Deep-SeaResearchII93(2013)2–15 ContentslistsavailableatSciVerseScienceDirect Deep-Sea Research II journal homepage: www.elsevier.com/locate/dsr2 Sea change: Charting the course for biogeochemical ocean time-series research in a new millennium Matthew J. Churcha,n, Michael W. Lomasb, Frank Muller-Kargerc aDepartmentofOceanography,UniversityofHawaii,1000PopeRoad,Honolulu,HI96822,UnitedStates bBigelowLaboratoryforOceanSciences,60BigelowDrive,EastBoothbay,ME04544,UnitedStates cInstituteforMarineRemoteSensing,CollegeofMarineScience,UniversityofSouthFlorida,St.Petersburg,FL33701,UnitedStates a r t i c l e i n f o a b s t r a c t Availableonline31January2013 Ocean time-series provide vital information needed for assessing ecosystem change. This paper summarizesthehistoricalcontext,majorprogramobjectives,andfutureresearchprioritiesforthree Keywords: Time-series contemporaryoceantime-seriesprograms:TheHawaiiOceanTime-series(HOT),theBermudaAtlantic Oceanbiogeochemistry Time-series Study (BATS), and the CARIACO Ocean Time-Series. These three programs operate in HOT physicallyandbiogeochemicallydistinctregionsoftheworld’soceans,withHOTandBATSlocatedin BATS theopen-oceanwatersofthesubtropicalNorthPacificandNorthAtlantic,respectively,andCARIACO CARIACO situatedintheanoxicCariacoBasinofthetropicalAtlantic.Allthreeprogramssustainnear-monthly Carboncycling shipboard occupations of their field sampling sites, with HOT and BATS beginning in 1988, and CARIACO initiated in 1996. The resulting data provide some of the only multi-disciplinary, decadal- scaledeterminationsoftime-varyingecosystemchangeintheglobalocean.Facilitatedbyascoping workshop(September2010)sponsoredbytheOceanCarbonBiogeochemistry(OCB)program,leaders ofthesetime-seriesprogramssoughtcommunityinputonexistingprogramstrengthsandforfuture researchdirections.Themesthatemergedfromthesediscussionsincluded: 1. Shipboard time-series programs are key to informing our understanding of the connectivity betweenchangesinocean-climateandbiogeochemistry. 2. The scientific and logistical support provided by shipboard time-series programs forms the backbonefornumerousresearchandeducationprograms.Futurestudiesshouldbeencouragedthat seekmechanisticunderstandingofecologicalinteractionsunderlyingthebiogeochemicaldynamics atthesesites. 3.Detectingtime-varyingtrendsinoceanpropertiesandprocessesrequiresconsistent,high-quality measurements. Time-series must carefully document analytical procedures and, where possible, tracetheaccuracyofanalysestocertifiedstandardsandinternalreferencematerials. 4.Leveragedimplementation,testing,andvalidationofautonomousandremoteobservingtechnologies attime-seriessitesprovidenewinsightsintospatiotemporalvariabilityunderlyingecosystemchanges. 5.Thevalueofexistingtime-seriesdataforformulatingandvalidatingecosystemmodelsshouldbe promoted. In summary, the scientific underpinnings of ocean time-series programs remain as strong and importanttodayaswhentheseprogramswereinitiated.Theemergingdatainformourknowledgeof theocean’sbiogeochemistryandecology,andimproveourpredictivecapacityaboutplanetarychange. &2013ElsevierLtd.Allrightsreserved. 1. Preface play interactive roles in shaping climate, serve as reactive pools of bioelements, and comprise vast reservoirs of biodiversity. Despite 1.1. Time,water,andchange theirimportancetohumansandplanetaryhealth,wecurrentlylack confidentpredictiveunderstandingofhowoceanicecosystemsmay Oceans are vital to Earth’s habitability and are important socio- respond to global change. In part this derives from chronic under- economic resources. The massive and diverse oceanic ecosystems sampling of these remote and spatiotemporally complex habitats. There are currently only a few regions in the sea where we have sustained, decadal-scale observations on the interactions between ocean biogeochemistry, hydrography, and ecology. These few time- nCorrespondingauthor. E-mailaddress:[email protected](M.J.Church). series, built largely around shipboard sampling programs, provide 0967-0645/$-seefrontmatter&2013ElsevierLtd.Allrightsreserved. http://dx.doi.org/10.1016/j.dsr2.2013.01.035 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 3 evidencethatocean–climate–humaninteractionscanhavecascading Time-series (HOT), Bermuda Atlantic Time-series Study (BATS), and impacts on ecosystem processes across a wide range of time and theCARIACOOceanTime-Series.Alltheseprogramsmaintainnear- spacescales.Moreover,thesetime-seriesrecordsprovidethecritical monthly ship-based sampling programs in hydrographically and data needed to better understand the ocean’s and our planet’s biogeochemicallydistinctregions:theNorthPacificSubtropicalGyre sensitivitytochange. (NPSG),thesubtropicalNorthAtlantic,andtheCariacoBasininthe A major challenge facing ocean scientists is how to advance tropical North Atlantic, respectively (Table 1). These sites are com- interdisciplinary, multi-decadal time-series programs that provide plementedbysystematicobservationscollectedfromotherobserving highqualitydataatsufficientspatiotemporalresolutiontoinformour platformsincludingmoorings,satellites,floats,andgliders.Thedata understandingofmarineecosystemchange.Despitethesuccessesof fromthesetime-seriesprogramsprovidesomeoftheonlydecadal- ship-basedtime-seriesindocumentingoceanchange,samplingfrom scale records available for assessing seasonal- to interannual-scale researchvesselsisrelativelyrestrictiveinitsspatio-temporalresolu- changes in ocean hydrographic structure, biogeochemistry, and tion. Rising costs associated with operating and maintaining ships biology. require continual reassessment and justification of the cost–benefit Oceanbiogeochemistrytime-seriesprogramsremainvitalcom- scenariostoscienceandsocietybehindsustainingship-basedocea- munity resources, providing invaluable cross-disciplinary informa- nographic programs. Moreover, technology advances in the past tion on ocean ecosystem change. Discussions as part of this decade now allow for high-resolution autonomous sampling of workshopprovidedanopportunityforcommunityinputonscience variousoceanographicproperties.Suchfactorshavepartlymotivated priorities for the next decade of research at these sites, including the current expansion of ocean observational systems designed to exploringthestrengthsandlimitationsofexistingshipboardtime- relyonremoteandautonomoussensingplatforms,promisinghigher series programs. In this paper, we summarize some of these frequencyspatio-temporalsamplingofecosystemdynamicsdifficult discussionswithintheframeworkoftheresearchdoneatthethree to adequately sample from ships alone, at a potential cost savings. on-going time-series, define opportunities for new science built Satellites, floats, gliders, moorings, and remotely operated vehicles around these programs, and provide several recommendations for have all become increasingly central to such observing strategies. improvingthestateofglobaloceanobservingthatincludesalong- However,todate,welackthecapabilitiestodetectmanyofthekey termroleforshipboardtime-seriesresearch. climate-sensitivebiogeochemicalpropertiesandprocessesweknow are fundamental to ecosystem function from these autonomous sensingplatforms.Balancinginvestmentsrequiredtomaintaincom- 2. Background prehensiveoceancarbonandbiogeochemistrytime-seriesandmore autonomousobservatorysystemsiscentraltotheevolvingvisionof 2.1. Ifyoubuildit,theywillcome theoceansciences. FacilitatedbytheOceanCarbonBiogeochemistry(OCB)program, Thevisiontoestablishshipboardtime-seriessprungfromrecogni- threecontemporaryoceantime-seriesprogramsconveneda‘‘scoping tionthatdetectinghowanthropogenicandnatural-climatechanges workshop’’ in September 2010 to gather community input on the influence the biosphere is impossible without time-resolving mea- future direction and scope of ship-based time-series programs. The surementscollectedoverlongperiodsoftime(Keelingetal.,1995; workshop largely focused on three on-going programs whose Likensetal.,1996;Schindler,1988).Historicalrootsforcontemporary researchalignswiththeOCBprogramobjectives:theHawaiiOcean ocean time-series can be traced in part to use of weather ships, Table1 OceanCarbonBiogeochemistrytime-seriesprogramsandstudysitecharacteristics. Programand Periodand Generalsite Annualmean(and Samplinginfrastructure Programleadership studysite frequencyof characteristics range)primary shipboard productivityand observations carbonexport (molCm(cid:2)2yr(cid:2)1) BATS 1988–present SubtropicalNorthAtlantic 14(9.7–16), Shipboardobservations AnthonyKnap(1988–2012),Anthony (31.751N, (monthly) (SargassoSea),seasonally 0.87(0.67–1.1) (1988–present),bottom- Michaels(1989–1996),RobJohnson 64.161W) oligotrophic,moderate mooredsedimenttraps (1988–present),NickBates seasonality(largely (1978–present), (1995–present),DebbieSteinberg(1997– attributable profilingfloats(2009– 2001),CraigCarlson towintermixing) present), (1992–2001),MichaelLomas mooredplatform(1994– (2001–present) 2007) HOT—Station 1988–present SubtropicalNorthPacific, 14(9.4–18), Shipboardobservations DavidKarl,RogerLukas,RicardoLetelier, ALOHA (monthly) persistentlyoligotrophic, 0.84(0.64–1.2) (1988–present),bottom- JohnDore,RobertBidigare(all1988– (22.751N, lowseasonalityin mooredsedimenttraps present),EricFiring 1581W) hydrography (1992–present), (1988–1998),StephenChiswell(1988– andbiogeochemistry profilingfloats(2005– 1993),ChristopherWinn(1988–1997), present), MichaelLandry mooredplatforms(1997– (1992–present),LuisTupas present),cabledobservatory (1991–2000),DaleHebel (2010–present) (1988–2005),MatthewChurch(2009– present) CARIACO 1995–present TropicalCaribbeanSea 40(29–53),2.1 Shipboardobservations FrankMuller-Karger,MaryScranton, (10.51N, (monthly) (CariacoBasin),mesotrophic, (1.4–2.8) (1995–present),bottom- GordonTaylor,RobertThunell,Ramon 64.671W) highlyseasonalhydrography mooredsedimenttraps Varela,YreneAstor andbiogeochemistry (1995–present) (all1995–present),KentFanning,Luis (attributabletochangesin Troccoli(2000–present),BaumarMarı´n upwelling) (2000–present),RobertWeisberg(1996– 2006), JohnJ.Walsh(1995–2000) 4 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 deployedinremoteregionsoftheworld’soceanstogathermeteor- forcing of ocean physics, and with the support from the Office of ological data necessary to improve weather forecast models. NavalResearch,HenryStommel(WoodsHoleOceanographyInstitu- The value of such repeated occupations to otherwise inaccessible tion) and William Sutcliffe (Bermuda Biological Station) initiated a regions of the oceans attracted the interests of a diverse group of biweeklysamplingcampaigntophysicallycharacterizethe(cid:3)2600m scientists.The‘‘ancillary’’observationscollectedbyprocess-oriented water column at Hydrostation S in the northwestern Sargasso Sea scientific programs, conducted alongside the weather-ship hydro- (Jenkins,1982;Schroederetal.,1959).Providingphysicalcontextand graphic and meteorological sampling, led to important discoveries. regular access to the open sea, facilitated in large part by suitably Forexample,oceanproductivityandirradiancemeasurementsat equipped research vessels and infrastructure support from the Weathership ‘‘M’’ in the North Sea laid the framework for the Bermuda Biological Station, Hydrostation S rapidly attracted new formulationofthecriticaldepththeoryexplainingtheprerequisite anddiversifiedscienceprograms.Amongthefirsttocapitalizeonthe interactionsbetweenlight,mixing,andphotosynthesisincontrol- HydrostationStime-serieswereWoodsHolescientistsDavidMenzel ling formation of phytoplankton blooms (Sverdrup, 1953). The and John Ryther. By placing measurements of productivity and legacyoftheseprogramscontinuestoday.Forexample,measure- nutrientcyclingintothetime-resolvedhydrographiccontextafforded mentsconductedatStationP(501N,1451W)continuethe obser- byHydrostationS,theirworkhelpeddefinefactorsshapingbiogeo- vations started in the 1940s from a weather-ship outpost in the chemistryinthesubtropicaloceangyres(MenzelandRyther,1960, subarctic North Pacific (Freeland, 2007; Whitney and Tortell, 1961). By the 1970s and 1980s several important biogeochemical 2006). The resulting data highlight complex, climate-forced bio- studies had coalesced around the HydrostationS, including a study geochemical interactions, including long-term changes in dis- that continues today aimed at quantifying temporal relationships solved oxygen concentrations within the main thermocline of betweenupper-oceanproductivityandthedownwardfluxofmate- theNorthPacificOcean(Whitneyetal.,2007). rialtothedeepsea(Conteetal.,2001;DeuserandRoss,1980;Deuser In addition to ocean weather ships, economic interests in et al., 1983, 1990), and geochemical approaches to estimate net commercially important fisheries motivated the establishment of communityproductivity(Jenkins,1982;JenkinsandGoldman,1985). time-seriesmonitoringprograms.ThecollapseofthePacificsardine Thesestudiesrevealedthatthevastsubtropicaloceangyresserveas fishery in the mid 1940s resulted in the creation of the California globally important carbon reactors, where productivity was greater Cooperative Fisheries Investigations Program (CalCoFI) program in than historically recognized and whose carbon storage potential 1949. This program has maintained a near-continuous record of depended on complex biological–physical couplings. By embedding climaticinfluencesonecosystemvariabilityintheCaliforniaCurrent suchmeasurementsintothetime-resolvinghydrographicframework, ecosystem (Field et al., 2006; Rebstock, 2002; Rykaczewski and Hydrostation S became one of the few places in the world where Checkley,2008)withafocusonfisheriesstocksandmanagement. quantitativeinformationonseasonaltointerannualscaleinteractions The establishment of Hydrostation S (321100N, 641300E) off Ber- betweenplanktonecology,hydrographicforcing,andbiogeochemical mudain1954marksamajorhistoricalentrypointformodernocean cyclesintheopenseawasavailable. time-series research (Fig. 1). Motivated by the desire to better These early efforts solidified a recurrent theme among con- understand seasonal to interannual variability in meteorological temporary ocean time-series: regular access to the open sea Fig.1. Timelineofresearchconductedatvariousoceantime-seriessitesaroundtheglobe.Coloredlinesindicatetypesofsamplingandmeasurementactivities(ships, moorings,gliders,floats,andsedimenttraps).Solidlinesreflectsustainedmeasurementprogramonatleastmonthlytimescales;dashedlinesindicatemeasurements havecontinuedbutatlowerthanmonthlyfrequency;andgapsintime-seriesrecordsareshownbybrokenlines.Siteabbreviationsare:StationS,HydrostationS;OSP, OceanStationPapa;Pal.LTER,PalmerLongTermEcologicalResearchprograminAntarctica.FigureadaptedfromKarletal.(2003).(Forinterpretationofthereferencesto colorinthisfigurecaption,thereaderisreferredtothewebversionofthisarticle.) M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 5 afforded by a ship provides unique opportunities for studying initiated with support from the Venezuelan government, NSF, and oceanprocesses,developingnewmethods,trainingstudentsand NASA. The project seeks to examine linkages between upper ocean technicians, and testing hypotheses. Encapsulating these attri- productivity,terrigenousmaterialinputtotheenclosedbasin,mate- butesinarichhistoricalcontextofphysicalandbiogeochemical rial fluxes from the shelf into deep water, and the preservation of measurements further strengthens the allure of such programs. climate signals in the sediment accumulating at the bottom of this These efforts inspired questions that remain at the core of anoxic tropical ecosystem (Muller-Karger et al., 2010; Taylor et al., contemporaryoceanography,includingidentifyingprocessescon- 2012).Theregionhasarichhistoryofpaleo-oceanographicresearch trolling rates of primary production; examining how plankton (Black et al., 1999; Hughen et al., 2004) that provides additional communitystructureinfluencesproductivityandmaterialexport; motivationforcontemporarytime-seriesstudieslinkingupperocean quantifying nutrient supply to the upper ocean; and defining biogeochemical processes to climate signals stored in seafloor seasonal to interannual scale variability in stocks of oxygen, sediments. carbon, and nutrients essential for life on Earth. The time-series Thesuccessesofthetime-seriesprogramsmadethemamong sampling approach allows studying ecosystem behavior that themosttransformativeaccomplishmentsofJGOFS.Bytheendof wouldbeotherwiseobscuredbyhigherfrequency‘‘noise’’ortime the JGOFS program in the early 2000s, HOT, BATS, and CARIACO lagsbetweenperturbationsandresponses(Magnuson,1990). found themselves lacking a unified programmatic base to facil- The scientific and logistical successes of these early studies itate exchange of ideas on science priorities pursued by these stimulatedrecognitionthatoceanecosystemspreviouslyconsidered programs.Theinitiationof theOCBprogramin2007 provideda toberelativelystaticexhibitsignificanttemporalvariabilityovera scientific support framework whose research interests aligned varietyofscales.Moreover,theseprogramshighlightedthatefforts well with these on-going time-series efforts. HOT, BATS, and todocumentchangestothesearequireawell-formulatedplanfor CARIACO remain focused on studying processes that control the sustained, long-term ocean time-series research. By 1984, ocean distributions and cycling of elements in the sea, with specific scientistsworkingundertheumbrellaoftheScientificCommitteeof focus on carbon, in sufficientdetail to provide predictiveunder- OceanicResearch(SCOR)formulatedplansforanewoceanographic standing on how global scale perturbations to ocean-climate programtodocumentanthropogenicallyinducedchangestothesea. might influence biogeochemical transformations and feedbacks. TheGlobalOceanFluxStudy(GOFS)soughtbetterunderstandingof Toachievethisbroadobjective,theprogramsseekunderstanding the processes controlling ocean biogeochemistry at regional to ofthefollowing: global scales (Global Ocean Flux Study, 1984). With international partnerships in place, the Joint Global Ocean Flux Study (JGOFS) emerged as one of the first core projects of the International 1) The linkages between seasonal, interannual, and long-term Geosphere-Biosphere Program (IGBP). Central to the objectives of (multi-decadal)variabilityandtrendsinoceanphysics,chem- JGOFSwastheneedtodeterminetheinteractionsamongelemental istry,andbiology. cycles in the sea and understand the processes controlling time- 2) Processes underlying physical and biogeochemical temporal varying fluxes of carbon and associated bioelements. Ocean time- variability. series were included as an essential component of this interdisci- 3) Theroleofphysicalforcingoncarbonfluxes,includingratesof plinaryprogram(Brewer,2003).AtthesametimethatJGOFSwas biologically mediated carbon transformations, air–sea CO 2 formulatingitsscienceagenda,theWorldClimateResearchProgram exchange,andcarbonexport. (WCRP) was developing plans for the World Ocean Circulation 4) The response of ocean ecosystems and biogeochemistry to Experiment (WOCE), a program centered on observations and planetarychange. models of ocean-climate change. The programs collectively recog- nizedtheneedforanintegrated,interdisciplinaryapproachtoocean observing that included relying on time-series sampling. In 1987, Thescientificandlogisticalsupportaffordedbytheseprograms three separate proposals (two to JGOFS and one to WOCE) were continues to generate activities that serve as focal points for new submitted to the US National Science Foundation (NSF) for the science,education,andpublicoutreach.Thetimes-seriessiteshave establishmentofHOTandBATS.StartinginOctober1988bothHOT proven fertile grounds for improving existing methodologies and and BATS were in the water, undertakingnear-monthly shipboard implementing novel sea sensing technologies (Fig. 1). The short time-series sampling in the NPSG and Sargasso Sea, respectively duration (o1 week) cruises in globally significant but remote (KarlandMichaels,1996;MichaelsandKnap,1996). habitatscontinuestoattracttheinterestsofdiversescienceprojects HOT and BATS were born from the idea that time-series are thatbenefitfromatime-resolvedsamplingapproach.Thecoretime- essentialtounderstandtime-varyingfluxesofcarbonandassociated seriesbenefitfromtheknowledgeofecosystemdynamicsprovided vitalelementsintheoceans.ThecoreelementsofbothHOTandBATS by such ancillary research projects, while the ancillary projects fellundertheauspicesofJGOFSandWOCEthroughoutthelifetimes benefit from time-resolved scientific context and logistical, infra- of these larger programs. The time-resolved physical and biogeo- structural, and technical support provided by the core programs. chemical context and regular access to the open sea stimulated As a result, HOT, BATS, and CARIACO have attracted numerous numerous‘‘ancillary’’researchprojectsthatcontinuetocollectawide scientists,students,teachers,andvolunteersfromallovertheworld range of observations at both locations. Within 5 years it became seekingopportunitiestoparticipateintime-seriesresearchorjustto apparentthatoneofthemajorstrengthsoftheseprogramswasthe experience science from aboard a research vessel. Between 1988 interdisciplinary,multi-investigatorapproachtostudyingecosystem and 2009, more than 320 scientists and their staff have been dynamics. The resulting high-quality measurements, together with involved in process studies that build on the ‘‘core’’ time-series the scientific and ship-based infrastructure fueled numerous colla- programs,andanadditional 4140‘‘core’’time-seriesscientistsand borativescientificinteractions.Betweenthelate1980sandthemid- staffhaveparticipatedinHOT,BATS,andCARIACOcruises.Over420 1990s, many time-series programs were initiated in various undergraduate and graduate students and (cid:3)50 teachers marine ecosystems around the globe (Fig. 1) including Monterey (elementary to university level) from around the globe have and Bay(MBARI,361430N,1221240W;1989–present),intheMediterra- continue to participate in these cruises for education and training nean (DYFAMED, 431250N, 71 520E; 1991–present), in the northeast opportunities.Theseprogramshavebecomemodelsbywhichother Atlantic(ESTOC,291100N,151300W;1994–present),andinthesemi- nationsaredevelopingtheirowntime-seriesprogramstoassistin enclosed Cariaco Basin (CARIACO; 1995–present). CARIACO was understanding local ecosystems and responses to impacts and 6 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 change, and which will ultimately help us to understand global- hydrographic, biogeochemical, and ecological characteristics at HOT scaleoceanchange. and BATS, and both sites experience variability in upper-ocean dynamicsthatalterbiogeochemicaldynamicsandplanktoncommu- nity structure across a range of time scales. BATS features higher- 3. Programhighlights salinity waters from the surface to the ocean bottom, compared to thosefoundatHOT(Fig.2).Moreover,themid-anddeepwatersat 3.1. Fromthepredictabletotheunexpected HOT (41000m) have depleted concentrations of dissolved oxygen and enriched concentrations of nutrients compared to BATS, both Allthreetime-seriesprogramsmeasureacoresetofphysicaland signatures consistent with greater time-integrated organic matter biogeochemicalpropertiesoneachcruise(http://hahana.soest.hawaii. remineralization characteristic of older deep waters in the North edu/hot/; http://bats.bios.edu/; http://www.imars.usf.edu/CAR/index. Pacific(Fig.2). html);thesemeasurementswereselectedtoprovideacomprehen- The physical and biogeochemical characteristics of CARIACO sive and interdisciplinary framework from which to view contrastthoseobservedatHOTandBATS(Fig.2).WhileHOTand time-varying changes in these oceanic ecosystems. The long list of BATS both sample deep-ocean locations (44700m), CARIACO highlights emerging from the time-series data records include doc- lies on a geological fracture on the continental shelf (bottom umentationofprogressivechangesinoceaniccarboninventoriesand depth (cid:3)1400m) off the coast of Venezuela in an permanently fluxes(Astoretal.,2005,2013;Bates,2001,2007;Doreetal.,2003, anoxicbasin.Thephysicalandbiogeochemicalconditionsatthis 2009; Keeling et al., 2004); unexpected variability in the elemental site are defined by factors influencing oceanography of the stoichiometryofseawaternutrientpools(Karl,2002;Michaelsetal., tropical and subtropical Atlantic, the Caribbean Sea, and the 1994,1996;Thunelletal.,2008);complexclimate-linkedinteractions continental margin. High near-surface biological productivity betweenplanktonecologyandbiogeochemistry(Cornoetal.,2007; and settling particulate organic matter flux leads to anoxic Karl, 1999; Lomas et al., 2010; Taylor et al., 2012); elucidation subsurfacewatersintheCariacoBasin.CARIACOisamesotrophic of variability associated with pools and fluxes of organic matter tropical ecosystem that experiences seasonally dynamic trade- (Carlson et al., 1994; Emerson et al., 1997); and the importance of wind forced upwelling (January through May), and episodic plankton community structure in controlling time-variability in delivery of terrestrially derived organic matter in the autumn carbonsequestration(Dore etal.,2002;Lomasetal.,2009;Thunell (Hoetal.,2004).Anoxiabelowapproximately250miscausedby etal.,2007). physical isolation of the deep waters in the basin and the Onthesurface,HOTandBATSsamplesimilaroceanhabitats:both relatively high biological activity. The lack of oxygen in this sites are located in relatively warm and isolated subtropical regionofthewatercolumn(4250m)promotesfinelystructured gyres where Ekman downwelling associated with the anticyclonic verticalredoxgradients(Scrantonetal.,2001;Tayloretal.,2001), rotationofthegyresresultsindeeppermanentpycnoclinesandnear- includingcompleteremovalofnitratebydenitrification. surfaceoceanchlorophyllconcentrationsarepersistentlylow(Fig.2). HOT, BATS, and CARIACO all rely on near-monthly ship- Despitethesebroadsimilarities,therearefundamentaldifferencesin boardsampling.Theresultingtime-seriesdatacaptureseasonally Fig.2. StudylocationsofHOT(redcircle),BATS(bluediamond),andCARIACO(greentriangle)superimposedover6-yearcompositeofsatellite-derivednear-surfaceocean chlorophyllconcentrations.Bottompanelsdepicttemperature–salinityrelationships,andverticalprofilesofdissolvedoxygenandnitrateþnitritefromthethreetime- seriessites.SatellitedatacourtesyoftheOceanBiologyProgram(NASAGoddardSpaceFlightCenter).(Forinterpretationofthereferencestocolorinthisfigurecaption, thereaderisreferredtothewebversionofthisarticle.) M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 7 recurringpatternsinbothhydrographicforcingandbiogeochem- decrease to levels similar to those measured at HOT during the ical dynamics, withboth higher-and lower-frequencyvariability warm,stratifiedsummermonths(Fig.3).Nutrientconcentrations superimposedon that seasonaldynamic. Although allthree sites inthenear-surfacewatersatCARIACOaretypicallyrelativelylow, undergo relatively weak seasonal variability in sea surface tem- withNO3(cid:2)þNO2(cid:2) generally o1mM(Astoretal.,2003);however, peratures (SST), the amplitude and timing of SST variations at the strong vertical gradient in nutrient concentrations, together BATS are notably different than observed at HOT or CARIACO. with seasonal upwelling, yields upper ocean (0–100m) nutrient Near-surface ocean temperatures at HOT generally vary o51C inventories that are several hundred times greater than those over the course of the year, while SST at CARIACO varies about observedatHOTorBATS(Fig.3). 5–81C, and at BATS SST can vary by 491C (Fig. 3). All three Ratesofprimaryproduction(asestimatedfrom14C-bicarbonate regionsalsodemonstrateseasonalityinthedepthofupperocean assimilation) and particulate matter export at all three sites mixing. ThemixedlayeratHOTandCARIACOdisplays relatively demonstrate variability on seasonal to interannual scales weakseasonality(Fig.3);atHOTthemixedlayerisalmostalways (Figs. 4 and 5). Both production and export (and hence rates of restrictedtotheupper120m(andhencewithintheeuphoticzone newproduction)aresensitivetochangesinplanktoncommunity (cid:3)125m; Letelier et al., 2004) and at CARIACO the perennially structure and to interannual variations in hydrographic forcing warmandsalineupperoceanwatersrestrictsmixingto o50m, (Chavezetal.,2011;Cornoetal.,2007;Letelieretal.,1996;Muller- whichisoftheorderoftheeuphoticzonedepth(Lorenzonietal., Karger et al., 2001; Saba et al., 2010). Moreover, the emerging 2011).Incontrast,mixed-layerdepthsatBATScanexceed400m seasonalclimatologiesinproductionandparticulatecarbonexport inthelatewinter(andthuscanseasonallyexceedthedepthofthe at these sites highlights several patterns reflective of the unique euphoticzone (cid:3)100m;Siegeletal.,2001),thenshoalrapidlyto biological responses to annually recurring ecosystem dynamics o30mbythelatespringandsummer(Fig.3). (Fig.4).Consistentwiththerelativelyquiescentphysicalnatureof These seasonal differences in hydrographic forcing imprint the NPSG, primary production (0–100m) at HOT typically varies unique biogeochemical signatures on each region. The relatively (cid:3)2-foldover theyear((cid:3)30and50mmolCm(cid:2)2d(cid:2)1)(Fig. 5). A weak seasonal mixing combined with rapid plankton growth weak but predictable seasonal dynamic is observed where rates results in an upper ocean at HOT that is consistently starved of increaseduring the summerwhen irradiance ismaximaland the nutrients.Depth-integrated(0–100m)inventoriesofNO(cid:2)þNO(cid:2) upper ocean is well stratified. Particulate matter flux (150m) 3 2 at HOT are low throughout the year (Fig. 3), but become more increases o2-foldduringthemoreproductivespringandsummer variable during the winter periods when mixed-layer depths months (Fig. 5), resulting in a weak seasonal-scale coupling increaseandincidentirradiancereachesitsannualminima(Karl betweenproductivityandexport.Incontrast,upper-oceanproduc- etal.,2008;Letelieretal.,2004).Incontrast,nutrientinventories tivity(0–100m)andexport(150m)atBATSincreaseintheearly atBATSshowmoreprominentseasonality(Steinbergetal.,2001); spring when nutrient inventories are at their annual maximum. upperocean(0–100m)NO(cid:2)þNO(cid:2) concentrationsincreaseshar- The amplitude of the seasonal cycle in productivity at BATS is 3 2 ply during periods of late winter mixing, and then rapidly largerthanobservedatHOT,withprimaryproductionduringthe Fig.3. Meanmonthlyseasurfacetemperatures,mixed-layerdepths,anddepth-integrated,upperocean(0–100m)inventoriesofnitrateþnitriteatHOT,BATS,and CARIACO.Symbolsrepresentmonthlymeans,errorbarsarestandarddeviationsofthemonthlymeans. 8 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 Fig.4. Time-seriesdeterminationsofnetprimaryproduction(0–100mdepth-integratedrates)andparticulatecarbonexportatHOT(red),BATS(blue),andCARIACO (green).NotethatcarbonexportatHOTandBATSweremeasuredbyparticleinterceptortrapsat150m;fluxesatCARIACOweredeterminedusingbottommoored sedimenttrapcollectionsat225m.Solidlinesdepictmeanproductivityorexportforthetime-series;dashedlinesrepresent7onestandarddeviationofthemeanfluxes. NoteY-axesscalesdifferforCARIACOdata.(Forinterpretationofthereferencestocolorinthisfigurecaption,thereaderisreferredtothewebversionofthisarticle.) spring bloom periods sometimes 480mmol C m(cid:2)2 d(cid:2)1. By late relativelystrongseasonalityinupperoceanproductivityseason- spring and early summer production decreases sharply, often alityinparticleexportislesswelldefinedthanateitherHOTor remaining o30mmolCm(cid:2)2d(cid:2)1throughoutthesummerandfall BATS(Fig.5). (Fig.5).Despitenotableseasonaldifferencesinupper-oceannutrient Someofthemostimportantcontributionstoemergefromthe availabilityandgreatervariabilityinprimaryproductionandexport, ocean time-series programs are reconstructions of biogeochem- onanannualbasisratesofproductionandparticulatematterexport ical rate processes based on annual mass balances of properties at HOT and BATS are comparable, with net primary production suchasdissolvedoxygen,dissolvedinorganiccarbon,nitrate,and averaging (cid:3)14mol C m(cid:2)2yr(cid:2)1 and particulate carbon export nitrogen and carbon isotopes. Among other processes, such (150m)averaging (cid:3)0.8molCm(cid:2)2yr(cid:2)1,respectively(Table1). approacheshaveprovidedinsightintovariabilityassociatedwith Upper-ocean primary production (0–100m) at CARIACO is annualratesofnetorganicmatterproduction(ornetcommunity considerably greater than that observed at either HOT or BATS, production–NCP).AccuratedeterminationsofNCPrequiretime- withratestypicallyrangingbetween (cid:3)80–145mmolCm(cid:2)2d(cid:2)1 resolved measurements, and hence there are few places in the butpunctuatedbyperiodsofveryhighproductivity(4200mmol world’soceanswheresuchmeasurementsexist.Overthehistory Cm(cid:2)2d(cid:2)1)inthelatewinterandearlyspring(Figs.4and5).The ofHOTandBATSavarietyofsamplingandanalyticalapproaches observed seasonal cycle of primary production at CARIACO have been used to determine NCP, yielding some of the most reinforces the importance of seasonal-scale changes in trade- comprehensive estimations of NCP anywhere in the world’s winddrivenupwellingasamajorcontrolonecosystemvariability oceans (Table 2). Such efforts have been strengthened by use of inthisregion(Muller-Kargeretal.,2004;Thunelletal.,2000).The instrumented remote and autonomous sampling platforms origin of material sustaining particulate material export at CAR- (moorings,floats,gliders),providinginformationonthecoupling IACO derives from both autochthonous productivity and terres- betweenhigh-frequencyphysicaldynamicsandvariabilityinnet trial material introduced through riverine discharge (Lorenzoni organic matter production and export. Together these time- etal.,2011;Montesetal.,2012;Muller-Kargeretal.,2004).Both resolved sampling approaches have helped transform our view these processes dominate export at different times of the year, oftemporalvariabilityassociatedwithorganicmatterproduction with productivity peaking in the late winter and spring during and carbon sequestration in this open-ocean habitat. The periodsofstrongupwelling,andriverineinputgenerallyincreas- emerging data suggest annual NCP at BATS and HOT may be ing in the summer (Thunell et al., 2007). As a result, despite similar,ranging2.3–4.7molCm(cid:2)2 yr(cid:2)1 and1.1–4.1molCm(cid:2)2 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 9 Fig.5. Meanmonthlyprimaryproductionandupper-oceanparticulatecarbonexportatHOT,BATS,andCARIACO.Symbolsrepresentdepth-integrated(0–100m)ratesof productivityorparticleflux(150mforHOTandBATS,225mforCARIACO);errorbarsarestandarddeviationofthemonthlymeans.NotedifferencesinY-axes. Table2 RangeofannualestimatesofnetorganicmatterproductionandexportatHOTandBATS.Whereavailableuncertaintiesassociatedwiththedeterminationsareshownin parentheses. Timeseries Method Rates Periodof References site molCm(cid:2)2 measurements yr(cid:2)1 HOT MixedLayerO2þArbudgets 1.1–3.7(7 1992–2008 Emersonetal.(1997),HammeandEmerson(2006),JuranekandQuay 1.0) (2005),Quayetal.(2010) DICþDI13Cbudgets 2.7–2.8(7 1988–2002 QuayandStutsman(2003),Keelingetal.(2004),Brixetal.(2004) 1.4) MooringO2/N2 4.1(71.8) 2005 Emersonetal.(2008) Sub-mixedlayerfloatprofiles 1.1–1.7 2003–present RiserandJohnson(2008),Johnsonetal.(2010) (70.2) Sub-mixedlayerglidersurveys 0.9(70.1) 2005 Nicholsonetal.(2008) 150msedimenttrapsþmigrant 0.9–1.1 1989–2011 HOTcoredata;Emersonetal.(1997),Landryetal.(2001),Hannidesetal. zooplanktonflux (2009) BATSand SubmixedlayerO2þAr,Hebudgets 4.3(70.7) 1985–1987 SpitzerandJenkins(1989) StationS DICþDI13Cbudgets 2.3(70.9) 1991–1994 Gruberetal.(1998) 3HeandNO3(cid:2)fluxes 2.9–4.5 1985–1987, JenkinsandDoney(2003,1992) Sub-mixedlayerO2profiles 3.3–4.7 1961–1970 JenkinsandGoldman(1985),JenkinsandWallace(1992) 150msedimenttrapsþmigrant 0.6–2.0 1989–2011 BATScoredata;Lomasetal.(2002),Steinbergetal.(2000,2001,2012)) zooplanktonflux yr(cid:2)1, respectively (Table 2). Such results are intriguing given (Laws et al., 2000). Such discrepancies provide sobering remin- differences in the timing and magnitude of vertical delivery of ders that there remain a number of unresolved issues central to nutrients to the upper ocean observed at these sites (Fig. 3). ourunderstandingofoceanecosystemfunctioning. Moreover,thesegeochemicalmassbalancesappearasmuchas4- Among the most recognized biogeochemical measurements fold greater than predicted based on sediment trap estimates of conducted by these programs are those documenting time varia- carbon export (Table 2), and estimates derived from ocean bilityassociatedwithseawaterCO (Fig.6).Thesemeasurementsat 2 circulation (Schlitzer, 2004) and satellite ocean-color models HOT, BATS, and CARIACO, together with measurements conducted 10 M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 Fig.6. Meanannualnear-surfaceoceanpCO2(redcircles)andseawaterpH(bluediamonds)atHOT,BATS,andCARIACO.Errorbarsdepictstandarddeviationsofannual means.(Forinterpretationofthereferencestocolorinthisfigurecaption,thereaderisreferredtothewebversionofthisarticle.) as part of the ESTOC and MBARI time-series programs, comprise high-qualitymeasurementsoverlongperiodsoftimeisthefounda- some of the most robust decadal-scale datasets available for tion of success for these programs, and retaining well-trained describing the response of the ocean to progressive increases in scientists and technical staff are essential to these efforts. While atmospheric CO . These measurements indicate that over annual certifiedreferencematerialsprovideameanstotracethequalityof 2 time scales, HOT and BATS are both weak to moderate sinks for an oceanographic measurement, for many of the biogeochemical atmospheric CO (Bates et al., 1996; Winn et al., 1998), while measurements (particularly most rate measurements), no certified 2 CARIACO is a net source of CO to the atmosphere (Astor et al., standards exist. Furthermore, developing confidence that measure- 2 2005,2013).Time-seriesmeasurementsatHOT,BATS,andCARIACO mentsconductedatdifferentsitesareintercomparabledemandsthat documentprogressiveincreasesinthepartialpressureofCO (pCO ) theseprogramscontinuetoregularlyparticipateincommunity-wide 2 2 in the near-surface ocean with concomitant decreases in seawater effortsdirectedtowardstandardizingmethodologiesandanalyses. pH(Fig.6).Thelong-termincreaseinseawaterpCO atHOT,BATS A significant scientificchallengehas beento design sampling 2 andCARIACOappearsimilartotherateofCO accumulationinthe schemesappropriatetocapturingimportantmodesofecosystem 2 atmosphere ((cid:3)1.7matm yr(cid:2)1), with concomitant long-term variability. Despite near-monthly sampling schedules these pro- decreasesinseawaterpHranging(cid:2)0.0017to(cid:2)0.0019y(cid:2)1.Despite grams spendo20% of the year on site observing ecosystem similarlong-termtrends,thereissubstantialinterannualtoseasonal dynamics, implying that some important processes will be scale variability in upper ocean CO attributable to local- and under-sampled in both space and time (Wiggert et al., 1994). 2 regional-scale ecosystemdynamics(Astor et al., 2013; Bates etal., Episodic events and processes that exert high frequency varia- 1996; Dore et al., 2003; Gruber et al., 1998, 2002; Keeling et al., bility can be missed with this sampling strategy (Levin, 1992; 2004). Interannual variations in pCO and pH at all three sites Munk,2000).Moreover,limitedsamplinginthespatialdomainis 2 depends on regional to basin scale fluctuations in ocean-climate also an issue that can lead to aliasing of Eulerian processes connectivityandbiologicalactivity,withvariationsintemperature, attributable to horizontal advection, thereby hindering differen- evaporation–precipitation, and upper ocean mixing all imparting tiating time-dependent local changes from those attributable to characteristic signatures onCO system dynamics in these regions spatialvariabilityontheregionalscale. 2 (Astoretal.,2005;Bates,2007;Doreetal.,2003,2009). 4.2. Transformativetechnologies 4. Chartingthefuturecourse The science of observing the ocean has made huge advances since the 1980s. While the disciplinary expansion of oceanogra- 4.1. Wherethefuturemeetsthepast phy in the 20th century was largely propelled by ship-based expeditionaryscience,thenewgenerationofoceanobservations Ocean time-series programs provide critical long-term records capitalizesontechnologicaladvancesinremoteandautonomous needed for assessing the interactions between chemistry, biology, sensingplatforms.FollowingtheendofWorldWarII,duringthe physicsandgeology.Thescientificvalueoftheseprogramscontinues eraoftenreferredtoas‘‘goldenageofoceanography’’,shipboard toincreasethroughsustainedobservations.However,theexpansion researchstoodastheprimarymeansofgatheringinformationon ofoceantime-seriesresearch,bothindurationofindividualprograms the ocean. However, the ‘‘golden age’’ of oceanography rapidly and the number of programs, during the 1990s taught important yielded to the ‘‘era of the electron’’ where improved platform lessons,somerelatedtoscience,othersontherealitiesandlogisticsof engineering, data storage, sensor stability and durability, and sustaining long-term science programs (Karl, 2010). Time-series communications technologies have thrown the door wide open programs are resource intensive and thus maintaining ocean com- tonewtechnologiesforsensingthesea.Anewinternationalwave munity support for these programs requires strong and positive of contemporary ocean observatories rely on measurements leadershipandadedicatedteamofpeoplethatunderstandthevalue conducted from instrumented remote sensing platforms includ- of long-term observations. The requirement to maintain consistent, ing satellites, moorings, floats, and various autonomous vehicles M.J.Churchetal./Deep-SeaResearchII93(2013)2–15 11 (e.g. http://www.eurosites.info/index.php; http://www.oceanob ideal for developing, testing, and implementing novel ocean- servatories.org/). Advances in sensor technologies now provide observing technologies. Moreover, such projects demonstrated small, low-power, stable instrumentation that can be outfitted that enhancing the observational capacities of the time-series onto diverse ocean platforms (Johnson et al., 2009; Perry and programs does not necessarily require expansion of the existing Rudnick, 2003). Such technological advances therefore have shipboard programs. Rather, the shipboard programs and their provided new opportunities to enhance observational capacity long-termdatarecordsserveastheunifyingcorestructurefrom aroundshipboardprograms. which new science directions and observational technologies There are numerous examples of studies that have leveraged arebuilt. ship-based time-series programs with higher frequency, spatio- Overthenextdecade,theshipboardprogramsmustcontinue temporally resolving, autonomous and remote measurements to be proactive about promoting the implementation of new (Conte et al., 2003; Emerson et al., 2002, 2008; Johnson et al., ocean sensing technologies at these sites. However, despite the 2010; Letelier et al., 2000; McGillicuddy et al., 1998; Muller- growinglistofpotentialapplicationsforremoteandautonomous Karger et al., 2004; Nicholson et al., 2008). Such studies yield sensingofoceandynamics,thereremainseverallargehurdlesto further insightinto scalesof variability associated with plankton be overcome before ship-based time-series become obsolete. metabolism and biomass, carbon export, nutrient fluxes, water- Currently no combination of autonomous or remote sensing massventilation,andair–seagas exchanges,toname a few. Ina technologies could be employed to replace the full suite of recentexample,withsupportfromNSFandtheNationalOceano- high-qualitymeasurementsroutinelyconductedaspartofinter- graphicPartnershipProgram,SteveRiser(UniversityofWashing- disciplinary shipboard time-series programs. In many cases the ton) and Ken Johnson (MBARI) instrumented quasi-Lagrangian, long-termaccuracyofsuchsensorsremainsunknown,challenged vertically profiling floats with nitrate, oxygen, fluorescence, and in part by non-trivial issues associated with biofouling and backscatter sensors (http://www.mbari.org/chemsensor/floatviz. instrumentstability(Dickey,1991;Johnsonetal.,2009).Perhaps htm). Several of these floats have now been deployed at HOT, most importantly, there are currently a limited set of sensors BATS,andStationP,providingnewtoolsforexaminingsimilarities readily available for detecting many of the key biological and anddifferencesinecosystemprocessessuchasnutrientsupplyto chemicalpoolsandfluxesknowntobeclimatesensitiveandplay theeuphoticzone(andhighlightingpotentiallyunderappreciated roles in the ocean carbon cycle. While numerous ‘‘in water’’ mechanismssuchphytoplanktonverticalmigration),netcommu- sensors are currently available and widely used for detecting nityproduction,andlengthscalesoforganicmatterremineraliza- ocean hydrographic variability, sensors for autonomous and tion(Martzetal.,2008;RiserandJohnson,2008).Inaddition,with remote detection of ocean biogeochemistry, beyond nutrient support from the Gordon and Betty Moore Foundation and NSF, and oxygen dynamics, have proven more difficult to develop ocean gliders have been in service at or around Station ALOHA andimplement.Forexample,todate,therearefewtoolsavailable since 2008, providing insight into spatiotemporal variability in for remote quantification of plankton community structure temperature, salinity, oxygen, backscatter, and fluorescence (although see Scholin et al., 2009). Moreover, although sensors (http://hahana.soest.hawaii.edu/seagliders/index.php). These pro- foropticalbaseddeterminationsofnitrateareavailable(Johnson grams highlight a few examples where remote or autonomous andColetti,2002),sensorsfordetectionofphosphate,silicate,and sensing approaches leveraged the shipboard programs to test dissolvedorganicorinorganiccarbonarenotyetwidelyavailable. hypothesesgeneratedfromthehistoricaltime-seriesdata. Thereare stillfewerinstrumentsthat canmakedirectmeasure- Hydrographic and biogeochemical moorings have been ments of ecosystem rate processes, versus the time-derivative deployed for substantial periods of time at each of the OCB geochemical estimates of rate processes (e.g., Nn and related stations.TheBermudaTestbedMooring(BTM)operatedformore variables). Such measurements are fundamental to informing than a decade (1994–2007) near the BATS site, forging new, our understanding of plankton ecology and ultimately biogeo- collaborative science partnerships that informed understanding chemicalcontrolsoncarbonsequestrationinthevastoceangyres. of episodic physical forcing on ocean biogeochemistry (Bates etal.,1998;Dickeyetal.,1998;McNeiletal.,1999).Since1997, 4.3. Improvedprocess-levelunderstandingandlinkagesto varioushydrographic,biogeochemical,andmeteorologicalmoor- ecosystemmodels ings have been maintained at or near Station ALOHA, including HALE ALOHA (1997–2000), MOSEAN (2004–2007), and WHOTS The existing ocean time-series records clearly demonstrate (2004–present). These moorings have provided critical observa- biogeochemical and hydrographic variability in ecosystem tions for understanding high-frequency variability in ecosystem dynamicsoccurringoveralargerangeoftimescales.Concentra- processes (Church et al., 2009; Emerson et al., 2002; Karl et al., tionsofchlorophyll,ratesofprimaryproduction,nutrientinven- 2003). Recently, scientists and engineers from the University of tories and stochiometries, export of organic material from the Hawaii led the successful installation of the ALOHA Cabled upper ocean, and stocks of organisms and plankton community Observatory (http://aco-ssds.soest.hawaii.edu/ACO/index.php). structure have all been shown to vary over seasonal to sub- By takingadvantage of anexisting seafloor fiber optic cable, the decadal time scales. However, in many of these examples, the ACO provides a seabed node for powering instruments and mechanisms underlying the observed time-varying changes transmitting data. Such infrastructure has the promise to trans- remain obscure. Long-term increases (HOT, BATS) or decreases form thesea sensing capabilities at StationALOHA. At CARIACO, (CARIACO) of primary production and inventories of chlorophyll subsurface hydrographic and current moorings have been reportedatthesesiteshavebeenattributedtobasin-scaleclimate deployed sporadically over periods spanning several years fluctuations (Chavez et al., 2011; Corno et al., 2007; Saba et al., (Alvera-Azca´rate et al., 2008), and moored sediment traps have 2010; Taylor et al., 2012); however, our understanding of the beensamplingroutinelysince1996(Benitez-Nelsonetal.,2007; processes linking ocean–climate to changes in seawater biogeo- McConnelletal.,2009;Tedescoetal.,2007;Thunelletal.,2007). chemistry remains rudimentary. Alteration in phytoplankton These previous and ongoing efforts to enhance the observa- productivityandbiomass atthesesites couldstem frombottom tionalcapacitiesatthetime-seriessiteshaveallreliedheavilyon up processes such as changes in light (as a consequence of the scientific and logistical infrastructure afforded by the ship- changes in upper ocean stratification) or changes in nutrient board programs. The many years of developing highly skilled supply (Bidigare et al., 2009). Alternatively, temporal variations workforcesandcapableinfrastructuremakethetime-seriessites in various top-down processes could control plankton biomass