Social Network and Content Analysis of the North American Carbon Program as a Scientific Community of Practice

SocialNetworks44(2016)226–237

ContentslistsavailableatScienceDirect

Social

Networks

j ourna l h o me pa g e :w w w . e l s e v i e r . c o m / l o c a te / s o c n e t

Social

network

and

content

analysis

of

the

North

American

Carbon

Program

as

a

scientific

community

of

practice

Molly

E.

Brown

_,

_Monica

_Ihli

_,

_Oscar

_Hendrick

_,

_Sabrina

_{Delgado-Arias}

_,

Vanessa

M.

Escobar

_,

_Peter

_Griffith

a_{DepartmentofGeographicalSciences,UniversityofMaryland,CollegePark,MD20740,UnitedStates} b_{UniversityofTennessee,Knoxville,TN37996,UnitedStates}

c_{ScienceSystemsandApplicationsInc.,NASAGoddardSpaceFlightCenter,Code618,Greenbelt,MD20771,UnitedStates}

a

r

t

i

c

l

e

i

n

f

o

Keywords:

NorthAmericanCarbonProgram Carboncycle

Knowledgedomain Knowledgevisualization Communitiesofpractice Co-authorshipnetworkanalysis

a

b

s

t

r

a

c

t

TheNorthAmericanCarbonProgram(NACP)wasformedtofurther thescientificunderstandingof sources,sinks,andstocksofcarboninEarth’senvironment.Carboncyclescienceintegrates multidis-ciplinaryresearch,providingdecision-supportinformationformanagingclimateandcarbon-related changeacrossmultiplesectorsofsociety.Thisinvestigationusestheconceptualframeworkof com-munitiesofpractice(CoP)toexploretherolethattheNACPhasplayedinconnectingresearchersintoa carboncycleknowledgenetwork,andinenablingthemtoconductphysicalsciencethatincludesideas fromsocialscience.ACoPdescribesthecommunitiesformedwhenpeopleconsistentlyengageinshared communicationandactivitiestowardacommonpassionorlearninggoal.WeapplytheCoPmodelby usingkeywordanalysisofabstractsfromscientificpublicationstoanalyzetheresearchoutputsofthe NACPintermsofitsknowledgedomain.Wealsoconstructaco-authorshipnetworkfromthepublications ofcoreNACPmembers,describethestructureandsocialpathwayswithinthecommunity.Resultsofthe contentanalysisindicatethattheNACPcommunityofpracticehassubstantiallyexpandeditsresearch onhumanandsocialimpactsonthecarboncycle,contributingtoabetterunderstandingofhowhuman andphysicalprocessesinteractwithoneanother.Resultsoftheco-authorshipsocialnetwork analy-sisdemonstratethattheNACPhasformedatightlyconnectedcommunitywithmanysocialpathways throughwhichknowledgemayflow,andthatithasalsoexpandeditsnetworkofinstitutionsinvolved incarboncycleresearchoverthepastsevenyears.

©2015ElsevierB.V.Allrightsreserved.

1. Introduction

Climatechangehasemergedasasignificantscientific,socialand

economicchallengetosociety(IPCC,2014).Understandinghow

climatechangemayevolveoverthecomingdecadesrequires

sig-nificantinvestmentinresearchaboutcarbonandhowit cycles,

throughbothlivingand nonlivingstates(Smil,1996).Scientists

frequentlystudy thesebiogeochemical cyclesin the context of

subsystems suchas theterrestrial biosphere(land-based living

systems),oceanic systems(bothorganicandinorganicforms of

carbon),andtheatmosphere(Falkowskietal.,2000).These

investi-gationsmayalsoincludethespecificrolehumansplayinthecarbon

cycle,suchas theimpactof human-generatedemissionsorthe

consequencesofclimatechangetoagricultureandfoodsystems

∗ Correspondingauthor.

E-mailaddress:[email protected](M.E.Brown).

(Berthelotetal.,2002;BradbearandFriel,2013;Dempewolfetal., 2014;Shindelletal.,2012).Carboncyclescienceisrelevanttoa

greatmanyaspectsoflifeasweknowit:theconditionofour

envi-ronment,thequalityofairwebreathe,waterresources,thefood

thatweeat,andtheenergyweconsume.

Engagingacrossthesocialandphysicalsciencestoembraceall

aspectsofthecarboncycleisverychallenging,particularlywhen

theimplicationsoftheresearcharebothpoliticalandeconomic.

TheNorthAmericanCarbonProgram(NACP)isoneofthefew

pro-gramsonthistopictohostcollaborativeactivitiescuttingacross

allcarboncyclesciencedisciplines,andpromotingopportunitiesto

fosterinterdisciplinaryandintramuralcollaborationwhose

objec-tiveitistodointerdisciplinaryresearchthatresultsininformation

that can be directly relevant to critical social decision making

(Michalaketal.,2011).Centraltotheprogram’sscienceagendais

theengagementofsocial,economicandpolicy-relevantresearchin

ordertoimprovehowcarboncyclescienceisconductedtoensure

policy-relevantfindings.Thispaperusescommunitiesofpractice

http://dx.doi.org/10.1016/j.socnet.2015.10.002 0378-8733/©2015ElsevierB.V.Allrightsreserved.

asaconceptualmodelforexploringhowtheNACPhasfaredin

cre-atingsuchaknowledgenetwork,bothintermsofmeasuringthe

connectivityamongprogramparticipants,andintermsof

incorpo-ratingmeasuringsocial,economicandpolicy-relevanttopicsinto

carboncyclescienceresearch.

1.1. HistoryoftheNACP

TheNACPwasformallyrecognizedbytheUnitedStatesin2002

underthemantleofthenation’soverallclimatechange

manage-mentstrategy(WofsyandHarriss,2002).Thefirstimplementation

planfortheNACPputforwardaresearchagendathatwas

cen-teredonquantifyingandunderstandingcarbonsinksandsources

inNorthAmericaandsurroundingoceans,andtheintegrationof

suchinformationintosocially,economically,andpolitically

rele-vantdecision-supportsystems(SarmientoandWofsy,1999;Wofsy

andHarriss,2002).TheStateoftheCarbonCycleReportestablished

thatNorthAmericaisanetsourceofCO2totheatmosphere,dueto

fossil-fuelemissionsandthattherearegloballyimportantcarbon

sinkswhosefutureishighlyuncertain(Kingetal.,2007).

Under-standinghowhumansbothexperienceandinfluencethecarbon

cycleandclimatechangeiscriticaltotheinterestsofdecision

mak-ers(Bernabo,1995;FeldmanandIngram,2009),suchasthosewho

confersupportupon thememberagenciesoftheNACPthrough

fundingandotherresources.

In 2007, The US North American Carbon Program (NACP)

sponsored itsfirst “all-scientist” meeting toreview progress in

understandingthedynamicsofthecarboncycleofNorthAmerica

andadjacentoceans,andtochartacourseforimprovedintegration

acrossscientificdisciplines,scales,andEarthsystemboundaries

(Birdsey et al., 2007). Following this meeting, a 2011 US

Car-bonCycleSciencePlanwaspublishedthatsetforthprioritiesfor

researchincarboncyclescienceforthecomingdecade(Michalak

etal.,2011).Inadditiontoreaffirmingtheneedforbasicresearch

and for continuingtraditional researchin carbon cyclescience,

theplanrecommendedsubstantialexpansioninresearchonthe

efficacyandenvironmentalconsequencesofcarbonmanagement

policies,strategies,andtechnologies;prioritizationofresearchon

humanelements of thecarbon cycle; anincreased exploration

ofthedirectimpactofrisinggreenhousegasconcentrationsand

carbon-managementdecisionsonecosystems,species,andnatural

resources;andresearchonhowtoexpressuncertaintyinallaspects

oftheglobalcarboncycleaswellasimprovedwaysofconveying

thoseuncertaintiestopolicyanddecisionmakers,aswellassociety

atlarge. Toachievetheseobjectives,thereportauthors

recom-mendedasubstantialfocusonconductingresearchthatintegrates

humandimensionswiththe biologic,atmospheric, and oceanic

sciences.Socialprocessesthatdrivelanduseandfossilfuel

emis-sionsshouldbequantitativelyintegratedintolanduse/coverand

emissionsmodelingtopromotetheintegratedcarbon,climate,and

socialmodelingneededtoprovidescienceandanalyticaltoolsfor

climateactionprogramsatvariouslevelsofgovernment(Michalak

etal.,2011).

The challenges facing the North American Carbon Program

bringtolightthelargerissueastohoworganizations,agencies,

andnationsatanylevelcancultivatethedevelopmentof

inter-organizational and interdisciplinary networks targeted toward

creationofspecifickindsofknowledgeresources.Tothateffect,this

paperseekstoapplyasystemicapproachforassessingthe

knowl-edgecreationthattakesplacewithinaresearchprogramsuchas

theNACP.Howmightwecomparetheprofessedknowledgegoals

oftheNACP,orsimilarprograms,totheactualknowledgecreated

byparticipants?Wealsoconsiderhowtodescribethestateof

col-laborationsbetweenparticipantswithinsuchacommunity.How

docollaborationsamongstcoreparticipantsgrowandchangeover

time?Aretherechangesinresearchers’tendenciestocollaborate

acrossinstitutionalboundariesoverthesameperiodoftime?These

arethequestionsweseektoanswerinanalyzingtheNACPasa

communityofpractice.

2. Theoryandrationale

Acommunityofpracticeisdefinedas“agroupofpeoplewho

shareacommonsetofproblems,orapassionaboutatopic,andwho

deepentheirknowledgeandexpertiseinthisareabyinteracting

onanongoingbasis”.Wengeretal.(2002)describethree

struc-turalelementstotheCoPmodel:domain,community,andpractice.

Domainreferstotheknowledgeconcernsandissuesaroundwhich

theCoPisstructured.Awell-definedknowledgedomaintranslates

toastrongsenseofpurpose,guidingtheactivitiesofmembers.It

alsoimpliesasharedcompetenceandcommitmenttothesubject

matter. Domain manifests as thespecific knowledge the

com-munitydevelops,shares,andmaintains.Thecommunityelement

referencesthesocialenvironmentitself:thepeopleand

relation-shipsthroughwhichlearning,knowledgetransfer,andknowledge

creationtakesplace.Practiceconcernsalloftheirrituals,systems

ofmeaning,andchannelsofcommunication.

TheCoPmodelprovidesatheoreticalfoundationuponwhich

tobasediscussionandanalysisofthescientificcommunityand

itsresearch.Structuralelementsofthemodelaidusin

commu-nicating fundamental assumptions aswell aslimitations of the

study(Wengeretal.,2002).Additionally,itaidsusin

understand-ingandexpressingtherelationshipsand distinctionsbetweena

community,individualmembers,andseparatebutparticipatory

institutions that provide support to scientists. Other examples

oftheCoPmodelbeingemployedtostudyknowledgenetworks

via shared resources and sustained interaction, including an

ethnographicstudyofclimatechangeadaptationprojectsatthe

science-policyinterface(Iyalomheetal.,2013).

2.1. CommunityoftheNACP

Thereisnoasingleformofsocialstructurewhichqualifiesasa

communityofpractice,andmembershipisnotbyvirtueof

tradi-tionalorganizationalordepartmentalboundaries.Thesizeofthe

community couldbeless thanten,or itcouldnumber intothe

hundreds. Itmightbecommunityof individualswho allliveor

workincloseproximitytooneanother,oritcouldbedistributed

acrossawiderangeofgeographicallocationsandorganizational

boundaries.TheCoPmodeldoes,however,predictanapproximate

distributionofmemberparticipationwhichcorrespondstothree

broadlevelsofinvestmentwithinthecommunity:(1)asmallgroup

of coremembers whoboth attendmeetings regularlyand who

alsooverseefunctionaltasks,(2)activememberswhoregularly

attendmeetings,and(3)peripheralmemberswhoonly

occasion-allyparticipateinthecommunity(Wenger etal.,2002).Aspart

ofconceptualizingtheNACPasacommunityofpractice,wewill

considerifthedistributionofparticipationfrequencyinmeetings

showsanyagreementtothedistributionsuggestedbythemodel.

We willdescribe and analyzethecommunity in terms of

rela-tionshipsbetweencoreparticipantsusingsocialnetworkanalysis

methods.AstheNACPalsoseekstoincreasecollaborationbetween

differentinstitutionsstudyingthecarboncycle,wewill

further-morelookathowtherelationshipsbetweenindividualstranslate

toconnectionsbetweentheinstitutionstheyrepresent.

2.2. KnowledgedomainoftheNACP

Adomainis“astatementofwhatknowledgethecommunitywill

steward”and“acommitmenttotakeresponsibilityforanareaof

expertise”(Wengeretal.,2002).Wehavenotedthattheknowledge

Plan(Michalaketal.,2011),theScienceImplementationStrategy

fortheNorthAmericanCarbonProgramof2005(Denning,2005),

andsimilarsupportingdocuments.ThedomainoftheNorth

Amer-icanCarbonProgramistostudythesourcesandsinksofcarbon

withtheexpectationthatresultingknowledgeshouldultimatelybe

accessibleandsalienttostakeholdersatavarietyoflevels.Although

scientificresearchdoesnothaveasimplecause-and-effect

rela-tionshipwithimprovedsocietaloutcomes,someresearchhasled

toreducedexposuretoextremeevents,improvedforest

manage-ment,andeffective policiestoensureagriculturalsustainability

(Rosenzweigetal.,2014).Rather,scienceisoneamongmany

com-plexfactorsthatinfluenceimprovedsocietaloutcomes(Pielkeetal.,

2010).Itisinherenttothedomainthatnaturalsciencesbe

inte-gratedwiththestudyofhumanprocesses.

However,the integrationof social and humanaspects in to

carbon science is challenged by the need for translation and

cooperationbetweendifferentkindsofstakeholders.Researchers

tendtointeractmorecloselywithotherresearchersintheirown

fields,whichcanfrustrateinterdisciplinarycooperationamongst

thosewhostudynaturalsciences,socialsciences,andeconomics

(Feldman and Ingram, 2009).This is what is calledfor

accord-ingtotheNACPstrategicplan,butaredifferentdisciplinestruly

integratedintotheknowledgecreatedbytheNACP?Thecontent

analysissectionofourresearchmethodswillcomparethedomain

oftheNACPagainsttheactualknowledgeproducedinpractice,

inorderthatwemayevaluateifthehumanprocessesaretruly

representedasnecessarytofulfilltheprogram’sgoals.

2.3. PracticeoftheNACP

A CoP is best characterized by consistent engagement in a

domain-drivenpracticeovertime.Sharedpracticerefersnotonly

totheactionsandchannelsofcommunicationbetweenmembers,

butalsototheactualcodifiedknowledgethatiscreatedasaresult

oftheinteractionsbetweenmembers.Forthisreason,theexistence

ofinterpersonalrelationshipsisakeyfeature–researchdoesnot

simplyhappentoco-occurwhenpeoplearephysicallypresentat

thesamemeetings.Interactionsoccurthroughandinthecourseof

sharedpracticesovertimeandspace.

IftheNACPistrulyacommunity,thenmembersshouldbe

inter-actingwithoneanother,andtheseinteractionsshouldmanifestas

knowledgeoutputs.Weshouldbeabletoobservesomethingabout

thestateofthecommunity,aswellasitsevolution,bydescribing

theserelationships.Knowledgeproductsofrelationshipsbetween

scientiststypicallyoccurintheformofpublishedresearch(Wenger

etal.,2002)andcoauthored projects.NACPmembersalso

con-tributetoknowledgeoutputsintheformofofficialreports(King

etal.,2007;Michalaketal.,2011).

Thecontinuedgrowthanddevelopmentofacommunityrests

ontheabilitytogaugeitseffectivenessincultivatingthe

knowl-edgedomainbywhichitisbound(Wengeretal.,2002).Itwould

bethereforeofinteresttotheNACPtohaveamethodtoassess

thecommunity’sprogressofintegratingcarboncyclesciencewith

humanprocesses toprovidesupportfordecision making.

Peer-reviewedpublications andfunded projectsprovide aconsistent

andaccessibleformofknowledgeoutput.Herewewillusecontent

analysisandsocialnetworkanalysistodescribetheknowledge

out-putsandmember-to-memberrelationships,respectively.Tothat

effect,akeywordanalysisisperformedontheabstractsofNACP

outputsinordertoderivethetopicsofresearchbeingproduced

sothatitcanbecomparedtotheknowledgegoalsoftheprogram.

Co-authorshiprelationshipsarealsoderivedfromthebibliographic

dataofarticlesandofprojectsaspartofoursocialnetworkanalysis.

Wewilldemonstratethevalueofthesetoolsforhelpinga

commu-nityofpracticesuchastheNACPtocontinuallyassessandre-orient

itselftowarditsknowledgecreationgoals.

3. Data

WeusedtwodatasetstoexplorethedegreetowhichtheNACP

isacommunityofpracticethathasfosteredthegoalstoexpandthe

program’sdisciplinaryfocus,assetoutinthe2011NACPplan:the

NACPprojectdatabaseandabibliographicdatasetfromtheWebof

Science.Thesearedescribedbelow.

3.1. NACPprojectdatabase

WeusedNACP’sownauthordatabase(foundatwww.nacarbon.

org)to identify 1070 individuals who are engaged in 408

sci-entificprojectsregisteredwiththeNACP,includingauthor,title,

abstractandkeywordsandusedthesetoanalyzethe

connectiv-itybetweenNACPmembers.Wenarrowdownto1007members,

includingonlymemberswhohaveworkedwithatleastoneother

person in a project. Scientists and projects are includedin the

databaseifaprojectisfundedthroughextramuralawards(grants,

cooperativeagreements,contracts,andinteragencytransfers)and

designatedaspartoftheNACPatthetimeofselectionandaward,

oraredesignatedaspartoftheNACPsubsequenttotheaward

beingmade.Theseprojectsaddressthegoalsandobjectivesofthe

NACPandcontributerelevantresearchproductswithinthe

time-frameoftheprogram.Coreprojectsalsoincludestandingagency

activities,includingthosethatmayhaveanotherprimarymission

but agencyprogram leadscommit toparticipatein NACPorto

providespecificmeasurements,datasets,orotherproducts.

Affil-iatedprojectsarethosethatareidentifiedafterselectionorafter

completionbyagencyprogrammanagersbecausetheyare

investi-gations,researchprojects,oroperationalprogramsorprojectsthat

arerelevantorprovidespecializeddatasetsorservicesimportant

totheNACPcommunity.Weusedtheprojectdatabasetoidentify

connectivitybetweenNACPscientiststhatparticipatedonprojects

together,andindeterminingthetopicsrepresentedbytheirstudies

usedtheabstractsprovidedintheNACPdatabase.

3.2. BibliographicinformationfromtheWebofScience

BasedonmeetingregistrationrecordsexportedfromtheNACP’s

online database, we found that 808 unique participants had

attendedthefourNACPmeetingsbetween2007and2013.Forty

ofthese(approximately5%)attendedallfourmeetings,79people

attendedthreemeetings(10%),184people(23%)participatedin

twomeetings,and505people(63%ofparticipants)attendedone

meeting.Tolimitthebibliographicanalysistoamanageable

sam-ple,weusedthe15%(113individuals)whoattendedbetweenthree

andfouroftheNACPmeetingsinthepastsixyears,obtainingthe

familyname,firstinitial,andorganizationalaffiliationfromNACP

records.Bibliographicdataandabstractsforarticleswerecollected

bysearchingtheMultidisciplinary ISIWebof Science(accessed

between12-06-2013and01-03-2014).We locatedpapers

writ-tenbyallindividuals,resultinginadatasetof2447peer-reviewed

articlespublishedin511journalsbetween2007and2013.Boththe

registrationrecordsofindividualstakenfromthewww.nacarbon.

orgwebsiteandthebibliographicarticledatawereimportedinto

ourprojectdatabaseandlinkedinarelationalschema.Additional

preprocessingincludedthestandardizationoforganizational

affil-iationdataintheregistrationrecords.

Thenetworkanalysisportionofourresearchrequiredseveral

additionalstagesofpreprocessingofthearticledata.Forreasons

furtherdiscussedintheappropriatesectionofourmethods,we

firstreducedthedatasettoonlyincludethose articlesthat had

beenpublishedbygreaterthanoneNACPcorememberfromour

sample.Thisleftuswith363articlesrepresentingtheworkof99

persons.Foranarticletobeincludedinoursample,eacharticle

Table1

Keywordlistsbycategory.

Economy Wordssignifyingtopicsofinformationoneconomy-relatedissues,whichconnecttotheevolutionofhuman

carbonsystemdriversandalsorelatetoissuesimportanttodecisionmakers.

Commandeconomy Food Foodsource

Consumer Foodavailability Foodsources

Costbenefit Fooddemand Foodtrade

Economic Fooddemands GDP

Economicactivity Fooddistribution Goodsandservices

Economicforecasting Foodinsecurity Grossnationalproduct

Economicmodel Foodpolicies Industrial

Economicmodels Foodpolicy Industry

Economicprojections Foodproducing Inflation

Economicsectors Foodproduction Integratedassessment

Economical Foodsecurity Investment

Economies Foodshortage Markets

Economy Foodshortages Socioeconomic

Fiscalpolicy Wastemanagement

Energy Wordsassociatedwithinformationneededintheenergysector,suchassourcesofenergy,energy

managementconcerns,andsustainabilityofsources.

Bioenergy Energypolicies Energyuses

Biofuel Energypolicy Fossilfuel

Biofuels Energyportfolio Fuel

Dam Energyproduction Fuels

Electrical Energyresources Hydroelectric

Electricity Energysource Oilandgas

Energybudget Energysources Renewableenergy

Energyconsumption Energysupplies Sustainableenergy

Energyefficiency Energysupply Transportation

Energyforecasts Energysustainability Windfield

Energyharvesting Energyuse Windfields

Energymanagement Windpower

LandUse&Water Wordsassociatedwithhumanfactorsthatdeterminecarbonemissionsfromlanduse&watercycle.

Agricultural Forestmanagement Landuses

Agriculturally Forestpolicies Maize

Agriculture Forestpolicy Ownership

Agroforestry Grazing Pasture

Corn Harvest Soybean

Crop Harvested Urban

Cropland Harvests Wateravailability

Cropping Landmanagement Watermanagement

Crops Landownership Waterquality

Farming Landproductivity Watersupplies

Fisheries Landuse Watersupply

Fishing Watertreatment

HumanImpacts Wordssignifyingcontentdiscussinghowhumansinteractwithandinfluencethecarboncycle.

Anthropogenic Emissions Humanactivities

Deforestation GHG Humanactivity

Disturbance Greenhousegas Pollute

Disturbances Greenhousegases Polluting

Emission Pollution

Society&DecisionMaking Wordswhichalsogeneralsignifiersoftopicsthatmakecarboncyclesciencerelevanttodecisionmakersand managerswhomustconsiderthevaluesandneedsoftheirconstituents.

Adaptation Operations REGULATION

Attitudes Planning REGULATIONS

Cultural Policies RESOURCES

Culture Policy RISK

Decision Political SOCIAL

Decisions Practice SOCIOPOLITICAL

Internationalpolicy Practices STAKEHOLDERS

Manage Regulating STRATEGIES

Management STRATEGY

Managers ZONING

Strategies Wordssignifyingspecificresponsesandstrategiesformanaginghumanimpactstothecarboncycle.

Carbonaccounting Conserve Offsetting

Carbonbudget Emissionstrading Renewable

Carbonoffset Mitigation Sequestered

Carbonoffsets Monitoring,accounting,and Sequestration

Carbonprogram Reporting Sustainability

Carbontrade MRV Sustainable

anotherthroughinclusionoftwonamesintheauthorlist. Con-nectionstopeopleoutsideoursamplewerenotrepresented.Next, weprogrammaticallyrestructuredthedatasothatanexportof allrecordsrelatingpeopletoarticleswouldbeconvertedintoa tablerepresentingallpossiblecombinationsofpeopleconnected tootherpeople,wheretheyhadauthoredthesamedocument. 4. Methods

4.1. Contentanalysisofpublicationsandprojects

Scientometricmappingandanalysisusingbibliographicdatais awell-establishedmethodology,forwhichavarietyofapproaches, techniques, and automated tools have been developed (Cobo

etal.,2011).Existingresearchhasdemonstratedthatanalysisof

abstractsandtitlesinbibliographicdatacanyieldinsightintothe

knowledgedomainsrepresentedwithinthedataset(Chen,2006).

Contentanalysismaybeapproachedeitherinductively,inapurely

exploratorycontext,ordeductivelywhenseekingtotestknown

ideasorcomparechangesincontentovertime(EloandKyngäs,

2008).Our keyword analysiswas conductedonboth theNACP

databaseandthebibliographicdatabyimportingtheauthor,title

andabstractinformationfrombotharticlesandprojectsintoatext

analysissoftware.Becausewedesiredtocomparetheknownand

publishedknowledgegoalsoftheprogramtoresearchoutputs,we

assembledadictionaryoftermsandphrases(Table1)that

refer-encedifferentaspectsofhumanprocessesandexperiencesrelevant

tointegrationwithcarboncyclescience.Theinitialsourceof

key-wordsweretermsandideasusedinMichalaketal.(2011),and

thesewerethenevaluatedinthecontextofabstracttextstoconfirm

theirappropriatenessforreportingoninakeywordanalysis.The

topthirtyjournalsforallarticles(2447)versusarticlesreturning

atleastonekeywordfromthehumanprocessescategories(1397)

aresummarizedinTable2.

Somewords,suchas“biofuel”,areveryexplicitintheir

mean-ingandrelevance.Otherwords,suchas“policy”or“decision”,are

inherentlyassociatedintheEnglishlanguagewithhumanagency

orperspective.Wehavelittlereasontobeconcernedforfalse

posi-tivesbecausewedonotspeakoftrees“deciding”thingsorclouds

developinga“policy”.However,thesewordscanstillbepairedwith

otherwordstocreatemoreexplicitmeaning,suchaswith“energy

policy”.Ontheotherhand,“energy”alonewouldriskconfusion

withanynumberofscientificprocessesdescribedinourabstracts

thathave nothingtodo withhumans.Therefore, we couldnot

includeawordsuchas“energy”withoutfurtherqualificationin

ordertoensureourresultsaremeaningful.

Thecategoriesareneithermutuallyexclusivenorcollectively

exhaustive.Somekeywordswouldhavecausedanarticletofall

intomorethanonecategory.Forexample,thetext“energypolicy”

willflaganabstractintothegeneralcategoryof“Social&Decision

Making”categoryduetothewordpolicy,aswellasthemore

spe-cificEnergycategory.Thiswasdeliberate;itwasdeterminedour

analysiswouldbemoreinformativeifwedidnotenforcearuleof

mutualexclusivity.

4.2. Socialnetworkanalysisofco-authorshipties

WeconductedanetworkanalysisonboththeNACPdatabase

andthebibliographicdata,andpresentnetworkgraphsresulting

fromtheanalysis.Networkanalysisisa researchapproach that

prioritizesrelationshipsbetweensocialunitsasopposedto

focus-ingonattributesoftheindividualunitsthemselves(Wasserman

and Faust, 1994). Here we report on co-authorship networks

thatdescriberelationshipsbetweenpeoplerather thanideas.A

document“is co-authored ifit hasmore than one author. It is

Table2

TopthirtyjournalsforallarticlesintheWebofSciencearticledatasetandforthe sub-setofarticlescodedwithhumanprocesskeywordsinthecontentanalysisportion ofthestudy.

Rank Journals Articles %oftotal

Top30journalsinarticlesdataset

1 RemoteSens.Environ. 148 6

2 J.Geophys.Res.Biogeosci. 123 5

3 Agric.For.Meteorol. 101 4

4 Glob.ChangeBiol. 89 4

5 J.Geophys.Res.Atmos. 85 3

6 Biogeosciences 71 3

7 Atmos.Chem.Phys. 54 2

8 Geophys.Res.Lett. 45 2

9 For.Ecol.Manage. 41 2

10 Glob.Biogeochem.Cycle 39 2

11 Environ.Res.Lett. 38 2

12 Ecol.Model. 37 2

13 Ecol.Appl. 35 1

14 Proc.Natl.Acad.Sci.U.S.A. 31 1

15 Int.J.RemoteSens. 29 1

16 Agron.J. 28 1

17 TellusSer.B:Chem.Phys.Meteorol. 26 1

18 Can.J.For.Res.Rev.Can.Rech.For. 24 1

19 Ecosystems 23 1

20 J.Geophys.Res.Oceans 22 1

21 PLoSOne 20 1

22 Atmos.Meas.Tech. 19 1

23 J.Environ.Manage. 18 1

24 Atmos.Environ. 18 1

25 IEEETrans.Geosci.RemoteSensing 16 1

26 Agric.Ecosyst.Environ. 16 1

27 SoilSci.Soc.Am.J. 16 1

28 Clim.Change 15 1

29 Bull.Amer.Meteorol.Soc. 15 1

30 Biogeochemistry 15 1

Top30journalsamongarticlescodedw/humanprocesseskeywords(n=1397)

1 RemoteSens.Environ. 87 6.2

2 J.Geophys.Res.Biogeosci. 62 4.4

3 Agric.For.Meteorol. 53 3.8

4 J.Geophys.Res.Atmos. 52 3.7

5 Glob.ChangeBiol. 49 3.5

6 Biogeosciences 47 3.4

7 Atmos.Chem.Phys. 36 2.6

8 For.Ecol.Manage. 35 2.5

9 Geophys.Res.Lett. 25 1.8

10 Ecol.Appl. 25 1.8

11 Environ.Res.Lett. 25 1.8

12 Agron.J. 24 1.7

13 Glob.Biogeochem.Cycle 23 1.6

14 Proc.Natl.Acad.Sci.U.S.A. 22 1.6

15 Ecol.Model. 22 1.6

16 TellusSer.B:Chem.Phys.Meteorol. 17 1.2

17 Can.J.For.Res.Rev.Can.Rech.For. 16 1.1

18 Agric.Ecosyst.Environ. 15 1.1

19 PLoSOne 15 1.1

20 Clim.Change 14 1.0

21 Ecosystems 12 0.9

22 SoilSci.Soc.Am.J. 12 0.9

23 Environ.Sci.Technol. 12 0.9

24 Bioscience 11 0.8

25 GCBBioenergy 10 0.7

26 Int.J.RemoteSens. 10 0.7

27 Atmos.Environ. 9 0.6

28 J.Clim. 9 0.6

29 Nature 9 0.6

30 J.Environ.Manage. 9 0.6

institutionallyco-authoredifithasmorethanoneauthoraddress, suggestingthattheauthorscomefromvariousinstitutions, depart-ments,orotherkindsofunits”(MelinandPersson,1996).

Manystudieshaveusedtherelationshipsbetweenauthorsof

adocument tostudybroaddomainssuchasphysics(Newman,

2001a)and mathematics(Barabási et al., 2002). Morerecently,

studieshavealsoemployednarrowersamplingapproaches,such

orbasedonattendancerecordsofparticipationfortheactive

mem-bersofacommunityofbusiness(Morlacchietal.,2008).Herewe

transformbothourdatasetsintoco-authorshipnetworkto

repre-sentinterpersonalrelationshipspresentintheNACP.Peopleare

generallyexpectedtocommunicateandinteractwithoneanother

in order to publisha document or conduct a projecttogether

(Newman,2001a).

WetookbibliographicdatafromtheWebofScienceand

con-vertedintoanetworkformatbyproducingallpossiblepairsoftwo

authorsforeacharticle.Thenumberofnetworktiesproducedby

eacharticlemaybeexpressedintermsofcombinationsasC(n,r),

where r=2 and nis the number of authorsfor a given article.

Wethencreatedabinarycodeforeachrowofnetworkdata

pro-ducedbythismethodtoidentifywhetherornottheconnection

wasbetweentwoauthorsofthesameinstitution.Asecond

co-authorshipnetwork wasalsocreatedfromthis data,butonein

whichwe representedauthors’institutionsasthenodes,and in

whichconnectionswithinthesameinstitutionswereignoredas

self-loops.

Weinvokedasimilarstrategyforthedataextractedfromthe

NACPprojectdatabase.Dataherewasrepresentedasa

relation-shipbetweenaprojectandoneormorepeople.Weusedthesame

methodto convertthis information to all possibleconnections

betweenpairsofresearcherswhoweresaidtohaveparticipated

onagivenproject.Cross-institutionalcodingwasnotperformed

ontheprojectdataset.

Networkscanbeanalyzedtolookforinformationaboutthe

net-workasawhole,ortolookforindividualsofinterest,suchaswhich

personismostcentralormostconnectedinanetwork.Wetended

toevaluateourgraphsforinformationaboutaveragebehaviorof

nodesasawhole,oratthenetworklevelofanalysis.Forexample,

modularityisthefractionoftheconnectionswithingivengroups

minustheexpectedsuchfractionifconnectionsweredistributed

atrandom (Newman, 2006).The modularityscore giventothe

networkscanbeinterpretedasdescribinghowwellanetwork,in

ourcasetheNACP,canbeseenasacollectionofsub-networks,or

isascoreofhowwellorganizedthenetworkis.Sincethe

modu-larityscoremeasuresthequalityofthepartitioningofthegraph

intosmallercommunities,thefactthatthenetworkshavecloseto

similarscoresmeansthattheyshareasimilarinternalcommunity

structure.

Theaverageclusteringcoefficientisameasureofhowconnected

aneighborhoodofnodesare.Fora particularnode(scientist),a

neighborhoodisallnodesconnectedtothisparticularnode.Ifall

thenodesintheneighborhoodofascientist-nodeareconnected

tootherscientist-nodesofthatneighborhood,thentheclustering

coefficientofthatneighborhoodwillbe1;ifnoneoftheothernodes

intheneighborhoodareconnectedtoeachotherthentheclustering

coefficientwillbezero(Latapyetal.,2008).Inotherwords,the

clusteringcoefficientofa nodethathasatleastoneothernode

connectedtoit,istheprobabilitythatanytworandomlychosen

neighborsareconnected.Thisprobabilityiscalculatedbydividing

thenumberoftrianglescontainingourparticularscientistbythe

numberofpossibleconnectionsbetweenhisneighbors(31).The

averageclusteringcoefficientistheaverageofthesescoresoverall

theneighborhoods.

Inordertomeasuretheconnectednessoftheparticipantsof

theNACP,wetakeourtwodatasets,theNACPprojectdatabase

andjournalarticleswrittenbythemembersoftheNACP,anduse

themodularityframeworktounderstandnetworkswithinthe

pro-gram.Ifacommunityisconnectedbymultipleco-authorships,then

thecommunityisinterpretedtohavemultiplelevelsof

communi-cation,interactionandscientificengagement.Ifacommunityhas

networksthatareisolatedfromoneanotherandaresparse,then

thecommunityisassumed tobelessengagedand tohave less

collaboration.

Table3

Theoreticalandobservedparticipationdistributions.ThetheoreticalmodelforCoP (a)proposesthree levelsofcommunityinvolvementandsuggestsdistribution rangesfortheproportionofthecommunitywhichwillfallintoeachlevel;(b) summarizestheattendancefrequencydistributionforindividuals(n=808)who attendedthemostrecentfourNACPmeetingsbetween2007and2013.

(a)CoPmodelproposeddistribution (b)Observedattendancedistribution (n=808)

Core 10–15% 3–4Meetings 14.7%

Active 15–20% 2Meetings 22.8%

Periphery 65–75% 1Meeting 62.5%

OneofthepropertieswelookforinthenetworkofNACPcore co-authorsistheexistenceofagiantcomponent.Componentsrefer togroupsofnodes,whicharehighlyinter-connectedtoeachother. Wherefewconnectionsexist,anetworkwillseemtobea scat-teredcollectionofisolatedgroupsofscientists.Thiscorresponds toa lackof socialpathwaysthroughwhich ideascanflow. But givenenoughconnections,isolatedgroupsmayconvergeintoa sin-glemass,whichencompassesthemajorityofnodes,andinwhich everyoneispotentiallyconnectedtoeveryoneelsebysomepath (Newman,2001b).

5. Results

Table3showsthethreelevelsofinvolvementandtheir

distri-butions,asproposedbytheCoPmodel,alongwiththesummary

distributionofmeetingattendancefrequenciesintothree

corre-spondingclasses.Thenearlytwo-thirdsofthe808individualswho

onlyattendedonemeetingbetween2007and2013arecompared

totheperipheralcomponentoftheCoPmodel.Thealmost23%of

individualswhoattendedhalfthemeetingsarecomparedtothe

activecomponent.Thenearly15%whoattendedthemajorityorall

meetingsarecomparedtotheprimarycomponentofacommunity

ofpractice.Thisdistributionofmembersbetweenthecore

mem-bers,activemembers,andperipheralmembersreflectsthetheory

ofaCommunityofPractice.

We identified 215 institutions, organizations, and agencies

acrossthe808participants.Althoughmanyoftheinstitutionsdid

notaccountformorethanasingleparticipant,31%ofthe

insti-tutionscouldaccountfor74%ofthepeople.Amongthesubsetof

113coremembers,wecounted79institutions,25%ofwhichcould

accountfor58%ofthepeopleinthesubset.

5.1. Resultsofcontentanalysis

Ofthe2447articleswrittenbyatleastoneNACPcoremember

from2007to2013,57%or1397containedatleastone

human-relatedcarboncyclesciencekeyword.Approximately63%(228)of

the363articlesco-authoredbyatleasttwoNACPcoremembers

containedatleastonekeyword.Wefoundageneralagreementin

thehuman-relatedcarboncyclescienceissuesdiscussedbetween

2007and2013intheresearchwrittenbyatleastoneNACPcore

memberandtheresearchco-authoredbyatleasttwoNACPcore

members.Overall,topics onSociety&DecisionMaking,Land&

WaterUse,andHumanImpactsweremoreprevalentthanthose

topics dealing withhuman-related issues pertaining to Energy,

StrategiesandEconomy(Fig.1aandb).

Thesameprevalencewasfoundintheanalysisoftheproject

database.Ofthe408totalprojectabstractsanalyzedfor

human-related carbon cycle science keywords, 66% or 268 projects

containedatleastonekeyword.Keywordsassociatedwith

Soci-etyandDecisionMakingwerefoundin60%oftheprojects,those

relatedtoLand&WaterUsein46%ofprojects,andthose

Fig.1.(a)ResultsfromcontentassessmentbyyearandcategoryfortheNACPCoreSubsetarticles(N=363)containingatleastonehuman-relatedcarboncyclescience keywordfromTable1.(b)Resultsbyyearandcategoryforallarticles(N=2447).

Fig.2. Resultsshowingthepercentofarticlesfrombothsamplesreferencingat leastonekeywordandthepercentprojectsreferencingatleastonekeyword.

intheprojectswerehuman-relatedcarboncyclesciencekeywords

associatedwithEnergy(24%),Strategies(23%),andEconomy(10%).

Anoverallincreaseinengagementwithhuman-relatedcarbon

cyclescienceissuesisnotedforbothNACPcorearticlesamples.

Fig.2alsoshowsthat,after2010,thosearticlesco-authoredbyat

leasttwoNACPcoremembersreferencedmorehuman-related

sci-encetopicsthanthosearticleswrittenbyatleastoneNACPcore

member.Thistrendpeaksin2011at73%andmaybeexplained

bythereleaseofthe2011USCarbonCycleSciencePlan(14).The

planprovideslong-termdirectionforguidingcarboncycleresearch

basedonthreeoverarchingquestions,whichincludetheeffectsof

humansoncarboncyclingandtheconsequencesofcarbon

man-agementdecisions.

Sincetheprojectdatabaseanalysisdoesnotprovideinformation

ontheevolutionthroughtimebecausetherearenodates

asso-ciatedwithprojectsinthedatabase,wecomparethepercentage

projectscontainingatleastonekeyword(66%)tothepercentage

oftotalarticlesfromthebibliographicdatabasefrom2007to2013

containingatleastonekeywordforbotharticlesamplesforeach

ofthehuman-relatedcarboncyclesciencecategories(Fig.3).The

NACPprojectsshowahigherfocusonSociety&DecisionMaking

(60%),followedbyHumanImpacts(56%)andLand&WaterUse

(46%).Articlesco-authoredbyatleasttwoNACPmembersfocus

moreonHumanImpacts(58%),Land&WaterUse(56%),followed

bySocietyandDecisionMaking(48%).Articleswrittenbyatleast

oneNACPcoremembershowagreatertendencyforcontaining

wordsrelatedtoSociety&DecisionMaking(53%),Land&Water

Use(51%)andHumanImpacts(44%).

Overall,therewasa29%increaseinthebibliographicjournal

articlesreferencingahuman-relatedcarboncyclesciencekeyword

forarticlesco-authoredby morethantwo NACPcoremembers

from2007to2013Sincethesearticleswerefromphysicalscience

andinterdisciplinaryjournalsonly,the29%increaseinpublications

Table4

NACPcorearticleauthorshipandnetworkconnectionstatisticsbyyear.Thistabledescribestheconnectionsderivedfromthe363articles,whichconnect99ofthemost activeparticipantsintheNACPbiannualmeetings.Theproportionoftieswhichcrossedinstitutionalboundariesiscomparedtothequantityofallco-authorshiptiesfor eachyear.

Publicationyear Articlespublished Co-authorship

ties

Tieswhichalsoconnected differentinstitutions

Ratiooftiescrossing institutionstoallties

2007 24 43 29 67% 2008 37 70 59 84% 2009 39 100 78 78% 2010 57 330 304 92% 2011 62 257 226 88% 2012 78 563 524 93% 2013 66 192 162 84% Total 363 1555 1382 89%

consideringsocial and humanaspects ofsocial science is quite large,particularlyconsideringtheshortperiodoftimeandthefact thatthegroupofresearchersdidnotchange.Ourresultsshowthat the individuals whoconsistentlyparticipate inthe NACP meetings areintegrating,eitherintentionallyornot,socialandhumanaspects ofthe carboncyclescienceintotheirwork,andhavebeenableto publisharticlesthatconsidertheseelementsinthephysicalscience andinterdisciplinaryjournals.TheknowledgedomainoftheNACP hasgrownoverthepastsevenyearstoincludesocialandhuman aspectsofthecarboncycle.

5.2. Resultsforthesocialnetworkanalysis

Table4reportsannualarticlestatisticsfortheaveragenumber

ofcoreNACPauthorsperarticle,aswellastheaveragenumberof

institutionsandnationstheyrepresented.Fig.4chartsthesame,

showingbriefspikesofcollaborativeactivityin2010and2012.

However,thesespikesofactivityarenotmaintainedovertimeand

sotheoverallincreaseinNACPcorecollaborationsperarticlefor

bothindividualsandinstitutionsismoremoderate.Table4also

reportsthesizeofthearticlenetworkintermsofco-authorship

tiesandalsoquantifiesthenumberofthesethatconstitute

cross-institutionalco-authorships,brokendownannually.Theratioof

cross-institutionaltiestoalltiesisalsorepresented,showingan

increasefrom67%in2007toanoverallaverageof89%by2013.

Fig.5showsagraphvisualizingtheco-authorshipnetworkfor

theNACPprojectdatabasethatdisplays1007peoplewith16,518

connectionsand15,716uniqueedges.Thefigurewascreatedusing

theFruchterman–ReingoldlayoutinGephi.Thealgorithmrepels

thenodesfromeachotherandusestheconnectionsbetweennodes

topullthemtowardoneanother.Theresultisanetworkgraph

whereonecanvisuallyseewell-connectedgroupsandhowthey

Fig.4.NACPcoreauthorshipstatisticsbyyear.

comparetoothergroupsinthegraph.In thisgraph,each node

representsa scientist;wherethedarker theblueandthelarger

thenodethegreaterconnectedthenodesareinrelationtoeach

other.Fig.5alsoshowstheperipheryandthecoreoftheproject

network–thesmallcirclesontheoutsidearepeoplewhoworkin

verysmallgroups.Atthecenterofthegraphwecanfindthelarge

nodeswithsolidbluecolors–thesearescientistswhoparticipate

inprojectsandcommitteesidentifiedbytheNACPtoacceleratethe

researchandworktogethertowardacommongoal.Nearthecenter

ofthegraphwecanfindthesmallerbluenodeswhichdonotappear

tobeasconnectedasthelargercircles,butstillmoreconnected

thanthescatterednodesintheperiphery.Thesenodesbelongto

thoseparticipatinginmorethanonetypeofprojectorareinacore

projectfundeddirectlybyNASA.Thecolorsoftheedgesinthegraph

representthekeywordorkeywordsthathighlighttherelationship

betweentheworkoftwoscientists.Ifthemostfrequentlyfound

word,withintheprojectabstractsoftwoscientists,wasbiofuels,

thenthekeywordassociatedbetweenthesewouldbeenergy.If

tiesoccur,thenweassigntherelationshipashavingmorethanone

keyword.Thegraphshows‘HumanImpacts’asthemostfrequent

keywordshowingupinrelationshipsbetweenscientists.

Finally,theanalysisshowsthatthereare 67communitiesin

thegraphwithamodularityscoreof0.42outof1.The

modula-rityscoredescribeshowthenetworkcanbeseenasacollection

ofsub-networks.Thescoreindicatesthattheprojectnetworkis

moderatelyorganized,whichreflectstheaverageclustering

coef-ficientof0.896,wherethebestscoreis1.Theanalysisshowsthat

theprojectnetworkhasmanywell-integratedneighborhoods.

Fig. 6 shows a similar network analysis graph for the

bib-liographic data,with1555 connectionsresulting in630unique

connectionsbetween99nodes.Thearticlenetworkhasasimilar

structuretotheprojectnetworkinthatthereisacenteranda

visi-bleperiphery.Thenodesatthecenterrepresentthescientistswho

collaboratetogetherondifferentideas,moresothantherestof

thepeopleinthegraph.Theanalysisshowsthatthereare6

com-munitiesinthegraphwithamodularityscoreof0.329outof1,

whichindicatesthatthearticlenetworkismoderatelyorganized.

Theaverageclusteringcoefficientis 0.594where thebestscore

is1forthisnetwork.Althoughthescoreislowerthantheproject

network,theaverageclusteringcoefficientscoreforthearticle

net-workisstillquitehigh.Thisshowsthatthearticlenetworkandthe

projectnetworkhavesimilardegreeofintegratedneighborhoods.

TheFruchterman–Reingoldlayoutwasusedtovisualizethis

net-workinGephi,wherethecolorshadeandsizeofthenodesindicate

howconnectedtheyare.Theedgesarecoloredbythekeyword,

however‘Strategies’wasthemostfrequentkeyword,with‘Human

Impacts’beingadistantsecond.

We canevaluatethebibliographicdatatotella story about

thefinaloutcomeofthenetworkshowninFigs.5and6.In2007,

Fig.5. Co-authorshipnetworkusingprojectdatafromallNACPmembers.Thisnetworkhad1007peoplewith16,518connectionsand15,716uniqueedges.Thenodesof thegrapharecoloredbythedegreeofconnection,sothatadarkerbluerepresentsthedegreeofconnectednessofnodes.Thenodesizesarealsocategorizedbythedegree ofconnections.Theedgesarecoloredbythekeywordsassociatedbetweentwopairsofnodes,themeaningofwhichisprovidedinthelegend.

Fig.6.Co-authorshipnetworkusingarticledatabaseof1555connectionsresultingin630uniqueedgesbetween99coreNACPscientists.Thenodesofthegrapharecolored bythedegreeofconnection,sothatadarkerbluerepresentsthedegreeofconnectednessofnodes.Thenodesizesarealsocategorizedbythedegreeofconnections.The edgesarecoloredbythekeywordsassociatedbetweentwopairsofnodes,themeaningofwhichisprovidedinthelegend.

Table5

StatisticsforNACPcoreco-authorshipnetworkinwhichpeoplearerepresentedasnodes.

Evolutionofco-authorshipnetworkbetween2007and2013

Pointofanalysis Yearsinanalysis Nodes Edges Components Avg.numberofneighbors

2007 1 35 43 9 2.11 2008 2 53 113 5 3.17 2009 3 64 213 3 4.47 2010 4 76 543 4 8.16 2011 5 83 800 2 9.18 2012 6 93 1363 2 12.41 2013 7 99 1555 2 12.72

toeach other,andtheirconnectionsformedafairlyfragmented network(9fragments/components),withtheaveragenumberof neighbors(howmanypeople,onaverage,eachpersonisdirectly connectedto)at2.11.Overtime,however,weseetheemergence ofthehighlyinterconnectednetworkshowninFig.6.Weobserve

fromtheincreasingnumberofnodesthat,astheyearspass,moreof

thesehighlyactiveindividualswithinthecommunitybegin

form-ingcollaborativerelationshipswitheachother.Itisalsointeresting

toobservethat,asnewnodesenterthenetwork,thenumberof

disconnectedfragmentsdecreasesfrom9to2,and theaverage

numberofneighborsincreasesfrom2to12(Table5).

Thesimplefactthatthesizeofthenetwork,intermsofnodes,

grewovertimeisnotsufficienttocharacterizeitsevolution.IfNACP

investigators had only begun forming collaborations with each

otherinsmallgroupssuchaspairsortriplets,thenwewouldexpect

thattheaveragenumberofneighborswouldremainrelatively

sta-ble while thenumber of componentswould have substantially

increased.Thefactthatweseethenetwork initscurrentform,

asshowninFig.6,indicatesthatthoseindividualswhoweremost

activeinattendingNACPmeetingshaveformedawell-integrated

networkofcollaborations.

ThesummarystatisticscharacterizingtheNACPcommunityand

thenetworkanalysissuggestthatdifferentinstitutionswithinthe

NACParebecomingmoreinterconnected.Thedatashowedasteady

increaseintheratioofinstitutionalco-authorshiptiesovertime,

whichmaysuggestthatcontinuedparticipationinsharedpractices

benefitstheopportunityforacollectionofresearchersrepresenting

differentinstitutionstoengageincross-organizational

collabora-tion.

6. Discussion

We observed the existence of relationships encompassing

approximately98%of theco-authorshipnetwork.Only asingle

othersmallcomponentoftwonodesexistedoutsideofthemain

cluster.Ifthedatahaddescribedascatteringofislandsof

collab-oration,each groupunconnected totheothers,thenthelackof

socialpathwayswouldbeexpectedtoimpedetheflow ofideas

throughoutthecommunity.However,thetightlyclustered

net-workof coreNACPco-authorshipssupportsthehypothesisthat

theNACPisacommunitybothinnameandinpractice,andthat

myriadsocialpathwaysexistforknowledgesharingand

collabo-rationacrossthemajorityofmembers.Itmayalsomeanthatthe

communityisfairlyinsular,workingmorewitheach otherthan

withthoseontheperiphery.Thesameholdstrueevenwhenthe

networkisrepresentedasagraphofinstitutions.

Although the community of practice framework tends to

emphasizetheroleof interpersonalrelationshipsbetween

indi-viduals,otherstudies havefocused moreonunderstandingthe

roleofcross-institutionalcollaborationinknowledgecreation.For

example,previousstudieshavefoundthatcross-institutional

col-laborationsupports thediffusion ofinnovationsand new ideas

withinafield(ZuckerandDarby,1996).Institutionallyco-authored

papershavealsobeenfoundtobemorehighlycitedthanpapers

authoredwithinasingleinstitution.

OurresultssuggestthattheNorthAmericanCarbonProgramhas

cultivatedanincreasinglyconnectedcommunityofpracticewhose

networksofcollaborationspanawidevarietyofinstitutionsand

topics(Table5).Itiscomprisedofaknowledgenetwork,which

hasgraduallyextendeditstopicsbeyondtraditionalcarboncycle

science.ItisclearfromthesuccessoftheNACPthatencouraging

collaborationstoconnectisolatedfragmentsandcultivating

long-termcollaborativerelationshipsacrossinternationalandcultural

boundariesisimportantfortheimprovedfunctioningofascientific

communityofpractice.

However,simplywritingpapersonrelevanttopicsandforming

a tighterresearchnetwork arenotsufficient toproduce

policy-relevantresearch.Theprocessofconnectingscienceanddecision

makersmustbeundertaken.Itisnotclearfromouranalysisthat

suchworkisbeingdoneyetintheNACP(DillingandLemos,2011;

Lemosetal.,2002).Previousresearchhasshownthatjustbecause

researchispolicyrelevant,doesnotmeanthatpolicymakerswill

useit(Cashetal.,2006).Sciencehastobedoneinawaythatdirectly

engagesstakeholdersand involvestheminsetting theresearch

agenda,workingwiththemiterativelyoverlongperiodsoftime

toensurethattheresearchcanbemoreusableintheend.The

NACPabstractdatabasedoesnotreallyprovideevidenceoneway

oranotherthatthatishappening.

Ourresearchdoesdemonstratethevalueofcontentanalysisand

socialnetworkanalysisusingpublicationdataforassessingaCoP’s

knowledgeproduction againstitsprofessed knowledgedomain.

Thishasimportantimplicationsfortheabilityofthecommunity

torealignitselfwithitsgoals.Thecommunitygainsthe

opportu-nitytofurtheritsdevelopment,integratingnewmemberswhose

researchspecialtieswillcontributetotheoverallknowledgegoals

ofthegroupandfocusingitseffortsongapsinpracticeand

repre-sentation.

Ourresearchhassimilarfindingstootherstudiesthat

investi-gatehowtocultivatecollaborationinacommunityofscientists.

For example, one study exploredthe introduction of scientists

who werepreviouslyunacquainted and fromdifferentresearch

backgrounds using brief meetings during a conference (Vaggi

etal.,2014).Accordingtothestudy,manyscientistreported

pos-itiveexperiencesandpotentialnewcollaborationsasa resultof

participation.TheNACPcouldemployasimilarapproachby

char-acterizingthe researchof members using contentanalysis and

creatingnewopportunitiesforfruitfulcollaborations.Thesocial

networkperspectivealsoenablesthecommunitytofurther

sta-bilizeandexpandsitsconnections,byencouragingcollaborations

thatwouldconnectisolatedfragmentsofresearchergroupstothe

overallnetwork.

6.1. Limitations

Theoutcomeofouranalysisisrepresentativeofthecommunity

thatwasstudiedandisderivedfromdataontheofficialmembers

ofthecommunityintheNACPdatabase.Withoutadditionalstudy,

wecannotknowiftheresultscouldbegeneralized for

describ-ingthelargerpopulationof allcarboncyclescientists including

thosefromoutside theprogram. Also,aspreviouslymentioned,

ourkeyword analysisresultsareonlyasthoroughastheability

ofourselectedkeywordstocapturetheintendedmeanings.Itis

possiblethatotherresearchers mayhave categorizedkeywords

differentlythanwehavechosentodosohere.Otherresearchers

couldfinddifferentresultsdependingonwhatkeywordstheyuse

indefiningcategories.Additionallimitationsareinherenttothe

bibliographicdataset,whichwasanalyzed.Articlespublishedby

thesameauthorsbutusingdifferentinstitutionalaffiliationswould

haveprobablybeenexcluded,aswellasanymaterialnotarchived

inWebofScience.

6.2. Futureresearch

Thisresearchfocusedprimarilyonobservingchangesoccurring

intheoverallnetworkofindividualswhoeitherattendedNACP

meetingsregularly,orwhoparticipatedinprojectsrelatedtothe

program.Futureresearchshouldfocusonanalyzingmoreclosely

theattributesandbehaviorsofindividualswithinthecommunity.

Acombinationofnetworkanalysiswithmorequalitativemethods

forchangewithinacommunity,aswellasunderstandingofwhich

individualstendtodrivechangeandhow.

AstheNACPevolves,aprogramofunderstandingtheimpactof

theprogram’scollaborativeapproachonincreasingitsimpacton

policyshouldbeofgreaterinterest.Becausethis wouldinvolve

interviewing stakeholders, it is outside of the methodological

approachofthecurrent article,but shouldbeattemptedbythe

organizationinthecomingyears(Adamsetal.,2013).

7. Conclusions

Thisresearchfoundanincreaseintheuseofsocialandeconomic

topicsininterdisciplinarycarboncyclescienceresearchfrom2007

to2013associatedwiththeNACPmembersandthepapersthey

havewritten.Oneconclusionthatthisresultcouldmeanisthatthe

NACPcommunityisactivelyworkingtoincorporatehumanfactor

intotheirresearch,orthatthemembersintheNACPhaveincreased

theframingoftheirresearchtoincludethesetopicstoimprove

thepolicyrelevanceoftheirresearch,althoughtheresearchitself

remainsfairly similar.Althoughouranalysiscannotdistinguish

betweenthesetwo hypotheses,theNACPcommunity ispaying

moreattentiontothesocialandeconomicrelevanceoftheirwork.

WefoundthatcoreNACPmembersarewellconnectedwithone

another,formingatightlyclusterednetwork.Cross-institutional

collaborationhasincreased,butmoreneedstobedoneto

culti-vatelong-termcollaborativerelationshipsacrossinternationaland

culturalboundaries.

Itisdifficulttoclearlyconnectcauseandeffectwithregardto

networks,andnotethattheanalysispresentedherecannot

deter-minewhytheNACPnetworkhaschanged.Itmightbethatthe

networkandinstitutionaldiversityhasgrownandconnectivityhas

increased,butisitbecauseofthedevelopmentofacommunityof

practiceorbecauseoftheavailabilityoffunding?Wecannot

deter-minethisfromtheanalysispresented,butbelievethattheNACP

isworkinghardtoincreaseboththerelevanceandqualityofthe

researchitdoes.TheNACPhassuccessfullyfosteredacommunityof

practice,andisworkingtowardincreasingtheinclusionofsocietal

factorsintoitsresearch.Theorganizationcanimproveits

engage-mentwiththeinternationalcommunity,anditsconsiderationof

economicsinitsresearch.

Acknowledgements

ThisresearchwasfundedbytheNationalAeronauticsandSpace

AdministrationthroughagrantundertheCarbonMonitoring

Sys-temScienceArea,EarthScienceDirectorate.Thisstudywaspartof

theNorthAmericanCarbonProgram.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,

intheonlineversion,athttp://dx.doi.org/10.1016/j.socnet.2015.10.

002.

References

Adams,V.,Blankenship,T.,Burgess-Herbert,S.,Corley,W.,Coughlan,J.,Gelso,B., Hinds,E.,Hurley,E.,Hutson,M.,Li,J.,Wilson,D.,2013.MeasuringSocioeconomic ImpactsofEarthObservations.NationalAeronauticsandSpaceAdministration, Washington,DC.

Barabási,A.L.,Jeong,H.,Neda,Z.,Ravasz,E.,Schubert,A.,Vicsek,T.,2002.Evolution ofthesocialnetworkofscientificcollaborations.PhysicaA311,590–614. Bernabo,J.C.,1995.Communicationamongscientists,decisionmakersandsociety:

developingpolicy-relevantglobalclimatechangeresearch.Clim.ChangeRes.: Eval.PolicyImplic.65,103–117.

Berthelot,M.,Friedlingstein,P.,Ciais,P.,Monfray,P.,Dufresne,J.L.,LeTreut,H., Fair-head,L.,2002.GlobalresponseoftheterrestrialbiospheretoCO2andclimate

changeusingacoupledclimate–carboncyclemodel.Glob.Biogeochem.Cycles, 16.

Birdsey,R.A.,Cook,R.,Denning,S.,Griffith,P.,Law,B.,Masek,J.,Michalak,A.,Ogle,S., Ojima,D.,Pan,Y.,Sabin,C.,Sheffner,E.,Sundquist,E.,2007.Investigatorsshare improvedunderstandingoftheNorthAmericanCarbonCycle.EOSTrans.Am. Geophys.Union88,255.

Bradbear,C.,Friel,S.,2013.Integratingclimatechange,foodpricesandpopulation health.FoodPolicy43,56–66.

Cash,D.W.,Adger,W.N.,Berkes,F.,Garden,P.,Lebel,L.,Olsson,P.,Pritchard,L., Young,O.,2006.Scaleandcross-scaledynamics:governanceandinformation inamultilevelworld.Ecol.Soc.11,1–8.

Chen,C.,2006.CiteSpaceII:detectingandvisualizingemergingtrendsandtransient patternsinscientificliterature.J.Am.Soc.Inf.Sci.Technol.571,359–377. Cobo,M.J.,López-Herrera,A.G.,Herrera-Viedma,E.,Herrera,F.,2011.Science

map-pingsoftwaretools:review,analysis,andcooperativestudyamongtools.J. Assoc.Inf.Sci.Technol.62,1382–1402.

Dempewolf,H.,Eastwood,R.J.,Guarino,L.,Khoury,C.K.,Müller,J.V.,Toll,J.,2014. Adaptingagriculturetoclimatechange:aglobalinitiativetocollect,conserve, andusecropwildrelatives.Agroecol.Sustain.FoodSyst.,4.

Denning,S.A.,2005.ImplementationStrategyfortheNorthAmericanCarbon Pro-gram:ReportoftheNACPImplementationStrategyGroupoftheU.S.Carbon CycleInteragencyWorkingGroup.U.S.CarbonCycleScienceProgram, Wash-ington,DC.

Dilling,L.,Lemos,M.C.,2011.Creatingusablescience:opportunitiesandconstraints forclimateknowledgeuseandtheirimplicationsforsciencepolicy.Glob. Envi-ron.Change21,680–689.

Elo,S.,Kyngäs,H.,2008.Thequalitativecontentanalysisprocess.J.Adv.Nurs.62, 107–115.

Falkowski,P.,Scholes,R.J.,Boyle,E.,Canadell,J.,Canfield,D.,Elser,J.,Gruber,N., Hib-bard,K.,Hogberg,P.,Linder,S.,Mackenzie,F.T.,Moore,B.,Pedersen,T.,Rosenthal, Y.,Seitzinger,S.,Smetacek,V.,Steffen,W.,2000.Theglobalcarboncycle:atest ofourknowledgeofearthasasystem.Science290,291–296.

Feldman,D.,Ingram,H.,2009.Makingscienceusefultodecisionmakers:climate forecasts,watermanagement,andknowledgenetworksweather.Clim.Soc.1, 9–21.

IPCC.,2014.WorkingGroupIIContributiontotheFifthAssessmentReportofthe IntergovernmentalPanelonClimateChange:Impacts,Adaptationand Vulner-ability.CambridgeUniversityPress,Cambridge,UK.

Iyalomhe,F.,Jensen,A.,Critto,A.,Marcomini,A.,2013.Thescience–policy inter-faceforclimatechangeadaptation:thecontributionofcommunitiesofpractice theory.EPG23,368–380.

King,A.W.,Dilling,L.,Zimmerman,G.P.,Fairman,D.M.,Houghton,R.A.,Marland, G.,Rose,A.Z.,Wilbanks,T.J.,2007.TheFirstStateoftheCarbonCycleReport (SOCCR):TheNorthAmericanCarbonBudgetandImplicationsfortheGlobal CarbonCycle.NationalOceanicandAtmosphericAdministration,National Cli-maticDataCenter,Asheville,NC,pp.p242.

Latapy,M.,Magnien,C.,Vecchio,N.D.,2008.Basicnotionsfortheanalysisoflarge two-modenetworks.Soc.Netw.30,31–48.

Lemos,M.C.,Finan,T.,Fox,R.,Nelson,D.,Tucker,J.,2002.Theuseofseasonalclimate forecastinginpolicymaking:lessonsfromCeara,northernBrazil.Clim.Change, 55.

Martin,T.,Ball,B.,Karrer,B.,Newman,M.E.J.,2013.Coauthorshipandcitation pat-ternsinthephysicalreview.Phys.Rev.E88,012814.

Melin,G.,Persson,O.,1996.Studyingresearchcollaborationusingco-authorships. Scientometrics36,363–377.

Michalak,A.M.,Jackson,R.B.,GreggMarlandSabine,C.L.,2011.AU.S.CarbonCycle SciencePlan.UniversityCorporationforAtmosphericResearch,Washington,DC, pp.81–89.

Morlacchi,P.,Wilkinson,I.F.,Young,L.C.,2008.SocialnetworksofresearchersinB2B marketing:acasestudyoftheIMPGroup,1984–1999.J.Bus.-to-Bus.Market. 12,3–34.

Newman,M.E.J.,2001a.Scientificcollaborationnetworks.I.Networkconstruction andfundamentalresults.Phys.Rev.E64,016131.

Newman,M.E.J.,2001b.Thestructureofscientificcollaborationnetworks.Proc.Natl. Acad.Sci.U.S.A.98,404–409.

Newman,M.E.J.,2006.Modularityandcommunitystructureinnetworks.Proc.Natl. Acad.Sci.U.S.A.103,8577–8696.

PielkeJr.,R.,Sarewitz,D.,Dilling,L.,2010.SciencePolicyAssessmentandResearch onClimate.CenterforScienceandTechnologyPolicyResearch,Boulder,CO,pp. p24.

Rosenzweig,C.,Horton,R.,Bader,D.,Brown,M.E.,DeYoung,R.,Fellows,M.,Friedl, L.,Gotwals,K.,Graham,W.,Hall,C.,Higuchi,S.,Hudson,C.,Jedlovec,G.,Kaye,J., Lowenstein,M.,Mace,T.,Milesi,C.,Patzert,W.,StackhouseJr.,P.W.,Toufectis, K.,2014.EnhancingclimateresilienceatNASAcenters:theClimateAdaptation ScienceInvestigatorworkgroup(CASI)forscienceandstewardship.Bull.Am. Meteorol.Soc.95,1351–1363.

Sarmiento,J.L.,Wofsy,S.,1999.AU.S.CarbonCycleSciencePlan.U.S.GlobalChange ResearchProgram,Washington,DC,pp.69–79.

Shindell,D.,Kuylenstierna,J.C.I.,Vignati,E.,Dingenen,R.V.,Amann,M.,Klimont, Z.,Anenberg,S.C., Muller,N., Janssens-Maenhou,G.,Raes,F.,Schwartz,J., Faluvegi, G.,Pozzoli, L.,Kupiainen, K.,Höglund-Isaksson,L., Emberson, L., Streets, D., Ramanathan, V., Hicks], K., Oanh, N.T.K., Milly, G., Williams, M.,Demkine,V.,Fowler,D.,2012.Simultaneouslymitigatingnear-term cli-matechangeandimprovinghumanhealthandfoodsecurity.Science335, 183–189.

Vaggi,F.,Schiavinotto,T.,Lawson,J.L.,Chessel,A.,Dodgson,J.,Geymonat,M.,Sato, M.,Salas,R.E.,Csikász-Nagy,A.,2014.Anetworkapproachtomixingdelegates atmeetings.eLife3,e02273.

Wasserman,S.,Faust,K.,1994.SocialNetworkAnalysis:MethodsandApplications. CambridgeUniversityPress,Cambridge,UK.

Wenger,E.,MacDermott,R.,Snyder,W.,2002.CultivatingCommunitiesofPractice: AGuidetoManagingKnowledge.HarvardBusinessPress,Boston.

Wofsy,S.,Harriss,R.,2002.TheNorthAmericanCarbonProgram(NACP).AReportof theNACPCommitteeoftheU.S.CarbonCycleScienceSteeringGroup.University CorporationforAtmosphericResearch,Boulder,CO.

Zucker,L.G.,Darby,M.R.,1996.Starscientistsandinstitutionaltransformation: pat-ternsofinventionandinnovationintheformationofthebiotechnologyindustry. Proc.Natl.Acad.Sci.U.S.A.93,12709–12716.