Science in support of the Deepwater Horizon responseJane Lubchencoa,1, Marcia K. McNuttb, Gabrielle Dreyfusa,2, Steven A. Murawskia,3, David M. Kennedya, Paul T. Anastasc,4,Steven Chud, and Tom Huntere,5aNational Oceanic and Atmospheric Administration (NOAA), Department of Commerce, Washington, DC 20230; bUS Geological Survey,Department of the Interior, Reston, VA 20192; cOffice of Research and Development, US Environmental Protection Agency, Washington, DC20460; dUS Department of Energy, Washington, DC 20585; and eSandia National Laboratory, Albuquerque, NM 87111

Edited by Paul G. Falkowski, Rutgers, State University of New Jersey, New Brunswick, NJ, and approved October 26, 2012 (received for review April 26, 2012)

This introduction to the Special Feature presents the context for science during the Deepwater Horizon oil spill response, summarizeshow scientific knowledge was integrated across disciplines and statutory responsibilities, identifies areas where scientific informationwas accurate and where it was not, and considers lessons learned and recommendations for future research and response. Scientificinformation was integrated within and across federal and state agencies, with input from nongovernmental scientists, across a diverseportfolio of needs—stopping the flow of oil, estimating the amount of oil, capturing and recovering the oil, tracking and forecastingsurface oil, protecting coastal and oceanic wildlife and habitat, managing fisheries, and protecting the safety of seafood. Disciplinesinvolved included atmospheric, oceanographic, biogeochemical, ecological, health, biological, and chemical sciences, physics, geology, andmechanical and chemical engineering. Platforms ranged from satellites and planes to ships, buoys, gliders, and remotely operated vehiclesto laboratories and computer simulations. The unprecedented response effort depended directly on intense and extensive scientificand engineering data, information, and advice. Many valuable lessons were learned that should be applied to future events.

science-based decision making | Gulf of Mexico | Spill of National Significance | Macondo | Oil Pollution Act

“We are fighting an omnidirectional, al-most indeterminate threat here. We aretrying to protect the entire Gulf Coast at thesame time.”

Coast Guard Commandant Thad Allen, May18, 2010 before the Senate Committee onCommerce, Science, and Transportation

The Deepwater Horizon (DWH)disaster, which began on April20, 2010 with a blowout of BPExploration and Production,

Inc.’s Macondo well located in lease blockMC252 in water ∼1,500 m deep and 84 kmfrom Venice, Louisiana, resulted in thelargest mobilization of resources to ad-dress an environmental emergency in thehistory of the United States. From day 1 towell shut in on day 87 and the present(timeline in Fig. S1), we oversaw or as-sisted the tactical and strategic responsesof our agencies and supported the overallUS Government’s effort. The papers inthis Special Feature focus on how scien-tific information was used to informthe response.Although the US Coast Guard (USCG),

the National Oceanic and AtmosphericAdministration (NOAA), and the Envi-ronmental Protection Agency (EPA) werewell-versed in oil response and remediationand the National Response Team respon-ded quickly, the level and scope of DWHtaxed our organizations in unprecedentedways, requiring substantial mobilizationof resources and people inside and outsideof the government and extraordinaryinteragency coordination.As Admiral Allen’s quote emphasizes,

the situation of the Macondo blowout wasunprecedented, with oil spewing forth intoan extreme ocean environment—deep,cold, dark, and high pressure—but rapidlyspreading to midwaters, the surface, and

the atmosphere. These circumstancesresulted in a constantly changing set oflogistical and policy challenges (1) (Fig. 1).Although predicted as likely (2, 3), thepresence of deep suspended microscopic oildroplets was not part of BP’s federally ap-proved spill response plan and had neverbefore been confronted as an operationalresponse priority. Assessing the quantity ofsubsurface oil and its likely environmentalconsequences (e.g., threats to biota andpotential for inducing hypoxia) requiredconsiderable human and materialresources.Experience and response methods ap-

plicable for other oil spills in many casesproved either impossible to apply or in-effective. New science needed to be de-veloped and delivered rapidly to allowappropriate decisions and actions to betaken—day in and day out for months. Weapplied lessons from the Exxon Valdez OilSpill (EVOS), such as using caution inchoice of shore cleanup techniques toavoid doing more harm than good andpaying attention to the vulnerability ofjuvenile stages of living organisms. FromEVOS, we know that oil exposure has bothacute toxicity and chronic effects (4),especially for juvenile stages. However,DWH was different from EVOS; the latteroccurred in a cold environment near thesurface and the shore, and it entaileda different type and known quantity of oil.This introduction to the Special Feature

integrates much of the key science mobi-lized or conducted during DWH. Wehighlight ways in which science was usedin real time to inform the response andthe public. We draw on our experiences,papers in this Special Feature, and otherpublished literature. We highlight areaswhere scientific information used was laterdetermined to be accurate and areas whereit was not. We identify lessons learned

about use and communication of scientificinformation and priority areas for research.This Special Feature focuses on the

physical and environmental science andengineering that guided the response. Wedo not address environmental health, hu-man health, social impacts, or economicimpacts; as of November of 2012, they arestill being documented and evaluated aspart of the Natural Resource DamageAssessment (NRDA) and the Gulf LongTerm Follow-up Study of the NationalInstitute of Environmental HealthSciences, for example.To understand how science informed the

response, knowledge of the responsestructure and process is essential. The basiclegal authorities and responsibilities inplace to deal with a major oil spill stemfrom the Oil Pollution Act, which wasamended in 1990 (OPA90) and imple-mented through the National ContingencyPlan (NCP). Oil spill provisions of the NCPspecify a role in response for ResponsibleParties (RPs; here, BP): specifically, theresponsibility for costs associated with

Author contributions: J.L., M.K.M., G.D., S.A.M., D.M.K.,P.T.A., S.C., and T.H. analyzed data; and J.L., M.K.M., G.D.,S.A.M., P.T.A., and S.C. 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:jane.lubchenco@noaa.gov.

2Present address: Office of Policy and International Affairs,US Department of Energy, Washington, DC 20585.

3Present address: College of Marine Science, University ofSouth Florida, St. Petersburg, FL 33701.

4Present address: School of Forestry and EnvironmentalStudies, Yale University, New Haven, CT 06520.

5Retired.

This article contains supporting information online atwww.pnas.org/lookup/suppl/doi:10.1073/pnas.1204729109/-/DCSupplemental.

20212–20221 | PNAS | December 11, 2012 | vol. 109 | no. 50 www.pnas.org/cgi/doi/10.1073/pnas.1204729109

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

removal, cleanup, and claims resultingfrom discharge or threat of discharge andcosts of much of the response under thedirection of the federal government. TheUSCG, as the On-Scene Coordinator(OSC) for maritime spills, is charged withensuring that the RP takes appropriateaction. NOAA is designated by Congress tobe the scientific advisor to the OSC and actas the Scientific Support Coordinator forscientific issues that include “expertise inenvironmental chemistry, oil slick track-ing, pollutant transport modeling, naturalresources at risk, environmental tradeoffsof countermeasures and cleanup, and in-formation management.” (5)The NCP provides for an extraordinary

determination that the spill is one of na-tional significance on the basis of severalfactors, including a substantial threat topublic health or welfare. In this case, theUSCG established a National IncidentCommand that assumes specified duties ofthe OSC in directing the response. Underthe NCP, on day 10, DWH was declareda Spill of National Significance, giving theRP significant responsibilities under thedirection of the OSC and the newly namedNational Incident Commander (NIC),USCG Commandant Admiral Thad Allen.The NCP has been used to respond to

a number of oil spills but never one of thismagnitude and complexity. Moreover,DWH was unlike conditions envisioned inthe tanker-centric OPA90 (passed in thewake of EVOS). Drilling technology hadevolved significantly since OPA90 was

passed, but comparable advances in re-sponse capacity had not been achieved.A key role of the NIC was to coordinate

the whole of government response. Theresponse included multiple interagencycommand centers in Washington, DC andthree states (National Incident Command,Unified Area Command Center, and In-cident Command Centers in Houma, LA,Mobile, AL, and Miami, FL—moved toSt. Petersburg, FL). In addition, the NIC,a Senior Advisor to President Obama,and the US Government Principals (Sec-retaries of the Departments of Interior,Homeland Security, and Energy and theAdministrators of EPA and NOAA) metat least daily throughout most of the event.

Scientific Responsibilities and Timelineof Scientific SupportSome scientific responsibilities are pre-scribed in legislation or regulations (de-scribed above). However, the uniquenature of DWH posed unanticipated needsfor scientific information. In response tothese novel challenges, new scientificworking groups were formed. These groupswere established during the first severalweeks as the scope and scale of the accidentand the resulting oil spill began to emerge(Fig. 1 and Fig. S1). Most teams wereconstituted by the NIC under the auspicesof the Federal Interagency SolutionsGroup and drew on expertise within andoutside the federal government. They re-viewed and evaluated scientific and tech-

nical information, devised solutions tospecific problems, and advised the NIC,the Principals, and the President. Not well-integrated initially, these teams becamebetter coordinated in time through theNIC and the Principals’ group (6). Newscientific teams created for DWH includedthe following:

i) The Flow Rate Technical Group(FRTG), chaired by US GeologicalSurvey (USGS) Director MarciaMcNutt, was composed of more thansix research groups and independentexperts that developed assessmentsof the rate of escaping oil from theMacondo blowout.

ii) The Oil Budget Calculator Scienceand Engineering Team (Oil BudgetTeam), co-led by Bill Lehr of NOAA,Sky Bristol of USGS, and AntonioPossolo of the National Institute ofStandards and Technology and com-posed of government, academic, andindustry experts, estimated the amountof oil in different categories to enableeffective oil recovery efforts.

iii) The Government-Led Science Team(GLST), led by the Secretary of En-ergy Steven Chu and Tom Hunter(Sandia National Laboratories, retired)and consisting of Department of EnergyNational Laboratories and USGS, andoutside scientists developed technicalanalysis and paths forward for flow con-trol in close collaboration with BP andprovided an independent determinationof leak rate.

iv) TheOperationalScienceAdvisoryTeam,chaired initially by Carl Childs (NOAA)and later by Lieutenant CommanderKenneth Boda (USCG), directed andthen synthesized information on theamount of residual oil in the subsurface,on the bottom, and in the water columnto devise response and potential recov-ery strategies.

v) The Joint Analysis Group (JAG),chaired by NOAA Fisheries ChiefScientist Steven Murawski, analyzed theplethora of data produced by govern-ment, academic, and industry-sponsoredmonitoring activities to determine theconcentration, distribution, and im-pacts of subsurface oil, particularlythe evidence for dissolved oxygen de-pletion in subsurface waters as a resultof the DWH spill.

This paper provides an overview of theoverall response and emphasizes in-formation from the last two teams, otheragency’s scientific efforts, and papers inthis Special Feature. Results from effortsof the first three groups are presented inthe work by McNutt et al. (7).The interagency scientific coordinating

mechanisms relied heavily on new intra-agency coordination bodies. For example,

Fig. 1. Timeline of various science-related issues during the DWH oil pollution event. Some additionalscience-related issues continue beyond 2010.

Lubchenco et al. PNAS | December 11, 2012 | vol. 109 | no. 50 | 20213

SPECIA

LFEATURE:PERSP

ECTIV

ED

ownl

oade

d by

gue

st o

n M

arch

24,

202

0

NOAA stood up coordinating bodies tointegrate scientific efforts across its sat-ellite, weather, oceanographic, atmo-spheric chemistry, fisheries, seafoodsafety, protected species, habitat, oil spillcleanup expertise, research, mapping,impacts, and restoration units. Withpartners at the University of NewHampshire, NOAA adapted its spatiallyexplicit management tool, the EmergencyResponse Management Application(ERMA) (SI Text), for broad interagencyuse, populating it with relevant data andinformation to enable efficient evaluationof assets and needs as well as rapid de-cision-making. Consistent with policieson openness and transparency, the ObamaAdministration decided to make largeportions of ERMA available to the publicon day 56. On the first day of going public,ERMA (http://gomex.erma.noaa.gov) re-ceived 3.5 million hits—a powerful in-dicator of the intensity of public interestin DWH.Nongovernmental scientists played a vi-

tal role during the response. Traditionally,academic and private institutions providescientific input primarily through con-tracting mechanisms on various aspects ofspills, which are determined by needs ofscience agencies (e.g., documenting dam-age to natural resources through theNRDA process as outlined in OPA90).DWH triggered an unprecedented re-sponse by the academic community. Notonly were numerous academics undercontract by federal and state agencies andBP, but the National Science Foundationalso awarded 88 grants totaling over $11million in its Rapid Response program tostudy various aspects of the spill. Addi-tionally, on day 35 (May 24, 2010), BPcommitted $500 million over 10 y to funda broad independent research program, theGulf of Mexico Research Initiative, al-though the first funds were not available formonths. Additionally, numerous academicand private institutions launched in-dependent investigations. As describedabove, numerous academic scientists alsoparticipated in the ad hoc scientific workinggroups that advised the NIC.Given the enormous interest by the ac-

ademic community, there was difficulty inmobilizing unified approaches and com-munication systems. The White HouseOffice of Science and Technology Policy,federal agencies, universities, and partners,such as the Consortium for Ocean Lead-ership, convened a series of meetings withgovernment, academic, and independentresearch scientists (SI Text). Each meetingwas productive, but collectively, they wereinsufficient to satisfy the need for nearreal-time transmittal of new findings anddiscussions of priorities moving forward.Many independent and university scien-tists were frustrated, and misunderstand-ings resulted from scientists obtainingmost of their information from the newsmedia rather than scientific reports.

From the outset, scientific supportwas necessary to address a diverse andcontinually evolving suite of issues andpotential threats to worker safety, humanhealth, and ecosystems (Fig. 1). The re-sponse and mitigation of the spill grew toencompass over a dozen major sets ofscientific investigations, all requiring mul-tiple agencies, review mechanisms, andcollaboration with academic and in somecases, international partners. Balancingthe portfolio of resources (both humanand logistic) among these many diverseissues was a management challenge. Evencollectively, we did not possess all of theships, aircraft, laboratory access, or per-sonnel necessary to evaluate every issue tothe satisfaction of all.The following sections address major

questions that drove the scientific supporteffort. Where is the oil going? Shoulddispersants be used and if so, at the surface,at depth, or both? How should oil beremoved, captured, and cleaned up? Isseafood safe? How should wildlife andhabitats be protected? The work byMcNutt et al. (7) summarizes the parallelefforts to estimate the flow rate, controlthe source, test for well integrity, and de-termine the oil budget. We conclude byreflecting on improving the use of sciencefor future events with the hope that it willnever be needed.

Where Is the Oil? Where Will It Go?Where Did It Go?Tracking and predicting surface oil wasa clear priority, even before the extent ofthe Macondo blowout was known. Withinhours of the explosion, NOAA’s oil tra-jectory models and spot weather forecastsbegan providing at least daily guidance tofirst responders and then later, those in-dividuals involved in all aspects of the re-sponse. The location and amount of oil onthe surface of the water varied consider-ably from day to day (Movie S1). Earlymodeling of surface trajectories involvedthree separate models—by NOAA, TexasA&M University, and the Navy; thesemodels evolved into an ensemble of sixmodels providing short- and longer-termforecasts of surface trajectories for oil.These models used ever more sophisti-cated ground truthing to document distri-bution of surface oil and inform themodels with oceanographic data (oceano-graphic modeling for DWH is archivedat http://www.aoml.noaa.gov/phod/dhos/index.php, and detailed oceanographiccruise data are archived at http://www.nodc.noaa.gov/General/DeepwaterHorizon/ships.html). Satellite imagery (coordi-nated by the National Aeronautics andSpace Administration and NOAA) andimages and reports from fixed-wingand helicopter aircraft provided modelstarting conditions daily. Comparisonof results from the ensemble of surfacetrajectory models allowed evaluation offorecast uncertainty. SI Text and Fig. S2

have information about longer-termmodeling and the Loop Current.An oil budget was developed and

maintained by the NIC (8) specifically totarget efforts to recover oil (7). The oilbudget depended on estimates of flowrate, finally determined to be 62,000 ±10% barrels per day (bpd) at the beginningof the event and 53,000 ± 10% bpd onday 87 at well shut in (7, 9). Despite ag-gressive recovery and removal efforts, onlyaround one-quarter of the oil was removedby the federally directed response. Mostof the oil on the surface was too distrib-uted for efficient skimming and burning.The final tallies for recovered oil indicatethat around 5% was burned, 3% wasskimmed, and 17% was recovered directlythrough the riser pipe (8).Airborne, surface, and subsurface

chemical measurements (10) were used toindependently calculate a total hydro-carbon flow rate. Those analyses showedthat ∼5% by mass of the dischargedhydrocarbons evaporated to the atmo-sphere and 10% contributed to the sur-face slick; the balance dissolved ordispersed within the water column, withabout one-third directly detected in thedeep persistent plumes.Unexpected new methodologies to esti-

mate flow rate emerged (e.g., estimatingflow rate from air chemistry measure-ments). NOAA and academic scientistsflew a special mission to evaluate air qualityfor workers on vessels in the Gulf near thespill. They discovered that not all con-stituents in the hydrocarbon fluids ema-nating from the well reached the surface orevaporated. However, with a sample ofthe original well fluid (11), the study byRyerson et al. (12) was able to determinethose components that passed through theocean filter to the atmosphere unabatedand use the detection of those componentsto estimate the total flow rate from thewell. In addition, by observing the de-pletion of the other constituents, the studyby Ryerson et al. (12) estimated the frac-tion of hydrocarbons dissolved in theocean. Because this methodology is notdependent on access to the well or wellshut in, it shows considerable promise forfuture spills.Combined with the total flow rate and

hydrocarbons remaining on the oceansurface (13), a consistent picture emergedabout where the oil went. Only about one-half of the oil and none of the methane gasever reached the ocean surface (10). Athird remote-sensing estimate of flow rate,consistent with other values, was providedby imaging oil droplets in the water col-umn using a narrow-beam echo sounderdeployed from a surface ship sonar (14).This methodology is also independent ofdirect access to the well, but it is inherentlyless accurate than a direct measurement.Nonetheless, both indirect methods shouldbe pursued as useful tools during any fu-ture catastrophic spill.

20214 | www.pnas.org/cgi/doi/10.1073/pnas.1204729109 Lubchenco et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

What Remains of DWH Oil inthe Environment?Repeated sampling of offshore watersrevealed that, by August 3, 2010 (106d after oil began flowing and 19 d after shutin of the well), oil had dissipated to back-ground levels in offshore water samples(15). Sediment sampling in deep (15) andshallow (16) waters revealed groundedoil in deep areas around the wellhead,deep-water sites to the northeast andsouthwest of the well, and many shallowcoastal areas around oiled marshes andnear some beaches (SI Text).The assessment of oil in deep-water

benthic animals (15) concluded that therewere areas that experienced significant ac-cumulation of oil on sediments. Some deepcoral communities around the well havebeen impacted, mostly less than 20 kmfrom the wellhead (17). In the beach andnear-shore environment, weathered oilsamples showed 86–98% depletion of totalpolycyclic aromatic hydrocarbons (PAHs)(16). Models predict that PAH concen-trations in supratidal buried oil will de-crease to 20% of current levels within 5 y,except under isolated conditions (16). Twoparticular routes of exposure pose poten-tially elevated risks to aquatic and wildliferesources in the beach environment: in-gestion of tar balls by adult subsurface-probing shore birds and contact betweenburied oil and sea turtle eggs and hatch-lings (16). A comprehensive assessment ofDWH-associated injuries to natural re-sources is underway as part of NRDA.

Should Oil Be Dispersed Chemically?How?The decision to apply chemical dispersantsin response to the DWH oil spill was drivenby two initial goals: accelerate naturaldegradation of hydrocarbons by changingthe surface-to-volume ratio, thus exposingmore oil to naturally occurring bacteria,and keep oil from sensitive coastal eco-systems, fisheries, and estuaries. The sci-entific literature and lessons learned fromprevious spills provided some insights intothe conditions and circumstances underwhich dispersants would be more or lesseffective (sea-state conditions, geographiclocation of the spill, and physical propertiesof spilled material) (3) as well as the en-vironmental tradeoffs associated with sur-face dispersant application (18).Novel subsea application of dispersants

was raised early in the event as a potentialresponse strategy. Arguments in favor in-cluded (i) direct injection of dispersantinto the leaking wellhead 1,500 m belowthe water surface could maximize the ex-posure of oil to dispersant before it sig-nificantly weathered and emulsified withwater, (ii) compared with surface appli-cations to slicks, significantly less disper-sant would be required to achieve thesame goal, and (iii) potential exposure ofspill response workers to both airbornedispersants from surface application and

volatile organic compounds associatedwith the spill could be minimized (19).This latter concern was not trivial consid-ering the hundreds of workers on therigs drilling the relief wells and thedozens of support vessels on scene directlyabove the leaking well (SI Text). Argu-ments opposing subsea application ofdispersants included (i) lack of under-standing of potential consequences, (ii)potential to trigger severe hypoxia as mi-crobial action rapidly degraded oil drop-lets and methane gas in the water column,and (iii) potential for dispersed oil anddispersants to cause damage to subseaflora and fauna. Balancing these tradeoffswas not easy, but the potential for morerapid degradation of hydrocarbonswas compelling.Because of the unprecedented nature of

this technique, the EPA administrator re-sponsible for making the final decision, inconsultation with the NIC, decided thatimplementation should be contingent on(i) strict monitoring of dissolved oxygen,particle size, rotifer toxicity, and water andsediments, (ii) additional toxicity screen-ing of dispersants, and (iii) rapid commu-nication of data to responders and thepublic (20, 21). Results from daily toxicitytesting on rotifers using commerciallyavailable Rototox testing kits indicated nosignificant biological impacts on testorganisms (22).Repeated sampling in the waters sur-

rounding the spill detected an oxygen “sag,”but this reduction never approached levelsconsidered hypoxic (Fig. 2) (23–25). Dis-solved oxygen (DO) measurements col-lected along vertical profiles and in discretewater samples remained above the 2 mL/Lhypoxia threshold set by the EPA, theUSCG, and the NIC that would triggerpossible suspension of subsea dispersantinjection (25).An important technical issue also arose

during this monitoring. DO is customarilymeasured using electronic meters witha semipermeable membrane. Researchersand one instrument manufacturer raisedconcerns that the membrane might befouled in oiled waters andprovide unreliablemeasurements. Accordingly, the JAG ad-vised that the DO measurements be con-firmed by traditional Winkler titrations tosupplement the meter-derived observations(Fig. 2). In some instances, contemporane-ous observations with both methods con-firmed the advisability of dual samplingmethodologies (25, 26).Aerial surveys and shallow fluorometric

monitoring are typically used to measureparticle size of near-surface oil as an in-dicator of dispersant efficacy. Particle sizemonitoring in the deep sea requiredmodified approaches, such as verticalprofiling and sampling at the ocean floor(27, 28). Cumulative particle size datasuggest that the range of observed dropletdiameters was consistent with chemicallydispersed oil (25).

These measurements and others led torevisions of the relative amount of chem-ically vs. naturally dispersed oil in the FinalOil Budget (8). The amount of chemicallydispersed oil doubled from the initial es-timates (from 8% to 16% of the total).In response to concerns about potential

persistence of dispersants in the environ-ment, thousands of water and sedimentsamples from near-shore and offshore werecollected and tested for major dispersantconstituents, such as butoxyethanol,dipropylene glycol N-butyl ether, pro-pylene glycol, and dioctyl sodium sulfo-succinate (DOSS). Few water and sedi-ment samples showed detectable levels.None of the water samples showing de-tectable concentrations exceeded EPA’saquatic life benchmarks (22). Althoughthe NCP requires manufacturers to submittoxicity information for approved products(29), additional data on oil-dispersantmixtures, endocrine effects, and other pa-rameters were and are needed.The EPA conducted in vitro and whole-

animal toxicity tests on eight dispersantslisted on the NCP product schedule, in-cluding the product predominantly used inDWH (Corexit 9500A) as well as mixturesof dispersants with Sweet Louisiana Crude(SLC) oil and SLC oil alone.Whole-animal

Fig. 2. Comparison of DO profiles in the vicinityof the DWH site: conductivity–temperature–density(CTD)/O2 (SBE43) and Winkler O2 values for Re-search Vessel Ocean Veritas station 153 (occupiedon August 2, 2010), Research Vessel Brooks McCallstation 161 (occupied onAugust 5, 2010), andWorldOcean Atlas 2009 (WOA09) (56) annual mean andSD froma1°grid box centeredat 28.5°Nand88.5°W.The figure and other analyses (57) suggest that theCTD/O2 sensors on both ships were comparable andwithin ∼0.1 mL/L of the Winkler O2 analyses. The O2

depressions at ∼1,200 m shown by the SBE43 sensorare also reflected in the Winkler values, and theseprofiles are representative of measurements takenearlier in the summer in the vicinity of the wellhead(24). They also illustrate the importance of bothmeasurement techniques in accurately assessing DOlevels in the presence of oil.

Lubchenco et al. PNAS | December 11, 2012 | vol. 109 | no. 50 | 20215

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

tests were conducted on two test speciescommonly found in the Gulf: mysid shrimp(Americanmysis bahia) and silverside fish(Menidia beryllina). Results indicate thatnone of the dispersants tested displayedbiologically significant endocrine-disrupt-ing activity (30); dispersants alone wereless toxic than dispersant–oil mixtures.SLC oil alone was generally similarly toxicto both test species as dispersant–oil mix-tures, and the toxicity of Corexit9500Awas generally similar to the toxicities ofother available dispersants (31).Additional studies are required before

a more complete understanding of trade-offs with use of dispersants is known (SIText), including potential impacts of dis-persants, dispersed oil, and oil alone onthe plethora of other species in the Gulf,especially plankton and juvenile stages.

Is Seafood Contaminated withHydrocarbons?Before DWH, around one-fifth of theseafood caught commercially in US waterscame from the Gulf of Mexico (32).Keeping Gulf seafood safe was a crucialpart of the response (33). NOAA has au-thority to close federal waters to fishingduring an oil spill; states regulate fisheriesin their waters. The first step was to closeoiled or potentially oiled waters based onthe observed presence of oil or modelprojections of where oil was expected to bein the next 24, 48, and 72 h. The firstfishery closure in federal waters occurredon day 13 (May 2, 2010); at its peak inearly June of 2010, ∼88,500 square miles(∼37%) of federal waters in the Gulf wereclosed to fishing.New scientific protocols were developed

by NOAA, the US Food and Drug Ad-ministration (FDA), and the Gulf Coaststates to determine when waters were safeto be reopened to fishing or harvest ofoysters. Knowledge of differential uptake,metabolism, and disposition of PAHs infish, crustaceans, and bivalves informed thetests (34): fish were known to metabolizePAHs relatively rapidly, crustaceans me-tabolized PAHs less rapidly, and bivalvesmetabolized PAHs almost not at all. Thebivalves of greatest concern in the Gulfwere oysters. Once contaminated, theyshould not be eaten. (Oyster fisheries oc-cur in state, not federal, waters.) For anarea to be reopened to fishing, it had tomeet three criteria: be free of oil for atleast 30 d, be free of expected oil for thenext 72 h based on model trajectories, andpass repeated tests on different types ofseafood (fish and crustaceans) sampledover multiple days for PAHs and DOSS(the dispersant component of greatest po-tential concern). Of oil contaminants, thehigh molecular weight PAHs are of greatesthealth concern because of their environ-mental persistence and potential for toxicor carcinogenic effects in humans (35).Lessons learned from previous oil spills

guided efforts to ensure that consumers

would not be exposed to contaminatedseafood (33, 36, 37). Risk managementanalyses informed reopening protocols(37); specifically, a human health risk as-sessment was conducted to determinemaximum PAH exposure levels believedto be safe and associated with negligiblerisk for consumers. Because uncertainty isinherent in any risk assessment process,various assumptions are made to estimatehuman population risks, with some sub-populations being at higher risk because ofhigher seafood consumption per unit bodyweight or more susceptible to PAH effectsbecause of compromised health. The in-puts used in the health risk assessmentwere derived to be overly protective toaccount for uncertainty and variability.Subsequent testing of Gulf seafood re-vealed very low levels of PAHs or levelsbelow quantification. Thus, even whenhigher consumption values were applied tothe risk model and at-risk subpopulations(e.g., pregnant mothers), seafood testeddid not seem to pose a health risk (33).Extensive sampling across a large geo-

graphic area and a wide range of specieswas required. Testing for PAH and DOSSwas needed to respond to public concerns.In total, more than 8,000 seafood speci-mens were tested for PAHs and DOSS.Before DWH, tests for PAH were time

consuming, and no validated routine testsexisted for dispersants. During the re-sponse to DWH, a rapid method to mea-sure PAHs in seafood was developed andvalidated by the FDA (38); NOAA and theFDA developed and validated a sensitivemethod to measure DOSS (39, 40). Therapid PAH method allowed larger num-bers of samples to be analyzed faster. Indeveloping a more rapid PAH method,there were tradeoffs in both sensitivity andspecificity; however, these tradeoffs wereoffset by also using a well-validated com-prehensive GC/MS method (41) as thestandard for measuring PAHs in seafoodand confirming PAH screening results.Of the compounds in the dispersant

mixture, DOSS was the greatest potentialconcern because of its bioactivity, ex-tremely low volatility, and potential topersist in the environment longer thanother dispersant components. Althoughprevious research suggested that DOSSposed a low risk to seafood consumers (42),the new methods allowed quantification ofpresence in Gulf seafood. Developmentand application of analytical methodsduring DWH were key scientific achieve-ments. Beginning on day 130, federal wa-ters that met all criteria (absence of oil onthe water, and if oil had been present,seafood passing repeated laboratory testsfor PAHs and DOSS) were successivelyopened to fishing, progressing from least tomost oiled areas. By day 210 (November 15,2010), all but 0.4% of federal waters hadbeen opened. The final closed area, in theimmediate vicinity of the well, was openedon day 365 (April 19, 2011) (Fig. S1).

This extraordinary effort to protect theintegrity of seafood seems to have beensuccessful: no tainted seafood was reportedto have reached the market. An inde-pendent assessment arrived at the sameconclusion (43). Nevertheless, public per-ceptions about seafood safety continuedthroughout the event and beyond. Thisarea is ripe for social science research.After seeing images of oil and gas flowingfrom the riser pipe and images of oil-cov-ered shores and birds day after day andweek after week, many people had diffi-culty believing that oil was disappearingrapidly from open waters, fish could me-tabolize PAHs, and seafood testingwas reliable.

What Is the Process for Assessing theEnvironmental Impact of the Spill?What Is the Best Path to Restoration?What Is the Role of Science in DamageAssessment?Science was as important to assessment ofdamage and design and monitoring ofrestoration as it was to the response.Among the numerous potential economic,social, human health, and environmentalimpacts of DWH, injury to natural re-sources and the public’s access to them isregulated under OPA90 (specifically, theNRDA regulations) (44). [NRDA statuteis 15 CFR § 990 et seq. Trustees (or nat-ural resource trustees) means those offi-cials of the federal and state governments,Indian tribes, and foreign governmentsdesignated under 33 USC 2706(b) of OPA15 CFR § 990.30. There are seven primarytrustees involved in the DWH NRDA, thefive affected states (Texas, Louisiana,Mississippi, Alabama, and Florida), andtwo federal agencies (the Department ofInterior and the Department of Com-merce through NOAA). Baseline meansthe condition of the natural resources andservices that would have existed had theincident not occurred. Baseline may beestimated using historical data, referencedata, control data, or data on incrementalchanges (e.g., number of dead animals)alone or combined as appropriate.] Theseregulations designate federal, state, andtribal natural resource trustees to con-duct NRDAs on behalf of the public. Tomeet this mandate, the trustees seek torestore to baseline (the condition thatthey would have been in had the spill notoccurred) injured resources and servicesand compensate the public for interimlosses (i.e., the losses that occur duringthe time that it takes the resources torecover to baseline).In view of the size, duration, and four-

dimensional and complex nature of thespill in the Gulf, the DWH NRDA (in-cluding restoration; see below) may con-tinue for years. Federal and state trusteesare working together to determine how theoil spill affected the Gulf of Mexico’snatural resources and the human use ofthose resources. With potential natural

20216 | www.pnas.org/cgi/doi/10.1073/pnas.1204729109 Lubchenco et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

resource injury spanning five states andtheir waters as well as federal waters, thisdamage assessment is the largestever undertaken.Under OPA90, NRDA has three phases:

the preassessment phase, in which thetrustees determine whether impacts tonatural resources have occurred; the injuryassessment/restoration planning phase, inwhich the trustees quantify injuries andidentify possible restoration projects; andthe restoration implementation phase, inwhich the trustees implement restorationto baseline and monitor its effectiveness.Typically, the information in all threephases is initially retained outside thepublic domain. During DWH, in support ofa commitment to openness and trans-parency and after determining that doingso would not jeopardize their case, thetrustees decided to allow public access tomost of the data collected during phase onebefore assessment of injury (SI Text).

Conclusion, Lessons Learned, andResearch PrioritiesThe DWH event required real-time scienceto inform responders on varied aspectsof the spill—from directly collaboratingwith BP to mitigate risks associated withstopping the release and determining theflow rate to measuring and modeling thefate of the oil to assessing effects on the air,seafood, species, and habitats (Fig. 1). Sci-entists from government, industry, and thebroader research community respondedwith novel applications of existingmethods—e.g., reservoir modeling (45),echo sounder flow detection (14), en-hanced-resolution γ-ray imaging of the blowout preventer (BOP) (46), and use oftechnologies originally adapted for re-searching deep-sea geological processes(47). New methods were also developedsuch as testing for dispersant in seafood (33,42) and estimating flow rate from atmo-spheric measurements (12). The science ofdeep spill containment, mitigation, andimpacts has rapidly accelerated because ofthe necessity of responding to this event.Response to future deep spills globally willbenefit from the many scientific break-throughs applied to DWH and emergingafter the response as well as the lessonslearned. The scale of this spill and the sci-entific efforts directed to assessing its mag-nitude and containing it deserve attention,both for what was revealed about subsurfaceprocesses and the ability of public in-stitutions and industry to respond to sucha disaster and for what new approaches andtools that we now have at our disposal.Scientific revelations from the DWH

spill are many, and we continue to besurprised by numerous aspects, such as thediscovery of novel microbial communities(48) and the conditions that led to rapiddecomposition of hydrocarbons duringthis event (49). Full conclusions about theimpact of the oil on species, ecosystems,and people will necessarily await ongoing

analyses for detecting long-term impacts.Some results are beginning to emerge,such as effects on marsh fish (killifish)(50), but sweeping conclusions about im-pacts are premature. Similar to the EVOSspill, some effects may be unknown orunappreciated for years, if ever (4).Based on the DWH experiences, we

compiled a list of scientific priorities forfuture oil spill response preparedness (51).

• Gather adequate environmental base-lines for all regions at risk.

• Developnew technologies for rapid precisereconnaissance and sampling to supporta timely and robust response effort.

• Support the development of models anddecision support tools, such as scenarioplanning (52), to enhance response anddamage assessment.

• Fill large information gaps regardingbiological effects of oil, changing climate,and other simultaneous drivers of variabil-ity in coastal and aquatic ecosystems.

• Build coupled ecosystem-scale routinemonitoring/research/communications forevery large marine ecosystem (LME) inUS waters, including the coastal zone, toprovide integrated interdisciplinary un-derstanding of how the ecosystem worksand is changing, ideally as a partnershipwith academic institutions in the region.

• Put greater emphasis on social sciencedata collection, including adequatebaselines, to understand costs to the re-gion and the nation of oil spill disasters.

• Conduct research on impacts of disper-sants and dispersants plus oil on a broadarray of species and life stages.

• Develop more efficient methodologiesfor capturing oil at the surface.

• Conduct social science studies to un-derstand public perceptions about sea-food safety.

Additional lessons and suggestions forimprovements related to preparedness andcoordination with the scientific communityare below.Although there was much criticism of the

pace of response efforts, we witnessed atremendous effort on the part of our sci-entific institutions, various ad hoc andstanding committees, and individuals to benimble, think outside the box, and work incollaborative ways unanticipated underthe provisions of OPA90. Throughout thecrisis, we provided integrated and com-prehensive information to guide the re-sponse—to stop, contain, track, measure,and remove the oil, protect the integrity ofthe seafood supply from the Gulf, andsafeguard wildlife. Coordination of theseefforts required a monumental effort andcareful attention to scientific information.Throughout the crisis, we also shared

new information publicly as soon as we hadgood reason to believe that it was accurategiven the information at hand. Notably,we did not speculate. We were criticized

both for sharing information and notspeculating about consequences or signif-icance when pertinent facts were notavailable. Some thought that we wereminimizing the extent and severity of thespill. Therefore, it is relevant to ask: “howaccurate was the information releasedduring the event?” With the benefit ofpostcrisis information, much, although notall, of the information that we releasedduring the crisis is now known to havebeen accurate. Here is a short summary ofboth based on information provided in thispaper and the work by McNutt et al. (7).

i) The basic conclusions of the prelimi-nary oil budget released on August 4were correct (7, 8): approximately one-half the oil was, indeed, gone (recov-ered, burned, skimmed, evaporated, ordissolved).

ii) However, the science team underesti-mated the amount of oil that had beendispersed chemically (8% of the totalin the preliminary oil budget vs. 16%in the final budget) and slightly over-estimated the amount of naturally dis-persed oil (16%; later determined tobe 13%). (The total amount of dis-persed oil was originally estimated at24% and later revised to 29%.)

iii) As we reported in early August, to thedisbelief of many people, much of theoil that had been dispersed (either nat-urally or chemically) was rapidly beingconsumed by bacteria (23, 26, 48, 49).

iv) Fishery closures plus newly developedand rigorously implemented protocolsfor testing of seafood for the compo-nents of hydrocarbons and dispersantthat were of potential concern seem tohave been successful, because notainted seafood was reported to haveentered the seafood supply.

v) The lack of DOSS in tested seafood(fishes and crustaceans) seems to sup-port our expectation that either dis-persant degraded rapidly or it wasmetabolized quickly by exposed ani-mals. (This finding does not mean thatdispersants had no environmental im-pact but only that the information inhand is consistent with the expectationthat this component of dispersant de-graded or was metabolized rapidly withrespect to contamination in these taxa.)

vi) Luckily, the Loop Current (SI Text) didnot behave in an average fashion.Based solely on climatology, therewas concern early in the disaster (bynumerous scientists, including govern-ment scientists) that the Loop Currentmight transport oil to the Florida Keysand possibly beyond. Fortunately, 2010was an atypical year.

vii) Although there was much speculation inthe scientific community (governmentand academic) and some modelingresults predicted hypoxia caused by

Lubchenco et al. PNAS | December 11, 2012 | vol. 109 | no. 50 | 20217

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

oxidation of deep dispersed oil andmethane, such an event never oc-curred. The evidence suggests that thisresult is because of two factors: rapidbiodegradation of hydrocarbons by bac-teria (23, 26, 48, 49) and use of disper-sants where hydrocarbons were spewingforth. The use of deep injection of dis-persants was carefully monitored ona daily basis along with the levels of dis-solved oxygen in the 1,000- to 1,200-mdepths, where dispersed oil was knownto be accumulating. The full extent ofenvironmental damage caused by theuse of such dispersants is not yet fullyknown, but if we are faced with sucha choice in the future, information avail-able to us now would lead to a similarrecommendation to proceed with cau-tion and an abundance of monitoring.

viii)Although the flow rate announced as5,000 bpd on day 9 was suspectedby government scientists as well asothers of being a significant underes-timate (7), the absence of a singlecredible methodology available to de-rive an accurate flow rate quickly ledthe NIC to establish the FRTG andcharge it with devising the best possi-ble estimate using a combination ofmethodologies. Successive estimateswere announced on days 37 (12,000–19,000 bpd), 53 (20,000–40,000 bpd),and 57 (35,000–60,000 bpd) as newinformation came to light, with a finalestimate derived from pressure meas-urements of 62,000 bpd at the outsetand 53,000 bpd at well shut in for a to-tal of 4.9 ± 10% million barrels (7)(Fig. S1). Note that it is misleadingto compare the FRTG flow rates di-rectly with higher flow rates fromnongovernment researchers, becausethe FRTG rates are oil flow rates on-ly, whereas the early nongovernmentflow rates are total discharge: oil plusliquid natural gas. The total discharge(oil + liquid gas) numbers should bemultiplied by 0.29 before being com-pared with days 37 and 53 FRTG es-timates and 0.41 to compare with day57 FRTG flow rate to account for onlythe oil fraction.

The lack of reasonable estimates of flowrate early on was problematic from theperspectives of both communications andresponse, but the lack was caused by realuncertainty rather than any attempt to hideinformation or underestimate numbers.It is true that much of the response did notdepend on knowing the exact rate, butsome of it did, particularly the capacity tocapture oil directly from the well. It isreasonable to suggest that future permitsbe conditional on having mechanisms torapidly assess flow rate to ameliorate theproblem in the future.

Still unresolved are issues having todo with environmental impacts (impact ofoil, dispersants, and dispersed oil oncoastal, midwater, and benthic taxa andcommunities, marine mammals, turtles,and birds), human health, and economicand social impacts.Based on these conclusions and our

broader experiences, some of the lessons thatwe learned include the following:

i) The importance of preparedness can-not be overstated. The consequencesof lack of investment over the lastfew decades in scientific understand-ing and technological developmentwere obvious during DWH. Despitesignificant advances in technology thatallowed drilling in deep waters, com-parable progress had not been made indevising methods that would have en-abled us to stop the flow from deepwells or deal with a spill of the magni-tude seen in DWH. Both could andshould have been anticipated. Scien-tific and technological expertise hadto be mobilized de novo to createsolutions. Cleanup technologies hadadvanced little since EVOS. The Her-culean efforts expended during theevent were admirable, but planningfor events like those that began onApril 20, 2010 and investing in devel-oping the capacity to deal with themwould likely have made a significantdifference. In view of how importantit is to know flow rate to mobilize partof the response effort, preparednessshould include devices installed on ex-traction equipment to provide flow rateinformation if needed as well as redun-dant mechanisms in case of failure.

ii) Preparedness also extends to acquiringa basic understanding of the placeslikely to be affected by a spill at theLME scale (e.g., the Gulf of Mexico).Accidents during exploration, extrac-tion, transport, or offloading takeplace within coupled human and natu-ral systems. Basic understanding of thedynamics of the ecosystem and conse-quences of changes to people requiresa comprehensive, integrated monitor-ing/research/communication effort fo-cused on an LME, ideally throughthe development of regional scientificcollaboration networks. This under-standing must be more than spatiallyexplicit descriptions of the speciespresent. It should include an inte-grated understanding of the physicaland ecosystem dynamics sufficient toknow where oil is likely to flow (alongthe shallow and deep inner shelf andnot just open surface waters), whichspecies and life stages would be af-fected at different times of the year,and how impacts to those species would

affect other species, the functioningof the ecosystem, the provision of eco-system services, and other impacts onpeople. This knowledge is neededfor every LME in the US Exclusive Eco-nomic Zone (and adjacent waters, whererelevant), and it would vastly enhanceeffective response and understanding ofimpacts. Moreover, it has the added ben-efit of significantly enhancing a variety ofother management efforts—water qual-ity, invasive species, fisheries, shipping,recreation, and conservation. Achievingthis integrated knowledge and sharingit publicly require stable funding andmechanisms to integrate monitoring,research, and communication activitiesacross a region and the nation.

iii) The DWH spill highlighted the needfor enhanced capacity to respond tospills and conduct training and otherpreparedness activities before spillsoccur. Capacity includes trained peo-ple, technical knowledge, equipmentfor oil removal, and protocols and net-works that can be activated quickly.The spill pushed our agencies to thelimit. Had another significant spill ornatural disaster (such as a major hur-ricane) occurred at that time, our abil-ity to respond would have beenseverely limited. The government andindustry have taken some steps to in-crease response capacity, such as withthe establishment of two containmentconsortia, but more progress is neededon an adequate funding mechanismfor research and development focusedon improving oil spill response, espe-cially in frontier areas such as theArctic (53). The DWH incident alsosaw the willingness of the academiccommunity to act in disaster-responsemode. New arrangements for trainingand funding need to be developed toenable greater participation from aca-demic and other sectors.

iv) Mechanisms are needed for rapid mo-bilization of more funding for researchduring a spill, especially early in anevent. Although some funds wereavailable through mission agenciesand the National Science Foundationearly in the response, they were insuf-ficient to enable the broader array ofknowledge acquisition that researcherswere ready to tackle and that couldassist in providing a more completeunderstanding of DWH impacts aswell as better response to future events(54). Mission agencies rapidly mobi-lized numerous preexisting relation-ships (e.g., with university or inde-pendent scientists) through ongoingresearch relationships financed throughcompetitive grants and preexistingcontracts to provide services in theevent of a spill. For example, all of

20218 | www.pnas.org/cgi/doi/10.1073/pnas.1204729109 Lubchenco et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

the seven NOAA vessels deployed forDWH response in the Gulf had aca-demic researchers on board doing re-search at one time or another. Inparallel, the National Science Founda-tion quickly provided funds to manyother researchers. However, taken to-gether, these mechanisms were insuffi-cient to provide adequate funds withthe rapidity required. Moreover, legalconstraints stipulated in OPA90 onfunding provided by the RP duringthe event meant that only response ac-tivities, not research, could be sup-ported with those funds.

v) Effective mechanisms are needed toenable rapid two-way communicationwith the broader scientific community.No single mechanism existed for us tocommunicate easily with the large, un-defined, and interdisciplinary commu-nity of scientists. The US Governmentset up daily calls with governors, mem-bers of Congress from Gulf Coaststates, parish presidents, and journal-ists. Unlike those easily identifiedgroups, “scientists interested in thespill” were a challenging group toidentify quickly and communicate withfrequently and in the depth requiredfor meaningful exchanges. New ve-hicles for that communication neededto be created. Universities in the Gulfregion and the Consortium for OceanLeadership were helpful but did notbegin to represent the universe of in-terested scientists. Meetings and work-shops were valuable but did not beginto meet the ongoing desire for crediblescientific information. For the mostpart, many scientists could get andshare updates only through informa-tion in the public press, which led toconsiderable misunderstandings andgreat frustration. Solutions include thedevelopment of regional scientific col-laboration networks (discussed in les-son ii) that could serve as a startingpoint and better use of web-basedcommunication tools.

vi) A new dialogue within the scientificcommunity and possible new mecha-nisms are needed to resolve the ten-sions around the appropriate time toshare preliminary findings with thepublic. The public demand for infor-mation created a challenging dynamic.A clash of three cultures emerged: themedia/public appetite for instanta-neous information regardless of accu-racy, the need for rapid but accurateinformation to inform the response,and the scientific convention of wait-ing for journal peer review and publi-cation before sharing results. Someacademics waited for results of thejournal peer review before talkingabout any of their results. Agencies

and some academic scientists heldthemselves accountable for ensuringquality assurance and quality control(QA/QC) of results (but not peer re-view) before releasing data or inter-preting findings; then, there was peerreview for final results. Other academ-ics talked about new findings with themedia well before any QA/QC, muchless any peer review. Our rationale forsharing new data and preliminary re-sults after they had gone through theQA/QC process reflected a commit-ment to rapid but accurate resultsand openness and transparency in themidst of a crisis and a recognition ofthe additional, sometimes significanttime required for peer review. One op-tion is to explore ways for very rapidpeer review.A parallel problem concerns intellec-tual property. The community cur-rently lacks effective processes toprovide rapid but accurate informationfor decision-making while still respect-ing many journals’ interests in report-ing new knowledge not previouslyshared. During DWH, some nongov-ernmental researchers did not wanttheir findings widely disseminated inthe press, because they thought itwould prevent subsequent publication.Some prestigious journals, like Sci-ence, sent clear signals that, in lightof the need for all credible informa-tion to inform the response, the jour-nal would make an exception (55).Responders need to make decisionsin a timely way that takes advantageof diverse science and views. Protect-ing intellectual capital is also impor-tant. However, some scientists com-municated misleading or wronginformation and conclusions in thepress before results had been substan-tiated, leading to massive and avoid-able confusion, waste of resources,and loss of public confidence. Thereis a need for a vigorous debate amongscientists, editors, and agencies to findcommon ground to act on best sci-ence in a crisis while still protectingscientific discovery.

vii) During a crisis, scientists must respectthat the priority needs of the responsemust come before acquisition of newknowledge when the two are in directconflict. In parallel, responders mustsupport gathering new data, unlessthose activities interfere with the re-sponse. Another culture clash ensuedin the immediate vicinity of the well-head. Academic and independent re-searchers wanted access to the wellsite at depth, but their presence hadstrong potential to interfere with oper-ations to stop the flow of oil. Remotelyoperated vehicles (ROVs) controlled

from the surface with acoustic com-mands were attempting delicate ma-neuvers at depth in the dark. ScientificROVs had strong potential to inter-fere physically or acoustically with re-sponse ROVs. After one responseROV accidentally bumped into anddislodged the riser insertion tube tool(an early device for collecting oil in-side the riser), the NIC declareda “no go” zone in the critical areaaround the well. Permission to enterthat zone was then allowed by theNIC only if activities would not inter-fere with response operations. Al-though some researchers understoodand respected the “no go” zone, otherscomplained that a heavy-handed gov-ernment was preventing science fromproceeding.

viii)Although a new standard for transpar-ency and rapidity of data sharing wasset with DWH, it was not enough tosatisfy everyone. The federal agenciesmade an unprecedented effort—in themidst of the response—to providedata about the spill in raw and synthe-sized form using the web and othervehicles. (i) A spatially explicit man-agement tool previously available onlyto responders, the innovative ERMA,was rapidly transformed into a publiclyaccessible platform that could servethe millions of hits per day expected(and received). This tool made verylarge datasets available as soon as theycould be go through the QC process.In some cases, private businesses, suchas ESRI, Inc. and Google, assisted indeveloping data visualization capabil-ities specifically for oil spill communi-cation. (ii) New web vehicles weredeveloped to communicate both spilland restoration efforts (SI Text). (iii)In light of the keen interest in thequestion of “where did the oil go?,”the Administration chose to sharea preliminary version of the oil budget,consistent with its commitment totransparency. (iv) The NRDA Trust-ees, at the urging of NOAA, took theunprecedented step of releasing muchof the preassessment data collected forthe NRDA. Notwithstanding thissignificant and ongoing effort at trans-parency, public and media interestin the event created a demand for al-most instantaneous collection and syn-thesis of information far beyond thecapabilities of existing science and in-stitutions to collect, process, and verifyas accurate.

ix) The scientific teams (FRTG, Oil Bud-get, GLST, Operational Science Advi-sory Team, and JAG) created duringDWH were highly successful in trou-ble-shooting, designing solutions, ana-lyzing and synthesizing data, and

Lubchenco et al. PNAS | December 11, 2012 | vol. 109 | no. 50 | 20219

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

evaluating options. Similar mecha-nisms should be used for future Spillsof National Significance or other ma-jor crises, where interagency, interdis-ciplinary, broad-based scientific inputis needed. Equally important is thecommitment of leadership during theresponse to use information from sci-ence teams. These teams ensured thatstrong scientific expertise was avail-able to inform the response. An im-portant element of these teams wasthe inclusion of outside experts. Equallyimportant were scientists at the helmsof agencies who articulated the needfor these scientific teams, supportedthe inclusion of expertise within andoutside government, and ensured thegroups’ reports and recommendationswere understood and incorporated in-to the response. These teams providedan excellent complement to the Scien-tific Support Coordinators advisingthe OSC in the region. The newly cre-ated science teams during DWH pro-vided broader interdisciplinary expertiseto deal with the many new challengesfaced in the DWH disaster. Strongsupport by the NIC for the scienceteams and a firm commitment to usetheir input in responding was fortuitousand essential. Planning for future eventshould include similar mechanisms andspecific roles for scientific informationand scientists in the process at multiplelevels of decision-making.

x) Intimate engagement with industry isessential. The government did nothave the equipment needed to per-form the mechanical operations re-quired near the sea floor, but itplayed an unprecedented role in help-ing BP gain control of the well. Al-though formal guidance was issued toBP through directives from the NIC,intimate engagement with BP inHouston allowed independent assess-ments of events to support timely de-cisions in support of the government’srole in the crisis (7). In the process ofunderstanding the issues of the welland well control, the GLST found that

deep-water drilling technologies wouldhave benefited from an improved sys-tems perspective. Instrumentation builtinto the BOP capable of providing ac-curate and redundant data that wouldallow cross-verification of the situationwas missing. For example, a system ofROV-accessible electronic and visualmeasurements of the position and lockstatus of the BOP’s rams would havebeen greatly helpful. There was nopressure instrumentation installed onthe Top Hat, and several pressuregauges on the BOP and capping stackfailed during the DWH incident. Otherexamples of needed deep-water tech-nologies include autonomous underwa-ter vehicles capable of station-keepingin the deep sea with methods for de-livering data for satellite transmissionat the ocean surface to remotely mon-itor the well during storms. Another im-portant contribution of the GLST wasits insistence that relevant data be takenwhenever possible. These data provedto be critical to the most significant eval-uation of the crisis—the determinationthat the capped well was not leaking (7).The GLST required enhanced ROVmonitoring, more seismic monitoringand analysis, and the addition of uniqueacoustic monitoring provided by theNOAA vessels. This information ulti-mately proved to be invaluable in reach-ing agreement with BP to install thecapping stack, conduct the well integritytest, and proceed with the Top Kill.To engage wider participation fromthe scientific community, it would beadvisable to consider establishing legalprotocols and agreement with industrythat would allow those individuals in-volved in any future response access tonecessary proprietary data. The oil/gasindustry had information that was keyto response—video, access to ROVfeeds of imagery and data, oil samples,formulations of dispersants, detailedgeological maps of the region, and otherinformation. Although the GLST hadaccess to the information that it needed,it was often difficult—especially early

on—for other science teams to obtaininformation from BP in a timely man-ner. For academic scientists, it waseven more difficult (e.g., access toMacondo samples).

Despite these challenges, the responseeffort was effective, because in large part,high-quality scientific and engineering in-formation was available and used. It is ourhope that lessons learned from this disasterwill be implemented before and usedduring any future events.

ACKNOWLEDGMENTS.We extend condolences tothe families of the 11 workers who lost their liveson the Deepwater Horizon rig and heartfeltsympathy to the thousands of Gulf Coast resi-dents devastated by the spill. We acknowledgethe thousands of scientists, technicians, responseworkers, and managers inside and outside of ouragencies who devoted extraordinary efforts tosupport the spill response. The magnitude andmagnanimity of their efforts have gone largelyunrecognized, and the work continues. Signifi-cant efforts continue to be devoted to under-standing the full impacts of the spill and itsconsequences for the Gulf of Mexico ecosystem.Restoration efforts have been initiated, but thereis much more work to be done. We thank themany thousands of women and men, both insideand outside of government, who contributed tothe efforts outlined in this special volume andthe larger body of documentation on the Deep-water Horizon event. We thank Chris Belter, Na-tional Oceanic and Atmospheric Administration(NOAA) library, for collecting and maintaininga Deepwater Horizon bibliography and Joe Kli-mavicz, CIO for NOAA, and others responsible forthe new online repository done in conjunctionwith the National Technical Information Service/NOAA: http://noaa.ntis.gov/site/home.php. Weacknowledge the contributions from JenniferPizza, John Stein, and many more from NOAA,US Geological Survey (USGS), and EnviromentalProtection Agency (EPA) in the preparation ofthis introduction. G.D. acknowledges supportfrom the American Association for the Advance-ment of Science (AAAS) Science and TechnologyPolicy Fellowship program and the Oak Ridge In-stitute for Science and Education. Any use oftrade, firm, or product names is for descriptivepurposes only and does not imply endorsementby the US Government. The scientific resultsand conclusions, as well as any views or opinionsexpressed herein, are those of the authors anddo not necessarily reflect the views of USGS,the Department of the Interior, NOAA, theDepartment of Commerce, the EPA, or theDepartment of Energy. And finally, we thankAdmiral Thad Allen and President Barack Obamafor their trust in and support of science.

1. Anastas PT, Sonich-Mullin C, Fried B (2010) Designingscience in a crisis: the Deepwater Horizon oil spill. En-viron Sci Technol 44(24):9250–9251.

2. National Research Council (1999) Spills of NonfloatingOils: Risk and Response Committee on Marine Trans-portation of Heavy Oils (National Academy Press,Washington, DC).

3. National Research Council (2005) Oil Spill Dispersants:Efficacy and Effects (National Academy Press, Washing-ton, DC).

4. Peterson CH, et al. (2003) Long-term ecosystem responseto the Exxon Valdez oil spill. Science 302(5653):2082–2086.

5. National Oil and Hazardous Substances Pollution Con-tingency Plan, 40 Federal Register 300 (1994).

6. National Commission on the BP Deepwater Horizon OilSpill and Offshore Drilling (2011) DEEPWATER: A Re-port to the President, January 2011 (US GovernmentPrinting Office, Washington, DC), p 149.

7. McNutt MK, et al. (2012) Applications of science andengineering to quantify and control the DeepwaterHorizon oil spill. Proc Natl Acad Sci USA 109:20222–

20228.8. Lehr B, Bristol S, Possolo A (2010) Oil Budget Calculator

Deepwater Horizon, Technical Documentation, A Reportto the National Incident Command, November 2010.

Available at http://www.restorethegulf.gov/sites/default/files/documents/pdf/OilBudgetCalc_Full_HQ-Print_111110.pdf. Accessed January 21, 2012.

9. McNutt MK, et al. (2012) Review of flow rate esti-mates of the Deepwater Horizon oil spill. Proc Natl

Acad Sci USA 109:20260–20267.10. Ryerson TB, et al. (2012) Chemical data quantify Deep-

water Horizon hydrocarbon flow rate and environ-mental distribution. Proc Natl Acad Sci USA 109:20246–20253.

11. Reddy CM, et al. (2012) Composition and fate of gasand oil released to the water column during the Deep-water Horizon oil spill. Proc Natl Acad Sci USA 109:

20229–20234.12. Ryerson TB, et al. (2011) Atmospheric emissions from

the Deepwater Horizon spill constrain air-water parti-tioning, hydrocarbon fate, and leak rate. Geophys Res

Lett, 10.1029/2011GL046726.13. Clark RN, et al. (2010) A Method for Quantitative Map-

ping of Thick Oil Spills Using Imaging Spectroscopy, US

Geological Survey Open-File Report, Vol 2010–1167. Avail-able at http://pubs.usgs.gov/of/2010/1167/downloads/

pdf/OF10-1167.pdf. Accessed February 25, 2012.14. Weber TC, et al. (2012) Estimating oil concentration

and flow rate with calibrated vessel-mounted acousticecho sounders. Proc Natl Acad Sci USA 109:20240–20245.

20220 | www.pnas.org/cgi/doi/10.1073/pnas.1204729109 Lubchenco et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020

15. Operational Science Advisory Team (OSAT) (2010) Sum-mary Report for Sub-Sea and Sub-Surface Oil and Dis-persant Detection, Sampling and Monitoring. Availableat http://www.restorethegulf.gov/sites/default/files/documents/pdf/OSAT_Report_FINAL_17DEC.pdf. Ac-cessed January 21, 2012.

16. Operational Science Advisory Team (OSAT) (2011) SummaryReport for Fate and Effects of Remnant Oil in the BeachEnvironment. Available at http://www.restorethegulf.gov/sites/default/files/u316/OSAT-2%20Report%20no%20ltr.pdf. Accessed January 21, 2012.

17. White HK, et al. (2012) Impact of the Deepwater Hori-zon oil spill on a deep-water coral community in theGulf of Mexico. Proc Natl Acad Sci USA 109:20303–20308.

18. Etkin DS (1998) Factors in dispersant use decision-mak-ing process: Historical overview and look to the future.Proceedings of the 21st Arctic and Marine Oil Spill Pro-gram Technical Seminar (Environment Canada, Ot-tawa), pp 281–304.

19. Coastal Response Research Center (2010) Deepwater Ho-rizon Dispersant Use Meeting Report, 26–27 May 2010.Available at http://docs.lib.noaa.gov/noaa_documents/NOAA_related_docs/oil_spills/dwh_dispersants_use_meeting_report-2010.pdf. Accessed January 21,2012.

20. Dispersant Monitoring Directive EPA (2010) DispersantMonitoring and Assessment Directive for SubsurfaceDispersant Application. Available at http://www.epa.gov/bpspill/dispersants/subsurface-dispersant-directive-final.pdf. Accessed February 25, 2012.

21. Quality Assurance Sampling Plan EPA (QASP) (2010) BPSpill Quality Assurance Sampling Plan to Evaluate theEffects to Water and Sediment from Oil and Dispersantto Shoreline, Nearshore, and Far Off-Shore Areas. Avail-able at http://www.epa.gov/bpspill/samplingplanjune1/BP_Spill_QASP_May_30_v2.pdf. Accessed February 25, 2012.

22. BP (2010) BP Rotifer Toxicity Test Results as of Sep-tember 3. Available at http://www.epa.gov/bpspill/dispersants-bp.html. Accessed February 25, 2012.

23. Hazen TC, et al. (2010) Deep-sea oil plume enrichesindigenous oil-degrading bacteria. Science 330(6001):204–208.

24. Joint Analysis Group (JAG) (2010) Review of R/V BrooksMcCall Data to Examine Subsurface Oil. Available athttp://www.noaa.gov/sciencemissions/PDFs/JAG_Report_1_BrooksMcCall_Final_June20.pdf. Accessed January 21,2012.

25. Joint Analysis Group (JAG) (2010) Review of Prelimi-nary Data to Examine Oxygen Levels in the Vicinityof MC252#1, May 8 to August 9, 2010. Available athttp://www.noaa.gov/sciencemissions/PDFs/JAG_Oxygen_Report%20(FINAL%20090410).pdf. Accessed January 21,2012.

26. Camilli R, et al. (2010) Tracking hydrocarbon plumetransport and biodegradation at Deepwater Horizon.Science 330(6001):201–204.

27. SMART (2006) Special Monitoring of Applied Re-sponse Technologies (SMART), v.8/.2006. Available athttp://response.restoration.noaa.gov/sites/default/files/SMART_protocol.pdf. Accessed February 25, 2012.

28. Fingas M-F (2003) Review of Monitoring Protocols forDispersant Effectiveness in Report for Prince WilliamSound Regional Citizens’ Advisory Council (Environ-ment Canada, Ottawa).

29. National Oil and Hazardous Substances Pollution Con-tingency Plan (NCP) (1994) National Oil and HazardousSubstances Pollution Contingency Plan (NCP) in Codeof Federal Regulations Ch. I (7–1–03 Edition) 59 FR47453, Sept. 15, 1994, § 300.915, Subpart J. Availableat http://www.epa.gov/oem/docs/oil/cfr/900_920.pdf.Accessed February 25, 2012.

30. Judson R-S, et al. (2010) Analysis of eight oil spill dis-persants using rapid, in vitro tests for endocrineand other biological activity. Environ Sci Technol 44:5971–5978.

31. Hemmer MJ, Barron M-G, Greene RM (2011) Compar-ative toxicity of eight oil dispersants, Louisiana sweetcrude oil (LSC), and chemically dispersed LSC to twoaquatic test species. Environ Toxicol Chem 30(10):2244–2252.

32. Fisheries of the United States (FUS) (2011) NOAA, Na-tional Marine Fisheries Service. Available at http://www.st.nmfs.noaa.gov/st1/fus/fus10/02_commercial2010.pdf.Accessed April 25, 2012.

33. Ylitalo GM, et al. (2012) Federal seafood safety re-sponse to the Deepwater Horizon oil spill. Proc NatlAcad Sci USA 109:20274–20279.

34. Varanasi U, ed (1989) Metabolism of Polycyclic Aro-matic Hydrocarbons in the Aquatic Environment (CRC,Boca Raton, FL).

35. Bolger M, Carrington C (1999) Evaluating and Commu-nicating Subsistence Seafood Safety in a Cross-CulturalContext: Lessons Learned from the Exxon Valdez OilSpill, eds Field LJ, Fall JA, Nighswander TS, Peacock N,Varanasi U (Society of Environmental Toxicology andChemistry, Pensacola, FL), pp 195–204.

36. Hom T, Brown DW, Stein JE, Varanasi U (1999) Evalu-ating and Communicating Subsistence Seafood Safetyin Cross-Cultural Contexts: Lessons Learned fromthe Exxon Valdez Oil Spill, eds Field LJ, Fall JA,Nighswander TS, Peacock N, Varanasi U (SETAC Press,Pensacola, FL), pp 169–194.

37. Yender R, Michel J, Lord C (2002) Managing SeafoodSafety After an Oil Spill (Hazardous Materials ResponseDivision, Office of Response and Restoration, NationalOceanic and Atmospheric Administration, Seattle).

38. US FDA (2010) Protocol for Interpretation and Use ofSensory Testing and Analytical Chemistry Results forRe-Opening Oil-Impacted Areas Closed to SeafoodHarvesting Due to the Deepwater Horizon Oil Spill.Available at http://www.fda.gov/Food/ucm217601.htm. Accessed February 25, 2012.

39. Gratz S, et al. (2010) Screen for the Presence of Poly-cyclic Aromatic Hydrocarbons in Select Seafoods UsingLC-Fluorescence, Laboratory Information Bulletin No.4475, DHHS. Available at www.fda.gov/downloads/ScienceResearch/UCM220209.pdf. Accessed February25, 2012.

40. Flurer RA, et al. (2010) Determination of Dioctylsulfo-succinate in Select Seafoods Using a Quechers Extrac-tion with Liquid Chromatography—Triple QuadrupoleMass Spectrometry. US Food & Drug Administra-tion. Available at http://www.fda.gov/downloads/ScienceResearch/FieldScience/UCM231510.pdf. AccessedFebruary 25, 2012.

41. Sloan CA, et al. (2005) Techniques in Aquatic Toxicology,ed Ostrander GK (CRC, Boca Raton, FL), pp 631–651.

42. Benner RA, Jr., et al. (2010) Investigation of Corexit 9500Dispersant in Gulf of Mexico Seafood Species. Available

at http://www.fda.gov/downloads/Food/FoodSafety/Product-SpecificInformation/Seafood/UCM250307.pdf.Accessed February 25, 2012.

43. Gohlke JM, Doke D, Tipre M, Leader M, Fitzgerald T(2011) A review of seafood safety after the DeepwaterHorizon blowout. Environ Health Perspect 119(8):1062–1069.

44. 15 Federal Register 990.30 (2012).45. Hsieh PA (2010) Computer Simulation of Reservoir De-

pletion and Oil Flow from the Macondo Well Followingthe Deepwater Horizon Blowout. USGSOpen-File Report2010-1266. Available at http://pubs.usgs.gov/of/2010/1266/of2010-1266.pdf. Accessed September 30, 2012.

46. Hansen R (2011) Lending a hand to an oily problem.Science & Technology Review, March:10–17. Availableat https://str.llnl.gov/Mar11/sharpe.html. Accessed Sep-tember 29, 2012.

47. Camilli R, et al. (2012) Acoustic measurement of theDeepwater Horizon Macondo well flow rate. Proc NatlAcad Sci USA 109:20235–20239.

48. Redmond MC, Valentine DL (2012) Natural gas andtemperature structured a microbial community re-sponse to the Deepwater Horizon oil spill. Proc NatlAcad Sci USA 109:20292–20297.

49. Valentine DL, et al. (2012) Dynamic autoinoculationand the microbial ecology of a deep water hydrocar-bon irruption. Proc Natl Acad Sci USA 109:20286–20291.

50. Whitehead A, et al. (2012) Genomic and physiologicalfootprint of the Deepwater Horizon oil spill on resi-dent marsh fishes. Proc Natl Acad Sci USA 109:20298–20302.

51. National Science and Technology Council, Subcommit-tee on Ocean Science and Technology (2012) Deepwa-ter Horizon Oil Spill Principal Investigator Workshop:Final Report (National Science and Technology Council,Washington, DC).

52. Machlis GE, McNutt MK (2010) Disasters. Scenario-building for the Deepwater Horizon oil spill. Science329(5995):1018–1019.

53. Oil Spill Commission Action (2012) Assessing Progress:Implementing the Recommendations of the NationalOil Spill Commission. Available at http://oscaction.org/osca-assessment-report-april-2012/. Accessed Sep-tember 30, 2012.

54. Gagosian RB, D’Elia CF (July 27, 2010) Research on gulfoil spill shouldn’t take a backseat to litigation. Washing-ton Post. Available at www.washingtonpost.com/wp-dyn/content/article/2010/07/26/AR2010072604443.html.Accessed November 7, 2012.

55. Board of Editors (2010) Embargo Policies for Science.Available at http://www.sciencemag.org/site/feature/con-tribinfo/faq/index.xhtml#embargo_faq. Accessed April25, 2012.

56. Garcia HE, et al. (2010)World Ocean Atlas 2009, Volume3: Dissolved Oxygen, Apparent Oxygen Utilization, andOxygen Saturation, NOAA Atlas NESDIS 70, ed Levitus S(US Government Printing Office, Washington, DC).

57. Wanninkhof R, Park G-H, Berberian G (2011) Oxygenwinkler titrations by NOAA/AOML in support of DeepWater Horizon spill monitoring. NOAA technical mem-orandum OAR AOML-99, p 15. Available at www.aoml.noaa.gov/ocd/gcc/co2research/datareports/AOML%20Tech%20Memo-99.pdf. Accessed November 9, 2012.

Lubchenco et al. PNAS | December 11, 2012 | vol. 109 | no. 50 | 20221

Dow

nloa

ded

by g

uest

on

Mar

ch 2

4, 2

020