the plume material from below and the increased temperature.rvill redr-rce the velocity of elastic waves. In the vicinity of the present plume centre, only the lowest part of the lithosphere is heated ar.rd the lithospheric thickness does not change much rvith respect to that of a 90-100-Myr-o1d oceanic plate. As time increases, more and more material from the lower part of the lithosphere is warmed up and the seismic LAB becomes shallor.ver. After 3 4 Myr of heating time the thinnest lithosphere of 50-60 km is achieved beneath Oahu and Kauai-a location about 400 krn ar,r.av from the current plume centrel0. Further away, at Midway Island, which passed the plume centre about 28 Myr ago, the plume material begins to lose its temperature and the lithosphere thickens again"'''. We conclude that the Hawaiian plume has gradually reheated up to 50 km of the lower lithosphere lr.ithin the last 3 4 Myr in an area of 500 X 300km along the island chain. This n'reans that a sub, stantial amount olheat is trapped belorv tl.re lithosphere. The lateral extent of the asthenospheric updoming is sigr.rificantly smaller than the extent of the Hawaiian topographic swel1. A consequence of our resuit could be that the entire topographic srvell is caused by, for example, the dynamic support of the plume, and so rejuvenation rvould be contributing only in the central part. Data from longer- term broadband stations on Kauai, and on the ocean bottom northwest of Kauai, couid permit imaging of the 're-ageirg' of the rejuver.rated Hawaiian lithosphere. tr Receii,ecl 27 Scptember 2003i acccptecl I ,1 .lanuan' 100.1i cloi: I rl. I 03 3,/nature02i,i9. l. \{ilsoD, J. T. A possible origin olthc Harvaiian islancl. (.rn. I. I,ltys.4l, 86-l E68 (19{i.l). 2. l\4organ,\\i.l.Con\,ectileplLnlesiDthclorrcrmantle.\-,ilr,d230,111,1(i9/l). 3. Nalaf,II.C.Seisnicinagingol-nrantleplumes 1rrrr.Rdr.lld,.lrPlantt.ici.28,19t.11;(2000). 1. Rlbe, N. \f. & Christensen, Li. R. The dmamical orlgln ofthe Ilauaiian volcanism. Edrrft Pldr.t. -S., Lel. t7t, 517 531 ( 1999). 5. Detrick, R. S. & Crough, S. T. Ishnd subsiderce, hLrt spols. rnd lithospheric thinning. /. Ceop/r.1,s. ps5 83, 1236 r2,14 0978). 6. Sleep, N. H. Ilotspots ard mantle pluneir Somc phenonrenolog.r. l. Geophy' Rd 95,6715 6736 (1990). Jordan, Il H. in Thc llantLe Santple: 1u.lrr i.rl lr ,(ilrl,.r/it.s ontl OLher )'ol.canic (ecls Bovd, F. R. & Me1"er, IJ. O. A.) 1 l4 (Pro .lnd Int. Kimhcrljtc (-onterence, r\mericarl Ccoph).sical Union, 1979). Robiisor, l. lU. lhe topographic and grar itrtional erpression ofdcDsiil, anomrlies due Lo nrell crtractioninlheuppennostoceanicnentie-L,trtltPLtiltt 5.i.I.fi.90,221 228(1988). 9. Plipps trlorgan, I., luorsan, \\'. j., Zhelrg. \. S. & Smith, \\'. H. F ObserlariorJ hiits for a plumc fid suboccaricasthcnosphereanditsrolcinmrnrl onyectiof./.C.oprl,s.X.r.l00,l2753 l2;6;il99a) 10. Priest1e1,, K. & Tilmaill, f. Shear Iale cuurrure ofthc lirhosphere above the Ilarvaiian hot spot liom t\\,o st:1tion ravleigh ware ph.rse relorii! mearuremcnts. (;.oplrys. ,R \. ttit. 26, i.l9l-1.196 ( r999). 1 1. \{oods, NL, Levequc, I. I. & Okil. I:. .1. Tri o-station measurernents oI Ra,vieigh rvavc group velocity along the IIalvaiian swe1l. (ieopl,rr. ,R i.ii 18, 105 108 (1991 ). I 2. \\bods, \,1. T. & Oka1, E. A. Ravleigh-r ar e iti.persion rlong the Harvaiian Srvcll: a tcst of iithospherlc thinnirg by the thcrmal re]uvenation et r hr)ripor. C.dpl7l,s l. Iilt. 125, ]25 -l-19 (1996). l.l. Bock, G- Long period S to P conreneLl \\.1\ tj .tnd the olls(.t of pirLial rnel Li ng ben eath Oahu, Hat aii. Ceop,tl'5. p,5. Ie1l. 18, 869 [t7] { 1991 r. 14. Li, X. c1 a/. Nlapping thc Harraiian plum: rirh couyertcd seismic wa1.es. r\rdtrr.405, 9-33-9.11 (2000). I 5. Col1ins, j. A., \icrnon, L I ., Orcutt, I. -{. .\ S:cphen. R. A. Lppcr mantlc structure benerth the Hawaiian swell: constraints trcm the ocean seisntir n.ari ork pilot cxpcrimcnt . G$phfs. Rrs. 1.ttr. 29, doi: I01029/2001G101 3302 120021. 16. Laske, G., Phipps \lorgan, J. & Orcutt. l. 1. l:ir rL r :.ults ii,rnr tlle H:Liian S\\i[LL pilor crpcriment. Ceoplqs. Res. Lcu.26,3197 3100 (1999i. I 7. Fabcr, S. & trhlller, G. Sp phascs from the transiiion zonc br-heen Lhe upper and lorver rnantle. Ba//. Scisnol. .Soc. r\ar. 70, 4lt7 50iJ ( l9ll0). 18. Farra, \i & \rinnilr, L. Upper ilantle stratill.etion l)\ I ilJ S rccci|cr flmcrions. Gcophys. l. )nt. l4l, 699 7r2 (2000). Supplementary lnlolmation ;rccompanies the papcr on www,nature,c0m/nature. Acknowledgements \,\,'ethankC.Aschandf.X,Iechiefbrtheiraili.\\ial-\orhankthcpcopleofthe Kaieahala National Park for supporting our station I\IAUl. l-his rork hes bccn supporred bv rhe l)eLrtsche Forschungsgemcinschafl lvithin the ICDP trrroject and br the GeoForscltungszentrunl, Potsdam. \{aveform data havc been provided by the IRIS, GEOSCOfE and GLOFON dattr centres. Competing interests statement The authors declirre that thcl,haYe no colpering flnancial intercsts. Correspondence and requests fbr natcdals should be acldrcsscd to R.K. lkind6.gt pots.hn.clc). NATURE I VOL.127 26 FEBRUARY 2004 l nfl\$ iatrue.com/nature letters to nature lron Gorrosion by nouel anaerobic microorganisms Hang T. Dinhl, Jan Kuevert,z, Marc MuBmann,, Achim W. Hassel3, Martin Stratmann3 & Friedrich Widdelt tMax Planck InstitLtte .flr NIarilrc Microbiology, CelsiusstraJSe l, 28359 Brenten, Gerntany ).Ittstitute .fitr Material Testing, PatLL-Feller-Stra.{3e l, 28199 Brernert, Germar,y 'Max Planck Institttte for Iron llesearch, Max Plunck,stralJe 1, 402 j7 Di)ssetrlorf, ()ermany Corrosion of iron presents a serious economic problem. Whereas aerobic corrosion is a chemical processr, anaerobic corrosion is frequently linked to the activity of sulphate-reducing bacteria (SRB)'z 6. SRB are supposed to act upon iron primarily by produced hydrogen sulphide as a corrosive agent3,s,7 and by consumption of 'cathodic hydrogen' formed on iron in contact with water2-6'8. Among SRB, Desufovibrio species-with their capacity to consume hydrogen effectively-are conventionally regarded as the main culprits of anaerobic corrosion2 6,8-10; however, the underlying mechanisms are complex and insuffi- ciently understood. Here we describe novel marine, corrosiye types of SRB obtained via an isolation approach with metallic iron as the only electron donor. In particular, a Desulfobacter- ium-like isolate reduced sulphate with metallic iron much faster than conventional hydrogen-scavenging D esulfovibrio species, suggesting that the novel surface-attached cell q?e obtained electrons from metallic iron in a more direct manner than via free hydrogen. Similarly, a newly isolat ed, Methanobacterium-like archaeon produced methane with iron faster than do known hydrogen-using methanogens, again suggesting a more direct access to electrons from iron than via hydrogen consumption. Some l0% of al1 corrosion damages to metals and non-metals may result from microbial activities'r. A significant process in this respect is the anaerobic corrosion ofiron or steel, for instance in oi1 te^chnology'"' The primary dissolution (Fe e:] Fe2+ i 2e ; E" - -0.44V) can in principle be driven by numerous oxidants. In oxic humid surroundings, oxidrtion by 02 (El]s7: f0.82V) yields rustl. In anoxic surroundings, H+ ions frbm water yieid H1 (E$uz: -0.41V). Of the sequential sreps (2e +ZFi+- 2Hladso,bedy + H2(adsorbcd) * H:1"q*"urr)), the combination of the H irtoms is presumably rate-limitingr2 and is a main reason for the slowness of iron oxidation in anoxic sterile water (Fe * 2HrO Fet* + Hu * 2HO ). However, sulphate-reducing bacteria [SRB) promote the anaerobic oxidation of ironz tj. Their activity often occurs in biofilms and tends to pit the iront,.,, An indirect and a direct corrosion mechanism may be distinguished3,.: these may occur simultaneously at different extents, depending on the load of rvaters with biodegradable organic compounds. The indirect mechanism is a chemical attack by hydrogen sulphide (Fe -l H2S * FeS -1_ Hr) which is faster than that by rvater3'5'7 and also promotes so-called hydrogen embrittlement of the metalr'12 (see also http://nr,wv.corrosion,doctors.org). Because SRB commonly use organic compounds (showrr here as ICHrO]) and often aiso H2 for sulphate reduction, the net reaction of indirect corrosion (for complete carbon oxidation by SRB contmunttie:, can be written as: 2[cH2o] + rrlFe + 1rA soi t'lr]ts* l1) 2HCot + rf,res+ t%H2o In the direct mechanism according to the depolarization ineorr- (see Supplementary Infonnation)S, SIIB are suppr6531l rrr ltimulate corrosion by scar.enging 'cathodic hr,drogen' or a '}tldroqen Ii1m' 429 E letters to nature on water-exposed iron (often written with unspecified hydrogen as 8LHl + SOi- + 2H+ - H2S + 4H2O). The resulting ner reiction of direct corrosion is: aFefSO] F4H2O-FeS*3Fe2++BHO- (2) In addition to FeS, also Fe(OH)2 or FeCO3 can precipitate. The direct corrosion mechanism is commonly attributed ti Desulfoyi- brlo species3-6,E 10,13, the best-studied SRB, and to their efficient H, utilizationla'rs. Indeed, stimulating effects of Desulfovibrio cells on the current via iron cathodes have been observed''ln. On the other hand, stimulation of iron oxidation due to consumption of chemi- cally formed H2 has been questionedl^,r".,r; for instance, H2 did not inhibit its own formation on iron in sterile water6. It is true that Desulfotibrio speciesa'r0't8 and also methanogenic archaea,t,,ro formed sulphide or methane, respectively, with metallic iron in groMh media; however, this was apparently due to secondary consumption of chemically formed H, without stimulation of corrosionu''0. To search for possibly yet undetected SRB with poter.rtial for direct corrosion (equation (2)), we established enrichment cultures with iron specimens as the oniy electron donor and marine sediment as inoculum. Iron has been used as a reductant in a former enrichment technique2r, but has not been reported to yield cultures other than Desulfotibrio species that grow with organic substrates or Hr. For comparison, we enriched parallel cultures with H2 instead of iron. Within two weeks, sulphate reduction in cultures with iron exceeded the endogenous sulphate reduction with sedi- ment alone in iron- and H2-free controls (4mM versus 2mM). Consecutive subcultures with iron were inoculated with a part of the previous iron specimens. As carbon sources, cultures contained either CO2 alone, or CO2 plus acetate (1 mM). Sulphate reduction in subcultures became faster, and black layers of ferrous sulphide became visible. Microscopy revealed onl1. a few free-living 1k. Ir., contrast, the enrichment culture with H, yielded abundant free- living cells. Two representative strains, IS4 and IS5, were isolated from iron- grown sulphate-reducing enrichments (without or lvith acetate, respectively). A third strain, HS2, was isolated from the H2_grorvn enrichment (with acetate). Strain IS4 is rod-shaped (Fig. ta-l) anci affiliates with Desulfobacterium (Fig. 2), wheieas the other two strains are comma-shaped (Suppleme4tary Fig. 1) and represent Desulfovibrio species. A11 three strains also-grew by r;rlphut. reduction with lactate or H2. With H2 and CO2, strain IS4 did not depend on an organic carbon source, whereas strains IS5 and HS2 needed acetate (for cell synthesista). With lactate or H2, strain IS4 exhibited much slorver growth (doubling time, >2 days) than strains IS5, HS2 and the authenticated species, DestLlfotiirio sale_ xigens and D. tulgaris (doubling time, < 1 day). Probing of the enrichment culture with a fluorescent 165 ribo_ somal RNA-targeted oligonucleotide specific for strain IS4 revealed high n-umbers of precipitate-associated cells with shape similar to that of.strain ISa (Fig. ld, e); they represented the majority of the detectable ce1ls. A common oligonucreotide probe for besulfovibrio species22 did not hybridize. We also enriched marine methanogenic nicroorganisms rvith metallic iron in low-sulphate medium. Iron-dependent methane production first became obvious after 20 days. From the fourth subculture, a methanogenic strain, IM1, r,r,as isolated. Attempts to retrieve a 165 rRNA gene sequence from strain IM1 yielcled only a short fragment (1,000bp) revealing an alfiliation wtth Metharto_ bacterium and Methanobreyibacter (Suppiementary Fig. 2). In the absence of n.retaliic iron, strain IM1 grew slowly with frz f CO: if the pH was above 7.5. For direct measurement of corrosiveness, we follor.ved sulphate reduction or methanogenesis with metallic iron as the only electron donor by the newl1, isolated as well as by authenticated SRB and methanogenic archaea, respectively (Fig. 3a, b). The authenticated species are known hydrogen utilizers. Sulphate redr_rction to sul_ phide by strain IS4 with iron was fast, -r.h u, in the enrichment culture. Sulphate reduction gradually slowed down, but became faster again if fresh iron was added (not shown), suggesting that formed crusts (Fig. 1a) act as process barriers. Sulphate-r-educt[n by strain iS5 was slower than by strain IS4, but lasterihan by strain HSi and the authenticated D. salexigens and D. t,ttlgaris. Sulphate reduction rates of the latter three were in agreement with a secondary consumption of chemically formed H, (l mol sulphate requiring 4n-ro1 H2). Methanogenesis r,r.ith iron was also more pronotinced rvith the new strain IM1 than with the authenticated Methanococcrts maripaludis (Fig. 3b), Methanogenium organophi- lum and Methanosarcina mazei (not shown). The rate of methano_ genesis n,ith iron by the authenticated species .r,vas again in accordance with a secondary consllmption of chemicaly ?ormed H2 (1 mol CHa requiring 4 n.rol H2). T- pL L( el. i:r a::. nla el:: pe: nrl H. ICC: hoi tik; sur erpl. forr Figure 3 - electron c _ - Desulfovit - . strain ll\41 sterile iror - productior :. or methanc:, indicateci b- . NATURI,l\ r , '.* s-,- , .s#-,#{- iffi '** J' E';ffi;""' Archaeoglobus fulgidus Desu lfovi bri o desu lf u ricans Desu lfom i crobiu m apsheronu m Strain lS5 Desulfovibrio senezii Desu lfov i b rio p rofu nd us Desu lfovi brio caledoni e nsis Strain HS2 Syntrophobacter Geobacter metallireducens Desu/fobulbus spp. ll*J Figure 1 t,i i'isi!:, :' :- as a. Scanning electrof micrograph of an tron coupon ,-:ai:l',','l-s::a^ S*=;._:e, ,,,eeks.Flamentouscel sareembedcledinprecipitatecl . '.^a i;S b. P:ase::r as :.rgraphof vrablecells0f stranls4grownforl week 1 ' c. Ce s n the sa:re cL .r.e after three weeks. d, Genera (DApl) staining of : : t:'< crec pitate rn an enrchmer: arlture \,/rth metailic lron. e, The same sectjon '- :,: -:. cn of f uorescent (tyramide-i ,torescein) staining via a specific "' :, := c,obe foTstrain iS4. Bar'l0p,nr tapplcabeto all panels) 83,0 Figure 2 Phylogenetic reratlonships (based on 1 6s rRNA gene sequences) o{ new rsolates of sulphate-reducing llacteria from enrichrnent cuitures with metallic iron (strarns ls4 anci ls5) or hydrogen (strain HS2) and surphate. Bar indicates r 0o/o estimatecr sequence diverqence. NATLUT E \,'O I_ .12 7 | 26 FEIIRLAII\' 200.1 ] l\af (,.naturc.coir/Dature Desulfobacterium catecholicum Strain lS4 Desulforhopalus spp. Desu/foialea spp. Desuifocapsa spp. 10% Desulfofustis glycolicus li &qry* L# H_d =${.'".r' r$:gr"t ;\#- -; r F*- ' - *ts-#r- - '=#-{ ' = '.ir*; * '€'=-; , :il Desulfobacter s1o Because the speed of sulphate reduction by strains IS4 and IS5 and of methanogenesis by strain IMI with iron cannot be explained by mere consumption of the chemical\, formed hl.drogen, these organisms must obtain reducing equivalents more efliciently. In cultures of strain IS4 with iron, H, accumulated to high levels (Fig. 3c) and hence cannot be the rate-iimiting intermediate; this was not observed in incubations rvith strains IS5, HS2 and the authenticated Desulfot,ibrio species. An eflrcient use of metallic iron for sulphate reduction (or methanogenesis) rvould be possible by an electron uptake via a cell-surface-associated redox-active com- ponent, a principle known from aerobic iron(ll)-oxidizing bac- teria23. The inverse process, a deliven' of electrons (from organic electron donors) by cells to external soiid or dissolved matter occurs in iron(III)-reducing bacteriara. Follorving hydrogen formation assays with cell fractions of D. vulgaris, a cttochrome in the outer membrane has been suggested to participate in iron corrosion; an electron flow to the sulphate reduction enzymes via two types of periplasmic hydrogenase and H. as a direct intermediate was proposed, according to: Fe * c\.tochrome hydrogenase(1) * Hz - hydrogenase(2) * electron transport system sulphate reduction enzymes'' (arrorvs indicate electron flow). In our study, however, the same D. tulgaris did not rer.eal a corrosive potential like strains IS4 and IS5. Nevertheless, the assumption of a cell- surface-associated redox component is an obvious hypothesis to explain how the new isolates can circumvent the slow chemical formation of H2 (ref. 12) to obtain their reducing equivalents. The as a result of an imbalance between eiectron donation by fresh iron and electron consumption by sulphate reduction (see also Sup- plementary Fig. 3). The hydrogenase may otherwise function in growth with external hydrogen. Detailed models of electron flow are at present not possible because knowledge of the topology and function of redox proteins in various SRBIa (and rnethanogens ) is insufficient. The abundance of cells resembling strain IS4 in our enrichment culture with iron and the effective iron-dependent sulphate reduction by this strain suggests a thus-far overiooked involvement of such or physiologically similar SRB in anaerobic corrosion. proof of this assumption requires the examination of technical plants with corroding iron or steel. The natural significance ofthe cipacity for using metallic iron as electron donor is unknown. Apart from rare meteorites, metallic iron is a technical product and is thus a very 'recent' growth substrate. One may speculate that iron-using unu robes can also obtain electrons in cell contact with certain letters to nature pronounced H2 formation during growth of strain IS4 with metallic iron does not necessarily indicate that H2 is a direct biochemical intermediate connecting two hydrogenases,r. H2 may just as well be formed via a branch according to: Fe electron transport system+ suiphate reduction enzymes TI hydrogenase;l H2 mtcroorganlsms. Methods (3) tr L l t !' l e a e 'e d Strain lS4 Sterile + HrS '10 Time (d) Figure 3 lncubation experiments with iron granules (30 g in t50 ml medium) as sole electron donor. a, Sulphide formation (va sulphate consumpti0n) by strains lS4, lS5 and Desulfovibrio salexigens, and the orig nal eni'ichment cuiture. b Methane formatton by strain lMl and Methanococcus maripaludis. c, Hydrogen formation by strain lS4 and in sterileirontncubationswithoutorwithhydrogensulphide(4mM). lnthe atter,hydrogen production became s ower after binding offree su phide as FeS. a, b. Sulphate reducti0n 0r methanogenesis expected from mere consumption 0f chemically formecl hydrogen is indicated by dotted lines. NAIUIIE ]VOL 427 | 26 FEBIiUARY 2004 | umrnature.com/nature Enrichment, isolation and cultivation Marine sediment was collected near Wilhelmshaven, North Sea. Cultures were grown at 28 oC in anoxic searvater (marine organisms) or freshrvater (D. wlgaris) meclium with 28mM sulphate (0.1mM for methanogens) under N2 + CO, (90/10, vol.A,ol.),;. If inciicated, I mM sodium acetate was added as organic carbon source. Iron granules (99.g% Fe, size 2 mm, 20 g per 100ml) or mild steel coupons (l mm thickness, jitted to tubes or bottles) u'ere added as the electron soruce. For subculturing, I 0% ofthe culture liquid and part ofthe iron specimens rrere transferred e,eryse'en weeks. parallel enrichme,t cultures were carried out with H. + CO: (+1 nM acetate) without metallic iron. For isolation of enriched organisms, precipitates from the iron surface rverc homogenized anaerobicallv and sequentiarly diluted in medium rvith iron granuJes. The highest dil,tions shorving sulphirte recluction or methanoge,esis were again cliluted. sRR were finall). diluted in anoxic agar,. u,ith a mixture oflactate, propionate, but).rate, plruvate, ethanol (each 2 mM) and hydrogen. Colonies rqere transferred to liquid medium. ALltlrenticated strains were front the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). Chemical analyses Sulphate rras determined micro-gravimetrically as BaSO, precipitated from l-ml samples rvith the same volume of 0.2 M l3aC12 and 0.2 M HCi. CH. and H2 were quantified on agas chromatograph u,ith a flane ionization or thermal conductivity detector, respectively. N4inerals on the iron surface were analysed by X-ray photoeleciro, ,p*,-,.irr. ' Analyses or 165 rRNA genes I)NA was extracred rvith the DNeasl. Tissue hit (eiagen). Nearly full_length (i,500 bp) 16S IRNA gene sequences from SRB were amplified using general bacterial primers16. A partial (about 1,000bp) 165 IRNA gene sequence ofthe methanogenic isolate \sas amplified rvith archaeal primers:7. Sequences were analysed using the ARB programrs, and a piylogenetic tree lvas constructed via maximum parsimony, neighbour .joining and maximum likelihooci anal)'ses. The methanogenic strain IMr was positioned in the tree according to parsimony criteria r.ithout affecting the overall topology. Specific cell hybridization and unspecific staining Precipitates scratched from corroded iron surfaces were fixed for 12 h at 4.C in 4% formaldehyde, washecl tr.ice with PBS (10 mN4 sorliurr phosphate, pH 7; 130 nNI NaCl l stored in PBS-ethanol (1:1) at 20'C, and collected on polycarbonate filters (0.2 um pores; Millipore). A horseradish.peroxidase labelled probe (5,- CTCCTCCTGCTGCAGTAGCT 3' ) u,as specifically designed ancl srnthesized (Thermohybaid)tbrstrainlS4.Afterhybridizatiol.rtj5.Cinthepresenceof550,, r r. vol.) formamide and rvashing, the probe lvas reacted rrith tyramide fluorescein: . Df- I4 .h-diamidiro-2 phenylirrdolc \\d\ u\.(.1 tur gerrerrl .cll rtrining. Remarks on redox potentials and equations Indicatedredoxpotentialsareversusstandardhydrogenelectrode(E0:0\.1:: Fer+/Fe redox potcntial in cultures and sea lvater is much more negatile l1:: ::;:: standarci conditiotls ( 0.44\r); rvhereas Fe remains in standarcl state. salir ::: i'6 E E E o p ! _o f a2 ?4 E E E ;2 c 6 t o >0 )- 20 o E g c 3, to o ! I )- in :d 20 15 rates 1 and l 83t 6. 7. 8. letters to nature precipitation as sulphide and carbonate decrease the Fe2+ activitlf such that redox potentials around -0.53\rare realistic. Proton reduction to Hz (Eluz: -0.41 V) on metallic iron in such surroundings rould thus be far fron the thermodynamic equilibrium. Equation (1) is based on the equations for sulphate reduction with the organic compound (2JCHrOl + SOi- * 2HCO; + H2S) and hydrogen (4H, + SOi * 2H+ - H2S + 4HrO). If SRB reduce 1 mol SOi with an organic compound to 1 mol H25, the latter yields I mol H2 upon chenical reaction with Fe to FeS. Use of H2 for further sulphate reduction yields r/o mol HzS which ieads to r/4 mol H2. Continuation ad infinitum leads to a total of lr/3 (sum ofinfinite row 1 + r/4 + 1/r5 etc.) mol H2S that attacks the iron. (For other remarks on reactions in corrosion see Supplementary Information.) Received 25 April; accepted 23 December 2003; doi:10.103S/natureo232I. 1. 2. Uh1ig, H. H. Corrosion and Corrosion Conrol 3rd, edn (Wiley, Net,York, 1985). Hamilton, \\r. A. Microbially influenced corrosion as a model systeD lor the studv ofmetal microbe interactions: a unifuing electron transfer hlpothesis. BiolbLling 19, 65-76 (.2003). Lee, \\1, Les,andowski,2., Nielsen, P H. & Hamllton, Ul A. Role ofsulfate-reduci[g bacteria in corrosion of mild steel: a rcvie* Bioloulirry 8, 165-194 (1995). Pankhania, I. P Hydrogen nretabolism in sulphate reducing bacteria and its role in anaerobic corrosion. -Biofbaling l, 27-47 (1988). Widdel, F. in Bioierlrnlagy Fo.us 3 (eds Finn, R. K. el al.) 277 318 (Hanse! Munich, 1992). Cord-Ruwisch, It. ii ,firironfrefital Micrabe Metal Interaction (ed. Lor.lel', D. R.) 159-173 (ASM Press, Washington, DC, 2000). Costello,l.A.Cathodicdepolarizationbysulphate reducingbacterra.S.Afr.l.Sci.70,202 204(1974). von \\blzogen Kuehr, C. A. H. & ran der VILrgt, I. S. The grxphitization of cast iron as rr electrobiochemical process jn anaerobic soil. Wdter 18, I 47-l 65 ( 1 934). Gonuentional taxonomy obscurcs deep diueryenoe between Pacific and Atlantic Gorals Hironobu Fukamir'2, Ann F. Budd3, Gustav Paulaya, Antonio So16-Cara' Ghaolun Allen Ghen6, Kenji lwaoT & llancy l(nowltonI,2 1 Smithsonian Tropical Research Institute, Naos Marine Laboratory, Box Balboa, Republic of Panama 2Center for Marine Biodiversity and Conseruation, Scripps Institution oi Oceanography, University of California San Diego, La lolla, Calfornia 92093-0202, USA 3Department of Geoscience, Llniversity of lowa, Iowa City, lowa 52242, L-5.:- aFlorida Museum of Natural History, Llniversity of Florida, Gainesville, FL :.: 32611 7800, USA 5Department of Genetics, Federal Llniversity of Rio de Joneiro, Rio de Janeirt. . Brazil 6lnstitute of Zoology, Academia Sinica, Nankang, Taipei 115, Taiwan 'Akalima Marine Science Laboratory, Zamami son, Okinawa 901-3311, lap,;, Only l7o/o of lll reef-building coral genera and none of the 18 coral families with reef-builders are considered endemic to the Atlantic, whereas the corresponding percentages for the Indo- west Pacific are 760/o and 39o/or'2. These figures depend on the assumption that genera and families spanning the two provinces belong to the same lineages (that is, they are monophyletic). Here we show that this assumption is incorrect on the basis of analyses of mitochondrial and nuclear genes. Pervasive morphological convergence at the family level has obscured the evolutionan- distinctiveness of Atlantic corals. Some Atlantic genera conven- tionally assigned to different families are more closely related to each other than they are to their respective Pacific 'congeners'. Nine of the 27 genera of reef-building Atlantic corals belong to this previously unrecognized lineage, which probably diverged oyer 34 million years ago. Although Pacific reefs have larger numbers of more narrowly distributed species, and therefore rank higher in biodiversity hotspot analyses3, the deep evolution- ary distinctiveness of many Atlantic corals should also be con- sidered when setting conseryation priorities. We tested the assumption of inter-oceanic monophyly by exam- ining the relationships of Atlantic and Pacific members of the Faviidae and Mussidae, ecoiogically dominant families that com- prise one-third of all reef-building or zooxanthella-containing coral genera''t. We analysed mitochondrial cltochrome B (cytB) and cytochrome oxidase 1 (COI) genes as well as parts of two exons from a nuclear B-tubulin gene. We sequenced representatives of more than half the genera in both families, including most of the Caribbean species, as well as taxa in related families. Unexpectedly, most Atlantic lineages conventionally assigned to the Faviidae and Mussidae are not distributed within the more numerous Pacific lineages of these 'families', but instead represent a well-defined clade (Fig. 1). Notably, the Atlantic 'faviid' Favia and the Atlantic'mussid' Scolymia are more closely related to each other than they are to their respective 'congeners' in the Pacific (taxa in bold in Fig. 1). The only Atlantic corals closely related to the group containing Pacifi c'faviids' belong to the clearly polyphyletic'genus' Montastraea; seven other genera in this group that are now restricted to the Pacific once had Atlantic representatives (on the basis of conventional taxonomy)4. Moreover, the distinctiveness of this previously unrecognized Atlantic clade is greater than that of several conventionally recog- nized families that are restricted to the Indo-west Pacilic. Specifi- cally, the Merulinidae and Pectiniidae (which in our analyses are pollphyletic), as well as the monotypic Trachlphylliidae, are nested within and are more closely related to Pacific'faviids' and'mussids' 9. Ilooth, G. H. & filler A. K. Cathodic characteristic ofnild steel in suspension ofsulphate reducing bacteria. Corro-\. Sci. 8,583 600 (1968). 10. Pankhania, t. P, Moosari, A. N. & Hamilton, W. A. Utilization ofcathodic hydrogen by Desulfwibrio ralgaris (Hildenborough). I Gcr. Microbiol. 132, 3357-3365 (1986). 1 1. lvcrson, \\l P & O1son, G. l. in Currenr Perspectives in Microbial Erologl (cds K1ug, tr4. I. & Reddy, C. A. ) 623 627 (ASIU, \\'ashington, DC, 1984). 12. Bochis, l. O'M. & Reddy, A. K. N. M0derfl DLectrochemistry YoL2 (Plenum, Ne$, York, I 970). I 3. Beech, I. B. et dl. Stud,v ofparameters implicated in lhe biodeterioration ofmild steel in the presence of diflerent species ofsulphate-reducing bacterta. Int. Biodeter. Biodegrad.34,289 303 (1994). 14. Rabus, R., I{ansen, T. & Widdel, F. rnThe Prokaryotes: An Etobing Electronic Resource Jbr the Microbiological Commrrirr,(eds Dworkin, M., Falkou', S., Rosenberg, E., Schleifer, K H. & Stackebrandt, E.) (Springer, New York, 2000). I 5. Widdel, L & Bak, F. )r The Prokatyotes 2nd edn Vol. 6 (eds Balows, A., Tlriper, H. G., Dworkin, N,i., Harder, W. & Schleilir, K H.) 3352-3378 (Springer, New York, 1992). i6.Hard,Il.A.Utilisrrlonofcathodich,vdrogenbysulphatereclucingbacteda.BrCorroi./.18,190 193 (1e83). 17. Laishlcl E. I. & Bryant, R. D- in Biochemistry and Physiology of Anaerohic Bdcteria (eds Ljungdahl, L. G., Adams, M. \41, Barton, L. L., Feny, l. G. & Iohnson, M. K.) 252-260 (Springel New York, 2003). 1 8. (lord Ruwisch, R. & Widdel, F. Corroding iron as a hydrogen source for sulphate reduction in growing cultures ofsulphate-reducing bacteria. Appl. Microbiol. Biotechnol.25,169 174 (1986)- 19. Daniels, L., Belay, N., Rajagopal, 13. S. &\Vermer, P I. llacterial methanogenesis and gromh lrom CO2 withelementalironasthesolesourceofelectrons.S.ier.e23T,509 511(1987). 20. Dcckena, S. & Blotevogel, I(. H. Feo oxidation in the presen.e of nethanogenic and sulphate- reducingbacterlaanditspossibleroleinanaerobiccorrosior.Biofauling5,2ST 29311992)- 21. Schlegel, H. G. General MicrobiologyTth edn (Cambridge Univ. Press, Cambridge, 1993). 22. lvlanz, \\r., EisenbrecheL N{., Neu, T. R. & Szervzyk, U. Abundan.e and spaiiai organization ofCram- legative s ul fate-redu cing b acteria in activated sludge investigated by in situ probj n g wi th sp eci fi c I 65 rRNA targctcd oligonucleotides. -FEMS Mt.rohiol. Ecol. 25, 43 61 (1998). 23. r\ppia-Ayne, C., Guiliani, N., Ratouchniak, f. & Bonnefby, \1 Characterization ofan operon encoding trvo type cytochromes, an da r-type cfochrome oxidase, and rusticyanin ii ThiobocillLts lerrooxi dans .\ffCC, 3302.0. Appl E1*ian. Microbiol. 65, 17 81 47 87 (.1999). 24. Bond, D. R. & Lovley, D. R- Elecrricitv production bv GeoDartrr sulfurreducens attached io electrodes. Appl. Enriron. Microbiol. 69, 1548-I555 (2003). 25. Deppenmeier, U. The unique biochemistry ofmethanogcnesrs. Prog Nucleic Acid Res. Mo|. Biol.7l, 223 2Bl (2002). 26. Mu),zer, G., Teske, A., Wirsen, C. O. & Iannasch, H. \,\l Ph).logenetic relationships ofThiomicrosplra species and their identilication in deep-sea hydrothermal vent samples by denaturing gradient gel electrophoresis of 165 IDNA fragments. Ar.h. Microbiol. 164,165 172 (1995). 27. Huber, H. cr al. A new phylum ofArcllaea represented by a nanosized hlperthermophilic symbiont. Naturc 417,63 67 l20Al. 28. Ludwig, W. sr dl. ARBi a softNare enl,iron,nent for sequence data. (Department ofMicrobiologl, Technical Univ. N4unich, 2002); ar.ailable at (http://m.arb home.de/). 29. Pernthaler A., Pcrnthale! I. & Amann, R. Fluorescence in sitil hybridization and catalyzed reporter deposition for the identification of marine b acteia. Appl. Enrirofl. Microbiol. 68, 3094-3 t 0 I (2002 ). Supplementary lnlormation accompanies the paper on www.nature.com/nature. Acknowledgemenb \\re thank K. Nauhaus, M. Nellesen and G. Herz for technical help. This study ivas supported by the Gernan Acadenic Exchange Service, the Fonds cier Chemischen Industrie, and the Max Planck Society. Competing interests statement The authors declare that they have no competing financial interests. Correspondence and requests for matedals shoulcl be addressed to F.Ut (fivicldel@mpi-brenren.de). The nucleotide sequences have been deposited at EMBL GenBank unclcr accession numbers AY274.1,14 (strain HS2) Ay274449 (strain IS5), AY274450 (strain IS,1) and AY274451 (strain Itr41). Figun l\1ea sec-: one ::: a32 NAIURl VOL 42 lo I fBRl \R\ '004 |,,rrs.narrrre.iom nJtLrr( . I VOL.127 26 FEBRUARY 2004 l nfl$ iatrue.com/nature letters to nature lron Gorrosion by nouel anaerobic microorganisms Hang T. Dinhl, Jan Kuevert,z, Marc MuBmann,, Achim W. Hassel3, Martin. (Fig. 1a) act as process barriers. Sulphate-r-educt[n by strain iS5 was slower than by strain IS4, but lasterihan by strain HSi and the authenticated D. salexigens and. s1o Because the speed of sulphate reduction by strains IS4 and IS5 and of methanogenesis by strain IMI with iron cannot be explained by mere consumption of the chemical, formed