediment Water originating from the Arctic (Bickert and may be assumed to disperse with oceanic currents. Wefer, 1996). Noticeably, the Walvis Ridge has been shown to function as a barrier for the dispersal of Correspondence: J Harder, Department of Microbiology, Max somecrustaceanspeciesofPeracarida(Brandtetal., Planck Institute for Marine Microbiology, Celsiusstr. 1, Bremen 2005), but it is not known whether this physical D-28359,Germany. barrier also affects microbial dispersal. E-mail:[email protected] To analyze whether different deep water masses Received 11 May 2009; revised 6 August 2009; accepted 2 September2009;publishedonline15October2009 associated with the physical barrier of the Diversityandbiogeographyindeep-seasediments RSchaueretal 160 Walvis Ridge have significant structuring effects on cores were sliced on board in layers of 2cm and the microbial diversity, the bacterial diversity in three layers were subsampled top-to-bottom by sterile deep-sea basin surface sediments was determined 1- to 2-ml syringes at 41C. After storage at (cid:2)801C, by16SribosomalRNA(rRNA)genesequencingand DNA was extracted from 0.5g of the surface sedi- the community fingerprinting method terminal ment sample (0–2cm) of the Cap, Angola and restrictionfragmentlengthpolymorphism(T-RFLP). Guinea I areas (Figure 1, Table 2) after the protocol The relative contribution of environmental hetero- of the FastDNA SPIN Kit for Soil (Q-BIOgene, geneity and of historical events on microbial Carlsbad,CA,USA).Bacterial16SrRNAgeneswere biogeography were assessed for this data set in amplified using the primer pair GM3/GM4 (Muyzer concertwithearlierpublisheddataonbasalticlavas et al., 1995). The 100-ml reaction contained 30ng in the Pacific Ocean (Santelli et al., 2008), shallow DNA as template, 0.5mM of each primer, 10mM of permanently cold sediment of the Arctic Ocean dNTPs, 1(cid:3) buffer (Eppendorf, Hamburg, Germany) (Ravenschlag et al., 1999) and Antarctic continental and 5 U of the Takara-Taq DNA polymerase shelf sediment (Bowman and McCuaig, 2003). (TAKARA, Dalian, China). PCRs were performed in 10 replicates with 20 cycles to minimize PCR bias. Final extension was performed 60min at 601C to increase 30-A-overhang.Theamplicons were pooled Materials and methods and purified with a PCR purification kit (Qiagen, 16S ribosomal RNA gene clone libraries construction Hilden, Germany). Cloning of the amplicons was Sediment sampling was performed on the DIVA II performed using TOPO TA Cloning Kit for sequen- cruise by a multicorer (Barnett et al., 1984) in water cing(pCR4-TOPO,Invitrogen,Karlsruhe,Germany). depths ranging from 5032 to 5649m. The sediment Clones with a correct insert size of B1500bp were Figure1 SamplingareasintheSouthAtlanticOceanaswellastheWalvisRidgethatseparatestheCapeBasinfromtheAngolaand Guineabasins.Forthe16SribosomalRNA(rRNA)geneapproachsurfacesediment(0–2cm)oftheCape,AngolaandGuineaIareaswere usedandfortheterminalrestrictionfragmentlengthpolymorphism(T-RFLP)analysis3–5surfacesedimentsoftheCape,Angolaand GuineaI–IIIareaswereanalyzed. TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 161 sequencedusingthevectorprimersM13F(50-GGAA as 558F (FAM, 50-ATTGGGTTTAAAGGGTCCG-30) ACAGCTATGACCATG-30)andM13R(50-GTTGTAA (Abell and Bowman, 2005a,b) and 1390R (HEX, AACGACGGCCAGT-30). 50-GACGGGCGGTGTGTACAA-30)(Zhengetal.,1996), targeting the class Flavobacteria. Undigested and digested amplicons were identified by capillary Phylogenetic and sequence analyses electrophoresis to verify the absence of false-posi- Thequalityoftheobtainedsequenceswasmanually tive fragments in the undigested control and the checked using Sequence Analysis 5.2 (Applied completeness of the digestion. PCRs were carried Biosystems, Weiterstadt, Germany). Full-length se- out in a total volume of 25ml, including 12.5ml PCR quences were assembled with Sequencher (Gene MasterMix(PromegaGmbH,Mannheim,Germany), Code, Ann Arbor, MI, USA). No chimeras were 1mM forward and reverse primer, and 5–24ng DNA detected with Bellerophon (Huber et al., 2004) and template. PCR reactions were carried out in tripli- CHECK_CHIMERA (Maidak etal., 1996). Sequences catesandpurifiedonSephadexcolumns(Sephadex were imported into the ARB software package G-50 Superfine, Amersham Biosciences AB, Uppsa- (Ludwig et al., 2004) and aligned using the ARB la, Sweden). PCR amplicons (70-120ng) were FastAligner, then refined manually. The ARB soft- digested in a total volume of 10ml at 371C for 3h ware package was used to generate phylogenetic using5UoftherestrictionenzymeAluI(Fermentas, trees of 810 full-length sequences using the max- Burlington,Canada)forbacterialampliconsand5U imum likelihood algorithm with a 50% positional of the enzyme MspI (Fermentas) for Flavobacteria conservation filter and with 100 bootstrap repli- amplicons. The two restriction enzymes were cates. Sequences reported in this study were chosen based on high numbers of unique terminal deposited at EMBL under the accession numbers restrictionfragmentsassessedwithinsilicoanalyses AM997284–AM997988 for 705 full-length sequences using enzyme restriction power analysis (http:// andunderAM997989–998333andAM997283for346 mica.ibest.uidaho.edu/) as well as on best perfor- partial sequences. mance in laboratoryexperiments (that is, producing The software distance-based OTU and richness maximumnumbersofterminalrestrictionfragments (DOTUR) was applied to ARB distance matrices (TRFs)). After heat inactivation (651C, 25min) and generated with the Jukes-Cantor correction to esti- purification on Sephadex columns, detection of mateoperationaltaxonomicunits(OTU),rarefaction TRFs was performed on a ABI Prism 3130 XL curves of observed OTUs, richness estimators and Genetic Analyzer (Applied Biosystems, Foster City, diversity indices (Schloss and Handelsman, 2005). CA,USA)equippedwitha80-cmcapillary,aPOP-7 A sequence identity of 98% was used to define polymer and the filter set DS-30. The ROX-labelled OTUs, as this cut-off roughly corresponds to the MapMarker 1000 (Eurogentec, Seraing, Belgium) species level (Rossello-Mora and Amann, 2001; served as a size standard between 50 and 1000bp. Stackebrandt and Ebers, 2006). The statistical tool The fragment profiles were visualized and automa- R -LIBSHUFF was applied to genetic distance ma- tically analyzed with GeneMapper v. 3.7 Software trices to determine whether differences in library (Applied Biosystems), using standardized settings compositionwerebecauseofchanceortobiological with a peak detection cut-off set to 30 fluorescence effects, and significances were assessed by Monte units. The 50-end labelled TRFs were used as they Carlo permutations and further corrected for multi- produced a higher number of fragments in compar- ple comparisons (Schloss et al., 2004). The statis- ison with 30-end TRFs (Suzuki et al., 1998; Osborn tical tool SONS (Schloss and Handelsman, 2006) et al., 2000). wasusedonfull-length16SrRNAgenesequencesto A binning procedure was applied to the Gene- calculate Chao1 shared richness estimates, the J Mapper output to compensate for slight peak shifts class indexfortheratioofsharedtototalnumberofOTUs, between runs and for TRF size calling imprecision, and y for the estimated similarity in community in order to avoid artificial, technically derived yc structure between any two communities. differencesbetweenprofiles(HewsonandFuhrman, 2006). The technical variability of peak size calling in different replicates including runs conducted on Terminal restriction fragment length polymorphism different days was determined as of±0.25bp (win- Terminal restriction fragment length polymorphism dow size of 0.5bp). The binning function included analyses included three to five samples of surface two different starting points (50 and 50.25bp) and sediments (0–2cm) from several cores of each area, the binning strategy yielding higher correlation Cape, Angola and Guinea I–III (Figure 1, Table 2). between all samples was selected for further Genomic DNA was extracted from 0.5g sediment statistical analyses. The binning window was samples using the FastDNA Spin Kit for Soil adjusted to 1bp for samples amplified with Flavo- (Q-Biogene,Irvine,CA,USA).PCRamplification ofthe bacteria primers, because a window frame of 0.5bp 16S rRNA gene was carried out using the fluores- did not yield higher resolution. The computation cently labelled primers 27F (FAM, 50-AGAGTTTGA wascarriedoutwiththeInteractiveBinnerfunction TCCTGGCTCAG-30) and 907R (HEX, 50-CCGTCAAT (Ramette, 2009 http://www.ecology-research.com). TCCTTTRAGTTT-30), targeting all bacteria as well The output consisted of a table of TRFs with TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 162 corresponding relative fluorescence intensities, deep-sea (Bowman and McCuaig, 2003; Polymena- which are the individual peak area divided by the kou et al., 2005, 2009; Xu et al., 2005) and shallow totalareaofpeaksinagivenprofile.Masterprofiles sediments (Ravenschlag et al., 1999) had also found were generated by building a consensus table of the alargediversity,whichmaybebasedonaweakand binned TRF profiles for all samples from one basin, symmetriccompetition(Grant,2000).Thereciprocal averaging the respective relative fluorescence in- Simpson’s indices for all sites were above 50 tensities values of all samples. A TRF was consid- (Table 2), suggesting evenly distributed diversity ered present if it appeared in one or more PCR profiles as typical dominance profiles show 1/D parallels, therefore including all natural and tech- values below approximately 50 (Zhou et al., 2002). nical variability at this level of analysis. Total richness estimates (Chao1) (Table 2) and rarefaction curves (Supplementary Figure S1) based Statistical analyses on a 98% sequence identity showed that Cape, Non-metric multidimensional scaling (nMDS) and Angola and Guinea basin surface sediments con- analysis of similarity (ANOSIM) were carried out tained an equal bacterial richness at a significance with the program PAST (Paleontological Statistics, level of 0.05. ver. 1.47, http://folk.uio.no/ohammer/past). Simple Both analyses predicted a lower richness for the and partial Mantel tests were used to determine the South Atlantic sediments in comparison to the significance and correlation coefficients between Antarctic sediments and a higher richness in genetic-, spatial- and environmental distance matri- comparison to the Arctic sediment. The library- ces, using theR package vegan (http://vegan.r-forge. based equality of richness was supported by the r-project.org/) (Legendre and Legendre, 1998; T-RFLP analysis, as basin-specific master profiles Mantel, 1967). Spatial dissimilarities based on showed a comparable OTU richness (167, 190 and geographic distances between sites and environ- 182 TRFs for the Cape, Angola and Guinea Basin, mentaldissimilarities(temperature,salinity,pH,Eh, respectively) (Figure 4a). TOC, Chl a and grain size; Table 1) were used to explain genetic dissimilarity. To determine the strength of the relationship between genetic and Bacterial diversity of the 16S ribosomal RNA genes geographic distance linear models were fitted and The clone libraries contained 521 phylotypes with slope coefficients were calculated with their 95% 98% identity in 1051 sequences, containing 705 confidence intervals. full-length sequences. Applying a 100% identity threshold revealed 230 sequences, which were present at least twice, with a majority of 176 Results and discussion sequences (18 OTUs) present in all deep-sea sedi- Bacterial biomass and richness in sediments of the ments. The bacterial communities were dominated South Atlantic Ocean by Proteobacteria, which accounted for 64, 58 and The cell numbers of the suboxic surface sediments 63% of all sequences in the Cape, Angola and (0–2cm) in three eastern South Atlantic Ocean Guinea Basin, respectively, with the class Gamma- basins were 3.4–3.7(cid:3)109cellsg–1 sediment proteobacteria representing 45, 37 and 40% of all (Table 1). The abundances were in the range found sequences in the respective basins (Figure 2). The inotherdeep-seasediments(9.2(cid:3)108cellsg–1(Dem- class Gammaproteobacteria comprised 116 phylo- ing and Colwell, 1982), 1.5(cid:3)109cellsg–1 (Guezen- types (98% identity, 427 sequences), of which 39 nec and Fiala-Medioni, 1996) and 5(cid:3)108cellsg–1 phylotypes (138 sequences) were related to known (Harvey et al., 1984)). The 16S rRNA gene libraries cultivated species. These belonged mainly to fa- showed a high diversity with up to 20 different milies of psychrophilic microorganisms including phylain theCape Basin and17 phylainthe Angola Enterobacteriaceae, Alteromonadaceae, Oceanos- and Guinea basins (Figure 1). Earlier described pirillaceae and Legionellaceae (Figure 3a). Among Table1 Sedimentdata(Tu¨rkeyandKro¨ncke,inpreparation)andcellnumbersofmicrobialcommunitiesintheSouthAtlanticOcean Basin Depth Temp. Salinity pH Eh TOC Chla Grainsize(%) Cellcounts MPN (mbsl)a (1C) (%) (mV) (%) (mgg(cid:2)1) (cellg(cid:2)1) (cellsml(cid:2)1) o63mm 463mm Cape 5032 1.14 34.6 7.74 177 0.83 0.017 92.89 6.87 3.5(cid:3)109 1.22(cid:3)104 Angola 5649 ND ND 7.72 96 0.9 0.069 83.84 16.4 3.4(cid:3)109 2.67(cid:3)105 GuineaI 5063 2.1 34.9 7.77 183 0.72 0.264 84.23 15.23 3.7(cid:3)109 2.67(cid:3)104 GuineaII 5225 ND ND 7.76 132 0.77 0.301 84.99 14.46 ND ND GuineaIII 5525 2.1 34.5 ND ND 0.76 0.152 86.45 13.34 ND ND Abbreviation:ND,notdetected. aMetersbelowsealevel. TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 163 Table 2 Sampling sites of sediments used for 16S rRNA gene sequencing or for T-RFLP analysis with corresponding richness and diversityindicesforbacteria Sampling Latitude Longitude Depth Stationsb Stationsb No.of OTU Richness Simpson Study area (mbsl)a T-RFLP 16S clonescDOTUR estimator 1/D rRNA 0.02 Chao1d gene seq. Cape 2810604200 S 712004800 E 5032 33,34,37,38 33 342FP 202 466(369,620) 77 Thisstudy 228F 145 508(351,785) 53 Angola 915600000 S 015304800 E 5649 46,48,50 46 354FP 183 256(227,305) 77 Thisstudy 219F 126 369(259,570) 59 GuineaI 010000000 S 212500600 W 506356,58,59,60,61 60 355FP 203 369(308,465) 125 Thisstudy 258F 155 489(348,735) 91 GuineaII 015000000 N 513500000 W 522574,75,76,77,79 — — Thisstudy GuineaIII 013701200 N 612800600 W 5525 95,97,99 — — Thisstudy Antarctic 6613108600 S 14313803000 E 761 — MERTZ 590P 322 899(713,1175) 125 Bowmanand continental 0–2cm McCuaig,2003 shelf Arcticocean 7914208100 N 1110501800 E 218 — StationJ 123P 84 125(104,167) 167 Ravenschlag Svalbard etal.,1999 EastPacific 912804800 N 10411304800 W 2516 — EPR 370F 239 601(475,796) 200 Santellietal., 2008 Rise (cid:2)915003800 N(cid:2)10411708600 W(cid:2)2674 Hawaii 1815201700 N 15511405300 W 888 — PV 472F 276 764(597,1017) 167 Santellietal., 2008 (cid:2)1815803100 N(cid:2)15515304200 W(cid:2)1714 Abbreviations:DOTUR,distance-basedOTUandrichness;OTU,operationaltaxonomicunits;rRNA,ribosomalRNA;T-RFLP,terminalrestriction fragmentlengthpolymorphism. aMetersbelowsealevel. bFordetailsseecruisereportDIVAII(M63/2). cNumberoffull-length(F)andpartial(P)sequences,full-lengthsequencesandvaluescalculatedfromthemarepresentedinbold. dChao1richnesswithlowerandupperboundof95%confidenceinterval. these phylotypes 11 OTUs (12 sequences) clustered phyla Chloroflexi (1, 10 and 4% for Cape, Angola withtheNOR5/OM60cladethatincludes‘Congregi- andGuineabasins,respectively),Planctomycetes(6, bacter litoralis’strain KT71, the first marine aerobic 4 and 10%), Acidobacteria (4, 7 and 5%) and anoxygenic phototrophic Gammaproteobacteria Bacteroidetes (10, 4 and 6%). in culture (Fuchs et al., 2007; Yan et al., 2009). Three phylotypes (5 sequences) were related to free living (Thiothrix) and endosymbiotic sulfur oxi- Bacterial diversity comparison ‘dizers and methylotrophic bacteria. A large por- The proportion of bacteria present in two or three tion of 77 phylotypes (289 sequences) clustered basins was high in the 16S rRNA gene sequences distinctly from cultured species to JTB255/BD3-6 analyses (23%) and in the T-RFLP analyses (58%) (38 phylotypes, 192 sequences), BD7-8/MERTZ (Figure 4a). A third of the fragments (93 TRFs) was (10OTUs,36sequences),JTB23/Sva0091(18OTUs, detected in the sediments of all basins and repre- 34 sequences) (Figures 3a and b) and to Cret-1F, sented 82% of the total relative fluorescence BD1-1, PWP and South Ionian groups (11 OUT, intensities. Among the 16S rRNA gene sequences, 27 sequences). These groups included only 16S a shared membership of 19 OTUs (98% identity) rRNA gene sequences that originated from other wasfoundinallthreebasinswiththestatisticaltool deep-sea or permanent cold marine habitats (Kato SONS. The manual assignment in ARB confirmed et al., 1999; Li et al., 1999; Ravenschlag et al., 1999; the small fraction of OTUs detected in all three Urakawa et al., 1999; Bowman and McCuaig, 2003; basins (29 OTUs, 347 sequences), but provided Polymenakou et al., 2005; Xu et al., 2005; Zhao and additional information regarding the sequence Zeng, 2005). abundance and identity of each OTU. These The Alpha-, Beta- and Deltaproteobacteria ac- were dominated by Gammaproteobacteria (76%; counted together for 18 to 23% of all sequences in Figure4b).Inthisclass,thecommonmemberswere the libraries. Deltaproteobacteria (11 to 14%) out- related to marine heterotrophic aerobic and faculta- numbered Alphaproteobacteria (6 to 8%) and tive anaerobic microorganisms (Alteromonadaceae Betaproteobacteria (1 to 3%) (Figure 2). Other and Oceanospirillaceae), photoheterotrophic aero- groups with a sequence abundance of over 5%, bicbacteria(NOR5/OM60clade)(Fuchsetal.,2007), which occurred in all three basins, were the and to groups consisting of uncultivated bacteria TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 164 candidate division WS3 candidate division TM6 Guinea Basin I candidate division OP11 Angola Basin candidate divison OP5 Cape Basin candidate division OP3 candidate division OD1 candidate "Chaldithrix" candidate division BRC1 Verrucomicrobia Sphingobacteria Flavobacteria Bacteroidetes s Acidobacteria s a cl Gemmatimonadates Fibrobacteres Planctomycetacia Actinobacteria Clostridia Deltaproteobacteria Gammaproteobacteria Betaproteobacteria Alphaproteobacteria Chlorobi Cyanobacteria Nitrospira Chloroflexi 0 10 20 30 40 50 relative abundance of sequences [%] Figure 2 Bacterial diversity in the Cape Basin (342 sequences), Angola Basin (354 sequences) and Guinea Basin (355 sequences). AlldetectedclassesinthedomainBacteriaareshown. (JTB255/BD3-6,JTB23/Sva0091/BD3-1,BD7-8/MERTZ, and significantly different Flavobacteria T-RFLP Gret-1F and South Ionian). profiles (Figure 5b) (analysis of similarity, R values Phylotypes present in two of three basins be- 0.869, Po0.001) (Supplementary Table S2). These longed to the Gamma- and Deltaproteobacteria and differences were consistent with a different to the Chloroflexi. A major group of Chloroflexi- chlorophyllacontentaswellasadifferentsediment OTUs were restricted to Angola and Guinea basin particle size in the Cape Basin (Table 1) (Etter R sediments (7 OTUs, 21 sequences). The -LIB- and Grassle, 1992), indicating that environmental SHUFF analyses revealed no significant difference factors seem to influence bacterial communities between the Angola and Guinea libraries as well as in deep-sea sediments of the eastern South Atlantic Cape and Guinea libraries (using a minimum Ocean. It is, however, needed to also take spatial P-value of 0.0012) (Supplementary Table S1). Thus, parameters into account in this analysis to common phylotypes dominate the communities of strengthen our interpretation concerning environ- thesebasins.ThelargestnumberofTRFscoveredby mentalorspatialeffectsontheobservedcommunity two basins was found for the Angola and Guinea shifts. basins (30 TRFs, 30.5 relative fluorescence inten- sity) (Figure 4a). High chlorophyll a contents were detected in the Angola and Guinea surface sedi- Biogeography: environmental and historical factors ments indicating a large fraction of fresh, recently In the eastern South Atlantic Ocean the Walvis arrived organic carbon (Table 1, Tu¨rkay and Ridge separates the Cape Basin from the Angola Kro¨ncke, in preparation). This probably originated and Guinea basins below a depth of about 3000m from a primary productivity in the surface waters and causes different deep water masses in these thatcanbelinkedtothedischargeofnutrientsfrom basins. The dominance of common phylotypes in the Congo and the Niger Rivers into the Angola and the 16S rRNA gene libraries and T-RFLP master Guinea basins, respectively (Schefuss et al., 2004). profiles suggested that microbial dispersal may Angola and Cape basins showed significantly not be influenced by the Walvis Ridge or by the R different communities ( -LIBSHUFF test, P¼0.008) presence of different water masses. This was TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 165 Figure3 Phylogenetictreebasedon16SribosomalRNA(rRNA)genesequencesoftheclassGammaproteobacteriashowingpositionof (a)marineheterotrophicaerobicandfacultativeanaerobicmicroorganismsandphotoheterotrophicaerobic(NOR5/OM60clade)bacteria and(b)potentialauto-ormixotrophicsulfuroxidizersandbacteriathatinhabitvariousgeographicregions(JTB255/BD3-6).Thetreewas calculatedusingthemaximum-likelihoodalgorithmwitha50%positionalconservationfilterandwith100bootstrapreplicates.Thebar represents10%estimatedsequencedivergence.Full-lengthsequences(Ca,AnandGu),partialsequences(cap,angandgui),thenumber ofOTUsinaclusterandthecorrespondingnumberofsequences(squaredbracket)areshown. supportedbythesignificantlysimilardistance–decay very low (0.003 to 0.07) (Table 3), as also found relationshipsoftheTRFsinthepairwisecomparison in taxa-area relationships for soil and salt marsh (Cape/Angola, slope coefficient 6.9(cid:3)10–5 and 95% communities (0.03 to 0.074) (Green et al., 2004; confidence interval (3.4(cid:3)10–5, 10.3(cid:3)10–5); Angola/ Horner-Devine et al., 2004; Fierer and Jackson, Guinea, slope coefficient 8.7(cid:3)10–5 and 95% confi- 2006), suggesting high dispersal rates and low dence interval (1.5(cid:3)10–5, 15.9(cid:3)10–5)). Phylotypes extinction rates because of vast population sizes common in the communities of the South Atlantic (Connor and McCoy, 1979). Ocean and the Pacific, Antarctic and Arctic Oceans From the clustering of TRF profiles by basins, as sediments (Supplementary Table S1) indicated that shown by non-metric multidimensional scaling somemicroorganismsdisperseeffectivelyoverahuge (Figure 5a) associated with large, significant distance and therefore are cosmopolitan, at least at R values for all pairwise comparisons between the resolution of 16S rRNA genes that is insufficient the deep-sea basins (analysis of similarity, 0.586 for the classification of microorganisms into species to 0.999, Po0.001) (Supplementary Table S2), and (Konstantinidis and Tiedje, 2005). from significant differences between the South To get more information regarding the amount Atlantic Ocean communities to all other commu- R of spatial structure present, we analyzed the rela- nities( -LIBSHUFFtests,SupplementaryTableS1), tive relationships between genetic diversity and itseemedobviousthatcommunitieswerestructured geographic distances. The 16S rRNA gene and either by the contemporary environment, spatial TRFs based distance–decay relationships for the distances (historical events) or by a combination of South Atlantic Ocean and for all sites were all both(Martinyetal.,2006;RametteandTiedje2007). TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 166 a total richness945 (Cl=736-1244) Guinea Basin I TRF 283/RFI 299.7 Chao 489 TRF 182 Chao 322 TRF 41/RFI 3.8 Chao 77 Shared Richness TRF 18/5.9 C-G=96 Chao 71 C-A=68 TRF 30/30.5 G-A=90 Chao 19 Chao 345 TRF 93/245.5 C+G/A=142 TRF 34/RFI 1.9 C+A/G=167 G+A/C=163 Chao 49 TRF 22/5.3 Chao 227 Angola Basin TRF 45/RFI 6.8 Chao 369 Cape Basin Chao 508 TRF 190 TRF 167 b 300 250 s Ca+An e c en 200 Ca+Gu I u q An+Gu I e s of 150 Ca+An+Gu I er b m 100 u n 50 0 ChloroflNietAxrliopshpaiCrprahlBoeottreaoGopbabriaomtctemeoaribpDaraeoclttteearbcipraalcoattseeroisAabcaticPtlneaorinbacGatecotmemriymacateitamcoinaAaciddatoebsBaactStecerrpioihaidnetgceoVasebnrardicutdcearoticeaa mi"ncdCrihodabalitdaei tdhirivixs"ion OP3 Figure 4 (a) Shared and basin-specific SONS generatedOTU and terminal restrictionfragment (TRF)and corresponding relative 0.02 fluorescenceintensities(RFI)valuesforthethreedeep-seabasinoftheSouthAtlanticOcean.(b)PhylogenyofthesharedOTUsbetween threeandtwodeep-seabasinsandthecorrespondingnumberofsequences. To disentangle the relative influence of environ- Butenvironment(r¼0.588,Po0.001)overwhelmed mental heterogeneity and spatial distance on the anyeffectofgeographicfactors(r¼0.278,P¼0.009) distribution of microbial deep-sea sediment com- for intermediate distances (1200–3500km), as munities, we used a combination of simple and also supported by significant partial Mantel tests partial Mantel tests. For distances of 0–1200km (Table 3). A higher correlation between spatial and T-RFLP results showed a comparable influence genetic distance for small spatial scales (o200m) of both factors (environment r¼0.636, Po0.001, was reported for other microbial groups in soil geography r¼0.651, Po0.001) (Table 2) (Figure 6b). (Cho and Tiedje, 2000), suggesting the existence of TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 167 Guinea II mberofmples 14 31119 7 multiple Guinea III Nusa or f Guinea I Cape Angola PartialManteltests GeographyEnvironment 0.054(0.306)0.131(0.134) 0.031(0.002)*(cid:2)0.021(0.981)(cid:2)0.138(0.86)0.768(0.001)*o(cid:2)0.227(0.992)0.573(0.001)*o NANA lengthpolymorphism.significantafterBonferronicorrection publishedstudies. assbdlsiIttgcCFwesr(Fuaqannihremuehaplisseinuoeviagreiv¼tedmdecnWMroatptdnuvhietnnheeltietoilgwerlewSut0psiymiactaMhebyetfamdo,.dileltinadde)e0scrB5lieeeicfcudsipigtalinn,r0¼ahdndstnaedletcIateticg8.asadhttlyat(rgoic0tautsSihdn,ns–dniiatooo)t.e.otACtfcGraPeid-0hrbtramfGnefNasfawuexeesuf1tfpans¼stnurowaoleafsripgies3ettioatcinenitro,r:cnnhier0ras,inse-etotni0naevnmoher,f.bstanPtn.bia0tleheg0afdci.eAsoyaa fcl0ecdd7frIr¼tiocuItlsnroesIa.e.6Iputhidrtgimeesrcpisdr0aISsvaerni.anufnh.imtemieirSona0narGcccgloeexri1tlmeereute0deurrncnssffasr6seltpoel1isutoivmoniseoStrifr1fbfa)dcssdorieiloci5nd¼:6mlirpeuriica(t1rla0meGosST dalet7t.rrIl0dtdtne1IitrheinauFRiahhvos.5d,nesimrb0etnieeN.nfeRsbtten2ald2tf(iaaccbene7iebheoANre4oniedoonFdf)napreelar,saf3CArgcen,acolt-traPalee1,MnaesmatItTeng(e+4topehsatFMg¼laD-idec5reagIciaaRvseeniain7SlInmerhtgaa09eel+nRFGdtiaeitcpu0lnvp.noawIeLpmiu0lr7IAlgretdaonnndeiiIRPh0eeonnftlngessei(asdfisl1dyn(uevrgcleeetsa’T6imuaa.a,sMbshoqeqovneoita andelerernoiIduucgsDeaf)obecidsnatTgordgeteSnwisolssalhwecawBtnnhe)upyiespyafweosermpbctrecseefianith3oedsifleesteorooyactomiienhd)yeenhssss–---flf.t- Table3SlopecoefficientsandMantelrstatisticsforgeneticdistancematricesderivedfrom16SrRNAgenesequencesorT-RFLP Slopecoefficient(geneticversusSimpleManteltestsgeographicdistance) (cid:2)5aLogtransformedGeographyEnvironment(cid:3)1010 transformedUntransformedLog10 SouthAtlantic,0–1200kmdistanceb0.074[0.053,0.095]0.651(0.001)*0.593(0.001)*0.657(0.001)*T-RFLP28.28[21.34,35.22]ooo SouthAtlantic,0–3500kmdistance16SrRNAgene0.19[0.17,0.22]0.0036[0.0028,0.0044]0.024(0.001)*0.012(0.001)*0.008(0.006)*T-RFLP(GuineaI)7.5[5.38,9.62]0.076[0.063,0.090]0.698(0.004)*0.841(0.001)*0.886(0.001)*T-RFLP(GuineaI+II+III)2.6[1.24,3.96]0.055[0.042,0.068]0.278(0.009)*0.532(0.001)*0.599(0.001)*o Allsites,0–18000kmdistance16SrRNAgene0.02[0.019,0.022]0.0030[0.0027,0.0032]0.013(0.001)*0.013(0.001)*NA Abbreviations:NA,noenvironmentalparametersavailableforallcomparisons;rRNA,ribosomalRNA;T-RFLP,terminalrestrictionfragmentForsimpleandpartialManteltests,P-valuesbasedon1000permutationsareindicatedinparentheseswith*whentheP-valuesarestillcomparisons.aForsimpleManteltests,geographicdistanceswereeitherleftuntransformedorlog10transformedtoallowforcomparisonswithpreviouslybLowerandupperboundsof95%confidenceintervalfortheslopecoefficientsareindicatedwithinsquarebrackets. TheISMEJournal Diversityandbiogeographyindeep-seasediments RSchaueretal 168 The influence of both factors at intermediate scales was already shown by other studies (Green et al., 2004; Reche et al., 2005; Yannarell and Triplett, 3 2005),butourstudysuggestaneffectofbothfactors es for large scales as well, as shown for soil microbial c n communities (Fierer and Jackson, 2006). Although a st thesmallsize,highdispersalrates,largepopulation di 2 e sizeandlowextinctionratesofmicroorganismssug- en gest a low effect of geographic barriers on micro- g A organisms (Staley and Gosink, 1999; Beja et al., N 2002; Finlay, 2002; Ramette and Tiedje, 2007), our R 1 S r studyshowsthatthedistributionofmicroorganisms 16 in deep-sea sediments is limited at intermediate (10–3000km) and large scales (43000km). 0 0 5000 10000 15000 20000 Acknowledgements Geographic distance [km] We thank M Tu¨rkay and P Mart´ınez Arbizu for the invitation to participate to the Meteor cruise DIVA II 0.7 (M63/2) and the captain, crew and shipboard scientific partyfortheirassistanceduringsampling.Wearegrateful to I Kro¨ncke and M Tu¨rkay for providing environmental 0.6 data of the deep-sea sediments. The project has been carriedoutintheframeworkoftheConsensusofDiversity s ce ofAbyssalMarineLife(CoML/CeDAMar)andtheMarBEF an 0.5 Network of Excellence ‘Marine Biodiversity and Ecosys- st tem Functioning’ (contract no. GOCE-CT-2003-505446). di F This work was financially supported by the Max Planck R 0.4 Society. T 0.3 References 0.2 Abell GCJ, Bowman JP. (2005a). Ecological and bio- 0 500 1000 1500 2000 2500 3000 3500 geographic relationships of class Flavobacteria in the Geographic distance [km] SouthernOcean.FEMS Microbiol Ecol 51: 265–277. Abell GCJ, Bowman JP. (2005b). Colonization and com- Figure 6 Scatterplots with loess curve presenting (a) genetic munity dynamics of class Flavobacteria on diatom dissimilarity plotted against geographic distance for genetic dissimilarity of distance matrices derived from 16S ribosomal detritus in experimental mesocosms based on RNA(rRNA)genesequencesofsamplesoftheSouthAtlanticand Southern Ocean seawater. FEMS Microbiol Ecol 53: (b) for terminal restriction fragment length polymorphism 379–391. (T-RFLP) profiles of samples from five different stations in the Baas-BeckingLGM.(1934).GeobiologieofInleidingTotDe SouthAtlantic(Cape,Angola,GuineaI,GuineaIIandGuineaIII). Milieukund. WP Van Stockum and Zoom (in Dutch): TheHague,the Netherlands. Barnett PRO, Watson J, Connelly D. (1984). A multiple corer for taking virtually undisturbed samples from provides information regarding randomly chosen shelf,bathyalandabyssalsediments.OceanolActa7: phylotypes (‘sampling communities’) where the 399–408. finding of an OTU is proportional to its abundance BejaO,KooninEV,AravindL,TaylorLT,SeitzH,SteinJL in the clone library (Bent and Forney, 2008). In etal.(2002).Comparativegenomicanalysisofarchae- contrast, the fingerprinting method T-RFLP screens algenotypicvariantsinasinglepopulationandintwo for all OTUs present above the detection threshold different oceanic provinces. Appl Environ Microbiol of the method (‘screening’ communities; Bent and 68: 335–345. Forney2008),typically46(cid:3)102–103DNAfragment BentSJ,ForneyLJ.(2008).Thetragedyoftheuncommon: understandinglimitationsintheanalysisofmicrobial copies per ml samples (Ramette 2009), but does not diversity.ISME J2:689–695. provide clear taxonomic distinction (Dunbar et al., Bickert T, Wefer G. (1996). Late quaternary deep water 2001). circulationintheSouthAtlantic:reconstructionfrom Although high dispersal rates were detected for carbonate dissolution and benthic stable isotopes. In: some groups in deep-sea sediments, both T-RFLP Berger WH, Wefer G, Siedler G (eds). The South and 16S rRNA-based analyses suggest barriers for Atlantic: Present and Past Circulation. Springer: the dispersal of microorganisms in the deep sea. Heidelberg, pp599–620. TheISMEJournal