Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen...

8
Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloa a,1 , Donald E. Caneld b , Edward F. DeLong c , Ricardo M. Letelier d , and Frank J. Stewart e a Departamento de Oceanografía, Universidad de Concepción, Concepción 4070386, Chile; b Institute of Biology and Nordic Center for Earth Evolution, University of Southern Denmark, 5230 Odense M, Denmark; c Departments of Biological Engineering and Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA 02138; d College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, Corvallis, OR 97331; and e School of Biology, Georgia Institute of Technology, Atlanta, GA 30332-0230 Edited by David M. Karl, University of Hawaii, Honolulu, HI, and approved August 16, 2012 (received for review April 26, 2012) Vast expanses of oxygen-decient and nitrite-rich water dene the major oxygen minimum zones (OMZs) of the global ocean. They support diverse microbial communities that inuence the nitrogen economy of the oceans, contributing to major losses of xed nitrogen as dinitrogen (N 2 ) and nitrous oxide (N 2 O) gases. Anaerobic microbial processes, including the two pathways of N 2 production, denitrication and anaerobic ammonium oxidation, are oxygen-sensitive, with some occurring only under strictly anoxic conditions. The detection limit of the usual method (Winkler titrations) for measuring dissolved oxygen in seawater, however, is much too high to distinguish low oxygen conditions from true anoxia. However, new analytical technologies are revealing vanishingly low oxygen concentrations in nitrite- rich OMZs, indicating that these OMZs are essentially anoxic marine zones (AMZs). Autonomous monitoring platforms also reveal previously unrecognized episodic intrusions of oxygen into the AMZ core, which could periodically support aerobic metabolisms in a typically anoxic environment. Although nitrogen cycling is considered to dominate the microbial ecology and biogeochemistry of AMZs, recent environmental genomics and geochemical studies show the presence of other relevant processes, particularly those associated with the sulfur and carbon cycles. AMZs correspond to an intermediate state between two end pointsrepresented by fully oxic systems and fully suldic systems. Modern and ancient AMZs and suldic basins are chemically and functionally related. Global change is affecting the magnitude of biogeochemical uxes and ocean chemical inventories, leading to shifts in AMZ chemistry and biology that are likely to continue well into the future. M idwater depths in the ocean typically contain reduced oxygen concentrations. This is due to the decomposition of surface-derived sinking organic material by aerobic respiration, combined with the introduction of high-latitude, oxygen-rich surface water to deeper zones (1). In some areas of the open ocean, high rates of phytoplankton productivity, coupled with poor ventilation and sluggish circulation, lead to an extensive, oxygen-decient in- termediate layer a few hundred meters in depth, where nitrite accumulates (2) and oxygen concentrations drop to low levels undetectable by the most sensitive modern techniques (3). Also, in this layer, nitrous oxide (N 2 O) does not accumulate but covaries inversely with nitrite concentra- tion (46). Henceforth, we will designate these essentially anoxic marine zones as AMZs to differentiate them from oxygen- containing oxygen minimum zones (OMZs) as commonly found in the global ocean. The nitrite-rich AMZs are prominent in the eastern tropical North Pacic (ETNP), the eastern tropical South Pacic (ETSP), and in the Arabian Sea (Fig. 1). Due to the vanishingly low oxygen con- centrations in these AMZ waters, higher eukaryotic taxa, such as sh, are normally excluded or persist due to special adapta- tions (7) and anaerobic microbial pro- cesses (i.e., those that do not depend on free oxygen) dominate (8). Indeed, signif- icant amounts of nitrate are converted to dinitrogen (N 2 ) gas in AMZs, generating large xed nitrogen(nitrate plus nitrite and ammonium) decits, thereby contrib- uting to an estimated 3050% of all the nitrogen lost to N 2 in the oceans (9). The extent of these AMZs, however, is un- certain due to the limitations of standard techniques for measuring very low oxygen concentrations. These limitations hinder our ability to delineate the spatial distri- bution and intensity of aerobic and an- aerobic processes in OMZ settings. Climate-induced warming of the upper ocean appears to have already contributed to deoxygenation of the global ocean (10, 11) and to an expansion of OMZ waters (12). This trend is expected to increase into the future (13), expanding the AMZ area and inuencing rates of xed nitrogen loss from the oceans. Here, we highlight how recent advances in oxygen sensing and observational tech- nologies (applied initially to the study of the AMZ of the ETSP) are redening our understanding of oxygen and its dynamics in OMZs. These results, combined with novel molecular and experimental ap- proaches, reveal unexpected complexity to elemental cycling in AMZs and show how modern AMZs represent an interme- diate state of a geochemical and micro- biological continuum to suldic basins like the modern Black Sea and ancient equivalents. How Much Oxygen Is in OMZs? It is well known that oxygen concentrations can reach very low values in some OMZ waters. Using the Winkler titration method to determine oxygen concentrations, Johs Schmidt (14) observed in 1921 that the water off the western coast of Panama contains practically no oxygen at allat depths between 400 and 500 m. In- troduced in 1888 by the Hungarian chemist Lajos W. Winkler (15), Winkler titrations remain a standard method today, and if carefully applied, they can reach a detection limit of about 1 μM. These concentrations are quite low compared with air-saturation values of about 250 μM in equatorial waters and with values in most of the open ocean, including abyssal depths. However, is 1 μM sufciently low in concentration to delineate where an- aerobic metabolism dominates and where aerobic metabolism becomes less rele- vant? Based on recent results enabled by newly developed advances in sensor tech- nology (see below), we argue that it likely is not and that the role of oxygen in structuring OMZ microbial metabolism is actually much more complex. The switchable trace oxygen (STOX) amperometric microsensor (16) has been developed recently for measuring ultralow oxygen concentrations. Its construction is based on the presence of a front guard secondary cathode that, when polarized, prevents oxygen from reaching the pri- mary internal sensing cathode. With this design, a zero calibration is obtained in each measuring cycle, bringing the Author contributions: O.U., D.E.C., E.F.D., R.M.L., and F.J.S. wrote the paper. The authors declare no conict of interest. This article is a PNAS Direct Submission. 1 To whom correspondence should be addressed. E-mail: [email protected]. 1599616003 | PNAS | October 2, 2012 | vol. 109 | no. 40 www.pnas.org/cgi/doi/10.1073/pnas.1205009109 Downloaded by guest on September 14, 2020

Transcript of Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen...

Page 1: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

Microbial oceanography of anoxic oxygenminimum zonesOsvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M. Letelierd, and Frank J. StewarteaDepartamento de Oceanografía, Universidad de Concepción, Concepción 4070386, Chile; bInstitute of Biology and Nordic Center for EarthEvolution, University of Southern Denmark, 5230 Odense M, Denmark; cDepartments of Biological Engineering and Civil and EnvironmentalEngineering, Massachusetts Institute of Technology, Cambridge, MA 02138; dCollege of Earth, Ocean, and Atmospheric Sciences, OregonState University, Corvallis, OR 97331; and eSchool of Biology, Georgia Institute of Technology, Atlanta, GA 30332-0230

Edited by David M. Karl, University of Hawaii, Honolulu, HI, and approved August 16, 2012 (received for review April 26, 2012)

Vast expanses of oxygen-deficient and nitrite-richwater define themajor oxygenminimum zones (OMZs) of the global ocean. They supportdiverse microbial communities that influence the nitrogen economy of the oceans, contributing to major losses of fixed nitrogen asdinitrogen (N2) and nitrous oxide (N2O) gases. Anaerobic microbial processes, including the two pathways of N2 production, denitrificationand anaerobic ammonium oxidation, are oxygen-sensitive, with some occurring only under strictly anoxic conditions. The detection limitof the usual method (Winkler titrations) for measuring dissolved oxygen in seawater, however, is much too high to distinguish lowoxygen conditions from true anoxia. However, new analytical technologies are revealing vanishingly low oxygen concentrations in nitrite-rich OMZs, indicating that these OMZs are essentially anoxic marine zones (AMZs). Autonomous monitoring platforms also revealpreviously unrecognized episodic intrusions of oxygen into the AMZ core, which could periodically support aerobic metabolisms ina typically anoxic environment. Although nitrogen cycling is considered to dominate the microbial ecology and biogeochemistry of AMZs,recent environmental genomics and geochemical studies show the presence of other relevant processes, particularly those associated withthe sulfur and carbon cycles. AMZs correspond to an intermediate state between two “end points” represented by fully oxic systemsand fully sulfidic systems. Modern and ancient AMZs and sulfidic basins are chemically and functionally related. Global change is affectingthe magnitude of biogeochemical fluxes and ocean chemical inventories, leading to shifts in AMZ chemistry and biology that are likely tocontinue well into the future.

Midwater depths in the oceantypically contain reducedoxygen concentrations. Thisis due to the decomposition

of surface-derived sinking organic materialby aerobic respiration, combined with theintroduction of high-latitude, oxygen-richsurface water to deeper zones (1). In someareas of the open ocean, high rates ofphytoplankton productivity, coupled withpoor ventilation and sluggish circulation,lead to an extensive, oxygen-deficient in-termediate layer a few hundred meters indepth, where nitrite accumulates (2) andoxygen concentrations drop to low levelsundetectable by the most sensitive moderntechniques (3). Also, in this layer, nitrousoxide (N2O) does not accumulate butcovaries inversely with nitrite concentra-tion (4–6). Henceforth, we will designatethese essentially anoxic marine zones asAMZs to differentiate them from oxygen-containing oxygen minimum zones (OMZs)as commonly found in the global ocean.The nitrite-rich AMZs are prominent in theeastern tropical North Pacific (ETNP), theeastern tropical South Pacific (ETSP), andin the Arabian Sea (Fig. 1).Due to the vanishingly low oxygen con-

centrations in these AMZ waters, highereukaryotic taxa, such as fish, are normallyexcluded or persist due to special adapta-tions (7) and anaerobic microbial pro-cesses (i.e., those that do not depend onfree oxygen) dominate (8). Indeed, signif-icant amounts of nitrate are converted todinitrogen (N2) gas in AMZs, generatinglarge “fixed nitrogen” (nitrate plus nitriteand ammonium) deficits, thereby contrib-

uting to an estimated 30–50% of all thenitrogen lost to N2 in the oceans (9). Theextent of these AMZs, however, is un-certain due to the limitations of standardtechniques for measuring very low oxygenconcentrations. These limitations hinderour ability to delineate the spatial distri-bution and intensity of aerobic and an-aerobic processes in OMZ settings.Climate-induced warming of the upperocean appears to have already contributedto deoxygenation of the global ocean (10,11) and to an expansion of OMZ waters(12). This trend is expected to increaseinto the future (13), expanding the AMZarea and influencing rates of fixed nitrogenloss from the oceans.Here, we highlight how recent advances

in oxygen sensing and observational tech-nologies (applied initially to the study ofthe AMZ of the ETSP) are redefining ourunderstanding of oxygen and its dynamicsin OMZs. These results, combined withnovel molecular and experimental ap-proaches, reveal unexpected complexity toelemental cycling in AMZs and show howmodern AMZs represent an interme-diate state of a geochemical and micro-biological continuum to sulfidic basinslike the modern Black Sea and ancientequivalents.

How Much Oxygen Is in OMZs?It is well known that oxygen concentrationscan reach very low values in some OMZwaters. Using theWinkler titration methodto determine oxygen concentrations, JohsSchmidt (14) observed in 1921 that thewater off the western coast of Panama

“contains practically no oxygen at all”at depths between 400 and 500 m. In-troduced in 1888 by the Hungarianchemist Lajos W. Winkler (15), Winklertitrations remain a standard method today,and if carefully applied, they can reacha detection limit of about 1 μM. Theseconcentrations are quite low comparedwith air-saturation values of about 250 μMin equatorial waters and with values inmost of the open ocean, including abyssaldepths. However, is 1 μM sufficiently lowin concentration to delineate where an-aerobic metabolism dominates and whereaerobic metabolism becomes less rele-vant? Based on recent results enabled bynewly developed advances in sensor tech-nology (see below), we argue that it likelyis not and that the role of oxygen instructuring OMZ microbial metabolism isactually much more complex.The switchable trace oxygen (STOX)

amperometric microsensor (16) has beendeveloped recently for measuring ultralowoxygen concentrations. Its construction isbased on the presence of a front guardsecondary cathode that, when polarized,prevents oxygen from reaching the pri-mary internal sensing cathode. Withthis design, a zero calibration is obtainedin each measuring cycle, bringing the

Author contributions: O.U., D.E.C., E.F.D., R.M.L., and F.J.S.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail:[email protected].

15996–16003 | PNAS | October 2, 2012 | vol. 109 | no. 40 www.pnas.org/cgi/doi/10.1073/pnas.1205009109

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 2: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

detection limit of the sensor to the nano-molar range (∼1–10 nM), depending onthe instrument’s configuration and elec-tronics (16).STOX sensors provide strong clues

about the oxygen levels supporting anaer-obic metabolism in AMZs. The first STOXmeasurements in the OMZ off the Peru-vian coast demonstrated oxygen concen-trations of less than the 3-nM detectionlimit of the sensor (16). These values areabout three orders of magnitude lowerthan those revealed by the Winklermethod and underscore Schmidt’s senti-ment of “practically no oxygen at all” inthe core of AMZs. Oxygen concentrationsbelow the detection limit of the STOXsensor were also measured off the coast ofIquique, Chile (17). Moreover, in a surveyof nearly 1,600 km of the ETSP, numerousSTOX measurements through OMZ wa-ters reveal that the observations made offPeru and Iquique generally apply and thatnitrite accumulates (>0.5 μM) only whenoxygen falls below 50 nM (3). This obser-vation, when coupled with known ocean-ographic nitrite distributions, suggests thatessentially oxygen-free waters occupy avolume of at least 2.4 × 1014 km3 in theETSP (3). Essentially oxygen-free waters

have also been found using STOX sensorsin the nitrite-rich regions of the ArabianSea (18). It is clear, therefore, that vastvolumes of nitrite-rich essentially anoxicwaters (AMZs) occur in the ETSP and theArabian Sea. STOX sensor measurementshave not yet been reported for the ETNP,but the combined distribution of oxygen,nitrite, and N2O (4) suggests that anoxicwaters are also present in this system. Incontrast, other oceanic OMZs, such asthose in the eastern tropical South Atlan-tic or the eastern subarctic North Pacific,do not show any signs of anoxia; here,anoxic waters, when present, are restrictedto coastal regions or fjords.Anaerobic microbes are known to be

oxygen-sensitive, but they differ in theiroxygen sensitivities depending on thespecific metabolic pathways involved. No-tably, sulfate reduction and methano-genesis are believed to require strictlyanoxic conditions in most environments(19). In contrast, based on experimentswith natural populations from the BlackSea and the Namibian and PeruvianOMZs, anaerobic ammonium oxidation(anammox) may occur at oxygen concen-trations up to about 10–20 μM (20, 21).Experiments with cultures of OMZ bac-

teria have shown a switch between oxygenand nitrate respiration at about 2–4 μMoxygen (22), whereas those with naturalpopulations from the Namibian and Pe-ruvian OMZs show that dissimilative ni-trate reduction to nitrite is still active at25 μM oxygen (21).It seems apparent, however, that al-

though some microbes may tolerate lowmicromolar levels of oxygen during theiranaerobic metabolism, they may not es-tablish naturally abundant populations atthese oxygen levels. The Bay of Bengaloffers an interesting test case in whichoxygen concentrations drop to about 3 μM(or possibly lower, given uncertaintiesof the Winkler method used to establishthese levels); however, there is no evi-dence for anaerobic metabolism or N2production through denitrification oranammox (23, 24). Therefore, despite theability of anaerobic bacteria to metabolizein up to 20–25 μM oxygen (21), they mayrequire significantly lower oxygen levelsto grow and to persist at high abundancein the environment.The STOX sensor (16) also has been

used to explore the low oxygen metabolismof Escherichia coli when grown as a strictaerobe (25). E. coli possesses a low-oxygenaffinity dioxygen reductase of the cyto-chrome bd superfamily, and it can grow atoxygen concentrations below the 3-nM ox-ygen detection limit of the sensor, witha half-saturation constant with respect togrowth value of 121 nM (25). If other low-oxygen–adapted aerobes behave similar toE. coli, the 1-μM detection limit of theWinkler method is clearly insufficient tomark the limits of aerobic metabolism.

How Does Aerobic Metabolism Varywith Time and Space in AMZs?Although recent STOX sensor measure-ments have demonstrated essentially an-oxic conditions in the core of the ETSP andArabian Sea AMZs, even the current res-olution of 3 nM oxygen cannot rule outthe possibility of aerobic metabolism insome of these waters. Indeed, aerobicmetabolism likely persists in the dimly litzones of the upper OMZ off the coast ofChile, where cyanobacteria, particularly ofthe genus Prochlorococcus, form a deepchlorophyll-a maximum (26) (Fig. 1B). Thissame feature is also observed in the AMZsof the Arabian Sea and the ETNP (27).Oxygen produced by oxygenic photosynthe-sis by cyanobacteria would provide oxygento feed aerobic processes at these depths.Deeper in the AMZ, however, and

outside of the photic zone, any aerobicmetabolism must be fueled by oxygensupplied from turbulent diffusion or by theepisodic injection and mixing of oxygen-rich water into the AMZ. Turbulent dif-fusion seems an unlikely oxygen source,because the oxygen gradients and mixing

A

B

Fig. 1. Location and characteristic biogeochemical profiles of AMZs. (A) Global map of major AMZsderived from the distribution of waters with ≤2 μM oxygen and ≥0.5 μM nitrite based on data from theUS National Oceanographic Database (NODC) and from Thamdrup et al. (3) and Canfield et al. (17)(observations from inland seas, such as the Baltic Sea, have not been included). Note that oxygenmeasurements reported in the NODC database are based on traditional standard techniques, with de-tection limits in the low micromolar range, and not the new highly sensitive STOX sensors. (B) Cartoon ofcharacteristic profiles in AMZs illustrate the accumulation of nitrite within the AMZ, due to the anaer-obic microbial process of nitrate reduction, and the high N2O concentrations at the boundaries (oxy-clines). The figure also shows the presence of high microbial cell abundance and of a secondarychlorophyll-a maximum due to picocyanobacteria within the AMZ waters.

Ulloa et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 15997

PERSPECTIV

ED

ownl

oade

d by

gue

st o

n S

epte

mbe

r 14

, 202

0

Page 3: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

rates necessary to drive it have not beenreported. In contrast, autonomous con-tinuously profiling floats deployed off thecoasts of Chile and Peru (28) have revealedepisodic intrusions of oxygen into theAMZ core (Fig. 2). The available data donot reveal the frequency or spatial distri-bution of such intrusions, but such oxygendynamics may play an important role inestablishing and maintaining aerobic mi-crobial communities in essentially anoxicwaters and in regulating biogeochemicalcycling in AMZs. More work with autono-mous sensing platforms, such as floats andgliders, coupled with highly sensitive oxygensensors, would help characterize these dy-namics in detail. More importantly, theimpacts of oxygen intrusions on AMZ bio-geochemical cycling remain to be assessed.The emerging picture suggests that

AMZs have extremely low oxygen levelsof <3 nM. In their upper sunlit portions,aerobic respiration may be maintained bylocal oxygenic photosynthesis or turbulentdiffusion, but below this layer, the waters

are essentially anoxic, save for periodicinjections of oxygen during mixing events.This picture is supported by recent mi-crobial community gene content (metage-nomic) and gene expression (metatran-scriptomic) surveys of the permanentAMZ waters off the Chilean coast (29).Regular patterns emerge when the genesor transcripts are sorted by key taxa ofspecific metabolic pathways (Fig. 3). Inparticular, sequences matching “Candida-tus Nitrosopumilus maritimus,” a memberof the archaeal phylum Thaumarchaeotaknown to oxidize ammonium to nitriteaerobically during the autotrophic nitrifi-cation process (Fig. 4), dominate the geneexpression libraries in the upper oxic wa-ters (50 m), in the lower oxic zone (85 m),and in the transition zone (110 m). Tran-scripts of these aerobic organisms, how-ever, are essentially absent in the core of theopen-ocean AMZ at 200 m, where se-quences matching the anammox bacterium“Candidatus Scalindua profunda” (30) areprominent (Fig. 3). These patterns are

consistent with the general vertical distribu-tion of archaeal groups in the ETSP AMZ(31) and with the segregation of aerobicammonia-oxidizing archaea and anaerobicanammox bacteria in the Arabian Sea (32).These observations, however, seem

to contradict other studies reporting ni-trification activity in the core of AMZwaters off the coast of Peru (33, 34). Inthese prior studies, nitrification was mea-sured in 2–3 μM oxygen, which was be-lieved to be the ambient oxygen level.These oxygen levels are clearly higher thanambient levels as revealed by recentSTOX sensor measurements, and elevatedoxygen may have stimulated nitrificationactivity not otherwise found in the AMZcore. Nevertheless, these earlier studies,combined with more recent studies alsoshowing nitrification activity (21, 35)as well as the expression of ammoniaoxidation genes (36) at very low oxygenconcentrations, demonstrate the signifi-cant potential for aerobic metabolism inthe heart of AMZs. As described above,

Fig. 2. Profiling float observations of oxygen, particle backscattering at 700 nm (an index of particle abundance), and salinity in the upper 800 m of the AMZof the ETSP. Oxygen-deficient conditions associated with a high particle load persist over several months within the AMZ but are occasionally interrupted byinstructions of waters with higher oxygen concentrations and lower salinities. (Inset) Trajectory of the float, which profiled each 3 d during the 9 mo in 2007.Data are from Whitmire et al. (28).

15998 | www.pnas.org/cgi/doi/10.1073/pnas.1205009109 Ulloa et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 4: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

this aerobic process may lay dormantuntil activated by recurrent oxygeninjection events.

How Are Microbial Communities andBiogeochemical Processes Coupledin AMZs?In the classic view, anaerobic microbialmetabolisms found in AMZs were thought

to be dominated by heterotrophic de-nitrification to N2 gas (Fig. 4). This viewhas changed substantially, in part, becausethe anaerobic oxidation of ammoniumwith nitrite, (anammox) has now beenidentified as a significant or even domi-nant pathway of N2 formation in manyoxygen-deficient marine systems (e.g., 8,37, 38). However, denitrification has re-

cently been shown to be important for theremoval of fixed nitrogen in the AMZsof the Arabian Sea and ETSP (39, 40).Modern measurements of anammox anddenitrification have not yet been reportedfor the ETNP AMZ. In addition, the dis-similatory reduction of nitrate to ammonia(DNRA; Fig. 4) has been shown to bea significant nitrogen cycling process in the

Pelagibacter sp. HTCC7211 (heterotrophy)

SUP05 (sulfur oxidation)

Scalindua profunda (anammox)

Nitrosopumilus maritimus (nitrification)

Nitrosopumilus-like (Thaumarchaea)Prochlorococcus-like (Cyanobacteria)Scalindua-like (Planctomycetes)SUP05-like ( -proteobacteria)Pelagibacter-like ( -proteobacteria)

0 5 10 15 20 25

Oxy

clin

eA

MZ

0 0.3 0.6 0.9 1.2 1.5 1.8

dsr (sulfide oxidation)

sox (thiosulfate oxidation)

apr (sulfite oxidation)

hzo (hydrazine oxidation, anammox)

narG (nitrate reduction, denitrification)

nirS/nirK (denitrification, anammox)

norB (denitrification)

nosZ (denitrification)

amo

0 1 2 3 4 5

hzo narG

amo (aerobic ammonia oxidation)

NH3 + O

2 NO

2- + 3H+ + 2e-

Tra

nsc

rip

t re

lati

ve a

bu

nd

ance

(%)

0.4

0.3

0.2

0.1

0.4

0.3

0.2

0.1

0.4

0.3

0.2

0.1

0.4

0.3

0.2

0.1

0

amo transcript abundance (%)

Transcript relative abundance (%)

Oxycline50 m

AMZ, top85 m

AMZ110 m

AMZ, core200 m

Transcript relative abundance (%)A

C

B

Candidatus Vesicomyosocius okutanii HA Chlorobium chlorochromatii CaD3 Candidatus Ruthia magnifica str. CmThiobacillus denitrificans ATCC 25259 endosymbiont of Bathymodiolus brevioruncultured prokaryoteChlorobium tepidum TLS uncultured -proteobacterium EBAC2C11Thiobacillus denitrificans

Thiolamprovum pedioformeRobbea sp. SB-2008 associated bacteriumuncultured SUP05 cluster bacteriumThiodictyon sp. f4Candidatus Kuenenia stuttgartiensissulfur-oxidizing bacterium DIII5endosymbiont of Bathymodiolus azoricusOTHER (32 taxa, < 0.0005% of total nr hits each)Putative S-reducers

Caldivirga maquilingensis IC-167 Desulfobacca acetoxidansDesulfonatronovibrio hydrogenovoransDesulfothermus naphthaeThermacetogenium phaeumThermodesulfatator indicusThermodesulfovibrio islandicusThermodesulfovibrio yellowstonii DSM 11347 uncultured sulfate-reducing bacterium

Putative sulfate-reducers:aprA genes

(AMZ core, 150 m)Oxycline50 m

AMZ, top85 m

AMZ110 m

AMZ, core200 m

D

HS- + NO3

- + H+ + H2O

SO4

2- + 4H+

NH4

+ + NO2

- N2 + 2H

2O

CH2O + O

2 CO

2 + H

2O

*

*

Protein coding genes

30

Fig. 3. Key patterns inmetabolic protein-coding transcripts and gene sequences in the AMZ off Iquique, Chile. (A) Metatranscriptome sequencing reveals diversetranscripts matching the genomes or metagenomes of functionally diagnostic taxa. (*Pelagibacter-like transcripts encoding enzymes of heterotrophic aerobicrespiration are mostly restricted to oxycline depths. Pelagibacter-like transcripts at the AMZ core primarily encode transport-related proteins.) (B) Relativetranscriptional activity of genera representing distinct functional clades varies with depth. Reactions (boxed) are based on the characterized metabolisms ofclosely related taxa. Colors match those inA. (C) Genes for dissimilatory sulfur and nitrogenmetabolism are transcribed in depth-specific patterns. amo, ammoniamonooxygenase (amoABC subunits combined); apr, APS reductase (subunits aprAB and membrane anchor aprM combined); dsr, dissimilatory sulfite reductasepathway (all dsr genes combined); hzo, hydrazine oxidoreductase; narG, nitrate reductase; nirS/nirK, NO-forming nitrite reductase; norB, NO reductase; nosZ, N2Oreductase; sox, sulfur oxidation pathway (all sox genes combined). The scale differs for the amo gene. (D) Metagenome (DNA) sequencesmatching the aprA genesuggest a diverse AMZ community of both sulfur oxidizers and sulfate reducers. Abundances are shown as percentages of the total number of protein-codingsequences in shotgun-sequenced community RNA (A–C) and DNA (D) datasets. Protein-coding genes are arrayed along the x axis in A, with per-gene transcriptabundance normalized to kilobases of gene length. Figures are based on samples collected from the ETSP AMZ in June 2008 (29) (A–C) and January 2010 (17) (D).

Ulloa et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 15999

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 5: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

Peruvian AMZ (36), as well as in watersover the Oman Shelf, where DNRA rep-resents an important ammonium sourcefor anammox (18).These insights are largely based on

biogeochemical studies using 15N tracerexperiments and on gene expression assayswith specific gene markers to explore thenitrogen cycle (8). However, recent envi-ronmental genomic approaches have alsoadvanced our understanding of both thenitrogen cycle in AMZ waters as well asthe cycling of other elements previouslyunrecognized. Regarding anammox, genesurveys of AMZ waters have documentedan anammox community of relatively lowdiversity, characterized by members ofthe marine genus “Candidatus Scalindua”of the bacterial Planctomycetes phylum(41). A metatranscriptome depth profilefrom the Chilean open-ocean AMZ(Fig. 3) revealed few anammox bacterialtranscripts in surface waters and throughthe oxycline, whereas anammox genetranscripts were a prominent feature inthe AMZ core. This is consistent withpredicted distributions of N2 productionby anammox bacteria in AMZ settings(42). Anammox rates, however, have alsobeen shown to be highest at the top of theAMZ and not in the core (36, 38, 40, 43,44), suggesting that other factors, such aswater circulation (e.g., upwelling in-tensity) and ammonium availability, arealso important in determining the verticalpartitioning of anammox rates at anygiven time.Despite the apparent dominance of

anammox in N2 production in many AMZsettings, gene surveys reveal a complexassemblage of diverse taxonomic groupspotentially capable of denitrification (39,45, 46). Sequencing of community RNA(29) from an AMZ site off Iquique, Chile,has uncovered all the genes required fora complete denitrification pathway, andthese are transcriptionally most active inthe core of the AMZ (∼200 m) (Fig. 3C).Although the capacity for classic hetero-trophic denitrification appears large inAMZs, it is unclear why denitrificationappears less significant than other pro-cesses in some cases, although appearingrelatively more important in other settings(39, 40). One possibility is that heterotro-phic denitrification is limited by the avail-ability of organic matter (38–40, 47), andhence much more variable in space andtime than anammox.The first step of the denitrification

pathway, the reduction of nitrate to nitrite,accounts for the nitrite accumulation inAMZ waters (Fig. 1B). Microbial com-munity transcriptomes sampled off theChilean coast (Fig. 3) revealed a diverseand abundant pool of nitrate reductasegene (narG) transcripts used in dissimila-tory nitrate reduction to nitrite in the

AMZ core (29). Heterotrophic nitratereducers might supply significant amountsof both ammonia (via organic matterdecomposition) and nitrite (via nitratereduction) for the anammox process (36).Transcripts from the anammox genehydrazine oxidoreductase (hzo), found atrelatively high abundance in the AMZcore (Fig. 3), support a possible linkbetween anammox and dissimilatorynitrate reduction to nitrite.Perhaps the most significant recent re-

direction of the classic view of AMZ bio-geochemistry and microbiology has beenthe recognition of a cryptic pelagic sulfurcycle in these regions. In traditional

thinking, the anaerobic organic mattermineralization process of sulfate reduction,the first step in the sulfur cycle, should notbegin until both oxygen and nitrate/nitriteare fully used (48). Therefore, a sulfurcycle was never envisioned for nitrate- andnitrite-rich AMZs (Fig. 1), although tran-sient accumulations of H2S have beenobserved in anoxic, nitrate- and nitrite-depleted, continental shelf waters (49, 50).Surprisingly, recent taxonomic, meta-genomic, and metatranscriptomic surveyshave uncovered an abundant and diversesulfur-oxidizing microbial community inthe AMZ water column. This communityis particularly enriched in γ-proteobacteriarelated to sulfur-oxidizing symbionts ofdeep-sea bivalves (17, 29, 51–54) (Fig. 3).Metagenomic analysis of this clade, re-ferred to as “SUP05,” reveals a versatilemetabolism with genes mediating carbonfixation, dissimilatory reduction of nitrateto N2O, and oxidation pathways for di-verse reduced sulfur species (sulfide, sul-fite, elemental sulfur, and thiosulfate)(55). SUP05 bacteria thus have the geneticrepertoire for carrying out autotrophicdenitrification coupled with sulfur oxida-tion (Fig. 4).The fact that sulfide-oxidizing micro-

organisms are abundant in AMZ watersimmediately raises the question, “Wheredoes the sulfide come from?” Radio-labeled sulfate tracer experiments fromthe AMZ off the Chilean coast revealedsignificant sulfate reduction in the upperreaches of the AMZ water column at ratesestimated to contribute up to 20% of theAMZ carbon mineralization in this region(17). Consistent with these results, meta-genomic and metatranscriptomic datasetsfrom the same AMZ revealed the pres-ence of dissimilatory sulfur respiratorygenes (aprA and dsrB) similar to thosefound in the genomes of known δ-pro-teobacterial sulfate reducers (17, 29) (Fig.3D). Further experiments revealed thatthe rapid removal of sulfide was coupledto the reduction of nitrate to nitrite, orN2O reduction to N2 (17). The rapidity ofsulfide oxidation in high backgrounds ofseawater sulfate, as well as its couplingwith denitrification processes typicallyascribed to heterotrophic carbon de-composition, partially explains why thispelagic sulfur cycle went previously un-recognized. Overall, these recent findingssuggest the presence of an active pelagicsulfur cycle in AMZ waters. This sulfurcycle fuels nitrate reduction, thereby sup-plying additional substrates (nitrite andammonia) for anammox bacteria (Fig. 4).Until now, an active planktonic sulfur cy-cle has only been measured in the ETSPAMZ, but we expect it to be present alsoin the other AMZs. Taxonomic molecularsurveys in the AMZs of the Arabian Seaand ETNP (51, 56) have suggested the

Fig. 4. Major microbial biogeochemical processesin the AMZ core and in the adjacent oxic waters.The heterotrophic processes within the core areanaerobic and include sulfate reduction, nitratereduction to nitrite, nitrate reduction to ammo-nium (DRNA), and denitrification to N2 gas. Theseprocesses oxidize organic matter and liberateammonia for use by anammox bacteria. The sul-fide produced by sulfate reducers is oxidizedagain to sulfate through autotrophic microbialmetabolisms with nitrate and nitrite as electronacceptors. Sulfur metabolisms are given with bluearrows, and nitrogen metabolisms are given withblack arrows, except those coupled to sulfur (sul-fide plus S-intermediate compounds), which aregiven by red arrows. Anammox is also an auto-trophic microbial process. Ammonium oxidation(nitrification) is a significant aerobic microbialprocess in the oxic waters surrounding the AMZcore and most probably occurring during oxygenintrusions into the core. Nitrogen fixation is theenergy-intensive fixation of N2 gas to organic ni-trogen (the oxidation state of ammonium) and isaccomplished by a wide range of microorganisms,including sulfate reducers, sulfur oxidizers, andcyanobacteria. It occurs in both the oxygenatedupper layers of AMZ settings and in the AMZ core.

16000 | www.pnas.org/cgi/doi/10.1073/pnas.1205009109 Ulloa et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 6: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

presence of sulfate reducers and sulfuroxidizers in the water column.

What Processes Control AMZChemistry?Despite complex interactions between thesulfur and nitrogen cycles in modernAMZs, nitrate and nitrite are normallypresent in excess. Why should AMZ waterchemistry be poised in this chemical state?Modeling has revealed that natural feed-backs operate to maintain a nitrate/nitrite-rich AMZ chemistry if nitrogen fixation islimited (57). With limited nitrogen fixa-tion, the nitrate reduced to N2 in the AMZwill not be replenished, ensuring nitrogen-limited primary production in the over-lying waters (57). The feedback control isquite simple: If denitrification becomes toointense, nitrogen will become further limit-ing, decreasing primary production. This willdiminish the organic carbon flux into theOMZ, reducing the rate of denitrificationand allowing nitrate (plus nitrite) to persist.Rather paradoxically, nitrate and nitrite-richOMZs persist in this chemical state due tonitrogen limitation in the overlying waters.This scenario changes, however, if nitro-

gen fixation balances the N2 loss in theAMZ. In this case, sulfidic conditions woulddevelop (57) when upwelling rates becomehigh enough to generate sufficient primaryproduction and a downward carbon fluxto remove all nitrate (plus nitrite) by de-nitrification and anammox. This does notnormally happen because nitrogen fixationinAMZregions appears to be insufficient to

replace the N2 lost in the AMZ core. Ni-trogen fixation, however, does occur, andhas been documented in the surface watersof theArabian Sea (58) and in the upwellingwaters off the coasts of Peru and Chile (59).Also, blooms of the nitrogen-fixing cyano-bacterium Trichodesmium, derived fromocean color satellite data, have been re-ported for surfacewaters of theArabian Seaand the ETNP off Central America (60). Inthe case of the ETSP AMZ (59), nitrogenfixation appears to be accomplished bya mixed population of prokaryotes, in-cluding sulfate reducers and sulfur oxidizersbut apparently not cyanobacteria, which arebelieved to be responsible for most of thenitrogen fixation in the oceans (61). Nitro-gen fixation is distributed in both the oxicupper layers of the water column and wellinto the AMZ at rates varying considerablyfrom year to year (59). Still, even at themaximum observed rate, nitrogen fixationwould only replenish about 5% of the ni-trate lost to N2 in the ETSP AMZ.Although the factors regulating nitrogen

fixation would be paramount in establish-ing AMZ chemistry, there is no consensusas to why nitrogen fixation is so limited inAMZ areas. One possibility is that theturbid highly productive waters typicallyoverlying AMZs are not well suited to thegrowth of Trichodesmium-like and unicel-lular nitrogen-fixing cyanobacteria, whichprefer open-ocean, oligotrophic, and calmconditions (62). It is not clear, however, whyAMZ settings do not support other nitro-gen fixers whose physiological traits are

more compatible with the ambient physi-cochemical conditions. As another possi-bility, iron limitation demonstrably hindersprimary production in the ETSP AMZ re-gion (63). Severe iron limitation could ex-clude nitrogen-fixing cyanobacteria whoseiron requirements are particularly high (64).Nevertheless, there is no inherent reason

why AMZs must maintain a nitrate/nitrite-rich chemistry or why such chemistry shouldhave always defined AMZs in the past.Indeed, there is evidence for sulfidic AMZsin theNorth Atlantic during the Cretaceousocean anoxic event 2 at the Cenomanian/Turonian boundary (65), at the Permian/Triassic boundary (66), and for time peri-ods in the Precambrian (67). Becausemodern AMZs support the same bio-geochemical cycles as sulfidic water bodiestoday, and presumably also sulfidic AMZsof the past, one can view them as a partic-ular chemical intermediate state betweentwo “end points” represented by fully oxicsystems and fully sulfidic systems (Fig. 5).Thus, one can picture periods in Earthhistory in which the ocean resembledmodern AMZs with a nitrate/nitrite-richchemistry but without the accumulation ofH2S. As nitrogen fixation increases, sulfidicAMZs become possible, and with basinrestriction or contact with underlying sedi-ments, sulfidic conditions can extend to thesediment–water interface (a conceptualmodel is provided in Fig. 5). Other factors,such as atmospheric oxygen concentrations(controlling oxygen availability to deepocean waters), iron availability, and large-scale ocean ventilation, may also influenceAMZ chemistry, and these factors shouldinform the further development of AMZbiogeochemical models (67, 68).

AMZs in the AnthropoceneOMZs are microbial reactors of globalsignificance. Models have predicted theirexpansion in response to global warming,both as a result of reduced oxygen solubilityat higher temperatures and from anincreased water-column stratificationresulting from stronger temperature gra-dients in the upper waters (69, 70). Ex-amination of a 50-y dissolved oxygendatabase suggests such an expansion (12),which should continue well into the future(13). The increasing deposition of atmo-spheric anthropogenically fixed nitrogenon the open ocean (71) is also expectedto contribute to AMZ expansion andintensification, and eventually to thedevelopment of sulfidic conditions. Ex-panded areas of reduced oxygen will altermarine ecosystems, including the dis-tributions and ecology of marine animals.Expanding AMZs should also increaserates of fixed nitrogen loss as N2 (57, 72),as well as production rates of N2O, an im-portant greenhouse gas, at their boundaries(73). It is unclear how these changes will

oxygennitratenitritesulfide

Concentration

N-fixation

O2-availability

AMZ “sulfidic” AMZ sulfidic bottom

Dep

th

Open ocean-OMZ

Primary Production

Bay of Bengal

NE Pacific

low oxygen-OMZ

ETNP

ETSP

Arabian Sea

No permanent

modern exampleExamplesBlack Sea

Baltic Sea basins

Cariaco Basin

Saanich Inlet

Namibian shelf

Much of the

global ocean

Fig. 5. Cartoon shows thedevelopment of different chemistries dependingon the relativemix of differentdriving parameters, including rates of primary production, rates of nitrogen fixation, and the availability ofoxygen. Oxygen availability could imply oxygen limitation as observed inmodern anoxic basins, such as theBlack Sea, or the limitation of oxygen availability as occurred during times in the geological past whenatmospheric oxygen concentrations were lower. If oxygen availability becomes limiting enough, rates ofprimary production and nitrogen fixation may take a secondary role in determining the development ofwater column chemistry. Based on information provided in the text, OMZs include those regions of theglobal ocean where oxygen is decreased as a result of respiration but where nitrite does not accumulate(except a nitrite maximum that typically occurs in well-oxygenated, near-surface waters as a result of ni-trification). Modernmeasurements suggest that nitrite accumulation occurs at oxygen levels of less than 50nM. AMZs occur as oxygen falls below about 50 nM and nitrite begins to accumulate.

Ulloa et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 16001

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 7: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

affect human welfare. One might assume,however, that a contraction of animalhabitat and alterations in the elementalcycles would produce negative effects,particularly if AMZs expand further ontothe continental shelf and compound theexpansion of shelf anoxia and euxinia fromincreasing nutrient inputs from land (74).For these reasons, recent scientific

progress in our understanding of the mi-crobiological function, dynamics, and dis-tribution of oceanic AMZ waters is mostfortunate. However, we view this progressas only a beginning. New oxygen-sensingtechnologies have barely been applied tooutlining the spatial distribution and dy-namics of oxygen accurately in AMZwaters. We still do not know the nature ofthe episodic oxygen intrusions, as revealedby autonomous platforms, or their role instructuring AMZ microbial communitiesand biogeochemical cycling. Accurate

budgets of fixed nitrogen loss in AMZs arestill under debate, as are the processescontrolling this loss, and we have just begunto understand the role of sulfur cycling inAMZ waters. We are also uncertain as tothe rates of nitrogenfixation inAMZwatersand as to what processes regulate theserates. Despite high rates of nitrogen lossthrough denitrification and anammox, withthe establishment of severe nitrogen limi-tation, high concentrations of unused ni-trogen still persist in surface watersoverlying many AMZ settings. Iron limi-tationmay be partly responsible for this, butis this the only reason? Will these nutrientsbecome more available to fuel primaryproduction under future global climatechange or anthropogenic emissions?New tools and methods have revealed

novel complexity in AMZ ecosystems, butwe anticipate that AMZs may supportan even more diverse assemblage of in-

digenous microbial species that remaincompletely uncharacterized genetically,physiologically, and ecologically. Al-though our knowledge of AMZs has beengrowing at a breathtaking pace, thepotential impacts of AMZs on marineecosystem structure and global geo-chemical cycling highlight a need to ac-celerate this rapid pace of exploration anddiscovery further.

ACKNOWLEDGMENTS. We thank all our col-leagues who have contributed ideas and datamaking this review possible and the Agouron In-stitute for encouragement. We also thankWinstonRojas for help with the preparation of some of thefigures. Funding was provided by the AgouronInstitute. Additional financial support was pro-vided by the Chilean National Commission forScientific and Technological Research through theFondap Program (O.U.), the Gordon and BettyMoore Foundation (O.U., E.F.D., and R.M.L.), theDanish National Research Foundation (D.E.C.), andthe European Research Council (D.E.C.).

1. Wyrtki K (1962) The oxygen minima in relation toocean circulation. Deep-Sea Res 9:11–23.

2. Kamykowski D, Zentara S-J (1991) Spatio-temporal andprocess-oriented views of nitrite in the world oceans asrecorded in the historical data set. Deep-Sea Res 38:445–464.

3. Thamdrup B, Dalsgaard T, Revsbech NP (2012) Wide-spread functional anoxia in the oxygen minimum zoneof the eastern South Pacific. Deep Sea Res Part I Ocean-ogr Res Pap 65:36–45.

4. Cohen Y, Gordon LI (1978) Nitrous oxide in the oxygenminimum of the eastern tropical North Pacific: Evi-dence for its consumption during denitrification andpossible mechanisms for its production. Deep-Sea Res25:509–524.

5. Nicholls JC, Davies CA, Trimmer M (2007) High-resolu-tion profiles and nitrogen isotope tracing reveal a dom-inant source of nitrous oxide and multiple pathways ofnitrogen gas formation in the central Arabian Sea. Lim-nol Oceanogr 52:156–168.

6. Farías L, Paulmier A, Gallegos M (2007) Nitrous oxideand N-nutrient cycling in the oxygen minimum zoneoff northern Chile. Deep Sea Res Part I Oceanogr ResPap 54:113–128.

7. Childress JJ, Seibel BA (1998) Life at stable low oxygenlevels: Adaptations of animals to oceanic oxygen min-imum layers. J Exp Biol 201:1223–1232.

8. Lam P, Kuypers MMM (2011) Microbial nitrogen cyclingprocesses in oxygen minimum zones. Annu Rev Mar Sci3:317–345.

9. Codispoti LA, et al. (2001) The oceanic fixed nitrogenand nitrous oxide budgets: Moving targets as we enterthe Anthropocene? Scientia Marina 65:85–105.

10. Keeling RE, Körtzinger A, Gruber N (2010) Ocean de-oxygenation in a warming world. Annu Rev Mar Sci 2:199–229.

11. Helm KP, Bindoff NL, Church JA (2011) Observed de-creases in oxygen content of the global ocean. Geo-phys Res Lett 38:L23602.

12. Stramma L, Johnson GC, Sprintall J, Mohrholz V (2008)Expanding oxygen-minimum zones in the tropicaloceans. Science 320:655–658.

13. Shaffer G, Olsen SM, Olaf J, Pedersen P (2009) Long-term ocean oxygen depletion in response to carbondioxide emissions from fossil fuels. Nat Geosci 2:105–109.

14. Schmidt J (1925) On the contents of oxygen in theocean on both sides of Panama. Science 61:592–593.

15. Winkler LW (1888) Die Bestimmung des im Wasser Ge-lösten Sauerstoffes. Berichte der Deutschen Chemi-schen Gesellschaft 21:2843–2855.

16. Revsbech NP, et al. (2009) Determination of ultra-lowoxygen concentrations in oxygen minimum zones by theSTOX sensor. Limnol Oceanogr Methods 7:371–381.

17. Canfield DE, et al. (2010) A cryptic sulfur cycle in oxy-gen-minimum-zone waters off the Chilean coast. Sci-ence 330:1375–1378.

18. Jensen MM, et al. (2011) Intensive nitrogen loss over theOmani Shelf due to anammox coupled with dissimilatorynitrite reduction to ammonium. ISME J 5:1660–1670.

19. Canfield DE, Kristensen E, Tamdrup B (2005) AquaticGeomicrobiology (Academic Press, San Diego).

20. Jensen MM, Kuypers MMM, Lavik G (2008) Rates andregulation of anaerobic ammonium oxidation and de-nitrification in the Black Sea. Limnol Oceanogr 53:23–36.

21. Kalvelage T, et al. (2011) Oxygen sensitivity of anam-mox and coupled N-cycle processes in oxygen mini-mum zones. PLoS ONE 6:e29299.

22. Devol AH (1978) Bacterial oxygen uptake kinetics asrelated to biological processes in oxygen deficientzones of the oceans. Deep-Sea Res 25:137–146.

23. Rao CK, et al. (1994) Hydrochemistry of the Bay ofBengal: Possible reasons for a different water-columncycling of carbon and nitrogen from the Arabian Sea.Mar Chem 47:279–290.

24. Howell EA, Doney SC, Fine RA, Olson DB (1997) Geo-chemical estimates of denitrification in the ArabianSea and Bay of Bengal during WOCE. Geophys Res Lett24:2549–2552.

25. Stolper DA, Revsbech NP, Canfield DE (2010) Aerobicgrowth at nanomolar oxygen concentrations. Proc NatlAcad Sci USA 107:18755–18760.

26. Lavin P, González B, Santibáñez JF, Scanlan DJ, Ulloa O(2010) Novel lineages of Prochlorococcus thrive withinthe oxygen minimum of the eastern tropical South Pa-cific. Environ Microbiol Rep 2:728–738.

27. Goericke R, Olson R, Shalapyonok A (2000) A novelniche for Prochlorococcus sp. in low-light suboxic envi-ronments in the Arabian Sea and the Eastern TropicalNorth Pacific. Deep Sea Res Part I Oceanogr Res Pap 47:1183–1205.

28. Whitmire AL, Letelier RM, Villagrán V, Ulloa O (2009)Autonomous observations of in vivo fluorescence andparticle backscattering in an oceanic oxygen minimumzone. Opt Express 17:21992–22004.

29. Stewart FJ, Ulloa O, DeLong EF (2012) Microbial meta-transcriptomics in a permanent marine oxygen mini-mum zone. Environ Microbiol 14:23–40.

30. van de Vossenberg J, et al. (2012) The metagenome ofthe marine anammox bacterium ‘Candidatus Scalinduaprofunda’ illustrates the versatility of this globally im-portant nitrogen cycle bacterium. Environ Microbiol,10.1111/j.1462-2920.2012.02774.x.

31. Belmar L, Molina V, Ulloa O (2011) Abundance andphylogenetic identity of archaeoplankton in the per-manent oxygen minimum zone of the eastern tropicalSouth Pacific. FEMS Microbiol Ecol 78:314–326.

32. Pitcher A, et al. (2011) Niche segregation of ammonia-oxidizing archaea and anammox bacteria in the Ara-bian Sea oxygen minimum zone. ISME J 5:1896–1904.

33. Ward BB, Glover HE, Lipschultz F (1989) Chemoauto-trophic activity and nitrification in the oxygen mini-mum zone off Peru. Deep-Sea Res 36:1031–1051.

34. Lipschultz F, et al. (1990) Bacterial transformations ofinorganic nitrogen in the oxygen-deficient waters ofthe Eastern Tropical South Pacific Ocean. Deep-Sea Res37:1513–1541.

35. Füssel J, et al. (2012) Nitrite oxidation in the Namibianoxygen minimum zone. ISME J 6:1200–1209.

36. Lam P, et al. (2009) Revising the nitrogen cycle in thePeruvian oxygen minimum zone. Proc Natl Acad SciUSA 106:4752–4757.

37. Kuypers MMM, et al. (2005) Massive nitrogen loss fromthe Benguela upwelling system through anaerobic am-monium oxidation. Proc Natl Acad Sci USA 102:6478–6483.

38. Thamdrup B, et al. (2006) Anaerobic ammonium oxi-dation in the oxygen-deficient waters off northernChile. Limnol Oceanogr 51:2145–2156.

39. Ward BB, et al. (2009) Denitrification as the dominantnitrogen loss process in the Arabian Sea. Nature 461:78–81.

40. Dalsgaard T, Thamdrup B, Farias L, Revsbech NP (2012)Anammox and denitrification in the oxygen minimumzone of the eastern South Pacific. Limnol Oceanogr 57:1331–1346.

41. Schmid MC, et al. (2007) Anaerobic ammonium-oxidiz-ing bacteria in marine environments: Widespread oc-currence but low diversity. Environ Microbiol 9:1476–1484.

42. Chang BX, Devol AH, Emerson SR (2010) Denitrificationand the nitrogen gas excess in the eastern tropicalSouth Pacific oxygen deficient zone. Deep Sea Res PartI Oceanogr Res Pap 57:1092–1101.

43. Galán A, et al. (2009) Anammox bacteria and the an-aerobic oxidation of ammonium in the oxygen mini-mum zone off northern Chile. Deep Sea Res Part II TopStud Oceanogr 56:1021–1031.

44. Lam P, et al. (2011) Origin and fate of the secondarynitrite maximum in the Arabian Sea. Biogeosciences 8:1565–1577.

45. Jayakumar DA, Francis CA, Naqvi SWA, Ward BB (2004)Diversity of nitrite reductase genes (nirS) in the deni-trifying water column of the coastal Arabian Sea.Aquat Microb Ecol 34:69–78.

46. Castro-González M, Braker G, Farías L, Ulloa O (2005)Communities of nirS-type denitrifiers in the water col-umn of the oxygen minimum zone in the easternSouth Pacific. Environ Microbiol 7:1298–1306.

47. Ward BB, et al. (2008) Organic carbon, and not copper,controls denitrification in oxygen minimum zones of

16002 | www.pnas.org/cgi/doi/10.1073/pnas.1205009109 Ulloa et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020

Page 8: Microbial oceanography of anoxic oxygen minimum zones · Microbial oceanography of anoxic oxygen minimum zones Osvaldo Ulloaa,1, Donald E. Canfieldb, Edward F. DeLongc, Ricardo M.

the ocean. Deep Sea Res Part I Oceanogr Res Pap 55:1672–1683.

48. Froelich PN, et al. (1979) Early oxidation of organicmatter in pelagic sediments of the eastern equatorialAtlantic: Suboxic diagenesis. Geochim Cosmochim Acta43:1075–1090.

49. Dugdale R, Goering J, Barber R, Smith R, Packard T(1977) Denitrification and hydrogen sulfide in the Peruupwelling region during 1976. Deep-Sea Res 24:601–608.

50. Naqvi SWA, et al. (2000) Increased marine productionof N2O due to intensifying anoxia on the Indian conti-nental shelf. Nature 408:346–349.

51. Fuchs B, Woebken D, Zubkov M, Burkill P, Amann R(2005) Molecular identification of picoplankton popu-lations in contrasting waters of the Arabian Sea. AquatMicrob Ecol 39:145–157.

52. Stevens H, Ulloa O (2008) Bacterial diversity in the ox-ygen minimum zone of the eastern tropical South Pa-cific. Environ Microbiol 10:1244–1259.

53. Lavik G, et al. (2009) Detoxification of sulphidic Africanshelf waters by blooming chemolithotrophs. Nature457:581–584.

54. Wright JJ, Konwar KM, Hallam SJ (2012) Microbial ecol-ogy of expanding oxygen minimum zones. Nat RevMicrobiol 10:381–394.

55. Walsh DA, et al. (2009) Metagenome of a versatile che-molithoautotroph from expanding oceanic deadzones. Science 326:578–582.

56. Podlaska A, Wakeham SG, Fanning KA, Taylor GT(2012) Microbial community structure and productivityin the oxygen minimum zone of the eastern tropical

North Pacific. Deep Sea Res Part I Oceanogr Res Pap 66:77–89.

57. Canfield DE (2006) Models of oxic respiration, denitri-fication and sulfate reduction in zones of coastal up-welling. Geochim Cosmochim Acta 70:5753–5765.

58. Capone DG, et al. (1998) An extensive bloom of the N2-fixing cyanobacterium Trichodesmium erythraeum inthe central Arabian Sea. Mar Ecol Prog Ser 172:281–292.

59. Fernandez C, Farías L, Ulloa O (2011) Nitrogen fixationin denitrified marine waters. PLoS ONE 6:e20539.

60. Westberry TK, Siegel DA (2006) Spatial and temporaldistribution of Trichodesmium blooms in the world’soceans. Global Biogeochem Cycles 20:GB4016, 10.1029/2005GB002673.

61. Zehr JP, Kudela RM (2011) Nitrogen cycle of the openocean: From genes to ecosystems. Annu Rev Mar Sci 3:197–225.

62. Hood RR (2004) Modeling the distribution of Tricho-desmium and nitrogen fixation in the Atlantic Ocean.J Geophys Res 109:C06006, 10.1029/2002JC001753.

63. Hutchins DA, et al. (2002) Phytoplankton iron limita-tion in the Humboldt Current and Peru Upwelling. Lim-nol Oceanogr 47:997–1011.

64. Falkowski PG (1997) Evolution of the nitrogen cycleand its influence on the biological sequestration ofCO2 in the ocean. Nature 387:272–275.

65. Hetzel A, Böttcher ME, Wortmann UG, Brumsack H-J(2009) Paleo-redox conditions during OAE 2 reflectedin Demerara Rise sediment geochemistry (ODP Leg207). Palaeogeogr Palaeoclimatol Palaeoecol 273:302–328.

66. Algeo TJ, et al. (2011) Spatial variation in sedimentfluxes, redox conditions, and productivity in the Perm-ian–Triassic Panthalassic Ocean. Palaeogeogr Palaeocli-matol Palaeoecol 308:65–83.

67. Lyons TW, Anbar AD, Severmann S, Scott C, Gill BC(2009) Tracking euxinia in the ancient ocean: A multi-proxy perspective and Proterozoic case study. AnnuRev Earth Planet Sci 37:507–534.

68. Poulton SW, Canfield DE (2011) Ferruginous condi-tions: A dominant feature of the ocean throughEarth’s history. Elements 7:107–112.

69. Bopp L, Le Quéré C, Heimann M, Manning AC,Monfray P (2002) Climate-induced oceanic oxygen fluxes:Implications for the contemporary carbon budget. GlobalBiogeochem Cycles 16:1022, 10.1029/2001GB001445.

70. Matear RJ (2003) Long-term changes in dissolved oxy-gen concentrations in the ocean caused by protractedglobal warming. Global Biogeochem Cycles 17:1125,10.1029/2002GB001997.

71. Duce RA, et al. (2008) Impacts of atmospheric anthro-pogenic nitrogen on the open ocean. Science 320:893–897.

72. Schmittner A, Oschlies A, Matthews HD, Galbraith ED(2008) Future changes in climate, ocean circulation,ecosystems, and biogeochemical cycling simulated fora business-as-usual CO2 emission scenario until year4000 AD. Global Biogeochem Cycles 22:GB1013,10.1029/2007GB002953.

73. Codispoti LA (2010) Oceans. Interesting times for ma-rine N2O. Science 327:1339–1340.

74. Diaz RJ, Rosenberg R (2008) Spreading dead zones andconsequences for marine ecosystems. Science 321:926–929.

Ulloa et al. PNAS | October 2, 2012 | vol. 109 | no. 40 | 16003

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

14, 2

020