Acomparativeassessmentofdecision-supporttoolsforecosystem servicesquanti cationandvaluationKennethJ.Bagstada ,n,DariusJ.Semmensa,SisselWaageb,RobertWinthropcaU.S.GeologicalSurvey,Geosciences&EnvironmentalChangeScienceCenter,Denver,CO,USAbBSR,SanFrancisco,CA,USAcSocioeconomicsProgram,USDI Â— BureauofLandManagement,Washington,DC,USAarticleinfoArticlehistory: Received3January2013 Receivedinrevisedform 20June2013 Accepted8July2013 Availableonline30July2013 Keywords: Decisionsupport Ecosystemservices Modeling Valuation ComparativetoolsassessmentabstractToenterwidespreaduse,ecosystemserviceassessmentsneedtobequanti able,replicable,credible, exible,andaffordable.Withrecentgrowthinthe eldofecosystemservices,avarietyofdecisionsupporttoolshasemergedtosupportmoresystematicecosystemservicesassessment.Despitethe growingcomplexityofthetoollandscape,thoroughreviewsoftoolsforidentifying,assessing,modeling andinsomecasesmonetarilyvaluingecosystemserviceshavegenerallybeenlacking.Inthisstudy,we describe17ecosystemservicestoolsandratetheirperformanceagainsteightevaluativecriteriathat gaugetheirreadinessforwidespreadapplicationinpublic-andprivate-sectordecisionmaking.We describeeachofthetools intendeduses,servicesmodeled,analyticalapproaches,datarequirements, andoutputs,aswelltimerequirementstorunseventoolsina rstcomparativeconcurrentapplicationof multipletoolstoacommonlocation Â– theSanPedroRiverwatershedinsoutheastArizona,USA,and northernSonora,Mexico.Basedonthiswork,weofferconclusionsaboutthesetools current Â‘ readiness Â’ forwidespreadapplicationwithinbothpublic-andprivate-sectordecisionmakingprocesses.Finally, wedescribepotentialpathwaysforwardtoreducetheresourcerequirementsforrunningecosystemservices models,whichareessentialtofacilitatetheirmorewidespreaduseinenvironmentaldecisionmaking. PublishedbyElsevierB.V. Contents 1.Introduction.........................................................................................................28 2.Studycontext........................................................................................................29 2.1.Toolsreview...................................................................................................29 2.2.Evaluativecriteriatosupporttoolselection..........................................................................29 3.Findings:analyticalandmodelingapproaches.............................................................................30 3.1.Aspatialecosystemservicesimpactscreening:ESR....................................................................30 3.2.Independentlyapplicable,generalizable,landscape-scalemodeling:ARIES,Co$tingNature,EcoServ,InVEST,LUCI,MIMES,SolVES....32 3.3.Independentlyapplicable,place-speci c,landscape-scalemodeling:Envision,EPM,InFOREST.................................33 3.4.Proprietary,generalizable,landscape-scalemodeling:EcoAIM,ESValue...................................................34 3.5.Site-scalemodeling:EcoMetrix,LUCI...............................................................................34 3.6.Monetaryvaluation:Bene tTransferanduseEstimatingModeltoolkit,EcosystemValuationToolkit,NAIS......................34 3.7.ApplicationofselectedtoolstotheSanPedro........................................................................34 4.Conclusions.........................................................................................................35 4.1.General ndings................................................................................................35 4.2.Feasibilityforwidespreaduse.....................................................................................36 4.3.Implicationsforpublic-andprivate-sectorresourcemanagement........................................................36 4.4.Loweringbarrierstoecosystemservicemodelparameterizationandapplication............................................37 Disclosurestatement......................................................................................................38 Acknowledgments........................................................................................................38 AppendixA.Supportinginformation......................................................................................38 References..............................................................................................................38 Contentslistsavailableat ScienceDirect journalhomepage: www.elsevier.com/locate/ecoserEcosystemServices2212-0416/$-seefrontmatterPublishedbyElsevierB.V. http://dx.doi.org/10.1016/j.ecoser.2013.07.004 nCorrespondingauthor.Tel.: + 13032361330. E-mailaddress: firstname.lastname@example.org(K.J.Bagstad) . EcosystemServices5(2013)e27 Â– e39
1.Introduction Alargeandrapidlygrowingbodyofresearchseekstoidentify, characterize,andvalueecosystemgoodsandservices Â– thebene ts thatecosystemsprovidetopeople( MillenniumEcosystem Assessment(MA),2005 ).However,thedevelopmentofdecisionsupporttools(hereaftertools)thatintegrateecology,economics, andgeographytosupportdecisionmakingisamorerecent phenomenon( Ruhletal.,2007 ; Dailyetal.,2009 ).Currenttools rangefromsimplespreadsheetmodelstocomplexsoftware packages.Unlike adhoc methodsforquantifyingecosystem services(e.g., Egohetal.,2012 ; Martinez-HarmsandBalvanera, 2012 ),thisnewgenerationofanalyticaltoolsisintendedtoenable replicableandquanti ableecosystemservicesanalyses.Assuming thattoolsarewell-documentedandtested,theycanaddcredibilityandtrusttothedecisionprocess,increasingstakeholder con denceintheiruse.Iftheyare exibleenoughforusein diversedecisioncontextsandcanbeaffordablyapplied,theycould reasonablybeincorporatedintopublic-andprivate-sectorenvironmentaldecisionmakingonaroutinebasis. Numerousgroupsoftooldevelopersarenowdevelopingnew approachesforintegratingecosystemservicesintobothpublicandprivate-sectordecision-makingprocesses.Whileaspirations toaiddecisionmakersarecross-cutting,thetoolsvarygreatly. Somearedesignedtobegeneralizabletoanylocationintheworld whileothersareplace-speci c.Thetoolsdifferintheirapproaches toeconomicvaluation,spatialandtemporalrepresentationof services,andincorporationofexistingbiophysicalmodels. Despitetheproliferationoftools,therehasbeenlittlesystematic reviewandevaluationofecosystemservicestools,inorderto determinetoolstrengths,weaknesses,andapplicabilitytovarious settingsandconcurrentlyapplymultipletoolstoacommonstudy area.Thescopeofmostotherreviewshasbeenlimited,providing detaileddescriptionsof2 Â– 3toolsandreferencestoanother2 Â– 4 tools( NelsonandDaily,2010 ; VigerstolandAukema,2011 ).Aside fromtherapidevolutionofecosystemservicetools,amajorreason whythoroughreviewshavebeendif culttocompletehasbeenthe challengeincircumscribingwhatconstitutesanecosystemservice toolamidstthevarietyofemergingtoolsforconservation,land-use planning,andhydrologicandecologicalmodeling.Additional reviewshaveaddressedsomeoftheseothertypesoftools,aswell asone-offmodelingapproachesnotintendedtoforbroader applicability( InstituteforEuropeanEnvironmentalPolicy(IEEP) etal.,2009 ; Ambrose-OjiandPagella,2012 ; Egohetal.,2012 ; Martinez-HarmsandBalvanera,2012 ; Smartetal.,2012 ). Indeed,abroadtradeoffexistsbetweenusingnewecosystem servicetools,manyofwhichareintendedtobetransferrableto newgeographicanddecisioncontexts,versususingexisting mappingormodelingapproachesthatarelocallyknownand trustedbydecisionmakersbutrequiretheadditionofanecosystemservicescomponent.Emergingecosystemservicetoolsoffer thepotentialfor Â“ standardizing Â” assessmentstofacilitatetesting andcomparisonacrossbroadgeographiccontexts,andprovided thatmodelsareclearlydocumented,user-friendly,andeasily parameterized,theymayfacilitatewidespreadadoptionofecosystemservicesfordecisionmaking.However,thesemodelsareoftenlesswellknowntodecisionmakers,sotheyfacethecritical stepofachievingstakeholdertrustandbuy-in.Otherwellacceptedmodelsmayalreadyhavesuchbuy-in,butlackan ecosystemservicescomponent.Suchtools,then,mustseekto addcomponentsthataccuratelyquantifyecosystemservices.The lackofcomparabilitybetweensuchlocallyadaptedmodelsmay havetheaddeddisadvantageoflimitingthecomparabilityoftheir resultsandtheirusewithincommondecisionframeworks.This tradeoffisalsopartlyrelatedtoscale:whilesomegeneralized modelsmaybehighlyeffectiveatthenationaltoregionallevel, theymaybeineffectiveatthelocalleveliftheycannotincorporate accurate,high-resolutiondatawhileaccountingforlocalin uencesonecosystemservicesupply,demand,andvalue.Insuch caseslocallydevelopedmodelsmaybetteraccountfor ne-scale analysis( Smartetal.,2012 ).However,animprovedunderstanding ofgeneralizedmodelswasgenerallypreferredbytheU.S.-based public-sectorresourcemanagementagenciesandmultinational corporationsinvolvedinthisreview.Theseentities,whichare makingdecisionsacrossabroadrangeofgeographies,agreedthat uniformprocessesandprotocolswouldbeeasiertouse;however, forlocalizeddecisionmaking,adaptationanduseoflocalmodels mightbeapreferredstrategy( Smartetal.,2012 ). Whiletherelativevalueofthesetwoapproachesisaworthwhiledebateinthe eldofecosystemservicemodeling,theintent ofthisreviewistoqualitativelycatalogandevaluatemethodsthat arealreadygeneralizableorareintendedbytheirdevelopersto becomeso.Inexploringthispartofthetoollandscape,itisbeyond thescopeofthepapertoaddresstheadaptabilityofother biophysicalmodelstoecosystemservicesandwhetherthat approachortheuseofgeneralizableecosystemservicemodelsis amoreappropriatecourseofaction. Thispaperisbasedonastudythatwasundertakenin2010 through2011,whichwasspurredbythegrowingdemandfor morecomprehensiveanalysesoftheecologicalandsocioeconomic consequencesofland-managementdecisions,particularlywithin theU.S.government spolicydirectionforenvironmentaland naturalresourcemanagement( President sCouncilofAdvisorson ScienceandTechnology(PCAST),2011 ; CouncilonEnvironmental Quality(CEQ),2013 ).Inresponse,theU.S.DepartmentofInteriorBureauofLandManagement(BLM)launchedapilotprojectwith theU.S.GeologicalSurvey(USGS)toassesstheusefulnessand feasibilityofecosystemservicesvaluationasaninputinto decision-making.TheBLMmanagesnearly100millionhectares oflandacrossthewesternU.S.fromAlaska Â’ sNorthSlopetothe Mexicanborder.Underitsmultiple-usemission,BLM Â’ sresponsibilitiesrangefromfacilitatingthedevelopmentofoil,gas,coal, solarenergy,andothercommoditiestoprovidingmanyformsof recreation,restoringhabitat,andpreservingscenicvalues,archeologicalheritage,andenvironmentalquality( BureauofLand Management(BLM),2005 ). BLM Â’ sgoalsforthecomparativetoolsassessmentwereto(1) determinewhich,ifany,methodsforvaluingecosystemservices areripeforoperationaluseacrosstheagency,and(2)evaluatethe utilityofecosystemservicevaluationforitsresourcemanagement decisionprocesses.The rstphaseofthiseffortusedastudyarea Â– theSanPedroRiverwatershedinsoutheastArizonaandnorthern Sonora,Mexico(hereafterSanPedro) Â– thathadalegacyof biophysicalresearchtodrawuponandavarietyofecological stressorsrelevanttofederalresourcemanagement. TheBLM-USGSinitiativewascoupledwithcomparativeapplicationofadditionalecosystemservicetoolsandanalysisoftheir relevancetotheprivatesector Â– throughengagingthesame technicalspecialisttoconducttheassessment,whichwasconcurrentlycoordinatedbyBusinessforSocialResponsibility(BSR), anindependentnongovernmentorganization(NGO)focusedon sustainabilityissuesandtheirapplicationtotheprivatesector.The BSRinitiativeaskedofalltoolswhereahypotheticalresidential developmentwithintheSanPedroshouldbesitedtominimize impactsontheprovisionand owsofecosystemservices( Waage etal.,2011 ).Basedonthiscomparativeapplication,wesummarize the ndingsfromthesetwolinkedstudiesinthisarticlethrougha reviewofecosystemservicessoftwareandmodelingtools. Toourknowledgethisisthe rstefforttoevaluatemultiple ecosystemservicetoolsandtheirapplicabilitytoenvironmental decisionmakingacrossbothpublic-andprivate-sectorcontexts.Our analysisincludesboth(1)place-speci ctools Â– customizedforK.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e28
applicationinaparticulargeographiccontextbutthatcouldbeapplied elsewhere Â– and(2)generalizabletoolsintendedtobeapplicablein diversecontextswhenlocallyappropriateinputdataareavailable. Quanti edbiophysicalandmonetaryanalysisofecosystem servicevaluesfortheSanPedroarepresentedelsewhere( Waage etal.,2011 ; Bagstadetal.,2012 , inthisvolume ).Inthisarticle,we cataloganddescribe17existingtools,evaluatingthemintermsof eightevaluativecriteriausedtogaugetheirutilityinpublic-and private-sectordecisionmaking.Inaddition,wedescribethetime requiredtocompleteanassessmentforsevenofthesetools,which wereappliedtothecasestudyontheSanPedro. 2.Studycontext 2.1.Toolsreview Throughliteraturereviewsanddiscussionswith77colleagues acrosstheacademic,public,private,andNGOsectors(seesupportingonlinematerialforafulllistofprojectparticipants),we identi ed17toolsthatassess,quantify,model,value,and/ormap ecosystemservices( Table1 ),excluding adhoc ecosystemservice mappingefforts( Egohetal.,2012 ; Martinez-HarmsandBalvanera, 2012 ).Numerous Â“ ecosystem-basedmanagementtools Â” exist;for example,theEcosystem-BasedManagement(EBM)Toolsdatabase contained183toolsasofNovember2012( Ecosystem-Based ManagementToolsDatabase,2012 ).Welimitedthisreview, however,totoolswithanexplicitfocusonmultipleecosystem services,ratherthanthoseecological,hydrologic,orotherbiophysicalprocessmodelsthatlackacentralfocusonecosystem services.Wethus,forexample,excludetoolsforconservation planningoroptimization(e.g.,C-Plan, Presseyetal.,2005 ;NatureServeVista, NatureServ,2013 ),integratedmodelsnotexplicitly linkedtoecosystemservices(e.g.,LandscapesToolkit(LsT, Bohnet etal.,2011 )),andhydrologicprocessmodels(e.g.,SoilandWater AssessmentTool(SWAT, ArnoldandFohrer,2005 )).Wealso excludedfromourreviewone-timeapplicationsthatarenot readilyunderdevelopmentfornewlocations(e.g., Maesetal., 2012 ,AdvancedTerrestrialEcosystemAnalysisandModelling (ATEAM, Schroteretal.,2005 )),andtoolsintendedforsingle landscapetypes,whoseoutputscouldnotinformchangeanalyses (e.g.,CITYGreen, AmericanForests,2002 ).Finally,weincluded threevaluationdatabasesthatincludefunctionalityforusersto constructvaluationportfolios Â– theNaturalAssetsInformation System(NAIS),EcosystemValuationToolkit,andBene tTransfer andUseEstimatingModelToolkit.Weexcludefromourreview thosevaluationdatabasesthatsimplyprovideuserswitha locationtosearchthroughnon-marketvaluationstudies. We,orinsomecasesthetooldevelopersthemselves,applied sevenofthesetoolstotheSanPedro,including:Arti cialIntelligenceforEcosystemServices(ARIES),EcoAIM,EcoMetrix,EcosystemServicesReview(ESR),ESValue,IntegratedValuationof EcosystemServicesandTradeoffs(InVEST),andtheBene tTransferandUseEstimatingModelToolkit.Fortheremainingtentools thatwewereunabletoruninthepilotstudy Â– duetobudgetandtimelimitationsorbecausetheywereunderdevelopmentand unabletobeindependentlyrunatthetimeofthisassessment Â– weinterviewedthetooldevelopersinordertounderstandtheir tool Â’ sintendeduse,approach,andlevelofdevelopment.We includedescriptionsofthesetoolsinthisarticle.Wedonot, however,includefurtherdiscussionofprimaryvaluation(various techniquesfornon-marketvaluationofecosystemservices)or secondaryvaluation(varioustypesofbene ttransfers),asthese aredescribedindetailelsewhere( Farberetal.,2006 ; Wilsonand Hoehn,2006 ; Bagstadetal.,2012 ). 2.2.Evaluativecriteriatosupporttoolselection Basedondiscussionswith77stakeholdersandscientists involvedintheBLMandBSRprojects,includingacademicand agencyscientistsandprivate-sectorpractitionersconductingecosystemservicesanalysis,wedevelopedandreviewedasetofeight evaluativecriteriathatdescribeimportanttoolcharacteristics whichdecision-makersassertedwouldbekeyelementsofselecting analyticalecosystemservicestools(seesupportingonlinematerial forafulllistofprojectparticipants).Thesecriteriaqualitatively gaugeeachtool Â’ sabilitytosupportecosystemserviceassessments thatarequanti able,replicable,credible, exible,andaffordable. Weappliedthesecriteriatoeachtoolinordertoassessitsrelative strengthsandweaknesses.Theevaluativecriteriainclude 1. Quanti cationanduncertainty. Quanti edoutputsareessential formeasuringecosystemservicetradeoffs,thoughqualitative toolsmaybeusefulininitialscreening,scoping,orcoarse-grain rankingprocesses.Reportingasinglevaluecaninspirefalse con denceinthecertaintyofresults,souncertaintyestimates areavaluableadditiontothesetofmodeloutputs.Although anymodelcanproducearangeofoutputvalueswhentheuser suppliesmultiplepossibleinputvalues( Kareivaetal.,2011 ), yieldingsomeinformationaboutuncertainty,somemodels moreexplicitlyaccountforuncertaintyusingapproacheslike MonteCarlosimulationorBayesiannetworkmodeling. 2. Timerequirements. Asthetimerequiredtoapplyatool decreases,itbecomesincreasinglypracticalforwidespreaduse. 3. Capacityforindependentapplication. Toolsthatareinthepublic domain,orforwhichasoftwarelicensecanbepurchasedto allowthetooltobeindependentlyapplicable,wereastrong preferenceofarangeofagencystakeholdersinvolvedinthe BLMpilotstudyaswellasprivate-sectordecision-makersin theBSRcomponentofthestudy.Thiscontrastswithtoolsthat requirecontractingwithacademicorconsultinggroupsfor eachapplicationofthetool. 4. Levelofdevelopmentanddocumentation. Ideallytoolswouldbe suf cientlydevelopedtorunreliably,usevalidatedmodels, producereplicableresults,andtohavetheirmethods,assumptionsandkeyalgorithms,strengthsandlimitations,andapplicationsiteswelldocumentedinusermanualsand/orpeerreviewedjournalarticles,whichmayalsoincludevalidation exercises.Toolsthatarewell-developedanddocumentedhave greatertransparencyandcredibility,andarethusmorelikelyto engendertrustwithdecisionmakersandthepublic. 5. Scalability. Toolsmaybeapplicablefromparceltoglobalscales. Toolsthatareapplicableacrossmultiplespatialscalesare attractivetomanagersbecauseitiseasiertolearnonetool thanmany;however,notoolislikelytohandleanalysesatall scaleswell,whichmaynecessitateuseofmultipletools. 6. Generalizability. Tosupportwidespreaduse,toolswouldideally bebroadlyapplicableacrossavarietyofecoregionaland socioeconomicsettingswhileprovidingsomedegreeofcustomizabilitytoaccountfordifferinglocalconditions.Mosttools areeitherplace-speci c,reducingtransferabilitybutaccountingforlocallyimportantprocesses,orbroadlygeneralizable, sacri cinglocaldetailfortransferability.Sometoolscurrently useplace-speci ccasestudiesbutareintendedtobemore generalizableinfuturereleases. 7. Nonmonetaryandculturalperspectives. Stakeholdersconsulted assertedthatitwouldbeidealiftoolscouldprovideinformationthatincorporatesmultiplevaluationsystems(monetary andnon-monetary)andculturalperspectives(includingindigenouspeople Â’ sandotherspiritualandculturalvalues). 8. Affordability,insights,integrationwithexistingenvironmental assessment. ToolsforquantifyingandvaluingecosystemK.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e29
servicesaremoredesirableiftheycancost-effectivelyprovide additionalinformationthatconformswithestablishedmanagementandplanningprocesses. Wedidnotincludeeachtool Â’ sbiophysicalandsocioeconomic complexityasanevaluativecriterion.Althoughcomplexityisan importantissueinecosystemservicemodeling( Seppeltetal., 2011 ),andassessmentsbasedonproxyinformationsuchasland coverhavebeenshowntosacri ceaccuracy( Eigenbrodetal., 2010 ),complexmodelscanleadtoafalsesenseofcon dencein modelquality,whichcanmakeerringonthesideofsimplicity moredefensible.Whileamodel Â’ spurposeshouldtypicallydictate itsneededlevelofcomplexity,themostcomplexmodelsdonot alwaysperformbetterthanlesscomplexmodels( Fultonaetal., 2004 ; Raickaetal.,2006 ),nordotheynecessarilyaddvalueto decisionmaking( TallisandPolasky,2011 ). 3.Findings:analyticalandmodelingapproaches Thissectionoffersanoverviewofeachofthe17tools Â’ performanceagainsttheevaluativecriteria( Table2 )followedby detaileddescriptionsofalltools analyticalandmodeling approaches,theirintendeduses,ecosystemservicesmodeled, modelingandvaluationapproaches,datarequirements,andoutputs.Descriptionsformodelsthatlackfulldocumentationare necessarilylesscomplete,sowedonotlisttheservicesmodeled bytheseapproaches.Finallywedescribethetimerequirements forapplicationofseventoolstoacommonstudyarea Â– the SanPedro. 3.1.Aspatialecosystemservicesimpactscreening:ESR TheESR( WorldResourcesInstitute(WRI),2012 )isastructured processtoidentifyecosystemservicesimpacts,dependencies,and Table1 Asurveyofecosystemservicestools. Tool , URL , andreferencesBriefdescriptionTested forthe San Pedro? RationaleforchoicetotestfortheSan Pedro EcosystemServicesReview(ESR), http://www. wri.org/ ( WorldResourcesInstitute(WRI), 2012 ) Publiclyavailable,spreadsheet-basedprocessto qualitativelyassessecosystemservicesimpacts YesAwell-documentedapproachtoquickly describeecosystemservicesandimpacts qualitatively IntegratedValuationofEcosystemServicesand Tradeoffs(InVEST), http://www. naturalcapitalproject.org ,( Kareivaetal.,2011 ; Tallisetal.,2013 ) Opensourceecosystemservicemappingandvaluation modelsaccessedthroughArcGIS YesWell-documented;canbeindependently appliedandtested;amenabletowidespread use Arti cialIntelligenceforEcosystemServices (ARIES), http://www.ariesonline.org ( Bagstad etal.,2011 ; Villaetal.,2011 ) Opensourcemodelingframeworktomapecosystem service ows;onlineinterfaceandstand-alonewebtools underdevelopment YesDataandmodelsavailableforseveralwestern U.S.states;globalmodelandonlineinterface underdevelopmentwouldenable widespreaduse LUCI(formerlyPolyscape), http://www. polyscape.org ( Jacksonetal.,2013 ) OpensourceGIStoolboxtomapareasprovidingservices andpotentialgainorlossofservicesundermanagement scenarios NoToolnotdevelopedbythetimeofthepilot study MultiscaleIntegratedModelsofEcosystem Services(MIMES), http://www. afordablefutures.org Opensourcedynamicmodelingsystemformappingand valuingecosystemservices NoRequirescommercialmodelingsoftware; modelconstructioncurrentlyrequires contractingwithdevelopmentgroup EcoServ( Fengetal.,2011 )Web-accessibletooltomodelecosystemservicesNoStillindevelopment;initialcasestudiesnot availableforSouthwest Co$tingNature, http://www1.policysupport.org/ cgi-bin/ecoengine/start.cgi? project = costingnature Web-accessibletooltomapecosystemservicesand conservationpriorityareas NoToolnotdevelopedbythetimeofthepilot study SocialValuesforEcosystemServices(SolVES), http://solves.cr.usgs.gov ( Sherrouseetal.,2011 ) ArcGIStoolbarformappingsocialvaluesforecosystem servicesbasedonsurveydataorvaluetransfer NoNosurveydataavailable;conditionsatstudy sitetoodifferentfrompaststudiestosupport valuetransfer Envision, http://envision.bioe.orst.edu ( Guzyetal.,2008 )Integratedurbangrowth-ecosystemservicesmodeling system;hasusedexternalmodels,includingInVEST,or creatednewecosystemservicemodelsasappropriate NoHasnotyetbeenappliedintheSouthwest; infeasibletorunfornewsiteswithouta substantialexternalresearcheffort EcosystemPortfolioModel(EPM),http:// geography.wr.usgs.gov( Labiosaetal.,2013 ) Web-accessibletooltomodeleconomic,environmental, andqualityoflifeimpactsofalternativeland-usechoices NoDevelopedforadjacentSantaCruzRiver watershedbut;infeasibletorunfornewsites withoutasubstantialexternalresearcheffort InFOREST, http://inforest.frec.vt.edu/ Web-accessibletooltoquantifyecosystemservicesin Virginia NoHasonlybeendevelopedforVirginia EcoAIM(Waageetal.,2011)Proprietarytoolformappingecosystemservicesand stakeholderpreferences YesDemonstrationcompletedbyExponent (Waageetal.,2011) ESValue(Waageetal.,2011)Proprietarytoolformappingstakeholderpreferencesfor ecosystemservices YesDemonstrationcompletedbyEntrix(Waage etal.,2011) EcoMetrix,http://www.parametrix.com ( Parametrix,2010 ) Proprietarytoolformeasuringecosystemservicesatsite scalesusing eldsurveys YesDemonstrationcompletedbyParametrix (Waageetal.,2011) NaturalAssetsInformationSystem(NAIS),http:// www.sig-gis.com( TroyandWilson,2006 ) ProprietaryvaluationdatabasepairedwithGISmappingof land-covertypesforpointtransfer NoProprietarymethod;limitedprimary valuationstudiestosupportapplicationto studysite EcosystemValuationToolkit, http://www. esvaluation.org ( EcosystemValuationToolkit, 2012 ) Subscription-basedvaluationdatabasepairedwithGIS mappingofland-covertypesforpointtransfer NoToolnotdevelopedbythetimeofthepilot study Bene tTransferandUseEstimatingModel Toolkit, http://www.defenders.org ( Loomis etal.,2008 ) Publiclyavailablespreadsheets,usefunctiontransferto valuechangesinecosystemservicesintheU.S. YesWell-documented;canbeindependently appliedandtested;amenabletowidespread use K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e30
Table2 Descriptionofallecosystemservicetoolsagainstkeyevaluativecriteria. ToolQuanti able, approachto uncertainty Time requirements Capacityfor independent application Levelof development& documentation ScalabilityGeneralizabilityNonmonetary &cultural perspectives Affordability, insights,integration withexisting environmental assessment ESRQualitativeLow,depending onstakeholder involvementin thesurveyprocess YesFullydeveloped anddocumented Multiple scales HighNovaluation component Mostusefulasalowcostscreeningtool InVESTQuantitative, uncertainty through varyinginputs Moderatetohigh, dependingondata availabilityto supportmodeling Yes Â“ Tier1 Â” models fullydeveloped anddocumented; Â“ Tier2 Â” documentedbut notyetreleased Watershed or landscape scale High,though limitedby availabilityof underlyingdata Biophysical values,canbe monetized Spatiallyexplicit ecosystemservice tradeoffmaps; currentlyrelatively timeconsumingto parameterize ARIESQuantitative, uncertainty through Bayesian networksand MonteCarlo simulation Hightodevelop newcasestudies, lowfor preexistingcase studies Yes,throughweb explorerorstandalonesoftwaretool Fully documented;case studiescomplete butglobalmodels andwebtool under development Watershed or landscape scale Lowuntilglobal modelsare completed Biophysical values,canbe monetized Spatiallyexplicit ecosystemservice tradeoff, ow,and uncertaintymaps; currentlytime consumingfornew applications LUCIQuantitative, currentlydoes notreport uncertainty Moderate;toolis designedfor simplicityand transparency, ideallywith stakeholder engagement Yes,thoughwebsite isunder developmentand moredetaileduser guidanceis presumably forthcoming Initial documentation andcasestudy complete;followupcasestudiesin development Siteto watershed or landscape scale Relativelyhigh;a stakeholder engagement processisintendedtoaidin Â“ localizing Â” thedataandmodels Currently illustrates tradeoffs between servicesbutdoes notinclude valuation Spatiallyexplicit ecosystemservice tradeoffmaps; designedtobe relativelyintuitiveto useandinterpret MIMESQuantitative, uncertainty through varyinginputs (automated) Hightodevelop andapplynew casestudies Yes,assuminguser hasaccesstoSIMILE modelingsoftware Somemodels completebutnot documented Multiple scales Lowuntilglobalor nationalmodelsare completed Monetary valuationvia input Â– output analysis Dynamicmodeling andvaluationusing input Â– output analysis;currently timeconsumingto developandrun EcoServQuantitative, uncertainty through varyinginputs Hightodevelop newcasestudies, lowforexisting casestudies Yes,pending releaseofweb explorer Under development,not yetdocumented Siteto landscape scale Lowuntilglobalor nationalmodelsare completed Biophysical values,canbe monetized Indevelopment,will offerspatiallyexplicit mapsofecosystem servicetradeoffs Co$ting Nature QuantitativeLowYesPartially documented Landscape scale HighOutputsindexed, bundled ecosystem servicevalues Rapidanalysisof indexed,bundled servicesbasedon globaldata,along withconservation prioritymaps SolVESQuantitative, noexplicit handlingof uncertainty Highifprimary surveysare required,lowif functiontransfer approachisused Yes,assuminguser hasaccesstoArcGIS Fullydeveloped anddocumented Watershed or landscape scale Lowuntilvalue transfercanbe shownto successfully estimatevaluesat newsites Nonmonetary preferences (rankings)of relativevalues forstakeholders Providesmapsof socialvaluesfor ecosystemservices; timeconsumingfor newstudiesbut lower-costforvalue transfer EnvisionQuantitativeHightodevelop newcasestudies YesDevelopedand documentedforPaci cNorthwest casestudysites Landscape scale Place-speci cAllows nonmonetary tradeoff comparison,also supports monetary valuation Cost-effectivein regionswhere developed;time consumingfornew applications EPMQuantitativeHightodevelop newcasestudies, lowforexisting casestudies Yes,throughweb browser Developedand documentedfor threecasestudy sites Watershed or landscape scale Place-speci cEcological, economic,and qualityoflife attributescould support nonmonetary valuation Cost-effectivein regionswhere developed;time consumingfornew applications InFORESTQuantitativeLow,accessed throughonline interface Yes,throughweb browser Developedand documentedonly forVirginia Siteto landscape scale Currentlyplacespeci c Designedasa creditcalculator, noeconomic valuation Cost-effectivein regionswhere developed;time consumingfornew applications EcoAIMQuantitativeRelativelylowfor basicmapping, greaterfor nonmonetary valuation NoPublic documentation unavailable Watershed or landscape scale HighIncorporates stakeholder preferencesvia modi edrisk analysis approach Spatiallyexplicit ecosystemservice tradeoffmaps; relativelytime consumingtorun K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e31
stakeholders.Althoughitisaqualitativetool Â– basedona structuredsetofquestionslaidoutinaspreadsheet Â– weinclude itinthisreviewduetoitsabilitytofunctionasalow-costscoping toolthatcanprovideanentrypointtoecosystemservicemapping, modeling,orvaluation.TheESRisafree,downloadablespreadsheetdescribing27ecosystemservicesderivedfromtheMillenniumAssessment( MillenniumEcosystemAssessment(MA),2005 ) ecosystemservicestypology.Itisfocusedonprivate-sectorecosystemservicesassessment Â– walkingusersthroughaprocessof identifyingbusinessdependencies,risks,andopportunitiesrelated toecosystemservices. 3.2.Independentlyapplicable,generalizable,landscape-scale modeling:ARIES,Co$tingNature,EcoServ,InVEST,LUCI,MIMES, SolVES Themajorityofecosystemservicetoolsseektoquantify servicesandtheirtradeoffsatalandscapescaleinordertosupport scenarioanalysisusingsimpli edunderlyingbiophysicalmodels or Â“ ecologicalproductionfunctions Â” ( Dailyetal.,2009 ).These toolsdifferintheirmodelingapproaches,generalizability,and whethertheyareinthepublicdomainorproprietary. InVESTandARIESareperhapsthebestknownofthegeneralizable,public-domaintools( VigerstolandAukema,2011 ).Bothuse avarietyofspatialdataasmodelinputsandencodeecological productionfunctionsindeterministicmodels(InVESTandARIES) andprobabilisticmodels(ARIES).Forprovisioningandregulating services,bothtoolsproducemapsdisplayingresultsinbiophysical units,towhichper-unitmonetaryvaluescanbeapplied;for culturalservicesandsomeaccompanyingmodels(e.g.,InVEST biodiversityandhabitatrisk)outputsareinrelativerankings. InVEST sunderlyingdeterministicmodelshavebeenmoreextensivelyvettedinthepeer-reviewedliterature,andmaybemore appropriateforuseincontextswhereecologicalprocessesarewell understood.ARIES Â’ probabilisticmodels,whichareencodedas Bayesianbeliefnetworks,maybemoreappropriateunderconditionsofdatascarcity( VigerstolandAukema,2011 ). ThecurrentInVESTreleaseincludesninemarineandseven freshwaterandterrestrialecosystemservicemodels(waveenergy, windenergy,coastalvulnerability,erosionprotection,marine sh aquaculture,estheticquality, sheriesandrecreationoverlap, habitatriskassessment,marinewaterquality,biodiversity,carbon storageandsequestration,hydroelectricpowerproduction,nutrientretention,sedimentretention,timber,andcroppollination ( Tallisetal.,2013 )).InVEST sTier1modelsrunwithinArcGIS ( EnvironmentalSystemsResearchInstitute(ESRI),2013 )oras stand-aloneexecutableprograms,uselandcoverandotherspatial datatoquantifyserviceprovisionviacoef cienttablesforeach land-covertype(e.g.,forcarbonstorage,evapotranspiration,or nutrient lteringcapacity);dataforthecoef cienttablesare typicallyderivedfrom eldexperiments.InVEST sTier2models havebeendescribedbutnotyetreleasedaspartofasoftware package.Tier2InVESTmodelshavetheabilitytoencodemore complexandpotentiallymorerealisticunderlyingprocessesbut aremoredata-andtime-intensivetoapply( Kareivaetal.,2011 ). ThecurrentARIESreleaseincludeseightecosystemservices Â– carbonsequestrationandstorage,riverineandcoastal ood regulation,freshwatersupply,sedimentregulation,subsistence sheries,recreation,estheticviewsheds,andopen-spaceproximityvalues,withadditionalservicemodelsinactivedevelopment. ARIESusesarti cialintelligencetechniquestopairlocallyappropriateecosystemservicemodelswithspatialdatabasedonasetof encodeddecisionrules,quantifyingecosystemservice owsand theiruncertaintywithinawebbrowserorstand-alonesoftware toolenvironment( Villaetal.,2011 ).ARIESusesagent-based modelstoquantifythe owofecosystemservicesbetween Table2 ( continued ) ToolQuanti able, approachto uncertainty Time requirements Capacityfor independent application Levelof development& documentation ScalabilityGeneralizabilityNonmonetary &cultural perspectives Affordability, insights,integration withexisting environmental assessment ESValueQuantitative, uncertainty throughMonte Carlo simulation Relativelyhighto support consultantstakeholder valuationprocess NoPublic documentation unavailable Watershed or landscape scale HighNonmonetary preferencesvia rankedanalysis oftradeoffsby stakeholders Stakeholder-based relativeecosystem servicevalue assessment; relativelytime consuming EcoMetrixQuantitativeRelativelylowto support eldvisits anddataanalysis NoPublic documentation unavailable SitescaleHigh,where ecological production functionsare available Designedasa creditcalculator, noeconomic valuation Onemethodforsitescaleecosystem servicesassessment NAISQuantitative, reportsrange ofvalues Variable dependingon stakeholder involvementin developingthe study NoDevelopedbut public documentation unavailable Watershed or landscape scale High,withinlimits ofpointtransfer Dollarvalues only Pointtransferfor Â“ ballparknumbers, Â” buildingawarenessof values Ecosystem Valuation Toolkit Quantitative, reportsrange ofvalues Assumedtobe relativelylow YesUnder development Watershed or landscape scale High,withinlimits ofpointtransfer Dollarvalues only Pointtransferfor Â“ ballparknumbers, Â” buildingawarenessof values Bene t Transfer andUse Estimating Model Toolkit Quantitative, uncertainty through varyinginputs LowYesFullydeveloped anddocumented Siteto landscapescale HighDollarvalues only Lowcostapproachto monetaryvaluation K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e32
ecosystemsprovidingtheserviceandtheirhumanbene ciaries, enablingquanti cationofactualserviceprovisionanduse,as opposedtotheoreticalserviceprovisionasestimatedbymany otherecosystemservicetools( Bagstadetal.,inpress ). Fouradditionaltools Â– Co$tingNature,EcoServ,LUCI,and MultiscaleIntegratedModelsofEcosystemServices(MIMES) Â– arealsospatiallyexplicit,public-domaintoolsthatbiophysically modelecosystemservicesbuthavenotyetbeenaswidely documentedandappliedasInVESTandARIES.MIMESisasystem dynamicsmodeldesignedtoaccountfortemporaldynamicsand feedbackloops,incorporateexistingecologicalprocessmodels intoecosystemservicemodeling,andeconomicallyvalueecosystemservicesviainput Â– outputanalysis( http://www.afordablefu tures.org ).MIMESwasdevelopedusingSimile,acommercial codingandsimulationsoftwarepackage( Simulistics,2013 ).Input dataincludevariedspatialdatasetsdependingontheservicesof interesttotheuseraswellasinformationthatauserappliesto parameterizethemodel sequations.MIMESoutputsinclude spatiallyexplicittimeseriesofecosystemservicevalues. LUCI,formerlyknownasPolyscape( Jacksonetal.,2013 ),is designedtousesimplealgorithmsandoutputstotransparently communicateecosystemservicetradeoffsinsettingswithstakeholdersanddecisionmakers.ItisaGIStoolboxthatcurrently includesmodelsforagriculture, oodregulation,carbonsequestration,sedimentregulation,andhabitatconnectivity,andquanti estradeoffsbetweenthose veservices.LUCIisdesignedfor applicationsrangingfromthefarm eldthroughthewatershedto landscapescale,withtheupperlimitsonanalysisextentdependingonthetradeoffbetweencomputationaltimeandtheneedfor presentingnear-real-timeresultsinpublicforums.LUCI sinputs includecommonlyavailabledatasetssuchaselevation,slope, hydrography,andlandcover,whichcanbemodi edbystakeholderstoimproveaccuracyathighspatialresolution.Itsoutputs showpartsofthelandscapethatcurrentlyprovideecosystem servicesandareaswheremanagementinterventionscould enhanceordegradeservices.InitialtestapplicationsforLUCIhave beenconductedintheU.K.,NewZealand,Ghana,andGreece. EcoServisaweb-basedtoolunderdevelopmentintheU.S.and CanadianPrairiePotholeregion,withtheintenttoeventually developadditionalcasestudiesthennationallyorgloballygeneralizedmodels.EcoServlinksexternalecosystemprocessmodels andspatialdataandwillmaketheseaccessibletothepublicviaa webtool( Fengetal.,2011 ).Itaccountsfortemporalclimate variabilityandcanprovideoutputmapsofserviceprovisionunder scenariosforclimateandland-usechange.EcoServdoesnot explicitlyuseproductionfunctionsinmodelingecosystemservices,andinsteadreliesonaseriesofexternalmodelstoproxya serviceofinterest.EcoServdoesnoteconomicallyvalueecosystem services,althoughmodeloutputscouldbeusedinexternal valuationefforts. Co$tingNatureisaweb-basedtoolthatjointlymapsecosystem servicesandconservationpriorities( Mulliganetal.,2010 ).Ituses pre-loadedglobaldatasetsat1km2or1haresolutiontoquantify wateryield,carbonstorage,nature-basedtourism,andnatural hazardmitigationforbaselineconditionsandclimateorland-use changescenarios.Co$tingNatureestimatesandaggregatesthese valuesintoa Â“ bundledservicesindex Â” (i.e.,withvaluesranging from0to1)forpotentialandrealizedservices,byaccountingfor ecosystemserviceprovision,bene ciarylocations,and ows. Whileitthusdoesnotsupportmappingofindividualservices, theirtradeoffs,orvaluation,Co$tingNaturecanbeusedto compareoverallservicegenerationwithbiodiversityandconservationpriorities,andcanberapidlyappliedinterrestrialenvironmentsglobally. Unliketheprevioustoolsthatusebiophysicalmodelsto quantifyecosystemservices,theSocialValuesforEcosystem Services(SolVES)tool( Sherrouseetal.,2011 )isintendedto quantifyandmaptheperceivedsocialvaluesforecosystem servicescalculatedfromacombinationofspatialandnon-spatial responsestopublicattitudeandpreferencesurveys.Thevalues thatarequanti eddependonthevaluestypologyprovidedwith thesurvey,whichhastypicallybeenbasedona Â“ forestvalues typology Â” ( BrownandReed,2000 ).ThistypologylargelycorrespondstoMAculturalservices(esthetic,recreation,spiritual, education,andculturalheritage)andnon-usevalues(option, existence,andbequestvalue),andhasbeenmodi edforusein diversesettingsrangingfromforeststocoastalecosystems. SolVESusesresponsestoavalue-allocationexerciseinthesurvey tocalculateaquantitative10-point Â“ ValueIndex. Â” Respondentmappedlocationsassociatedwitheachvaluetypearethenusedto calculatetherelationshipbetweenvaluesandphysicalattributes ofthelandscape(environmentaldatalayerssuchaselevation, distancetowater,land-covertype,etc.).Theserelationshipsand Â“ landscapemetric Â” datacanbeusedtotransfervaluestosites whereprimarysurveyworkhasnotbeencompleted.Inputdata alsoincludedemographicandattitudinalinformationaboutthe respondents,whichcanbeusedtoexploredifferencesinvalues acrossdifferentgroupsofrespondents. A nallandscape-scaletool,theUNEP-WCMCEcosystemServicesToolkit,identi esecosystemserviceimpactsandstakeholdersfor veservices:climateregulation,waterservices, harvestedwildgoods,cultivatedgoods,andnature-basedtourism andrecreation( UNEP-WCMC,2011 ).Ituses eldmeasurements, semi-structuredinterviews,expertconsultations,andpublished datatoquantifyecosystemservices.Giventhelackoffurther publishedinformationaboutthistoolkit,weareunabletodescribe itsperformanceagainsttheevaluativecriteriainthisstudy. 3.3.Independentlyapplicable,place-speci c,landscape-scale modeling:Envision,EPM,InFOREST Threeadditionalpublic-domaintools Â– Envision,Ecosystem PortfolioModel(EPM),andInFOREST Â– aredistinctlyplacespeci c,accountingfordetailedlocallyimportantecologicalprocessesandhumanpreferencesunderlyingecosystemservicesbut sacri cinggeneralizability.Thoughthesemodelscanbeadapted forapplicationinnewareas,thisisanexpensiveandtimeconsumingprocessthatislikelyimpracticalinmostcases;how-evertheymayprovidesubstantialinsightinregionswherethey havealreadybeendeveloped. Envisionisdesignedtoexplorehowdevelopmentpolicies affectland-useagentbehavioranddrivedevelopmentpatterns (i.e.,leadingtoalternativeurban-developmentscenarios),which yieldchangesinvariouslandscapemetrics,whichcaninclude ecosystemservicessuchasnutrientregulation,waterprovisioning,carbonsequestration,foodand berproduction,shoreline protection,andpollination( Guzyetal.,2008 ).Envisionisa modular,open-sourcemodelingframeworkthatcanincorporate externalecosystemservicemodelssuchasInVEST.Submodels quantifysocialpreferencesforeconomicdevelopment,landscape metrics,landvalue,andpopulationgrowthtolinkspatialdata withsetsofpoliciesthatachievecertainmixesofeconomicand environmentalgoals.Accompanyingeconomicvaluationisconductedusingmarketpricesoravoided/replacementcostmethods. EnvisionhaslargelybeenappliedintheU.S.Paci cNorthwest, thoughinternationalapplicationsinColombiaandNewZealand areunderdevelopment. TheEPMmodelsecological,economic,andquality-of-life values,offeringinsightintotheeffectsofland-usechange(includingdevelopment,conservation,andrestorationchoices)onthese values( Hoganetal.,2012 ; Labiosaetal.,2013 ).Somevaluesare monetized,likethepropertypremiumprovidedbyopenspace;forK.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e33
criteriathataredif culttovaluemonetarily,likebiodiversity, alternativeuserpreferencescanbecomparedusingamultiattributeutilityapproach.EPMcasestudieshavebeencompleted inMiami-DadeCounty,Florida,PugetSound,Washington,andthe SantaCruzRiverwatershed,Arizona.Whenanapplicationis complete,EPMfunctionsasaweb-basedtool,requiringtheuser tosimplychoosetheirareaofinterest,selectweightsforvaluation ofeachcriterion,andcompareresultsintheonlineviewer. TheInFORESTmodel( http://inforest.frec.vt.edu/ )isawebbasedassessmenttoolforquantifyingcarbon,watershednutrient andsedimentloading,andbiodiversity.Theuserenterstheonline interface,choosestheareaofinterest,and(ifdesired)entersland coverandagriculturalpracticesinformation.InFORESTisdesigned asanecosystemservicecreditcalculator;thusitdoesnotinclude economicvaluationasagoal.Itincorporatesaseriesofexisting carbonandhydrologicmodelsandhabitatmetrics.InFORESThas currentlybeendevelopedforapplicationonlyinthestateof Virginia. 3.4.Proprietary,generalizable,landscape-scalemodeling:EcoAIM, ESValue Twotools Â– EcoAIMandESValue Â– havebeendevelopedby private-sectorconsultantstomapandvalueecosystemservicesat thelandscapescale( Waageetal.,2011 ).EcoAIMisdesigned Â“ to(1) inventoryecologicalservicesandhelpinmakingdecisionsregardingdevelopment,transactions,andecologicalrestoration,(2) developspeci cestimatesofecosystemservicesinageographicallyrelevantcontext,and(3)offerthemeansforevaluating tradeoffsofecosystemservicesresultingfromdifferentlandor resourcemanagementdecisions Â” ( Waageetal.,2011 ).EcoAIMuses aseriesofpubliclyavailablespatialdatasetscombinedwitha weightingoraggregationfunctiontoderivespatiallyexplicit scoresforecosystemservicesofinterest.EcoAIMcanalsointegrate stakeholderpreferencesinconsideringecosystemserviceimpacts, usingamodi edrisk-analysisapproach. ESValuecombinesexpertandliterature-deriveddatato developecosystemserviceproductionfunctions( Waageetal., 2011 ).ESValuespeci estherelativevaluesthatsociety,managers, andstakeholdersplaceonecosystemservices,asdeveloped duringastakeholder-engagementprocess.TheESValuetoolthus facilitatesthecomparisonofwhatcanbeproduced(i.e.,the productionfunction)withwhatparticipantswanttobeproduced (i.e.,thevaluationfunction)toevaluatetradeoffsbetweennatural resourcemanagementstrategies. 3.5.Site-scalemodeling:EcoMetrix,LUCI Theabove-describedtoolsaregenerallydesignedtooperateat thelandscapescale,makingthemusefulformodelingwatersheds orlargeparcels,butaregenerallynotintendedforsite-scale analyses(i.e.,forareaslessthan 50ha),inpartduetotheir relianceonspatiallyexplicitdatathattypicallyhavealowerboundresolutionontheorderof30 30m.Site-scaleecosystem serviceinformation,evaluatingchangesonparcelsassmallas severalhectaresorless,couldbeusedtoselectbetweenrestorationordevelopmentalternativesat nespatialscalesafterappropriatemacro-levellocationsfortheseactivitieshavebeen identi edusingalandscape-scaletool.Asnotedpreviously,LUCI isonelandscape-scaletoolalsointendedforuseatthesitescale, assumingdataareavailableatanadequatespatialresolutionfor site-scaleanalysis. EcoMetrix,aproprietarytooldesignedforsite-scaleecosystem servicesassessment,combines eld-basedmeasurementswith spreadsheet-encodedproductionfunctionstoquantifysite-scale changesinecosystemservicesusingnon-monetary,servicespeci cmetrics( Parametrix,2010 ).Itsprimaryusehasbeento estimatethegenerationofenvironmentalcreditsformarket-basedtradingunderrestorationordegradationscenarios. 3.6.Monetaryvaluation:Bene tTransferanduseEstimatingModel toolkit,EcosystemValuationToolkit,NAIS Mostofthemodelingtoolsdescribedabovecanestimate monetaryvaluesbysupplyingaper-unitmarket,social,avoided, orreplacementcost(e.g.,socialcostpertonofcarbonoravoided costofdredgingatonofsediment).Thetoolsdescribedinthis sectionusevaluetransfertoestimatemonetaryvaluesforecosystemservices,independentlyoforinconjunctionwithother modelingtools. NaturalAssetsInformationSystem(NAIS)andtheEcosystem ValuationToolkitarevaluationdatabasesthatcombinealibraryof economicvaluationstudieswithGISanalysisoflandcover,which canbeusedforeconomicvaluationviapointtransfer( Troyand Wilson,2006 ; EcosystemValuationToolkit,2012 ).Spatiallyexplicitland-coverdata,classi edusingalocallyrelevantland-use/ covertypology,areusedasinputdata,andarethenmatchedto appropriatevaluationstudies.Outputsincludeper-hectaresummariesofecosystemservicevaluesforeachrelevantland-cover type.TheEcosystemValuationToolkitcanbeindependently appliedthroughasubscription,whileapplicationofNAISoccurs throughcontractingwithitsdevelopers. TheBene tTransferandUseEstimatingModelToolkit,by contrast,usesfunctiontransfer( Loomis,1992 ),withtransferfunctionsencodedinasetofpublic-domainspreadsheets( Loomisetal., 2008 ).Thetoolkitincludestransferfunctionsforrecreation,propertypremiums,andwillingnesstopayforthreatenedandendangeredspeciesrecovery.Theuserentersvaluesfortheindependent variablesrequiredbyagiventransferfunction(e.g.,open-space characteristicsoropen-waterarea),andthespreadsheetcalculates economicvalueperhouseholdorrecreationday. 3.7.ApplicationofselectedtoolstotheSanPedro We Â– orinthecaseoftheproprietarytoolsEcoAIM,EcoMetrix, andESValue,theirtooldevelopers Â– appliedseventoolstotheSan Pedrotodeterminethetimerequirementsofadaptingthesetools toanewstudyarea.TheSanPedro,atributaryoftheGilaRiver, owsnorthfromnorthernSonoraintosoutheastArizona.Itisa regionofhighecologicalsigni cance Â– oneofthelastfreeowing perennialriversintheU.S.Southwestandamajormigratorybird yway Â– butfacesseriouspressuresfromurbanizationand attendantgroundwaterdepletion( StrombergandTellman, 2009 ).ThestudyareaincludestheBLM sSanPedroRiparian NationalConservationArea,whichhasbeenafocalpointfor conservationandscienti cactivityinrecentdecades.Wequantiedservicesidenti edasimportantbyagroupof27localresource managersandscientistsincludingwater,carbonsequestrationand storage,biodiversity,recreation,andestheticvalues. Weevaluatedtheresponsivenessofecosystemservicestoolsto fourlocallyrelevantscenariosets,rangingfromlandscape-scale changetolocal-scalechangeontheorderofseveralhundred hectares.WeappliedtheInVESTandARIEStoolstomesquite management,wateraugmentation,watershed-scaleurbangrowth scenarios,andlocal-scalehousingdevelopmentscenarios( Bagstad etal.,2012 , inthisvolume ).ThedevelopersoftheEcoAIM, EcoMetrix,andESValuetoolsappliedthesetolocal-scalehousing developmentscenarios( Waageetal.,2011 ).WealsousedtheESR andBene tTransferandUseEstimatingModelToolkit,thoughthesetoolsareaspatialandinthecaseoftheESRproduced qualitativeresults.AlthoughtheBene tTransferandUseEstimatingModelToolkitisdesignedtosupportrapidmonetaryvaluationK.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e34
usingfunctiontransfer,itsvaluationfunctionswerenotwell suitedtovaluingthespeci cecosystemswithintheSanPedro ( Bagstadetal.,2012 ). Evengiventheircurrentverydistinctanalyticalapproachesand ecosystemservicemetrics,inthecomparativeapplicationofARIES andInVESTtotheSanPedrothetwotoolscametosimilarconclusions abouttheecosystemservicesimpactsfromavarietyofscenarios ( Bagstadetal.,inthisvolume ).However,evenafterthisapplication, furthermodeltestinganddevelopmentwouldbedesirable. Intheapplicationof vespatiallyexplicitmodelingtools Â– InVEST,ARIES,EcoAIM,EcoMetrix,andESValue Â– theirresults weretooincomparabletodrawdirectquantitativecomparisons, particularlyaboutwhetherthetools Â“ agreed Â” ontherelative impactsofdevelopmentatalternativesites( Waageetal.,2011 ). Unsurprisingly,theESR,ascreeningtool,andtheBene tTransfer andUseEstimatingModelToolkit,aspreadsheet-basedvalue transfertool,couldbecompletedmuchmorequicklythanthe spatiallyexplicitmodels( Table3 ).Oftheremainingmodels, EcoAIM,whichcalculatesaweightedaverageofpubliclyavailable GISlayersrelevanttotheserviceofinterest,andEcoMetrix,the site-scaleecosystemservicesscoringtool,couldbecompleted morequicklythantheremainingthreetools.However,publicdomainmodelssuchasInVESTandARIEScouldberunwith substantiallylowerresourcerequirementsifawiderarrayofinput dataandcontextuallyappropriatemodelswereavailabletothe user,asdiscussedfurtherin Section4.4 . 4.Conclusions 4.1.General ndings Thetoolsevaluatedinthisstudydifferedgreatlyintheir performanceagainsttheevaluativecriteria( Table2 ).Beyondthe keydistinctionbetweentheireaseofgeneralizability,different approacheswillbemoreappropriateindistinctgeographicand decisioncontexts,highlightingtheneedforfurthercomparative analysisofavailabletoolsindiversesettings(e.g., Bagstadetal.,in thisvolume ). Assumingthatatoolis exibleenoughtoquantifyecosystem servicesindiversecontextsandthatitsresultsarecredible Â– transparent,well-documented,andvalidatedwherepossible Â– a keytraitthatwillenhanceorlimititswidespreadadoptionisthe timerequiredtoapplyitrelativetothedepthandqualityof informationitaddstothedecision-makingprocess( Table3 ). Somecomplementarityexistsbetweentools,whichsuggests thatcertaintoolscouldbeusedtogetherto lldifferentecosystem serviceassessmentneeds,providedthatthetooloutputsare compatible(e.g.,thatmodeloutputscaneasilysupportvaluation; Fig.1 ).Forexample,theESRcanserveasa Â“ front-end Â” screening tooltoevaluateecosystemservicesofimportance,eitherinthe absenceoflocalstakeholderswhocanprovideinformedinput,or incollaborationwithstakeholdersasawaytostructuretheir input.Co$tingNaturecansimilarlyactasalow-cost,spatially explicitscreeningtoolforidentifyingpotentialecosystemservice hotspots,thoughitcannotdisaggregateservicesfortradeoff analysisandvaluation.Suchpreliminaryassessmentscouldthen beusedasabroaderanalytical Â‘ frame Â’ withinwhichtoconductmoregranularanalysesfrommappingandmodelingtoolsthat quantifylandscape-scaleecosystemservicestradeoffsusingbiophysicalmodels(e.g.,ARIES,EcoAIM,Envision,EPM,InVEST, InFOREST,LUCI,MIMES)and/orsurveystoelicitsocialvalues (SolVES).Ifneeded,EcoMetrixorLUCIcanbeusedforsite-scale modelingtocomparetradeoffsat nespatialscales.Forsome applicationsandecosystemservices,valuationcanbecompleted bysimplyapplyingaper-unitsocial,market,avoided,orreplacementcosttoresultsofbiophysicallymodeledservices.Inother cases,itmaybemoreappropriatetovaluemodeloutputsusing Table3 EstimatedtimetocompleteecosystemservicesassessmentsusingalternativemethodsfortheSanPedrocasestudy. Tool(servicesquanti ed)Estimatedperson-hoursInformationprovidedAdditionalcomments PilotstudyaWith improved data archivebBene tTransferandUseEstimating ModelToolkit(recreation) 1010Aspatialvaluation EcosystemServicesReview(27ecosystem services) 1010QualitativereviewTimetocompletioncouldbeseveraltimes greaterifalargenumberofstakeholders areinvolved InVEST(carbon,water,viewsheds,habitat quality) 27580SpatiallyexplicitoutputsTimetocompletecouldbedrastically reducedwithsystemforsharingdataand underlyingmodelassumptions ARIES(carbon,water,viewsheds,open spaceproximity,recreation) 80080Spatiallyexplicitoutputs,uncertainty, owdata Modelcustomizationanddebuggingwas extremelytimeconsumingbutwillbeless soforfutureapplications.Spatialdata managementsystemreducesdatainput needsinfutureapplications EcoMetrix(carbon,water,esthetic, recreation,culturalheritage, biodiversity) 8585RelativeservicescoresbysiteTimefor elddatacollection,dataentry, andanalysis EcoAIM(biodiversity)2525Spatiallyexplicitweightedaverage values TimetoprepareGISdataandrunoverlays ESValue(multicriteriaanalysisof22 ecosystemservices) 400400Relativepreferencesforalternative services Stakeholderengagementthemosttime consumingstepaAcommonanalystconductedanalysesusingtheBene tsTransferandUseEstimatingModelToolkit,EcosystemServicesReview,InVEST,andARIES.Thiswasnot possibleforthethreeproprietarytools Â– EcoMetrix,EcoAIM,andESValue.Inallcasestheanalystwasexperiencedinecosystemservicemodelingandvaluation,with graduate-leveltraining.bSuchanimproveddataarchive Â– includingspatialandaspatialdatatopopulatetheInVESTandARIESmodels,anddescriptionsofcontextsunderwhichsuchdataare applicable Â– couldsubstantiallyreducethetimerequirementsneededtoapplythesemodels,andcouldalsolikelybene ttheapplicationofproprietarytools. K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e35
multicriteriaanalysistools(e.g.,EcoAIM,ESValue)ormonetary valuationusingNAIS,theEcosystemValuationToolkit,orthe Bene tTransferandUseEstimatingModelToolkit. 4.2.Feasibilityforwidespreaduse Whileanyofthesetoolscanbeusedgivenadequateresources, theydifferintheirappropriatenessforwidespreaduseinpublicorprivate-sectorsettings,whererapidbutreliableassessmentsin diversegeographiccontextsaredesirable( Tables1 Â– 3 ).Basedon thisreviewandtheapplicationofseventoolstotheSanPedroin Arizona,andfollowingdiscussionswithadiversegroupofpublicandprivate-sectordecisionmakers,whodeemedasdesirable toolsthatarequanti able,replicable,credible, exible,andaffordable,wesummarizethecurrentreadinessofthesetoolsbelow. Withtheexceptionoflow-costscreeningtools,mostofthese decisionmakersfeltthatthetimeandcostrequirementstorun quantitativeecosystemservicemodelsremaintoohighforthese toolstobeusedinwidespreaddecisionmaking( Table3 ).Thisis particularlytrueastheiraddedvaluerelativetoexistingenvironmentalassessmentsremainstobeshowninpractice,evenafter multipleapplications( Waageetal.,2012 ).Forexample,qualitative reviewsofecosystemservicesorapplicationofspreadsheet modelstookonlyabout10h,whileapplicationofspatiallyexplicit modelingtoolsrequiredhundredsofhoursofworkbyanexperiencedanalyst.Whetherornottheseassessmentsyieldednew insightsrelativeto Â‘ businessasusual Â’ isthekeynextquestion. Basedonthecriteriade nedabovefortoolsconsideredinthis study,weconcludethefollowingabouttheirreadinessforwidespreaduseinpublic-andprivate-sectordecisionmaking.We presentusefeasibilityratherthanspeci callyfavoringoneormore toolsduetothediversityofdecisioncontexts,userneeds,ongoing evolutionoftools,andneedformorecomparativetesting:Feasibleforimmediatewidespreaduse :ESR,Bene tTransferand UseEstimatingModelToolkit,Co$tingNature.Potentiallyfeasibleforwidespreadusegivendevelopmentof supportingdatabasesforspatialandecologicaldata :InVEST.Potentiallyfeasibleforwidespreadusegivenimprovedguidance ontooluseandfeasibilityofconductingafullstakeholder engagementprocess :LUCI.Potentiallyfeasibleforwidespreadusegivenfuturedevelopment ofglobalmodelsorexpandedunderlyingdatasets :ARIES,EcoServ,SolVES.Proprietarytools,feasibleforuseinhigh-pro lecaseswhere contractingwithconsultantsordevelopers,orpayingfora subscriptionispossible :EcoAIM,EcoMetrix,EcosystemValuationToolkit,ESValue,NAIS.Public-domaintoolsthatareplace-speci c,requirealonglead timetodevelop,and/orrequirecontractingwithuniversitiesor consultants.Ifmodelshavebeenpreviouslydevelopedforanarea ofinteresttheycouldbeimmediatelyapplied :Envision,EPM, InFOREST,MIMES. Additionalmulti-toolreviewsandcomparativequantitative testingarethusdesirablefortwoimportantreasons:tobetter understandthetools timerequirementsandusefeasibilityin morediversegeographicanddecisioncontexts,andtotrackthe developmentofnewtoolsandexpandedcapabilitiesofexisting tools. 4.3.Implicationsforpublic-andprivate-sectorresource management Atpresent,fewtoolshavebeenpilottestedinagencyor corporatesettings,particularlyincomparativeassessments.In addition,noneofthesetoolsreadilymeshwithkeyexisting corporateprocessesand,withtheexceptionoflow-costscreening tools,requireconsiderableefforttoapply,whichservesasan impedimenttoimmediate,widespread,off-the-shelfbusiness application( Waageetal.,2011 ).Allofthetoolswouldrequire supplementaleffortforcorporateapplications,eitherintermsof assistancewithinterpreting ndingswithinacorporatesettingor customizationto tparticularcorporatedecision-makingcontexts. Public-andprivate-sectormanagerssimilarlyneedclarityonhow, when,andwhytoapplytoolstoparticulardecisioncontexts ( Waageetal.,2011 ). Quantifying,mapping,andvaluingecosystemservicesdoes offerthepublicandprivatesectoralikeapromisingwayto communicateresourcemanagementtradeoffs,particularlyfor developmentorextractiveresourceusethatcoulddegradeecosystemservices.However,dependingonthedecisioncontext, ecosystemserviceanalysismaybemoreorlessuseful,whichwill inpartbecontingentuponwhatadditionalnewinsightsan ecosystemservicesapproachoffersrelativetothe Â“ businessas usual Â” approachtoconductingenvironmentalimpactassessments. Inapublic-sectorsetting,suchasfortheBLM,analysisof ecosystemservicesorothernonmarketvaluesislikelytobemost usefulwhen:(1)aproposedactionislikelytohaveasigni cant directorindirecteffect(ecological,esthetic,historic,cultural, economic,social,orhealth),andthequalityormagnitudeofthe effectcanbeclari edbyconsideringsuchvalues,(2)the Fig.1. Potentialstepsinecosystemservicesassessmentprocess. K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e36
alternativeactionstobeconsideredpresentastrongcontrast betweenextractiveandnon-extractiveusesoflandandresources, or(3)themagnitudeoftheproposedchangeislarge( BLM,2013 ). 4.4.Loweringbarrierstoecosystemservicemodelparameterization andapplication AlthoughtheSanPedrowaschosenasastudyareaduetoits largebodyofpastresearch,muchofthisscienti cknowledgewas notusefulforparameterizingtheecosystemservicemodels,asit didnotoverlapwithmanyofthemodels Â’ inputdataneeds.Even forareaswithrichecologicalunderstanding,thisknowledgeisnot alwaysofthetypeneededtosupportecosystemservicemodeling, mapping,andvaluation( Norgaard,2010 ).Lookingforward,if ecosystemservicesapproachesaretobewidelyadoptedand applied,suchdatachallengeswillhavetobeaddressed.Inthe process,itwillbeessentialtofostercollaborationbetween ecosystemservicemodelersanddisciplinaryresearchersinorder tointegratepastworkintoecosystemservicemodelsanddevelop newresearchmethodsandidentifyindicatorstoquantifyecosystemserviceproductionfunctions.Participationofresourcemanagers Â– theendusersofthesetools Â– canhelpinformtool developersaboutwhichmetricsarelikelytobemosthelpfulin variousdecisioncontexts. Someoftheseissuesmayberesolvedinthenextgenerationof theecosystemservicesanalyticalmodels.Forexample,tool developersindicatedthroughdiscussionsthat,futureversionsof ARIES,EcoServ,Envision,InVEST,andothermodelsintendtolink morecompletelytoexisting,peer-reviewedecologicalandbiophysicalprocessmodels.Thiswouldbeamajorstepforwardfor ecosystemservicemodeling,butrequiressubstantialworkon modelsemantics,inputs,andoutputstobuildlinkagesbetween models. Althoughmodelerstypicallyrecognizetheneedformoredata, suchdataalsoneedtobebetterorganizedandaccessibletomodel userswhentheyseektochooseandparameterizeamodel. Althoughanambitiousgoal,semanticmetamodelingoffersa pathforwardinimprovingecosystemservicequanti cationinan eraof Â“ bigdata Â” ( FoxandHendler,2009 ; Villa,2010 ).Ecosystem servicepractitionerswouldbene tfromasystemofdatasharing for(1)spatialdata,(2)ecologicalstudiestoparameterizeecosystemservicemodels,and(3)economicstudiestosupportvaluation. Thetimespentonthispilotwouldhavebeensubstantially reducedifsuchresourceswereavailable,andtheywouldalso reducethelikelihoodthatpractitionerswilloverlookimportant datasources.Strategicinvestmentinsuchsystemscouldbe supportedbyFederalagencies,philanthropicfoundations,or industrygroupstosupportpublic-andprivate-sectorecosystem service-baseddecisionmaking.Althoughinsomecaseshigher qualitylocaldatamayexistandstakeholdersmaytrustlocally collecteddataover Â“ pre-wired Â” data,formanyothercaseswelldocumenteddataobtainedfromcrediblesourcescouldgive modelingeffortsalargeheadstart. WhileU.S.agenciesliketheUSGSandNaturalResources ConservationService(NRCS)houseabundantpublicdataonland cover,hydrology,andgeology,nosinglesitecontainedallthe spatialdataneededtorunecosystemservicesmodels.Collecting, storing,andpre-processingrelevantspatialdatainasingle locationcouldsavefutureuserssubstantialtimeandeffort.In thisregard,spatialdatamanagementthroughWebCoverage ServiceandWebFeatureService(WCS/WFS)thatcancallon annotatedspatialdatatosupportmultipleecosystemservice modelscouldbescaleduptosupportmultipletools.Emerging environmentaldatasharing,remotesensing,andvisualization toolsandpracticescanalsosupportnext-generationecosystem servicemodeling( EyeonEarth,2012 ; LifeWatch,2012 ; Geographic EcosystemMonitoringandAssessmentServiceProject,2013 ).These sourcescouldenhancethequalityandcredibilityofecosystem serviceassessmentsiftheycanimprovethecurrency,spatial resolution,andqualityofmodelinputandcalibrationdata. Ecosystemservicevaluationdatabaseshavebeendevelopedinthepastbuthavetoorarelyreceivedfundingformaintenanceand expansion( McCombetal.,2006 ; Curticeetal.,2012 ).TheBene t TransferandUseEstimatingModelToolkitisfree,NAISisa proprietarydatabaseandisnotavailableforpublicaccess,and theEcosystemValuationToolkitusesatieredsubscriptionranging fromfreeforcontributorstoanannualfeebasedonapplieduse. Otherdatabaseslackfunctionstoguideusersthroughtheprocess ofconstructingavaluationportfoliofortheirareaofinterest,but provideusefulrepositoriesofnonmarketvaluationstudies.Such databasesthathavebeenrelativelywellmaintainedinrecent yearsincludetheEnvironmentalValuationReferenceInventory (EVRI)database( EnvironmentalValuationReferenceInventory (EVRI),2011 ),MarineEcosystemServicesDatabase( Marine EcosystemServicesPartnership,2013 ),andtheTEEBValuation Database( vanderPloeganddeGroot,2010 ). Justasdatabasescatalogingeconomicstudiescansupport valuation,databasesofecologicalstudiesareneededtosupport modelingefforts.Aswebetterunderstandthedataneedsfor ecosystemservicemodels,itwouldbevaluabletodevelop databasesfortheecologicalparametersthatunderliesuchmodels. Forinstance,theTier1InVESTmodelslinkecosystemservice provisiontolanduse/coverviatables.Havingaccuratevaluesfor useinthesetables(e.g.,forcarbonstorage,rootingdepth,nutrient loading,andevapotranspirationcoef cientsbyland-use/cover type)iscriticaltorunningthemodelsandobtainingcredible results.Forothermodelingsystems,suchecologicalinformationis neededtoidentifyappropriatecontextstoapplyspeci cecological productionfunctions.TheU.S.EnvironmentalProtectionAgencyis beginningworkonan Â“ ecologicalproductionfunctionlibrary Â” that couldhelp llthisneedforfutureecosystemservicemodelers. Althoughsystemsmodelingmayremainthegoalofmore accuratelyrepresentingcomplexprocessesinquantifyingecosystemservices( Seppeltetal.,2011 ),thepotentialgainsinaccuracy associatedwiththisapproachmustbeweighedagainstthe increasedcomplexityandreducedgeneralizabilityinatimewhere ecosystemserviceassessmentsareincreasinglyseenasimportant inputstodecisionmaking( TallisandPolasky,2011 ).Simpler modelsmayalsogenerategreatertransparencyandtrustamong users,asmayincorporationoflocaldataandmodels( Smartetal., 2012 ).ModelingapproachessuchasLUCIhaveintentionally consideredthisissuethroughsimplifyingecosystemservicemodel outputstoimprovetheirintuitiveness( Jacksonetal.,2013 ). Comparativestudiesofsimpli edandcomplexmodels,which canhelpusunderstandthepotentialgainswhenaccountingfor complexity,areincreasinglycommoninecologicalandhydrologic modeling( Perrinetal.,2001 ; Fultonaetal.,2004 ; Raickaetal., 2006 ; Irmaketal.,2008 ).Theyhavebeenlesswellexploredin ecosystemservicemodeling(butsee TallisandPolasky,2011 ),and willbeanimportantareaoffutureresearchasdecisionmakers seektoidentifytoolsthatcanbeusedinavarietyofsettingswhile providingaccurateandusefulinformation. Asecosystemservicetoolscontinuetodevelop,additionalcase studiesmaysuggestmeanstobetterintegratewithinternal public-andprivate-sectordecisionprocesses,allowingtheecosystemserviceconcepttobetterdeliveronitspromiseofsupportingmoresustainabledecisionmaking( Dailyetal.,2009 ).Inan evolvingtoollandscape,public-andprivate-sectoractorsmust developanunderstandingnotonlyofusingecosystemservices conceptsandtools,butalsoofcostsandresourcesneededtodevelopandmaintainthetools,trainstaff,andintegratetheseinto planning,operations,andgovernance.K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e37
Disclosurestatement Theleadauthor(K.J.Bagstad),wholedcomparativeanalysisof ecosystemservicetools,hasworkedasaco-developeroftheARIES toolsince2007. Acknowledgments ThisprojectwasfundedbytheU.S.DepartmentofInteriorBureauofLandManagementandBSR.WethankTomDabbs, DistrictManageroftheBLMGilaDistrictforapprovingand supportingthisproject.JimBoyd,FrankCasey,BillyGascoigne, andLynneKoontzprovidedguidanceonecosystemservices valuationandreviewedprojectdocuments.DaveGoodrich,Delilah Jaworski,JoelLarson,MalkaPattison,andMarkRekshynskyj providedassistancewithprojectscopingandtechnicalreview throughouttheproject.BillLabiosaprovidedconstructivefeedbackonanearlierdraftofthismanuscript.Wealsothanka stakeholdergroupandexpertreviewpanelthatassistedwith projectscopingandmodelreview,respectively.Forafulllistof contributorstothesegroups,seethesupportingonlinematerial. Anyuseoftrade, rm,orproductnamesisfordescriptive purposesonlyanddoesnotimplyendorsementbytheU.S. Government. AppendixA.Supportinginformation Supplementarydataassociatedwiththisarticlecanbefoundin theonlineversionat http://dx.doi.org/10.1016/j.ecoser.2013.07.004 . ReferencesAmbrose-Oji,B.,Pagella,T.,2012.Spatialanalysisandprioritisationofcultural ecosystemservices:areviewofmethods,ForestResearch.AliceHoltLodge Farnham,Surrey . AmericanForests,2002.CITYgreen:CalculatingtheValueofNature,Version5.0 User sManual,AmericanForests,Washington,D.C. Arnold,J.G.,Fohrer,N.,2005.SWAT2000:currentcapabilitiesandresearch opportunitiesinappliedwatershedmodeling.HydrologicalProcesses19(3), 563 Â– 572 . Bagstad,K.J.,Villa,F.,Johnson,G.,Voigt,B.,2011.ARIES Â— Arti cialIntelligencefor EcosystemServices:AGuidetoModelsandData,Version1.0Beta.TheARIES Consortium,Bilbao,Spain . Bagstad,K.J.,Semmens,D.,Winthrop,R.,Jaworski,D.,Larson,J.,2012.Ecosystem ServicesValuationtoSupportDecision-MakingonPublicLands:ACaseStudy fortheSanPedroRiver,Arizona.(USGSScienti cInvestigationsReport2012 Â– 5251) . Bagstad,K.J.,Johnson,G.W.,Voigt,B.,Villa,F.,2013.Spatialdynamicsofecosystem service ows:acomprehensiveapproachtoquantifyingactualservices. EcosystemServices4,117 Â– 125.. Bagstad,K.J.,Semmens,D.J.,Winthrop,R.,2013.Comparingapproachestospatially explicitecosystemservicemodeling:acasestudyfromtheSanPedroRiver, Arizona.EcosystemServices . BureauofLandManagement(BLM),2005.BLMlanduseplanninghandbook,H1601-1.U.S.DepartmentofInterior-BureauofLandManagement,Washington, DC . BLM,2013.BureauofLandManagementSocioeconomicsProgramGuidance: EconomicMethodsforEstimatingNonmarketEnvironmentalValues.InstructionMemorandumNo.2013-131,U.S.DepartmentofInterior-BureauofLand Management;Washington,DC. Bohnet,I.C.,Roebeling,P.C.,Williams,K.J.,Holzworth,D.,vanGrieken,M.E.,Pert, P.L.,Kroon,F.J.,Westcott,D.A.,Brodie,J.,2011.LandcapesToolkit:anintegrated modellingframeworktoassiststakeholdersinexploringoptionsforsustainable landscapedevelopment.LandscapeEcology26,1179 Â– 1198 . Brown,G.,Reed,P.,2000.Validationofaforestvaluestypologyforuseinnational forestplanning.ForestScience46(2),240 Â– 247 . CouncilonEnvironmentalQuality(CEQ),2013.PrinciplesandRequirementsfor FederalInvestmentsinWaterResources.CouncilonEnvironmentalQuality, Washington,DC . Curtice,C.,Dunn,D.C.,Roberts,J.J.,Carr,S.D.,Halpin,P.N.,2012.Whyecosystembasedmanagementmayfailwithoutchangestotooldevelopmentand nancing.BioScience62(5),508 Â– 515 . Daily,G.C.,Polasky,S.,Goldstein,J.,Kareiva,P.M.,Mooney,H.A.,Pejchar,L.,Ricketts, T.H.,Salzman,J.,Shallenberger,R.,2009.Ecosystemservicesindecision making:timetodeliver.FrontiersinEcologyandtheEnvironment,7,21 Â– 28 . Ecosystem-BasedManagementToolsDatabase,2012.EBMToolsDatabase.Availablefrom: http://ebmtoolsdatabase.org/ (accessed28.11.12). EcosystemValuationToolkit,2012.EcosystemValuationToolkit.Availablefrom: http://esvaluation.org (accessed03.05.12). Egoh,B.,Drakou,E.G.,Dunbar,M.B.,Maes,J.,Willemen,L.,2012.Indicatorsfor MappingEcosystemServices:AReview.PublicationsOf ceoftheEuropean Union,Luxembourg.(ReportEUR25456EN) . Eigenbrod,F.,Armsworth,P.R.,Anderson,B.J.,Heinemeyer,A.,Gillings,S.,Roy,D.B., Thomas,C.D.,Gaston,K.J.,2010.Theimpactofproxy-basedmethodson mappingthedistributionofecosystemservices.JournalofAppliedEcology 47,377 Â– 385 . EnvironmentalSystemsResearchInstitute(ESRI),2013.ArcGIS:Mappingand SpatialAnalysisforUnderstandingOurWorld.Availablefrom: http://www. esri.com/software/arcgis . EnvironmentalValuationReferenceInventory(EVRI),2011.EVRI.Availablefrom: https://www.evri.ca/Global/HomeAnonymous.aspx (accessed03.05.12). EyeonEarth,2012.EyeonEarth.Availablefrom: http://eyeonearth.org/en-us/ pages/home.aspx (accessed03.05.12). Farber,S.,Costanza,R.,Childers,D.,Erickson,J.,Gross,K.,Grove,M.,Hopkinson,C., Kahn,J.,Pincetl,S.,Troy,A.,Warren,P.,Wilson,M.,2006.Linkingecologyand economicsforecosystemmanagement.BioScience56,121 Â– 133 . Feng,M.,Liu,S.,EulissJr.,N.H.,Young,C.,Mushet,D.M.,2011.Prototypinganonline wetlandecosystemservicesmodelusingopenmodelsharingstandards. EnvironmentalModelling&Software26,458 Â– 468 . Fox,P.,Hendler,J.,2009.Semanticscience:encodingmeaninginnext-generation digitallyenhancedscience.In:Hey,T.,etal.(Eds.),TheFourthParadigm: Data-IntensiveScienti cDiscovery.MicrosoftResearch,Redmond,WA, pp.147 Â– 152 . Fultona,E.A.,Parslowa,J.S.,Smitha,A.M.D.,Johnson,C.R.,2004.Biogeochemical marineecosystemmodelsII:theeffectofphysiologicaldetailonmodel performance.EcologicalModelling173(4),371 Â– 406 . GeographicEcosystemMonitoringandAssessmentServiceProject,2013.GeographicEcosystemMonitoringandAssessmentServiceProject.Availablefrom: http://www.space4ecosystems.com/projects/geographic-ecosystem-monitorin g-and-assessment-service-project/ (accessed06.05.13). Guzy,M.R.,Smith,C.L.,Bolte,J.P.,Hulse,D.W.,Gregory,S.V.,2008.Policyresearch usingagentbasedmodelingtoassessfutureimpactsofurbanexpansioninto farmlandsandforests.EcologyandSociety13(1),37 . Hogan,D.M.,Labiosa,W.,Pearlstine,L.,Hallac,D.,Strong,D.,Hearn,P.,Bernknopf, R.,2012.Estimatingthecumulativeecologicaleffectoflocalscalelandscape changesinSouthFlorida.EnvironmentalManagement49(2),502 Â– 515 . InstituteforEuropeanEnvironmentalPolicy(IEEP),AlterraWageningenUR, Ecologic,NetherlandsEnvironmentalAssessmentAgency,UnitedNations EnvironmentProgramme-WorldConservationMonitoringCentre,2009.ScenariosandModelsforExploringFutureTrendsofBiodiversityandEcosystem ServicesChanges.FinalreporttotheEuropeanCommission,DGEnvironment onContractENV.G.1/ETU/2008/0090r.Availablefrom: http://ec.europa.eu/ environment/enveco/biodiversity/pdf/Biodiversity_Scenarios_Models.pdf (accessed06.03.13). Irmak,S.,Istanbulluoglu,E.,Irmak,A.,2008.Anevaluationofevapotranspiration modelcomplexityagainstperformanceincomparisonwithBowenratioenergy balancemeasurements.TransactionsoftheASAE51(4),1295 Â– 1310 . Jackson,B.,Pagella,T.,Sinclair,F.,Orellana,B.,Henshaw,A.,Reynolds,B.,McIntyre, N.,Wheater,H.,Eycott,A.,2013.Polyscape:aGISmappingtoolboxproviding ef cientandspatiallyexplicitlandscape-scalevaluationofmultipleecosystem services.LandscapeUrbanPlan112,74 Â– 88 . Kareiva,P.,Tallis,H.,Ricketts,T.H.,Daily,G.C.,Polasky,S.,2011.NaturalCapital: TheoryandPracticeofMappingEcosystemServices.OxfordUniversityPress, Oxford . Labiosa,W.B.,Forney,W.M.,Esnard,A.-M.,Mitsova-Boneva,D.,Bernknopf,R., Hearn,P.,Hogan,D.,Pearlstine,L.,Strong,D.,Gladwin,H.,Swain,E.,2013. Anintegratedmulti-criteriascenarioevaluationwebtoolforparticipatory land-useplanninginurbanizedareas:theecosystemportfoliomodel.EnvironmentalModelling&Software41,210 Â– 222 . LifeWatch,2012.LifeWatchProject.Availablefrom: http://www.lifewatch.eu/web/ guest/home (accessed03.05.13). Loomis,J.B.,1992.Theevolutionofamorerigorousapproachtobene ttransfer: bene tfunctiontransfer.WaterResourcesResearch28(3),701 Â– 705 . Loomis,J.,Kroeger,T.,Richardson,L.,Casey,F.,2008.Abene ttransfertoolkitfor sh,wildlife,wetlands,andopenspace.WesternEconomicsForum7, 33 Â– 43 . Maes,J.,Hauck,J.,Paracchini,M.L.,Ratamaki,O.,Termansen,M.,Perez-Soba,M., Kopperoinen,L.,Rankinen,K.,Schagner,J.P.,Henrys,P.,Cisowska,I.,Zandersen, M.,Jax,K.,LaNotte,A.,Leikola,N.,Pouta,E.,Smart,S.,Hasler,B.,Lankia,T., Andersen,H.E.,Lavalle,C.,Vermaas,T.,HussenAlemu,M.,Scholefeld,P.,Batista, F.,Pywell,R.,Hutchins,M.,Blemmer,M.,Fonnesbech-Wulff,A.,Vanbergen,A.J., Munier,B.,Baranzelli,C.,Roy,D.,Thieu,V.,Zulian,G.,Kuussaari,M.,Thodsen,H., Alanen,E.-L.,Egoh,B.,Sorensen,P.B.,Braat,L.,Bidoglio,G.,2012.ASpatial AssessmentofEcosystemServicesinEurope:Methods,CaseStudies,andPolicy Analysis Â— Phase2,SynthesisReport.PEERReportno.4.Partnershipfor EuropeanEnvironmentalResearch,Ispra. MarineEcosystemServicesPartnership,2013.ExploreEcosystemServices.Availablefrom: http://mesp2.env.duke.edu/explore (accessed03.05.13). K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e38
Martinez-Harms,M.J.,Balvanera,P.,2012.Methodsformappingecosystemservice supply:areview.InternationalJournalofBiodiversityScience,Ecosystem Services,andManagement8(1 Â– 2),17 Â– 25 . McComb,G.,Lantz,V.,Nash,K.,Rittmaster,R.,2006.Internationalvaluation databases:overview,methods,andoperationalissues.EcologicalEconomics 60,461 Â– 472 . MillenniumEcosystemAssessment(MA),2005.MillenniumEcosystemAssessment:LivingBeyondourMeans Â— NaturalAssetsandHumanWell-Being.World ResourcesInstitute,Washington,DC . Mulligan,M.,Guerry,A.,Arkema,K.,Bagstad,K.,Villa,F.,2010.Capturingand quantifyingthe owofecosystemservices.In:Silvestri,S.,Kershaw,F.(Eds.), FramingtheFlow:InnovativeApproachestoUnderstand,ProtectandValue EcosystemServicesAcrossLinkedHabitats.UNEPWorldConservationMonitoringCentre,Cambridge,UK,pp.26 Â– 33 . NatureServ,2013.NatureServeVista:DecisionSupportforBetterPlanning.Availablefrom: http://www.natureserve.org/prodServices/vista/overview.jsp (accessed06.05.13). Nelson,E.J.,Daily,G.C.,2010.Modellingecosystemservicesinterrestrialecosystems.Facultyof1000BiologyReports2,53 . Norgaard,R.B.,2010.Ecosystemservices:fromeye-openingmetaphortocomplexityblinder.EcologicalEconomics69,1219 Â– 1227 . Parametrix,2010.AnIntroductiontoEcoMetrix:MeasuringChangeinEcosystem PerformanceattheSiteScale.Parametrix,Portland,OR . Perrin,C.,Michel,C.,AndrÃ©assian,V.,2001.Doesalargenumberofparameters enhancemodelperformance?Comparativeassessmentofcommoncatchment modelstructureson429catchments.JournalofHydrology242(3 Â– 4),275 Â– 301 . President sCouncilofAdvisorsonScienceandTechnology(PCAST),2011.Reportto thePresident:SustainingEnvironmentalCapital:ProtectingSocietyandthe Economy,ExecutiveOf ceofthePresidentoftheUnitedStates,Washington, DC. Pressey,R.L.,Watts,M.,Ridges,M.,Barrett,T.,2005.C-PlanConservationPlanning Software,UserManual.NSWDepartmentofEnvironmentandConservation. Availablefrom: http://www.edg.org.au/software/cplan/C-Plan_Manual_2005. pdf (accessed03.05.13). Raicka,C.,Soetaertb,K.,GrÃ©goire,M.,2006.Modelcomplexityandperformance: howfarcanwesimplify?ProgressinOceanography70(1),27 Â– 57 . Ruhl,J.B.,Kraft,S.E.,Lant,C.L.,2007.TheLawandPolicyofEcosystemServices. IslandPress,Washington,DC . Schroter,D.,Cramer,W.,Leemans,R.,Prentice,I.C.,Araujo,M.B.,Arnell,N.W., Bondeau,A.,Bugmann,H.,Carter,T.R.,Gracia,C.A.,delaVega-Leinert,A.C., Erhard,M.,Ewert,F.,Glendining,M.,House,J.I.,Kankaanpaa,S.,Klein,R.J.T., Lavorel,S.,Linder,M.,Metzger,M.J.,Meyer,J.,Mitchell,T.D.,Reginster,I., Rounsevell,M.,Sabate,S.,Sitch,S.,Smith,B.,Smith,J.,Smith,P.,Sykes,M.T., Thonicke,K.,Thuiller,W.,Tuck,G.,Zaehle,S.,Zierl,B.,2005.Ecosystemservice supplyandvulnerabilitytoglobalchangeinEurope.Science310,1333 Â– 1337 . Seppelt,R.,Dormann,C.F.,Eppink,F.V.,Lautenbach,S.,Schmidt,S.,2011.A quantitativereviewofecosystemservicestudies:approaches,shortcomings, andtheroadahead.JournalofAppliedEcology48,630 Â– 636 . Sherrouse,B.C.,Clement,J.M.,Semmens,D.J.,2011.AGISapplicationforassessing, mapping,andquantifyingthesocialvaluesofecosystemservices.Applied Geography31,748 Â– 760 . Simulistics,2013.SystemDynamicsandObject-BasedModelingandSimulation Software.Availablefrom: http://www.simulistics.com . Smart,S.M.,Pearce-Higgins,J.P.,Wright,L.,Comber,A.J.,Howard,D.C.,Maskell,L.C., Jones,M.L.M.,2012.ExploringtheFuture:Phase1 Â— ScopingCurrentandFuture UseofSpatialDecisionSupportTools(sDST)forIntegratedPlanningforLandUse,Biodiversity,andEcosystemServicesAcrossEngland.FinalReporttoDefra, ProjectcodeWC0794. Stromberg,J.C.,Tellman,B.(Eds.),2009.EcologyandConservationoftheSanPedro River.UniversityofArizonaPress,Tucson,AZ . Tallis,H.T.,Polasky,S.,2011.Howmuchinformationdomanagersneed?The sensitivityofecosystemservicedecisionstomodelcomplexity.In:Kareiva,P., Tallis,H.,Ricketts,T.H.,Daily,G.C.,Polasky,S.(Eds.),NaturalCapital:Theoryand PracticeofMappingEcosystemServices.OxfordUniversityPress,Oxford, pp.264 Â– 277 . Tallis,H.T.,Ricketts,T.,Guerry,A.,Wood,S.A.,Sharp,R.,Nelson,E.,Ennaanay,D., Wolny,S.,Olwero,N.,Vigerstol,K.,Pennington,D.,Mendoza,G.,Aukema,J., Foster,J.,Forrest,J.,Cameron,D.,Arkema,K.,Lonsdorf,E.,Kennedy,C.,Verutes, G.,Kim,C.,Guannel,G.,Papenfus,M.,Toft,J.,Marsik,M.,Bernhartdt,J.,Grif n, R.,2013.InVEST2.5.3User sGuide.TheNaturalCapitalProject,Stanford,CA . Troy,A.,Wilson,M.A.,2006.Mappingecosystemservices:practicalchallengesand opportunitiesinlinkingGISandvaluetransfer.EcologicalEconomics60, 435 Â– 449 . UNEP-WCMC,2011.MeasuringandMonitoringEcosystemServicesattheSite Scale:IntroducingAPracticalToolkit,UNEP-WCMC,Cambridge,UK. vanderPloeg,S.,deGroot,R.S.,2010.TheTEEBValuationDatabase Â— ASearchable Databaseof1310EstimatesofMonetaryValuesofEcosystemServices. FoundationforSustainableDevelopment,Wageningen,TheNetherlands . Vigerstol,K.L.,Aukema,J.E.,2011.Acomparisonoftoolsformodelingfreshwater ecosystemservices.JournalofEnvironmentalManagement92,2403 Â– 2409 . Villa,F.,2010.Bridgingscalesandparadigmsinnaturalsystemsmodeling. MetadataandSemanticResearch:CommunicationsinComputerandInformationScience108,1 Â– 7 . Villa,F.,Bagstad,K.,Johnson,G.,Voigt,B.,2011.Scienti cinstrumentsforclimate changeadaptation:estimatingandoptimizingtheef ciencyofecosystem servicesprovision.EconomiaAgrariayRecursosNaturales11(1),54 Â– 71 . Waage,S.,Armstrong,K.,Hwang,L.,2011.NewBusinessDecision-MakingAidsin AnEraofComplexity,Scrutiny,andUncertainty:ToolsforIdentifying,Assessing,andValuingEcosystemServices,BSR,SanFrancisco,California. Availablefrom: http://www.bsr.org/reports/BSR_ESTM_WG_Comp_ES_Tools_ Synthesis3.pdf (accessed23.05.13). Waage,S.,Hwang,L.,Armstrong,K.,2012.TheQuiet(R)evolutioninExpectationsof CorporateEnvironmentalPerformance:EmergingTrendsintheUptakeof EcosystemServices.BSR,SanFrancisco,California.Availablefrom: http:// www.bsr.org/reports/Quiet_Revolution_Env_Performance.pdf (accessed 23.05.13). Wilson,M.,Hoehn,J.,2006.Valuingenvironmentalgoodsandservicesusing bene ttransfer:thestateoftheartandscience.EcologicalEconomics60, 335 Â– 342 . WorldResourcesInstitute(WRI),2012.TheCorporateEcosystemServicesReview: GuidelinesforIdentifyingBusinessRisksandOpportunitiesArisingfrom EcosystemChange,Version2.0.WorldResourcesInstitute,Washington,DC. K.J.Bagstadetal./EcosystemServices5(2013)e27 Â– e39 e39
1 ASSESSMENT OF CLIMATE REGULATION, CARBON SEQUES T RATION, AND NUTRIENT CYCLING ECOSYSTEM SERVICES IMPACTED BY MULTIPLE STRESSORS By PASICHA CHAIKAEW A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
2 Â© 2014 Pasicha Chaikaew
3 To my loving parents and sisters
4 ACKNOWLEDGMENTS I would l ike to thank my advisor, Sabine Grunwald, for the opportunity to work with her and for her invaluable academic and professional coaching. I truly appreciate my supervisory committee, Alan Hodges, Howard Beck, Samira Daroub, and Timothy Martin, for their gu idance. I owe a great deal of thanks to the Royal Thai Government for financial support throughout the doctorate program and to the Chulalongkorn University, Department of Environmental Science for the opportunity to work with them after graduation. I also thank the University of Florida, Soil and Water Department for providing the utility resources. Funding for this research was provided by the National Institute of Food and d and by the Institute of Food and Agricultural Sciences at the University of Florida (UF/IFAS). The survey questionnaire was funded by Alan Hodges (UF/IFAS). To my Geographi c Information System and Pedometrics Laboratory colleagues, I thank Betty Cao, Wade Ross, Xiong Xiong, Chris Clingensmith, Carla Gavilan, Xu Yiming, Risa Patarasuk, Jongsung Kim, and Nichola Knox for their friendship and support. I appreciate Brandon Hoove r and David Depatie for their expertise in Information Technology assistance, and Carol Fountain for her professional editing. I would like to give special thanks to Seksit Niltub for his full support in my up and down moments during this journey. I am re ally grateful for the friendship and help from members of the Thai Student Association at the University of Florida. My most sincere thanks and gratitude go to my family, Sompong, Pikul, Pataree, and Pakinee, for their endless love, support, understanding , and encouragement.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ ................... 15 1 INTRODUCTION ................................ ................................ ................................ .... 17 Dissertati on Outline ................................ ................................ ................................ 18 Defining Ecosystem Services ................................ ................................ ................. 19 Quantification of Ecosystem Services ................................ ................................ ..... 24 Valuation of Ecosystem Services ................................ ................................ ............ 27 Stressors that Change Ecosystem Services ................................ ........................... 29 2 STUDY AREA ................................ ................................ ................................ ......... 34 Characteristics of the Suwannee River Basin ................................ ......................... 34 Location ................................ ................................ ................................ ............ 34 Topography/Physiography ................................ ................................ ................ 35 Geology ................................ ................................ ................................ ............ 36 Hydrology ................................ ................................ ................................ ......... 37 Soils ................................ ................................ ................................ ................. 39 Land cover/Land use ................................ ................................ ........................ 41 Climate ................................ ................................ ................................ ............. 42 Socio Economy ................................ ................................ ................................ 45 3 SPATIO TEMPORAL INTERACTIONS BETWEEN SOIL CARBON AND NUTRIENT CYCLES IN THE SUWANNEE RIVER BASIN ................................ .... 60 Overview ................................ ................................ ................................ ................. 60 M aterials and Methods ................................ ................................ ............................ 6 2 Historic Soil Carbon Dataset ( DSh ) ................................ ................................ .. 63 Sampling design ................................ ................................ ........................ 63 Laboratory analysis ................................ ................................ .................... 64 Conversion of SOC concentrations into stocks and reconstruction of the fixed depth ................................ ................................ .............................. 65 Current Soil Carbon Dataset ( DSc ) ................................ ................................ .. 66 Sampling design ................................ ................................ ........................ 66 Sample collection procedure ................................ ................................ ...... 67 Laboratory analysis ................................ ................................ .................... 67 Statistical and Geostatistical Analyses ................................ ............................. 67
6 Soil Organic Carbon Change and Sequestration Rates ................................ ... 70 Kriged estimates ................................ ................................ ........................ 70 Collocated sites ................................ ................................ .......................... 71 Surface Water Quality An alysis ................................ ................................ ........ 72 Total Carbon, Total Nitrogen, and Total Phosphorus Loading Trends ............. 73 Relationships between Soil Organic Carbon and Total Organic Carbon Loadings ................................ ................................ ................................ ........ 74 Spatial Variability of Organic Carbon to Total Nitrogen (C:N) and Organic Carbon to Total Phosphorus (C:P) Ratios across the Terrestrial and Aquatic Ecosystems ................................ ................................ ...................... 75 Results and Discussion ................................ ................................ ........................... 75 Historic Site specific Soil Organic Carbon Stocks ................................ ............ 75 Current Site specific Soil Organic Carbon Stocks ................................ ............ 77 Geospatial Patterns of DSh and DSc across the FL SRB ................................ 77 Quantificati on of Soil Organic Carbon Sequestration Rates ............................. 80 Soil organic carbon sequestration rates for kriged estimates ..................... 80 Soil organic carb on sequestration rates for collocated sites ...................... 81 Land cover/land use conditions and SOC change ................................ ..... 82 Long term Trend Analysis of Total Organic Carbon, Total Nitrogen, and Total Phosphorus ................................ ................................ .......................... 84 Relationships between Soil Organic Carbon and Total Organic Carbon in Surface Water ................................ ................................ ............................... 89 Variation in Carbon to Nutrient Ratios in Soils and Surface Water by Drainage Areas ................................ ................................ ............................. 90 Conclusions ................................ ................................ ................................ ............ 93 4 ESTIMATION OF THE TERRESTRIAL CARBON BUDGET ................................ 125 Overview ................................ ................................ ................................ ............... 125 Materials and Methods ................................ ................................ .......................... 130 Above Ground and Below Ground Carbon Data ................................ ............ 130 Environmental and Anthropogenic Covariates ................................ ............... 131 Modeling the Relationships bet ween SOC and STEP AWBH Factors ........... 132 Assessing Terrestrial Carbon Stocks ................................ ............................. 134 Results and Discussion ................................ ................................ ......................... 137 Variable Importance and Spatial Variation in Soil Organic Carbon ................ 137 Estimates of Terrestrial Carbon Budget ................................ ......................... 142 Model Calibration and Validation Results ................................ ....................... 146 Conclusions ................................ ................................ ................................ .......... 148 5 SOCIO ECONOMIC PERSPECTIVES OF ECOSYSTEM SERVICES ................. 164 Overview ................................ ................................ ................................ ............... 164 Valuation Approaches for Non market Goods and Services .......................... 165 Choice Experiments ................................ ................................ ....................... 168 Framework and Choice Experiment Modeling ................................ ................ 170 Random Utility Maximization (RUM) theory ................................ ............. 170
7 Choice Experiments (CE) and Willingness to Pay (WTP) ........................ 173 Materials and Methods ................................ ................................ .......................... 174 Survey Design ................................ ................................ ................................ 174 Choice Experiments Design ................................ ................................ ........... 174 Survey Sampling and Implementation ................................ ............................ 176 Analysis ................................ ................................ ................................ .......... 177 Demography, perceptions, and attitudes ................................ ................. 177 Choice experiment analysis ................................ ................................ ..... 178 Results ................................ ................................ ................................ .................. 179 Descriptive Statistics ................................ ................................ ...................... 179 Demographics ................................ ................................ .......................... 180 Perceptions and attitudes on ecosystem services ................................ ... 180 Socio economic factors influencing perceptions and attitudes ................. 182 Preferences ................................ ................................ ................................ .... 183 Ecosystem service preferences within the improvement group ............... 184 Ecosystem service preferenc es within the program administration group 185 ................................ ................................ ............... 186 Economic Estimates ................................ ................................ ....................... 187 Discussion ................................ ................................ ................................ ............ 188 Socio demographic, Geographic Factors and Ecosystem Services ............... 188 Scale Sensibility ................................ ................................ ............................. 191 Economic Valuation in Choice Experiments (CE) ................................ ........... 194 Conclusions ................................ ................................ ................................ .......... 196 6 SYNTHESIS OF ECOSYSTEM SERVICE VALUES BASED ON BIOPHYSICAL, ECOLOGICAL, AND SOCIO ECONOMIC MEASUREMENTS ... 222 Overview ................................ ................................ ................................ ............... 222 Bayesian Belief Networks (BBN) ................................ ................................ .... 223 Advantages and Shortcomings of the BBN Model in Ecosystem Service Modeling ................................ ................................ ................................ ...... 224 Materials and Methods ................................ ................................ .......................... 226 Bayesian Belief Network (BBN) Construction ................................ ................. 226 Scenario Features ................................ ................................ .......................... 228 Parameterizing the Model ................................ ................................ ............... 230 BNN GIS application ................................ ................................ ................ 230 Survey inco rporation ................................ ................................ ................ 231 Expert involvement ................................ ................................ .................. 231 Scores of ecosystem services ................................ ................................ .. 231 Model Assessment ................................ ................................ ......................... 232 Results and Discussion ................................ ................................ ......................... 233 Conclusions ................................ ................................ ................................ .......... 238 Summary and Synthesis ................................ ................................ ....................... 239 Summary of Findings ................................ ................................ ..................... 240 Synthesis ................................ ................................ ................................ ........ 243
8 APPENDIX SURVEY INSTRUMENT ................................ ................................ ............................. 261 LIST OF REFERENCES ................................ ................................ ............................. 269 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 310
9 LIST OF TABLES Table page 2 1 Generalized geological and hydrogeologic units in the Suwannee River Basin, Florida. ................................ ................................ ................................ ..... 46 2 2 Population and income in the 15 counties of the Suwannee River Basin, Florida. ................................ ................................ ................................ ............... 46 3 1 Datasets used and descriptive statistics of soil organic carbon stocks at observation sites in the top soil (0 20 cm) ................................ .......................... 98 3 2 Descriptive statistics of total nitrogen and total phosphorus stocks at observation sites in the to p soil (0 20 cm) in 2008/09. ................................ ........ 99 3 3 Selected monthly parameters of river flow, total organic carbon, total nitrogen, an d total phosphorus at the eight sub basins between water year 2 000 and 2010. ................................ ................................ ................................ . 100 3 4 Compariso n of soil organic carbon stocks in the top soil using calibration sets derived from o rdinary kriging and block kriging . ................................ ............... 10 2 3 5 Variogram parameters and validation prediction errors for SOC s tocks . . ......... 103 3 6 Soil organic carbon estimates (0 20 cm soil profile) derived from ordinary kriging and block kriging ................................ ................................ ................... 104 3 7 L and cover/land use confusion matrix from 1988/89 to 2006/08 ...................... 105 3 8 Drainage areas used in the hydrology analysis. ................................ ............... 106 3 9 Man n Kendall trend analysis for monthly unit area loads of total organic carbon by drainage area in harmonic mean between 2000 and 2010. ............. 107 3 10 Mann Kendall trend analysis for monthly unit are a loads of total nitrogen by drainage area in harmonic mean between 2000 and 2010. .............................. 108 3 11 Mann Kendall trend analysis for monthly unit area loads of total phosphorus by drainage area in har monic mean between 2000 and 2010. ......................... 109 3 12 Interpolation parameters and validation results derived by block kriging f or soil nutrient properties lnN and lnP in the top soil (0 20 cm). ........................... 110 3 13 Spatial distribution of mean carbon to nitrogen (C:N) and carbon to phosphorus (C:P) in soil and surface water. ................................ ..................... 111 4 1 Asse mbled environmental and human covariates representing STEP AWBH factors. ................................ ................................ ................................ .............. 151
10 4 2 Top forty three most important variables (the first quantile) derived by random forest model using STEP AWBH f actors to predict soil organic carbon stocks ................................ ................................ ................................ ... 154 4 3 Strong relevant variables derived by random forest model using STEP AWBH factors to predict terrestrial carbon stocks ................................ ............ 156 4 4 Descriptive statistics of terrestrial carbon stocks at observation sites. ............. 158 4 5 Descriptive statistics of terrestrial carb on stocks base d on observations and predictions under different scenarios. ................................ ............................... 158 5 1 Attributes and attribute levels in choice experiments. ................................ ....... 200 5 2 Optimal fractional factorial design survey of four 3 level attributes of ecosystem services survey ................................ ................................ ............... 201 5 3 Number of households sampled and number in population in the study area. . 202 5 4 Explanatory variables used in multinomial logit models. ................................ ... 203 5 5 Zip codes and R ural Urban Commuting Area Codes in the study a rea ............ 204 5 6 Variables for socio economic analysis. ................................ ............................. 205 5 7 Demographic statistics of survey respondents ................................ ................. 207 5 8 Descriptive statistics of respondent attitudes about ecosystem services .......... 208 5 9 Multinomial logit analysis of socio economic factors and resp about ecosystem services using the likelihood ratio test. ................................ . 210 5 10 Multinomial logit estimates of respondent familiarity about ecosystem services using Wald statistic tests. ................................ ................................ ... 211 5 11 importance using Wald statistic tests. ................................ ............................... 213 5 12 service using Wald statistic tests. ................................ ................................ ..... 216 5 13 Nested logit model results for choice experiments. ................................ .......... 218 6 1 Overview of variables used in the Bayesian Belief Network model, states, data sources, and justification technique. ................................ ......................... 247 6 2 Utility fu nction structure used for calculating impacts of socio ecological scenarios on the ecosystem service values. ................................ .................... 252
11 6 3 Sensitivity analysis results performed for the carbon sequestration ecosys tem service node. ................................ ................................ .................. 253 6 4 Sensitivity analysis results performed for the climate regulation ecosystem servic e node ................................ ................................ ................................ ..... 253 6 5 Sen sitivity analysis results performed for the nutrient cycling ecosystem services node. ................................ ................................ ................................ .. 253 6 6 Carbon content in the Suwannee River Basin, Florida. ................................ .... 254
12 LIST OF FIGURES Figure page 1 1 Connection between ecosystem structure and function, services, policies, and values . ................................ ................................ ................................ ......... 32 1 2 A framework of the ecosystem domains considered in this researc h study ........ 33 2 1 The Suwannee River Basin and su b basins in Georgia and Florida ................... 47 2 2 Digital Elevation Model and sub basins Suwannee River Basin. ..................... 48 2 3 Counties and major physiographic regions within the Suwannee River Water Management District. ................................ ................................ .......................... 49 2 4 Geological units by series in the Suwannee River Basin, Florida. ...................... 50 2 5 Spatial distribution of soil orders in th e Suwannee R iver Basin, Florida ............. 51 2 6 Map of soil suborder for the extent of the Suwannee River Basin, Florida ......... 52 2 7 Soil drainage class distribution in the Suwannee River Basin, Florida.. ............. 53 2 8 Soil leaching potential in th e Suwannee River Basin, Florida ............................. 54 2 9 Major land resource area in the Suwannee River Basin, Florida. ....................... 55 2 10 Map of land cover/land use for the extent of the Suwannee River Basin, Florida in 1989 and 1995 ................................ ................................ .................... 56 2 11 Map of land cover/land use for the extent of the Suwann ee River Basin, Florida in 2004 and 2008 ................................ ................................ .................... 57 2 12 Land cover/land use trends in 1989, 1995, 2004, and 2008 in the Suwannee River Basin, Florida. ................................ ................................ ........................... 58 2 13 Maps of 30 yr normal monthly average of precipitation, maximum temperature, and minimum temperature between 1981 and 2010 ..................... 59 3 1 Spatial distribution of historic soil organic carbon observations and soil organic carbon stocks in the top soil (0 20 cm).. ................................ .............. 112 3 2 Spatial distribution of current soil organic carbon observations and soil organic carbon stocks in the top soil (0 20 cm) ................................ ................ 113 3 3 Drainage area delineation from a digital elevatio n m odel and topographic attributes ................................ ................................ ................................ ........... 114
13 3 4 Kriged estimated of historic and current soil organic carbon stocks in the 20 cm depth derived by ordinary kriging and block kriging ................................ .... 115 3 5 Variance maps derived from historic and current soil organic carbon estimates classified by quantiles. ................................ ................................ ..... 116 3 6 Predictio n maps show s oil organic carbon gains and losses derived by subtracting historic prediction maps from current prediction maps ................... 117 3 7 Soil organic carbon gains and losses derived from historic and current collocated sites within 200 m radius.. ................................ ............................... 118 3 8 Spatial distribution patterns of total nitrogen and total phosphorus stocks in soil. ................................ ................................ ................................ ................... 119 3 9 Sp atial distribution patterns of soil organic carbon to total nitrogen (C:N) and soil organic carbon to total phosphorus (C:P) in the topsoil .............................. 120 3 10 Soil organic carbon stock variability in nitrogen limited and enriched areas and phosphorus limited and enriched areas. ................................ .................... 121 3 11 Fertilizer use by county between 1945 and 2010. ................................ ............ 122 3 12 Time series plots of monthly mean flows, concentration and loads for total organic carbon , total nitrogen , and total phosphorus between 2000 and 2010 wateryears ................................ ................................ ................................ ........ 123 3 13 Monthly mean precipitation and mean temperature with smoothing curves between 2000 and 2010. ................................ ................................ .................. 124 4 1 A work flow for soil organic carbon stocks prediction and the attainable terr estrial carbon prediction from the environmental human predictor s ............ 159 4 2 Spatial distributio n of terrestrial carbon stocks based on current soil organic carbon stocks (0 20 cm) and above gro und biomass stocks. ........................... 160 4 3 Terrestrial carbon stock estimates under different models derived by ordinary kriging ................................ ................................ ................................ ............... 161 4 4 Inde pendent validation of predicted soil organic carbon stocks using random forest model ................................ ................................ ................................ ...... 162 4 5 Independent validation of predicted actual terrestrial carbon stocks using random forest model ................................ ................................ ......................... 163 5 1 The Suwannee River and location of respo nses returned within study area .... 219 5 2 Degree of belief in global climate chang e caused by human activities ............. 220
14 5 3 Opinion of respondents on local government policies based on their beliefs in global climate change. ................................ ................................ ...................... 220 6 1 Conceptual model illustrating the key variables used to predict the overall ecosystem service scores ................................ ................................ ................ 255 6 2 Results of the Bayesian Belief Network when all scenarios are equally likely. . 256 6 3 Results of the Bayes . ....... 257 6 4 Results of the Bayesi an . ................................ ................................ ................................ ........ 258 6 5 Results of the Bayesian Belief Network for scena .... 259 6 6 Results of the Bayesian ....... 260
15 Abstract of Dissertation Presented to the Graduate School of The University of Florida In Partial Fulfillm ent of the Requirements for the Degree Of Doctor of Philosophy ASSESSMENT OF CLIMATE REGULATION, CARBON SEQUESTRATION, AND NUTRIENT CYCLING ECOSYSTEM SERVICES IMPACTED BY MULTIPLE STRESSORS By Pasicha Chaikaew August 2014 Chair: Sabine Grunwald Major: S oil and Water Science The importance of maintaining and enhancing ecosystem services is a vital basis for delivering benefits to human well being. There are still several research gaps in linking biophysical and socio economic characteristics, quantifying natural assets, and valuing services. The objective of this dissertation was to assess climate regulation, carbon sequestration, and nutrient cycling ecosystem services from the biophysical, ecological, and socio economic perspectives. Findings suggest th at top soils have acted as a carbon sink over the past decades in the Suwannee River Basin, Florida. The results coincided with an increase in total organic carbon (TOC) loads in surface waters of the Suwannee River, that were less pronounced than increase s in total nitrogen (TN) and total phosphorus (TP) loads of which only a few of the drainage areas showed impairment by TN and TP. The net mineralization of TN and TP in soils and surface waters pinpointed to potential risks for nutrient enrichment in aqua tic and terrestrial ecosystems. Biotic, soil, parent material, topographic, and water related factors played crucial roles in predicting soil organic carbon (SOC) storage, whereas climatic factors were of much less importance. The model derived from simula ted annealing and
16 random forest estimating actual and attainable terrestrial carbon suggested the limitation of carbon enhancement in some areas (e.g., wetlands), while others showed potential to sequester more carbon (e.g., under row/field crops). The bel iefs and perspectives of local residents identified nutrient cycling as the most important service, and climate regulation and carbon sequestration as the least important services, which somewhat contradicted the scientific based knowledge from the empiric al assessments. The willingness of the residents to pay for ecosystem services was extremely low (<$2/household/year). The socio ecological outcomes from this study and secondary data from the literature and expert knowledge were then integrated in the Bay esian Belief Network (BBN) model under four distinct scenarios. Besides the natural processes and services, awareness, and adaptation through management were identified as key factors in manipulating these benefits. This dissertation took a big step forwar d in developing an ecosystem service concept from theory to a novel implementation that engaged pedogenic, hydrologic, biotic, atmospheric, and anthropogenic domains together.
17 CHAPTER 1 INTRODUCTION This dissertation is focused on understanding, analyzin g, and synthesizing the interaction among selected ecosystem services and the benefits people derive from ecosystems. Since ecosystem services involve multiple domains, assorted challenges are inevitable. Ecosystem services go beyond the sole assessment of the physical characteristics of ecosystems and include the human component. This study engaged four interconnected ecosystem domains: pedogenic, hydrologic, atmospheric, and anthropogenic. These domains are interconnected and cannot be separated when asse ssing multiple ecosystem services, adding complexity to the study, and thus leading to the need for an interdisciplinary approach. This research focused on multiple ecosystem services: climate regulation, carbon sequestration, and nutrient cycling. The con cept of ecosystem services is a promising path for fostering sustainability. Ignoring or devaluing the importance of the services and being unaware of how to use the services wisely could result in adverse impacts to human well being in the future. In cont rast, improved management of ecosystems could promote better well being of societies in the long run. There are still substantial research gaps in the assessment of the services ecosystems provide that will be addressed in this dissertation research. For example, many ecosystem service studies often look at either the biophysical assessment of ecosystem properties and/or processes or the socio economic valuation, but not both concomitantly. A better connection between these domains is required. Effective f rameworks for bundling ecosystem services to derive their combined effects are still in their infancy. To optimize one service may degrade another service, and vice versa. In
18 addition, interactions among multiple services are still poorly understood and ar e confounded by geographic constraints. These identified gaps are the motivation for this research study. The following specific objectives were pursued in a large basin in the southeastern United States: 1. To conduct a spatially and temporally explicit asse ssment of carbon and nutrients from a biophysical perspective 2. states 3. To assess human perception and valuation of selected ecosystem services and 4. services Dissertation Outline Chapter 1 provides the overarching objectives and summarizes the re levant literature regarding ecosystem services. It includes general concepts, definitions, and approaches employed in the valuation of ecosystem services. Chapter 2 describes the important biophysical and socio economic characteristics of the study area, the Suwannee River Basin. Chapter 3 focuses on the biophysical assessment of selected ecosystem services, namely climate regulation, carbon sequestration, and nutrient cycling ecosystem services. This chapter investigates spatio temporal interactions betw een carbon, nitrogen, and phosphorus stocks in soils and loads in surface water. Chapter 4 elucidates on the major terrestrial carbon pools along the spectrum of actual and attainable carbon, the latter ranging from the minimum to maximum carbon that can be stored in the terrestrial ecosystem. The attainable carbon reflects the
19 realizations that can be achieved through a variety of envisioned human management activities providing a pluralistic perspective on the carbon regulation ecosystem service. The rat io of attainable carbon relative to actual carbon provides critical information on the type of management intervention that allows the optimization of the carbon sequestration ecosystem service. Chapter 5 investigates the perception, preferences, and valu ation of a variety of ecosystem services provided by residents in the study area. The chapter shows socio economic factors that influence different degrees of perception. An in depth analysis of how people value and rank different services is provided. Ch apter 6 presents a synthesis model that links correlative and probabilistic relationships among environmental and socio economic variables. Various scenarios, ecosystem se rvices. The networks and findings from Chapter 3,4, and 5 are harmonized and synthesized. Defining Ecosystem Services Various definitions of ecosystem services have been provided to clarify what they are and how they are classified based on scale of analy sis, system dynamics, and the (de Groot et al., 2002; National Research Council (NRC), 2004) . Because ecosystem services often are confused with ecosystem function and/or structure, the fun damental differences are discussed. According to the NRC (2004) , ecosystem structure signifies the composition and the physical and biological organization of an ecosystem. The structure defines how those parts are organized. Ecosystem function exposes a process that occurs in an ecosystem as a result of the interactions of plants, animals, and other organisms in the ecosystem with each other or
20 within their environment. Ecosystem function is conceived as a process that control subset the ecological structure; it naturally influences the ecosystem itself to provide services and goods for living and non living things that are critical to human survival (Kremen, 2005) . Although ecosystem services have been identified in economics since the 1980s (Fisher and Turner, 2008; Fisher et al., 2009) , there is still no consistent definition. There are three distinct defin itions of ecosystem services: 1. The conditions and process through which natural ecosystems, and the species within the ecosystem, sustain and fulfill human life (Daily, 1997) 2. The benefits human populations derive, directly or indirectly, from ecosystem functions (Costanza et al., 1997) 3. The benefits people obtain from ecosystems (Millennium Ecosystem Assessment (MA), 2005a) An additional three definitions have been proposed recently: 1. Services that are derived from natural elements of ecosystems (Wallace, 2007) 2. Components of nature directly enjoyed, consumed, or used to yield human well being (Boyd and Banzhaf, 2007) 3. The aspects of ecosystems utilized (actively or passively) to produce human well being (Fisher and Turner, 2008; Fisher et al., 2009) These definitions disagree on some points: (i) functions/processes are (not) services, (ii) services are (not) benefits, and (iii) services are direct (indirect) uses. Boyd and Banzhaf (2007) referred to services as ecological component s (i.e., things that are countable); therefore, functions or processes are not services. The reason that functions and processes are not services is that they are not end products; functions and processes are intermediate to the production of final servic es. This viewpoint is in direct opposition to that of other researchers (Costanza et al., 1997; Daily et al., 2000;
21 Fisher et al., 2009; MA, 2005a, 2005b; Wallace, 2007) who defined functions or processes in the ecosystems as services associated with human welfare beneficiaries. Costanza et al. (1997) , MA (2005a, 2005b) , and Wallace (2007) determined services are benefits since multi functional ecosystems that perform a variety of ecosystem service s are interconnected. Boyd and Banzhaf (2007) and Fisher et al. (2009) were concerned about the mix bet ween ends and means that often leads to confusion and double value of the services; thus, they viewed services as separate from benefits. Wallace (2007) and Boyd and Banzhaf (2007) argued that only the direct endpoints are ecosystem services, while Daily (1997) , Costanza et al. (1997) , MA (2005a, 2005b) , and Fisher et al. (2009) argued that ecosystem services can be utilized directly or indirectly as long as they affect net human benefits. Because ecosystems are complex and diverse, it is important to clarify the meaning of ecosystem services and to identify the appropriate classification schemes for implementation. There is broad agreement on the general ide a of ecosystem services, but no one classification scheme is adequate for the many contexts in which ecosystem service studies may be implemented (Fisher et al., 2009) . In this study, while we adopted the ideas outlined in the MA (2005b) framework for the biophysical assessment and ecosystem characteristics , we used the guidelines by Fisher et al. (2009) for the socio economic valuation. From an ethical ecosystem perspective, human domination of the biosphere has led to rapid alterations in the composition, structure, and function of ecosystems. On the other hand, modern ecology and other contemporary ecological movements in the late 1990s view humans as an integral part of ecosystems/biospheres where only a
22 synergistic functioning is achieved if humans and other organisms live in harmony with natur e (Christensen et al., 1996; Grumbine, 1994; Harwell et al., 1996; Mangel et al., 1996) . Destroying the biosphere (e.g., by maximizing human benefits) wou ld be detrimental from a contemporary ecological perspective because it destroys the physiological settings (e.g., soils and water) and biosphere on which human survival depends. Kidd (1992) outlined six ethical models that have emerged in response to ecological phenomena as the interrelationships among rates of population growth, resource use, and pressure on the environment. The models are (i) ecological carrying capacity, (ii) resource/environment, (iii) biosphere stewardship, (iv) critique of technology, (v) no growth/slow gr owth, and (vi) eco development. Contemporary ecological worldviews stress a holistic ethics that is rooted in sustainability concepts; spiritual connection to nature (deep ecology); and radical views, such as ecofeminism, environmental justice, and value e thics (Schmidtz and Willott, 2012) , as well as integral ecology (EsbjÃ¶rn Hargens and Zimmerman , 2009) . Common to all these contemporary environmental ethical worldviews is the idea that Earth and human dimensions are inseparable. This underpins the needs for ecosystem service assessments. Thus, it is decision making of the ecosystem management framework (Endter Wada et al., 1998; Grumbine, 1994) . The di agram in Figure 1 1 illustrates how human actions affect the structure, functions, goods, and services of ecosystems. These impacts do not occur only from the intentional, direct use of the ecosystem, but also from the unintentional, indirect impacts of o ther activities (NRC, 2004; Pascual et al., 2010) .
23 According to the MA (2005a, 2005b) , four categories of ecosystem services are distinguished: (i) provisioning services (such as food, water, timber, and fiber), (ii) regulating services (affecting climate, floods, diseases, wastes, and water quality), (iii) cultural services (providing recreational, aesthetic, and spiritual benefits), and (iv) supporting services (such as soil formation, photosynthesis, an d nutrient cycling). This study focuses on the assessment of climate regulation, carbon sequestration, and nutrient cycling, which are related to various environmental components and functions. To assess ecosystem services, ecosystem components and valu es cannot be separated. If the ecosystem is solely taken and simulated without taking human well being into account, the valuation of ecosystem services is not fully accomplished. In such a case, it would be considered an ecosystem or biophysical assessmen t. Besides quantifying the carbon and nutrient amount of services, it is also necessary to understand how humans perceive and value ecosystem services as benefits. Thus, underlying this research is the integration of quantification, valuation, and simulati on of selected ecosystem services. Figure 1 2 illustrates the ecosystem services considered, the relevant ecological processes, indicators used, and benefits. Climate regulation and carbon sequestration influence ecosystems through the terrestrial system that interacts with the climatic system. In this research, we identify atmospheric GHG emission s , organic carbon stocks, and sequestration rate in soil and terrestrial environment as indicators that influence climate regulation and soil quality. Nutrient s torage in soils and nutrient loading in surface water (total nitrogen (TN) and total phosphorus (TP) ) are indicators for nutrient cycling, which identify clean water provision benefits. The ratio of carbon to
24 nutrients indicates the balance of nutrient cyc ling across different types of ecosystems. The anthropogenic domain investigates the perception and beliefs of people and their valuation based on economic metrics pertaining to all three. Quantification of Ecosystem Services apital assets and they yield on going life sustaining services if properly managed. These beneficial services include the production of goods (such as seafood and timber), life support processes (such as pollination and water purification), and life fulfi lling conditions (such as beauty and serenity). However, ecosystems are rarely understood, compared to the other types of capital, and often are appreciated only upon their loss (Daily et al., 2000) . Due to limited understanding of the consequences of ecosystem change for human well being and the scientific knowledge for taking action to enhance th e sustainability, the MA was initiated in 2001 to provide information on the global scale. Although inspiring, it created concerns on how to apply the MA findings to national, regional, and local scales where most management decisions are made (Tallis and Polasky, 2009; Water Resources Institute (WRI), 2008) . Aided by U.S. government policy, the demand for more integrative and comprehensive analyses of ecological and socioeconomic aspects has grown tremendously (Council on Technology (PCAST), 2011) . As of 2014 , there is no single method that is hailed adequate for ecosystem services quantification. The tools that have been developed by numerous groups of researchers in both the public and private sectors vary greatly (Bagstad et al., 2013) . Some tools are generalized and can be applied to other areas, while others are specifically created for fine scale applications. They also differ in terms of spati al, temporal, and economic valuation approaches (Bagstad et al., 2013) .
25 According to Layke (2009) , ecosystem se rvice output can be defined as ecosystem service capacity measured in stocks and flows. The stock delivers a service, while the flow delivers a benefit to humans. It is often difficult to model or measure many of the ecosystem services (for instance, soil carbon sequestration and nutrient cycling services), thus proxy indicators are used for assessing changes in these services (Plummer, 2009) . Pol icy decision makers and practitioners, therefore, should be cautious in the use of these indicators (Vira and Adams, 2009) . Ecosystem services can be viewed and quantified at different spatial and temporal scales with respect to important area parameter s and data availability. For example, a European assessment of the provision of ecosystem services (Maes et al., 2011) dealt with different spatial resolutions and scale s of services related to soils (at resolutions < 1 km), atmosphere (at resolutions > 1 km), water (calculated on the basin and sub basin levels), and cultural context (a provincial level). Compromising between different spatial units was therefore a critic al step in order to convey meaningful information of ecosystem services and economic valuation. The spatial studies in ecosystem services are increasing because it is easier to acquire spatial data than temporal data and the outcome allows scientists to co nnect ecosystem services to human well being demand across space (Carpenter et al., 2009; Jaarsveld et al., 2005) . Temporal studies ar e more limited because they require historical data that are available through models, records, or local experts. A comparison between studies is difficult because numerous approaches can be taken to quantify ecosystem services. Remote sensing based techn iques were recommended for quantification ecosystem services (Dale and Polasky, 2007; Seidl and
26 Moraes, 2000; Verstraete et al., 1996) with comparatively low cost and frequent mon itoring. Between 1990 and 2012, the remote sensing approaches were used to quantify food and timber production, carbon storage, air quality regulation, erosion prevention, water purification, storm and flood regulation, mass flow protection, maintenance o f soil, and pest control (Ayanu et al., 2012) . In the same fashion, Raudsepp Hearne et al. (2010) utilized remote sensing data and heavily relied on the use of a geographic information system (GIS) with numerous statistical tests to bundle services. Along those lines, there have been attempts to develop built in ecosystem service tools. Among those models, the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) (Kareiva et al., 2011; Tallis et al., 2013) has been more extensively accepted in the peer reviewed process and may be more suitable when the ecological context in that area is well understood (Bagstad et al., 2013) . The Artificial Intelligence for Ecosystem Services (ARIES) (Bagstad et al., 2011; Villa et al., 2011) is an open source modeling framework to map ecos ystem services that is available in many western U.S. states. The ARIES developers are currently working to extend this tool on a global scale. Other quantification tools, such as Ecosystem Services Review (ESR) (WRI, 2012) , LUCI (Jackson et al., 20 13) , EcoServ (Feng et al., 2011) , Social Values for Ecosystem Services (SolVES) (Sherrouse et al., 2011) , and Ecosystem Portfolio Model (EPM) (Labiosa et al., 2013) , have been used to quantify various services. Each model has advantages and disadvantages, depending upon the usage. A detailed description of these ecosystem service tools can be found in Bagstad et al. (2013) . Despite having different quantifiable approaches across ecosystem service disciplines, the MA study (2005c) pointed out that the improvement of ecosystem
27 services manageme nt may be more imperative for researchers to consider within a social ecological context rather than across studies. Studies of mapping and quantifying trade offs among ecosystem services have been conducted using different approaches and at different sca les (Egoh et al., 2008; Nelson et al., 2009; Raudsepp Hearne et al., 2010; Swallow et al., 2009) , but there are still relatively few studies that analytically quantify multiple services across space and time based on both empirical and ancillary data. Valuation of Ecosystem Services m ay perceive and value climate, carbon, and nutrients in ecosystems differently. Thus, Humans may value ecosystem services based on the scale of perception. More than just a way of passively representing the intrinsic physical situation of ecosystems, this perception can be inherently active and exploratory (Braund, 2008) . For example, nutrient enrichment and drinking water degradation close to home may be more highly valued than a fertilizer spill in another part of the wo rld far away. Furthermore, ecosystem services provided at local, regional, and global scales may be perceived differently because of their immediate direct effect (Boissiere et al., 2 013; Howe et al., 2013) . For example, drinking water quality as an immediate physical environmental risk is easier to comprehend than global climate change or carbon cycling which occurs over a longer period of time with questionable consequences ( Carlton and Jacobson, 2013) . Regional, federal, or international regulated environmental services (e.g., nutrient and contamination standards by the federal Environmental Protection Agency) may be
28 perceived as more important by people compared to non regu lated services associated with higher uncertainties (e.g., climate change) for the simple fact that they are government regulated. In such case the scale of governance matters for ecosystem service management and valuation (Costanza and Liu, 2014) . Moreover, valuation of humans may differ by gender, education level (Peixer et al., 2011) or socio cultural background (Andersen et al., 2012) . Despite a controversy on whether non use values are appropriately commensurable in monetary measurement (Carson et al., 2001; Martinez Alier et al., 1998) , as pointed out by Costanza et al. (1997) , ecosystem ser vices should be valued, as valuation is useful to convey important information in terms of both the benefits that humans perceive from the ecosystem and the impacts that occur from changes in ecosystem services. Since the time that the MA advanced the conc ept of ecosystem services, economic valuation has become a core component of the ecosystem services science to assess the values of services and therefore is important (Hrabanski et al., 2013) . In general, there are two types of values under the concept of total economic value (TEV). On (Freeman, 1993; Randall, 199 1) . Use values are used by humans for consumption or production purposes (MA, 2003) . Tangible and intangible services that are being used or have the potential to provide future use are grouped into this category. On the other hand, when humans do not make use of ecosystem services, but know that the resources exist and wish to see them preserved, these values are called non use values, also known as existence values, passive values, or conservation values (Bishop and Welsh, 1992;
29 Smith, 1987) . Climate regulation, carbon sequestration, and nutrient cycling are considered both use and non use values (NRC, 2004; Pascual et al., 2010) . The methods of nonmarket valuation vary depending on types of services; nonetheless, stated preference approaches have been used to widely estimate non use values of provisioning, regulating, and supporting services (Dutton et al., 2010; Takatsuka et al., 2009; Uchida et al., 2007) . Besides having the ability to meas ure non use values, the stated preference approaches are also flexible. Since the stated preferences use hypothetical data to estimate ex ante (before the event) willingness to pay, these approaches allow ecologists, economists, and policy makers to const ruct realistic policy scenarios for most new policies (Whitehead et al., 2005) . However, the major weakness is their hypothetical nature in which respondents may be unfamil iar with situations (Whitehead et al., 2005) . Details of stated preference techniques and the method used in this study are discussed in Chapter 5. Stressors that Change Ec osystem Services According to the MA (2005c) , spatial scales are important to classify the drivers predominantly operating at five different scales (local, sub national, national, regional, and global). Examples are provided for drivers at the local and global scales. Local scale drivers most frequen tly mentioned are direct drivers, such as land use and land cover change, which can be classified as endogenous and appear to be related to management activities within local domain boundaries (e.g., watersheds). The local scale driver (which is more dire ct than indirect) can be controlled by changes in decision (MA, 2003) .
30 Global scale drivers are considered exogenous given that they were largely beyond the control of individual decision makers for a particular region (MA, 2005c) . Fo r example, global climate change and sea level rise are global scale drivers that may alter ecosystem services more than do local operations. Despite the scale dissimilarity, global and local scales are coupled in social ecological systems. In Carpenter (2009) , global biophysical and social stressors introduce impacts on regional systems through faster variables (e.g., animal behavior, soluble nutrients, floods, community i ncome, migration, and access to resources) and slower variables (e.g., soils, sediments, disturbance regime, property and use rights, wealth and infrastructure, and cultural ties to land). They further explained that it is unusual to find a linear relatio nship from changes in stressors biodiversity ecosystem functions/processes ecosystem services human well being social responses feedbacks to stressors (Carpenter et al., 2009) .When interactions among components of a system are not directly proportional, the system is reaches a threshold, and then shifts into a new state rapidly (MA, 2005a) . The Intergovernmental Panel on Climate Change (IPCC) (2013) attributed extreme floods in the past five centuries in some regions (i.e., northern and central Europe, w estern Mediterranean, and eastern Asia) and the occurrence of extreme droughts (greater magnitude and longer duration) in many regions since 1900s to climate change. Disturbance is one form of nonlinear ecosystem dynamics that is commonly anticipated, acc ommodated in planning, and even employed in assessment of ecosystem structure and management of ecosystem services (Burkett et al., 2005) .
31 Climate related disturbances may have already altered ecosystems and resulted in several species extinctions (Alan Pounds et al., 2006; Malcolm et al., 2006) . Regardless of environmental drivers, humans are part of the climate change equation and are anticipated to reach 9.6 billion in population by 2050 (United Nations (UN), 2013) . Challenges to connect climate, people, and ecosystems into t he ecological management strategy are inevitable. Although understanding the stressors of changes in ecosystem services is a major task for analyzing gains and losses from ecosystem change, we often lack the basic knowledge to understand the social context relevant to the stressors that change ecosystem services. Thus, there is a need to understand the factors and how drivers might cause nonlinear changes in ecosyst em service regulation.
32 Figure 1 1 . Connection betw een ecosystem structure and function, services, policies, and values [A dapted from the National Research Council, 2004 . Valuing Ecosystem Services: Toward Better Environmental Decision Making. The National Academies Press, Washington, DC]. ECOSYSTEM Functions Goods & Services VALUES Use v alues Nonuse v alues e.g., existence, species preservation, biodiversity, cultural heritage Consumption use e.g., harvesting , water supply (irrigation, drinking), genetic and medicinal resource Structure Nonc onsumption use Direct e.g., recreation, transportation, aesthetics, birdwatching Indirect e.g., UVB protection, habitat, support, flood control, pollution control, erosion prevention Human actions (private/public))
33 Figure 1 2 . A framework of the ecosystem domains considered in this research study, including indicators, services, and benefits.
34 CHAPTER 2 STUDY AREA Characteristics of the Suwannee River Basin The Suwannee River, a major river in the state of Florida, was desig nated an Outstanding Florida Water (OFW) by the Florida Legislature in 1979. The OFW signifies unique characteristics of both ecological and cultural values that the river provides to the state of Florida. This section describes the important characteristi cs of geography, geology, hydrology, biology, and climate of the Suwannee River Basin. The physical features influence the ways people live and benefit from the basin. Location The Suwannee River Basin extends over the Coastal Plain of Georgia, including the Okefenokee Swamp, and the lower drainage basin in north central Florida before draining into the Gulf of Mexico. The basin consists of the Alapaha, Withlacoochee, Aucilla, Coastal (Econfina Steinhatchee), Santa Fe, Upper Suwannee, Lower Suwannee, and Waccasassa River sub basins (Figure 2 1). The entire basin covers an area of approximately 32,914 km 2 . Part of the basin is regulated by the Suwannee River Water Management District (SRWMD), which was established to administer flood protection programs and develop water management plans in the north central Florida region. As part of its administration, the SRWMD encompasses all or parts of fifteen counties (Alachua, Baker, Bradford, Columbia, Dixie, Gilchrist, Hamilton, Jefferson, Levy, Lafayette, Madison, Putnam, Suwannee, Taylor, and Union), an area measuring approximately 19,665 km 2 in size (Figure 2 3). Throughout this study, the entire es.
35 Topography/Physiography Topographic relief and elevation in the entire SRB decrease with proximity to the Gulf of Mexico. The highest elevation is at the most upstream point of the Alapaha sub basin in Georgia, where elevations range up to 146 meters (m) above mean sea level (U.S. Geological Survey ( USGS), 2009) . The lowest elevation and least relief is along the Gulf of Mexico coastal line, where elevations are equal to the mean sea level (USGS, 2009) (Figure 2 2). The FL SRB lies in the Coastal Plain that consists of two major physiographic regions: the Northern Highlands and Gulf Coastal Lowlands (Florida Department of Environmental Protection (FDEP), 2001; Grubbs, 1998; Puri and Vernon, 1964; White, 1970) . The Northern Highlands region is generally characterized by altitude and thick, clayey strata. Overlying the Floridan aquifer system, the Northern Highlands includes gently rolling topography, with elevations ranging from 100 feet (30.5 m) to 230 feet (70.1 m) above mean sea level , msl (Hornsby and Ceryak, 1999; White, 1970) . S eparating the Northern Highlands and the Gulf Coastal Lowlands regions is an escarpment known as the Cody Scarp that constitutes the most prominent topographic break in the state of Florida (F DEP, 2001; Puri and Vernon, 1964) . T he Gulf Coastal Lowlands region has a low topographic relief with elevations ranging from sea level to about 100 feet (30.5 m) msl. The lowlands consist of carbonate rocks which make up the karst, sinkhole, and limestone natural springs topography in the a rea (Hornsby and Ceryak, 1999; White, 1970) (Figure 2 3). The Cody Scarp is a very significant physiographical feature as it relates to the hydrological system in the FL SRB. Except for the Suwannee River, all streams and rivers that originate in th e Northern Highlands vanish underground when they cross the
36 escarpment. The remaining streams in the Gulf Coastal Lowlands move downward into the upper Floridan aquifer and interconnect the unconfined areas in the upper Floridan aquifer (Hornsby and Ce ryak, 1999; White, 1970) . Geology The strata of carbonate rock platform or limestone as thick as 5,000 feet (1,524 m) in the subsurface of the basin are primarily Tertiary in age (Miller, 1986) . The Floridan geological structure found within these strata are similar to the strata found in Georgia. The geologic unit consists of the Pliocene, Miocene, Oligocene, Eocene, and Pleistocene/Holocene (Scott et al., 2001) . The Pliocene and Pleistocene/Holocene ( Quaternary aged) materials in most of the FL SRB area, are surficial sand deposits, which include undifferentiated formations. These deposits originated from marine terrace deposition and erosion, and chemical weathering. The sediments included in these d eposits are most often quartz sands, shell, and clay beds (Farrell et al., 2005; Fernald and Purdum, 1998) . The Miocene Series, in the northern and northeastern areas of the FL SRB, mainly consists of the Hawthorn group formation. As a result of unusual depos itional conditions during the M iocene epoch, the Hawthorn formation has an abundance of phosphate, palygorskite, opaline cherts, and other uncommon minerals, as well as an abundance of dolomite, which generally includes interbedded clay, sand, and carbonate strata (Scott, 1988) . These minerals affect groundwater quality during the weathering process in the uppermost port ion of the Floridan aquifer system or portions of the intermediate aquifer system (Fernald and Purdum, 1998) . The Oligocene Series with carbonate sediments is present in the western portion of the FL SRB and northern portion of the Suwannee River. These sediments form a l arge portion of the Floridan aquifer system. Suwannee
37 Limestone, which is the predominant formation, consists primarily of muddy carbonate limestone (Fernald and Purdum, 1998) . While Miocene and Plio Pleistocene strata are mainly composed of siliciclastic materials, the Eoc ene Series is predominantly composed of limestone and/or dolostone. The carbonate sediments in the Eocene Series form portions of the Floridan aquifer system. The predominant formations include Ocala Limestone, Avon Park Limestone, and Oldsmar Limestone. Figure 2 4 displays the geological series and Table 2 1 illustrates the generalized geological units and hydrogeologic units in North Florida. Hydrology The Suwannee River is the second largest river in the state of Florida in terms of water flow. The annu al average discharge is approximately 10,000 cubic feet per second (283.17 cubic meters per second) (Light et al., 2002) . The extent of the Hydrologic Unit Areas (identified by eight digit numbers) that make up the entire SRB is as follows: Waccasassa (03110101), Coastal River (03110102), Aucilla (03110103), Upper Suwannee (03110201), Alapaha (03110202), Withlacoochee (03110203), Suwannee River (03110205), and Santa Fe River (03110206). The major streams east of the Suwannee River include the Alapaha, Santa Fe, and Withlacooche e Rivers, and the major streams west of the Suwannee River include the Aucilla, Econfina, Fenholloway, and Steinhatchee Rivers. All drain into the Gulf of Mexico. The hydrogeological system in and nearby the FL SRB areas is divided into three principal un its: the surficial, intermediate, and Floridan aquifer systems. The surficial aquifer system is unconfined and is located throughout the Northern Highlands region and some parts of the Gulf Coastal Lowlands. The average water table depth varies from 15 fee t (4.5 m) along the Suwannee River and the coastal zone to greater than
38 100 feet (30.5 m) in the high topographic elevation of the areas (Denizman and Randazzo, 2000) . The intermediate aquifer structure is confined and lies beneath the surficial aquifer system in t he Northern Highlands (Planert, 2007) . In some areas, the intermediate aquifer obstructs water exchanges between the surficial and Upper Floridan aquifers. The Floridan aq uifer system, which is formed by the carbonate rocks from the Paleocene to early Miocene age, is subdivided into the Upper and Lower Floridan aquifer systems (Miller, 1986) . The Lower Floridan aquifer is present in the northern part of the FL SRB from Jefferson County east to Columbia County and sout h through Levy County (Miller, 1986) . The Upper Floridan aquifer is present in the northeastern FL SRB, mostly in the Gulf Coastal Lowlands at or near the land surface. The Upper Floridan aquifer is extremely permeable for transmitting a large amount of water volume. The high porosity of limestone allo ws the development of numerous karst features to take place (Planert, 2007) . The artesian spring flow from the Upper Floridan aquifer supplies the base flow to the Gulf Coa stal Lowland streams (Pittman et al., 1997) . The hydrology system in this region is predominantly influenced by the ground water system. The sub surface drainage sustains the flow of rivers, streams, and springs (Planert, 2007) . The water recharge of the Floridan aquifer system comes from precipitation and discharge comes from seepage from streams and wetlands (Hornsby and Ceryak, 1999) . The surface water in the Northern Highlands region reflects the water table level of the surficial aquifer system, while surface water in the Gulf Coastal Lowlands typically represents the water table within the Up per Floridan aquifer (Schneider et al., 2008) .
39 The FDEP reported in the 2004 State Water Quality Assessment 305(b) that the he lower river Maximum Daily Load (TMDL) establishment. Some portions of the upper Suwannee and Santa Fe River sub o low dissolved oxygen, high nutrients, and/or high levels of fecal coliform (FDEP, 2004a) . To comply with the legislation enacted in 1993 that was mandated to protect the functions of ecosystems in the state of Florida, the Suwannee River Partnership (SRP) was established in 1999 through a Memorandum of Understanding (Lubell, 2004) . The partnership was formed as a coalition of federal, state, and regional agencies; local governments; and privat e associations working together to limit increases in nitrate levels and to restore the water quality comparable at the time the river were designated as Outstanding Water Bodies in 1979 (SRP, 2002) . Soils While most s oils in Florida are hyperthermic, northern Florida soils typically have a thermic temperature regime. The soil temperature regimes in the north central Florida FL portion) soi l is hyperthemic (Natural Resource s Conservation Service (NRCS), 2006a) . Generally, soils in the FL SRB are loamy and sandy. They are poorly drained to very poorly drained in most of the area, except soils in the Lower Suwannee River Basin and some adjacent portions of the Santa Fe River Basin which are well drained to excessively drained (NRCS, 2006a) . While Spodosols are widely distributed in a statewide areal coverage, followed by Entisols and Ultisols (Grunwald, 2008; Stone et al., 1993) , the pred ominant soil orders
40 found in the FL SRB are Ultisols (33%), Spodosols (23%), Entisols (18%) (Figure 2 5) (NRCS, 2006 a ) . Soil suborders o f Ultisols in the study area are made up by Udults (humid) and Aquults (wet); Spodosols are made up by Aquods (wet) and Orthods (typical); Entisols are made up by Psamments (sandy) and Aquents (wet). Other suborders include Aqualfs, Saprists, Udalfs, Aquep ts, Aquolls, and other (Brady and Weil, 2008; NRCS, 2006a) (Figure 2 6). Alaquods formed in sandy marine sediments on flats and in depressions, whereas Paleudults formed in marine sediments on uplands. Quartzipsamments formed in mixed sandy eolian and marine sediments on uplands. Pale aquults formed in marine and fluvial sediments on terraces ( NRCS, 2006b) . In general, Entisols are among the most agriculturally productive soils in the world, while Spodosols and Ultisols are not naturally fertile. The latter two soils require fertilization to be productive (Brady and Weil, 2008) equire fertilization due to the sandy parent material in which most soils in Florida are formed. Poorly and very poorly drained soils predominate in the FL SRB (Figure 2 7). The sandy soils in this area tend to have medium to high potential for nutrient an d agrichemical leaching to groundwater as shown in Figure 2 8. Figure 2 8 shows the spatial distribution of leaching potential levels in the study area. High leaching Ultisols (excessively drained to well drained) are Apopka (loamy, siliceous, subactive, hyperthermic Grossarenic Paleudults) and Valdosta (siliceous, thermic, Psammentic Paleudults). High leaching Inceptisols (excessively drained to well drained) include Fort Meade (siliceous, thermic Humic Psammentic Dystrudepts); Fort Meade (siliceous, hyp erthermic Humic Psammentic Dystrudepts); and Orlando (sandy, siliceous, hyperthemic Humic Psammentic Dystrudepts). High leaching Entisols (excessively
41 drained to well drained) are Alaga (thermic, coated Typic Quartzipsamments); Alaga (thermic, coated Lamel lic Quartizipsamments); Arents (thermic Arents); Arents (hyperthermic Arents); Astatula (hyperthermic, uncoated Typic Quartzipsamments); Bigbee (thermic, coated Typic Quartzipsamments); Candler (hyperthermic, uncoated Lamellic Quartzipsamments); Gainesvill e (hyperthermic, coated Typic Quartzipsamments); Kershaw (thermic, uncoated Typic Quartzipsamments); Kureb (thermic, uncoated Spodic Quartzipsamments); Lake (hyperthermic, coated Typic Quartzipsamments); Lakeland (thermic, coated Typic Quartzipsamments); P aola (hyperthermic, uncoated Spodic Quartzipsamments); Penney (thermic, uncoated Lamellic Quartzipsamments); Quartzipsamments (thermic, uncoated Quartzipsamments); and Quartzipsamments (hyperthermic, uncoated Quartzipsamments) ( NRCS, 2006a; Soil Survey Staff, 2010) . Land cover/Land use Th e FL SRB area is within the Southern Coastal Plain, North Central Florida Ridge, Eastern Gulf Coast Flatwoods, and Atlantic Coast Flatwoods Major Land Resource Areas (MLRA) (Figure 2 9) (NRCS, 2006b) . Various kinds of land use are utilized within different MLRAs, yet the forests provide major coverage. Four land cover/land use ( LC/LU ) maps (1989, 1995, 2004, and 2008) of the study area are shown in Figures 2 10 and 2 11, respectively (ERDAS Inc., 1989; FDEP, 2009, 2004; SRWMD, 1995) . The predominant LC/LU s in 2008 are as follows: upland forest (46.4%), wetland (29.0%), agriculture (14.3%), urban and built up (5.9%); rangeland (2.4%); water (1.0%); transportation, communication, and utilitie s (0.7%); and barren land (0.3%).
42 Between 1989 and 2008, the predominant LC/LU in the FL SRB was upland forest, followed by wetland and agriculture, while barren land occupied the least portion of the area. In comparing different upland forest categories , tree plantations formed the majority. Interestingly, Kautz et al. (2007) pointed out that while most wetlands in Florida experienced a decline, the FL SRB stream corridors and larger wetlands remained relatively stable and have actually increased since 2000. The major wetland typ e is vegetated non forest (i.e., sawgrass, cattail, and broadleaved aquatic plants) (FDEP, 2009) . The scattered urban and built up areas have been observed to be markedly greater in the eastern area than in t he western area of the main Suwannee River region. This LC/LU category showed a noticeable increase (about 4.3%) between 1989 and 2008. On the other hand, agriculture is the only LC/LU class that presented a negative pattern, perhaps due to a shift towards intensive agricultural production (Figure 2 12). Agriculture activities in the FL SRB are diverse entailing pasture (55% improved, unimproved, and woodland pastures), hay fields (24%), field crops (2%), and dairies and horse farms (1%) (FDEP, 2009) . In terms of the number of farms, Alachua, Suwannee, Levy, Columbia, and Madison Counties were highest among the counties in the FL SRB (U.S. Department of Agriculture (USDA), 2007) . Climate The climate of the Suwannee River region can be described as hypothermic and temperate (NRCS, 2006a) with a mixture of warm temperate and subtropical conditions (Katz et al., 1998) . According to the Parameter elevation Relationships on Independent Slopes Model (PRISM) climate group, the 30 year normal mean annual temperature of the entire SRB from 1981 to 2010 was 20 o C. The average annual maximum and
43 minimum temperatures were 26.5 o C and 13.4 o C, respectively. The average monthly precipitation was approximately 1,288 mm (PRISM Climate Group, 2012) . Figure 2 13 (a to c) shows average monthly maximum/minimum temperatures and precipitation between 1981 and 2010 within the entire Suwannee River Basin. While an overall warming climate trend has occurred globally since the mi d twentieth century (Intergovernmental Panel o n Climate Change (IPCC), 2007) , the historical records of temperature and precipitation across the southeastern United States (includes the SRB) (PRISM Climate Group, 2012) , have not shown an explicit trend in surface temperature in the 20 th century (IPCC, 2007) . According to Konrad et al. (2013) temperatures in Florida have steadily increased since the 1970s due to increasing daily minimum temperatures from urbanization and irrigation (Christy et al., 2006) . Precipitation related to sea breeze circulation has increased along the Florida Panhandle and the northern Florida Gulf Coast (Keim et al., 2011; Misra et al., 2011) . Model simulations for climate scenarios in the southeastern United States have shown contrasting results of both increases and decreases of future precipitation patterns across the region. According to Konrad et al. (2013) , who focused on climate projections in the southeastern U.S., the mean annual precipitation is expected to decrease and temperature to increase through the first half of the 21 st century. T he Hadley climate model projected an increase in average annual pr ecipitation of up to nearly 25% for the twenty first century (Burkett et al., 2001) . In contrast, K onrad et al. (2013) projected proneness to drought in the southeast region with relatively short durations but high in magnitude (Konrad et al., 2013) . Future precipitation based on the
44 southeastern U.S . region, except during summer when a decreasing trend is expected in parts of Arkansas, Louisiana, and South Florida (Konrad et al., 2013) . The greatest changes are expected during summer months. Increases in the length of growing season, the number of cooling degree days, the n umber of consecutive hot days, and interannual temperature variability are projected through the end of the 21 st century (Konrad et al., 2013) . The IPCC model projected increases in precipitation by up to about 6 percent in North Carolina and Virginia, and decreases by 2 to 4 per cent in Louisiana and South Florida by the middle of the twenty first century (IPCC, 2007). In addition, Burkett et al. (2001) projected the southeast regional maximum temperature to increase by approximately 1.3 o C in summer and 0.6 o C in winter by 2030 and the mean annual temperature to increase by 1.0 o C by 2030 and by 2.3 o C by 2100. A recent study conducted by S ong et al. (2013) revealed that the precipitation and temperature have been projected to change dramatically in the sourtheast region during 2000 2099. According to three storylines, A1B, A2, and B1 scenarios from IPCC, Song et al. predicted the overall increase in temperature for all scenarios with the largest increase under the A2 climate scenario, followe d by the A1B, and the B1. Whereas the proj ected precipitation has insignificant trend with very high variations for all three scenarios (Song et al., 2013) . It should be noted that th rough the middle of the twenty first century, the range of temperature variability over a short time period of each model is quite large and contributes to high uncertainties (Hawkins and Sutton, 2011) . Therefore, instead of precise prediction numbers, climate trends should be interpreted as broadly indicative projections (Konrad et al., 2013) .
45 Socio Econom y The FL SRB includes nine co unties (i.e., Columbia, Dixie, Gilchrist, Hamilton, Lafayette, Madison, Suwannee, Taylor, and Union) and portions of six counties (i.e., Alachua, Baker, Bradford, Jefferson, Levy, and Putnam). It should be noted that the reported data in this section repre sent information on a county basis. The population of the fifteen counties was 656,338 in 2010 with nearly 40 percent of the total in Alachua Count y (U.S. Census Bureau, 2010a) . The most rapid population change between 2000 and 2010 occurred in Lafayette (26.4%) and Baker (21.8%) Counties, while Madison and Putnam Counties had the least pop ulation growth with 2.7 and 5.6 percent , respectively. The total number of households in 2010 in the fifteen counties was 246,707, with a median household income of $37,613 and per capita income of $18,649 (U.S. Census Bureau, 2010b) (Table 2 2). The percentage of individuals living below the poverty level was 21 percent between 2008 and 2012 (U.S. Census Bureau, 2014) . In 2012, about 54 percent of residents were males , 63 percent of residents were middle aged (45 64 years), and 17 percent were 65 years and over . Twenty on e percent of residents within 15 counties had less than a high school diploma or General Education al Development ( GED ) (U.S. Census Bureau, 2014) .
46 Table 2 1 . Generalized geological and hydrogeologic units in the Suwannee River Basin, Florida . System Series Formation Aquifer system Quaternary Pleistocene/Holocene Undifferentiated deposits Surficial Tertiary Plioc ene Undifferentiated deposits Surficial Miocene Hawthorne Group Intermediate aquifer system and confining beds Oligocene Suwannee Limestone Upper Floridan aquifer Eocene Ocala Limestone Avon Park Limestone Oldsmar Limestone Upper Floridan aquifer [Adapted from Farrell et al. 2005 . Technica l Report: MFL Establishment for the Lower Suwannee River and Estuary Little Fanning, Fanning and Manatee Springs. Suwannee River Water Management District in associate with Water Resource Associates, Inc. and SDII Global Corporation]. Table 2 2 . Populati on and income in the 15 counties of the Suwannee River Basin, Florida. County Population (2010) % change in population (2000 2010) Number of households (2010) Per capita income (2006 2010) Median household income (2006 2010) Alachua 247,336 20.0 10 0,516 $24,741 $40,644 Baker 27,115 20.4 8,772 $19,593 $47,276 Bradford 28,520 15.9 9,479 $16,997 $41,126 Columbia 67,531 32.6 24,941 $19,366 $38,214 Dixie 16,422 30.6 6,316 $17,066 $32,312 Gilchrist 16,939 49.3 6,121 $19,30 9 $37,039 Hamilton 14,799 21.9 4,617 $15,794 $37,613 Jefferson 14,761 14.2 5,646 $19,647 $41,359 Lafayette 8,870 25.9 2,580 $18,069 $46,445 Levy 40,801 33.0 16,404 $18,703 $35,737 Madison 19,224 13.1 6,985 $16,346 $37,459 Putnam 74,364 8.2 29,409 $18,402 $34,645 Suwannee 41,551 30.1 15,953 $18,782 $36,352 Taylor 22,570 12.5 7,920 $18,649 $37,408 Union 15,535 31.1 1,048 $13,657 $41,794
47 Figure 2 1 . The Suwannee River Basin and sub basins in Georgia and Florida. [ Sub basin boundary a dapted from National Hydrography Dataset (NHD). 1994. National Hydrography Dataset. Map scale 1:250,000. Accessible through http://datagateway.nrcs.usda .gov/GDGOrder.aspx ; Florida state adapted from Florida Department of Florida Protection. 1999. TIGER State and County Shapefiles. Map scale 1:24,000. Accessible through https://www.c ensus.gov/geo/maps data/data/tiger line.html ; Streams computed from a digital elevation model and topographic attributes ].
48 Figure 2 2 . Digital Elevation Model (DEM) and sub basins Suwannee River Basin. [ Sub basin boundary a dapted from National Hydro graphy Dataset (NHD). 1994. National Hydrography Dataset. Map scale 1:250,000. Accessible through http://datagateway.nrcs.usda.gov/GDGOrder.aspx ; Elevation adapted from National Elevation Datas et (NED). 2009. Map scale 1:250,000. Accessible through http://ned.usgs.gov/ ].
49 Figure 2 3 . Counties and major physiographic regions within the Suwannee River Water Management District. [Adapted from Puri , H.S. and Vemon , R.O. , 1964. Summary of the Geol ogy of Florida and a Guidebook to the Classic Exposures (Florida Geological Survey: Special publication, #5 ) . Florida Geological Survey, Tallahassee, FL; White, W.A., 1970. The Geomorphology of the Florida Peninsula, Published for Bereau of Geology, Divisi on of Interior Resources, Florida Department of Natural Reso urces. Designers Press, Orlando ; Study area boundary adapted from the Suwannee River Water Management District (SRWMD), 1999. SRWMD boundary. Map scale 1:24,000. Accessible through http://www.srwm d.state.fl.us/index.aspx?NID=319 ].
50 Figure 2 4 . Geological units by series in the Suwannee River Basin, Florida. [Adapted from Scott et al., 2001. Geological Map of the State of Florida, Florida Ge ological Survey Map Series 146].
51 Figure 2 5 . Spat ial distribution of soil orders in the Suwannee River Basin, Florida, derived from Natural Resources Conservation Service/Soil Survey Geographic Database. [Adapted from Natural Resources Conservation Service (NRCS), 2006. Soil Survey Geographic Database (S SURGO). United States Department of Agriculture (USDA). Map scale 1:24,000. Accessible through http://datagateway.nrcs.usda.gov/GDGOrder.aspx ].
52 Figure 2 6 . Map of soil suborder for the extent of the Suwannee River Basin, Florida . [Adapted from Natural Resources Conservation (NRCS), 2006. Soil Survey Geographic Database (SSURGO). United States Department of Agriculture (USDA). Map scale 1:24,000. Accessible through http://datagateway.nrcs.usd a.gov/GDGOrder.aspx ].
53 Figure 2 7 . Soil drainage class distribution in the Suwannee River Basin, Florida. [Adapted from Natural Resources Conservation (NRCS), 2006. Soil Survey Geographic Database (SSURGO). United States Department of Agriculture (US DA). Map scale 1:24,000. Accessible through http://datagateway.nrcs.usda.gov/GDGOrder.aspx ].
54 Figure 2 8 . Soil leaching potential in the Suwannee River Basin, Florida . [Adapted from Natura l Resources Conservation (NRCS), 2006. Soil Survey Geographic Database (SSURGO). United States Department of Agriculture (USDA). Map scale 1:24,000. Accessible through http://datagateway.nrcs.us da.gov/GDGOrder.aspx ].
55 Figure 2 9 . Major land resource area (MLRA) in the Suwannee River Basin, Florida. [Adapted from Natural Resources Conservation (NRCS), 2006. Soil Survey Geographic Database (SSURGO). United States Department of Agriculture (US DA). Map scale 1:2,000,000. Accessible through http://datagateway.nrcs.usda.gov/GDGOrder.aspx ].
56 (a) (b) Figure 2 10 . Map of land cover/land use ( LC/LU ) for the extent of the Suw annee River Basin, Florida, in (a) 1989 and (b) 1995. [Adapted from ESDAS Inc., 1989. Land Cover/Land Use. Map scale 1:500,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=31 9 ; Suwannee River Water Management District (SRWMD), 1995. Land Cover/Land Use. Map scale 1:40,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ].
57 (a) (b) Figure 2 11 . Map of land cover/land use ( LC/LU ) for the extent of the Suwannee River Basin, Florida, in (a) 2004 and (b) 2008. [Adapted from Florida Department of Environmental Protection (FDEP), 2004. Land Cover/Land Use. Map scale 1:12,000. Accessible through http://www.dep.state.fl.us/gis/datadir.htm ; Florida Department of Environmental Protection, 2009. Land Cover/Land Use. Map scale 1:5,000. Accessible through http://www.dep.state.fl.us/gis/datadir.htm ].
58 Figure 2 12 . Land cover/land use trends in 1989, 1995, 2004, and 2008 in the Suwannee River Basin, Florida. Note: Area in square kilometers derived by calculating spatial data (30 x 30 m resolution) . [ 1989 Land Cover/Land Use adapted from ESDAS Inc . Map scale 1:500,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ; 1995 Land Cover/Land Use adapted from Suwannee River Wa ter Management District (SRWMD) . Map scale 1:40,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ; 2004 Land Cover/Land Use adapted from Florida Department of Environmental Protection (FDEP) . Map scale 1:12,000. Accessible through http://www.dep.state.fl.us/gis/datadir.htm ; 2009 Land Cover/Land Use Florida Departm ent of Environmental Protection . Map scale 1:5,000. Accessible through http://www.dep.state.fl .us/gis/datadir.htm ]. 0 2,000 4,000 6,000 8,000 10,000 Urban and Built-up Agriculture Rangeland Upland Forest Water Wetland Barren Land Transportation, Communication & Utility Area (km 2 ) Land cover/land use type 1989 1995 2004 2008
59 a) (b) (c) Figure 2 13 . Maps of 30 yr normal monthly average of (a) precipitation, (b) maximum temperature, and (c) minimum temperature between 1981 and 2010 within the Suwannee River Basin, Florida. [Ada pted from PRISM Group, 2012. Precipitation, Maximum temperature, and Minimum Temperature 1981 2010. Spatial resolution 800 m. Accessible through http://www.prism.oregonstate.edu/ ].
60 CHAPTER 3 SPATIO TEMPORAL INTERACTION S BETWEEN SOIL CARBO N AND NUTRIENT C YCLES IN THE SUWANNE E RIVER BASIN Overview Biogeochemical fluxes occur along a continuum within the atmosphere, soil and water, and the cycling processes provide use and benefits to humans. One of the goals of this chapter is to investigate the connectivit y of selected ecosystem services: climate regulation, soil organic carbon (SOC) sequestration, and nutrient cycling in the terrestrial and aquatic ecosystems along spatial and temporal scales. Three elements carbon (C), nitrogen (N), and phosphorus (P) are primarily selected because of their essential services in the watershed. In Florida, there is a compelling evidence showing that soil has acted as a net carbon sink and has stored the highest organic carbon stock on a per unit area basis in the contermino us U.S. (Natural Resources Conserv ation Service (NRCS), 2006a) . This is largely due to favorable conditions such as high mean annual precipitation rates, high net primary productivity, and low relief topography that foster the accumulation of carbon in soils (Vasques et al., 2012; Vasques et al., 2010a) . The amount of SOC serves as a critical indicator in the carbon budget of an ecosystem and is interconnected with other biogeochemical cycles. The spatial distribution of historic SOC stocks in the state of Florida was estimated based only on historic and coarse soil taxonomic map units at a map scale of 1:250,000 (Guo et al., 2006) . Their rough estima tion did not provide adequate information of SOC dynamics and its uncertainty. Recently, the demand for soil property mapping across the region has increased in response to the rising concern for soil related conservation (Vasques et al., 2010a; Vasques et al., 2010b) .
61 The freshwater system is important in biogeochemical reactions involving carbon and nutrient cycles. Surface waters are highly dynamic and complex systems that receive carbon and nutrient influx es from surrounding areas within watersheds and are also characterized by internal transformations cycling elements. The Suwannee River and its tributaries have experienced impaired water quality for over a decade. Several water conservation programs have been implemented in the basin to improve water quality to sustain aquatic life and enhance human use. These practices can either provide co benefits to other services, such as soil carbon sequestration and climate regulation, or can result in trade offs. Often policies and legislation are disconnected to address soil, water, and air (Smith et al., 2013) . The lack of understanding the state of ecosystem servi ces and interactions among them limits the ability of policy makers and stakeholders to deliver a more integrative ecosystem management view. Attempts to quantify the value of ecosystem goods and services are growing rapidly including development of metho ds that range from simple spreadsheet to complex software such as InVEST, ARIES, LCUI, EcoServ, and ESValue (Bagstad et al., 2013) . This study aims to address one aspect of current research needs by quantifying concomitantly carbon and nutrient conditions in space and time, which involves the land water carbon and nutrient interactions. Currently, no s ingle complete model can provide the best answer to ecosystem management questions. A multi approach from different disciplines is required to improve our knowledge about ecosystem services. In this chapter the spatio temporal patterns of carbon and macro nutrients in soil and aquatic systems in the SRB are investigated from a biophysical perspective.
62 The objectives are: 1. Assess and compare spatially explicit changes in SOC stocks and organic (TOC) loads in the lotic system 2. Analyse temporal patterns of TO C, total nitrogen (TN), and total phosphorus (TP) loads in the water 3. Investigate relationships between organic carbon and nutrients (TN and TP) in soils across the basin and in the lotic system 4. Identify cross linkages between environmental factors (land co ver/land use change, climate properties, geographical and hydrologic characteristics) and changes in area standardized SOC stocks and TOC loads The hypotheses include: 1. The rate of TOC loading change in the drainage areas positively correlates with SOC sequ estration rate 2. The ratios of C:N and C:P are higher in soils than in river systems 3. The SOC stocks induced by external environmental factors are proportionally larger in nutrient enriched ecosystems than in nutrient limited ones. Materials and Methods Data used in this study consist of two parts: soil data and surface water data. Soil data contain two datasets (historic and current datasets) covering the time period 1965 to 2009. To quantify historic and current SOC stocks, the datasets were harmonized and nearly the same, due to the miniscule amounts of inorganic carbon ( IC ) in these soils. Surface water data contain discharge, concentration, and loads of TOC from 23 water gaugi ng stations which cover a time period of eleven years (2000 2010). Both SOC stocks and TOC loads were standardized on a unit area basis in order to compare organic carbon gains and losses in soils and in surface waters. Total nitrogen and TP observations i n soils and surface water characterize nutrient trends in the basin. The following subsections provide details for data collection and the analysis.
63 Total nitrogen and TP soil samples were collected at the same time as the soil carbon current dataset (20 08/2009) at 20 c m deep. A total of 234 site specific locations within the study are were used for statistical analysis. Total nitrogen and TP were derived by gas combustion analysis and inductively coupled plasma (ICP) methods, respectively. Historic Soi l Carbon Dataset ( DSh ) Florida Soil Characterization Database (FSCD) (http://flsoils.ifas.ufl.edu or http://TerrC.ifas.ufl.edu) (Grunwald and Harris, 2012). All soil sampling loca tions were projected in ArcGIS Â® using the Albers Conical Equal Area map projection. Soil samples from the FSCD that are located within the FL SRB (n = 234) were clipped to the study area. The FSCD included 1,300 soil profiles across the state over an appr oximate 30 year time period (1965 1996). Over 8,300 soil horizons were described. Data were collected for 144 physical, chemical, biological, morphological, and taxonomic soil properties including bulk density and soil organic matter (SOM). Data collection and lab Department of Agriculture (USDA) Natural Resources Conservation Service (NRCS) and Environmental Pedology Laboratory, Soil and Water Science Department, University of F lorida. Sampling design The FSCD samples were collected during the period 1965 to 1996 and soil sampling locations were determined based on tacit knowledge of soil surveyors. To the best of our knowledge, no strategic sampling design was employed (persona l
64 communication Dr. Harris and Dr. Grunwald). Figure 3 1 shows the spatial distribution of the DSh dataset and its SOC stocks in the FL SRB. Laboratory analysis The laboratory process was conducted by the Environmental Pedology Laboratory, University of Florida. Soils in general included mineral and organic soils. The SOM was measured using the Walkley Black modified acid dichromate extraction method (WB) (Walkley and Black, 1934 ) for mineral soils and loss on ignition (LOI) for organic soils. In order to obtain SOC content, a pedo transfer functions (PTF) was employed to convert SOM and LOI to SOC (Myers et al., 2011; Ross et al., 2013) . The PTF harmonized the SOC measurements and allowed comparison between DSh and the contemporary DSc (described below). The conversion of WB and LOI to SOC started by measuring total carbon (TC), inorganic carbon (IC) and LOI on a horizon stratified subset of DSh samples. Total carbon was measured by gas com bustion analysis, and IC was measured by acid reaction gases on a Shimadzu gas combustion analyzer (TOC V/SSM 5000). The PTF was developed for converting DSh mineral and organic soils separately to SOC using linear regression. A robust model was fitted us ing the iteratively re weighted least square to generate a model between soil organic matter (SOM) (%) and SOC (%) (Venables and Ripley, 2002) . A strong relationship between SOM and SOC (R 2 = 0.92) and low residual standard error (0.11) affirmed the robustness of the model (Equation 3 1). (3 1) where : Soil organic carbon representing combustion method (%)
65 : Soil organic carbon derived from Walkley Black dichromate extraction (%). The linear regression test between LOI and SOC i ndicated a possible segmented relationship. For non constant regression slope, the Davies test (Davies, 1987) suggested a change in slope around 75 percent LOI (p value < 0.001). A regression procedure in PTF showed a broken line on the slope which indicated two outcomes for converting LOI to SOC (%). Equation 3 2 is used to calculate samples with LOI less than 75.8 percent and Equation 3 3 is for samples with LOI greater than 75.8 percent. (3 2) (3 3) where : Soil organic matter derived from loss on ignition (%) Conversion of SOC concentrations into stocks and reconstruction of the fixed depth To convert SOC concentrations int o stocks, DSh data were standardized to a 0 20 cm depth. Carbon stocks were derived by using measured bulk density (BD) values multiplied with SOC (%) for the 20 cm soil depth (PD) (Equation 3 4). (3 4) where : Soil organic carbon stocks at 20 cm profile depth (kg C m 2 ) : Bulk density (g cm 3 ) : Soil organic carbon (%) : Profile depth (0.2 m)
66 DSh was converted to SOC stocks at fixed depth intervals (0 20 cm). This re construction of the depth allowed constant sample support for geostatistical analysis, which was not provided by soil horizons. The SOC stocks from the top horizon were assigned down to the 20 cm fixed layer, unless there were multiple horizons present in the top 20 cm profile. In these cases a depth weighted average of SOC weighted by the horizon depth was employed. Current Soil Carbon Dataset ( DSc ) ory Modeling of Changes in Soil Carbon Storage and (Grunwald et al., 2011a) . A total of 1,014 soil samples were collected from the topsoil (0 20 cm depth) across Flor ida in 2008/2009. The DSc coordinates were projected to Albers Conical Equal Area. After clipping, there were 138 sites within the FL SRB (DSc sa ) and an additional 96 soil samples inside a 20 km buffer. This made up a total of 234 site specific locations within the area of interest (DSc bf ) (Figure 3 2). Sampling design The sampling design was based on two conditions. First, it was a stratified random method. Second, about half of the samples coincided with the historic locations. The ground surface was s tratified by 13 land cover/land use (LC/LU) classes (Florida Fish and Wildlife Commission (FFWC), 2003) and 10 soil suborders (Natural Resources Conservation Service (NRCS), 2006a) . The classification scheme excluded the minor cove rage (<1%) of combinations. This resulted in a total of 63 LC/LU suborder The DSc data of this study consisted of 68 sites that coincided with the historic phase.
67 Sample collection procedur e The sample sites were located by differential global positioning system (GPS). Four 20 x 5.8 cm soil cores were collected (0 20 cm depth) from each location within a 2 m diameter distance. These four core samples were bulked into one sample in the field and were placed into a cooler for the laboratory process. Vegetation and LC/LU along with soil suborders were observed and confirmed in the field. Laboratory analysis Bulk density was measured in the laboratory based on gravimetric measurements. A 2 mm s ieve was used to retrieve the fine fraction of the air dried sample. After mixing the fractions thoroughly, these fine samples were stored in plastic containers. Subsamples were taken from air dried bulk samples and ball milled to determine TC and IC. A 5 percent replication for all samples was employed. TC and IC were measured by a Shimadzu TOC V/SSM 5000 gas analyzer using separate gas analysis procedures. TC was measured by CO 2 evolution using ball milled soil samples and combusted at 900 o C. IC was also measured by CO 2 evolution using ball milled samples with 42.5 percent phosphoric acid (H 3 PO) reaction in the gas analyzer at 200 o C. The SOC concentration was derived by subtracting IC from TC (mg kg 1 ). The SOC stocks (kg C m 2 ) were calculated using SOC c oncentration multiplied by measured bulk densities. Statistical and Geostatistical Analyses Descriptive statistics of SOC stocks describing the historic set are summarized in Table 3 1. Both the historic (DSh bf ) and current (DSc bf ) datasets were randomly split into a model calibration set (70% of data) and validation set (30% of data) in order to
68 independently evaluate model performances of the geostatistical analyses. The calibration sets were confirmed to have similar data distributions as the validation sets. This kriging approach has three main steps: (i) exploring observed data to characterize spatial continuity, (ii) developing a semivariogram model for spatial autocorrelation analysis, and (iii) calculating unknown values based on optimized weighted average of nearby locations within a defined search neighborhood (Goovaerts, 1997) . The kriging technique assumes stationarity of a random process in the stochastic view that the mean of the process is considered constant. Moreover, the variance of two different points is assumed to depend on the separation distance between two l ocations (Goovaerts, 1997) . The semivariogram plots the semivariance and dist ance between two locations (lags). Parameters, such as nugget, sill, and spatial autocorrelation range, can be derived from semivariograms. The nugget informs the measurement error and fine scale variability. The upper bound of second order stationarity th at remains after the initial increase is called the sill. The maximum distance of spatial autocorrelation is known as the range (Grunwald, 2006; We bster and Oliver, 2001a) . We used the standard geostatistical method of point ordinary kriging (OK) and block kriging (BK). The OK and BK approaches (Webster and Oliver, 2001a) were implemented in the ISATIS TM software (Geovariances Inc., France) using natural log transformed data of DSh cal and DSc cal . In krigi ng, the estimates are weighted linear combinations of the data using a known quantity from nearby locations (Webster and Oliver, 2001a) . The weights are allocated to the sample data within the neighborhood of the point (OK) or block (BK). The block size assigned in this study was 30 m by 30 m
69 with 5 m by 5 m esti mates averaged over a block. All observations and maps were projected to Albers Equal Area Conic map projection using ArcGIS Â® (Environmental Systems Research Institute (ESRI) Inc., Redlands, CA). After the kriging process was completed, the SOC stock esti mates were back transformed using Equation 3 5 to original units (Webster and Oliver, 2001a) . (3 5) where SOC is soil organic carbon stocks (0 20 cm) in kg C m 2 (of historic and current datasets, respectively), is the estimation of kriged values at location , and is the variance of kriged estimates. Finally, the back transformed SOC estimates were masked to the FL SRB boundary. High variance values from the kriging process that weakened the confidence of estimations were removed. This study followed the proto (2013) study suggesting that the areas greater than the third quantile of the variance maps of historic and current SOC estimates are masked out to increase confidence in the interpretation o f results. The calculation was conducted by a conditional statement in the raster calculator tool in ArcGIS Â® . Equation 3 6 provides the statement of raster calculation for eliminating the third quantile areas for estimates derived from BK. The same condi tional masking method was applied also to estimates derived from OK. (3 6) where is the variance of the kriged estimates. is a grid cell (30 m x 30 m). is the block kriged estimates of historic SOC stocks.
70 is the block kriged estimates of current SOC stocks. is the variance value of the third quantile (75 percentile) of the historic dataset. is the variance value of the third quantile of the current dataset. The validation sets (DSh val and DSc val ) were used to assess the kriging estimates. We report mean error (ME) and root mean square error (RMSE) (Equations 3 6 and 3 7) (Webster and Oliver, 2001a) . (3 7) (3 8) where = observed values = estimated values = point location Soil Organic Carbon Change and Seques tration Rates We applied two methods to quantify SOC stocks change and sequestration rate: (i) the SOC estimates derived by the kriged maps, and (ii) the SOC sequestration rate derived by the collocated sites. Kriged e stimates The SOC stock changes were derived by subtraction of DSh bf SOC estimates from DSC bf SOC estimates in ArcGIS (Equation 3 9). To quantify SOC sequestration rates from kriged estimates the following equations were used adopting the same
71 procedure as outlined by Ross et al. (2013) . The same formula was also applied to calculate the SOC stock changes derived from OK. (3 9) (3 10) where is SOC change (net gain and loss), refers to year of measurement (historic samples), is the number of years between historic and current observations, is a pixel with dimension width x leng th, here 30 m x 30 m (where the x and y coordinates are representing the center of each grid cell), and is the SOC sequestration rate based on kriged estimates (kg C m 2 yr 1 ) derived using BK. The was calculated for each dra inage basin and the whole SRB. Collocated s ites The collocated sites approach was adopted to assess the SOC sequestration rate on a per year basis. All historic observation sites (DSh bf ) located within 200 m of the current sites (DSc bf ) were used to asse ss the SOC sequestration rate (Ross et al., 2013) . This approach relies on direct measurements of historic and current SOC taken at a specific time. Overall, 68 paired sites matched the criteria. (3 11) (3 12) where is observed geographic coordinate (x and y coordinates) of point measurements, and is SOC sequestration rate (kg C m 2 yr 1 ) according to collocated sites.
72 Surface Water Quality Analysis The database used for hydrology analysis was obtained from the United States Geological Survey (USGS) and Suwannee River Water Management District (SRWMD) through the SRWMD water data portal (www.mysuwanneeriver.org/rivers.htm). Of the total 65 river gages, there were 23 that met our criteria for ecosystem service analysis within the FL SRB (Figure 3 3). The criteria required that the stations must (i) hav e discharge data available; (ii) contain total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) concentration; and (iii) cover long term data for the period 2000 to 2010. Total organic carbon was determined by the Environmental Protect ion Agency (EPA) Method 415.1, which is a combustion technique. Total nitrogen is the sum of nitrate+nitrite (NO x N), total Kjeldahl nitrogen (TKN), and ammonia nitrogen (NH 3 N). These three chemical components were measured using EPA methods 353.3 for NO x N, 351.2 for TKN, and 350.1 for NH 3 N, respectively (U.S. Environmental Protection Agency (USEPA), 1983) . To measure NO x N, reagents are added to a sample that forms a colored product when the reagents react with the nitrat e/nitrite, and then the intensity of colored product is measured. Total Kjeldahl nitrogen is measured using a strong acid and a metal ion catalyst solution to the sample. After the sample is taken to dryness, it is reconstituted with an alkaline solution. The ammonia is determined by indophenols colorimetry. Ammonia N nitrogen is an automated technique using a continuous flow analytical system with a phenate/hypochlorite color reagent that reacts with ammonia to form indophenols blue that is proportional t o the ammonia concentration. It has a measurement range limit of 0.01 mg/L (USEPA, 1983) . For all stations, discharge data were converted from cubic foot per second (ft 3 s 1 ) to cubic meter per day (m 3 d 1 ) (1 ft 3 = 0.02832 m 3 ) (USGS, 2014) . The
73 concentration and discharge datasets were s ummarized by averaging into monthly time intervals for the statistical analyses. Nineteen monitoring stations out of 23 contained data for the complete eleven year monitoring period, which is continuous from 2000 through 2010 (water years). One of the rema ining monitoring stations (Sntf.04) had continuous data from 2000 through 2009 (water years). Two of the remaining monitoring stations (Sntf.01 and Wacs.00) had continuous data from 2007 through 2010 (water years). Another remaining station (With.02) had a ten year continuous record from 2001 through 20 0 0 (water years). Please note that a water year defines the twelve month period between October September 2010 (USGS, 2014) . Total C arbon , T otal N itrogen, and T otal P hosphorus L oading T ren ds The total amount of TOC, TN, and TP that transferred to the water during a given (USEPA, 2014) . To account for the differences in sub basin areas the es timated TOC, TN, and TP loads were standardize by unit area. These unit area loads, expressed as loads per kg km 2 , was obtained by dividing sub basin TOC, TN, and TP loads by the respective drainage area. We realized that some old Hydrological Unit Codes (HUC) watershed boundaries from the Watershed Boundary Dataset (WBD) might not be accurate for loading analysis since some water monitoring gauges were no longer available. To eliminate the discrepancy, we re delineated the drainage areas and stream lines using the 1 arc second (~30 meter) digital elevation model (DEM) from the National Elevation Dataset (NED). After filling the sinks in the DEM the topographic attributes flow accumulation, flow directions, and total drainage area (watershed
74 boundaries) for each of the 23 water monitoring stations were calculated using the ArcGIS Â® hydrology tool. Drainage area delineation is shown in Figure 3 3 and information is illustrated in Table 3 4. Monthly harmonic means of TOC, TN, and TP loads standardized by draina ge area were analyzed from 2000 through 2010 (water years) to identify temporal trends. The nonparametric Kruskal Wallis test was used to determine seasonal variations (Kruskal and Wallis, 1952) . If a time series contained monotonic trend overtime, a nonseasonal Mann Kendall test was conducted (Mann, 1945) . If seasonality contained an overall trend component, a seasonal Mann Kendall test was implemented (Hirsch and Slack, 1984; Hirsch et al., 1982) . Relationships between Soil Organic Carbon and Total Organic Carbon Load ings Changes in SOC stocks and TOC loads were standardized by drainage area and number of years. This allowed us to quantify the SOC sequestration rate and TOC accumulation rate per area and per year. (3 13) where: is the total organic carbon change in water per drainage area in kg C m 2 is the number of water years between 2000 and 2010 (or the available time period provided for a specific water monitoring station) is the TOC accumulation rate in kg C m 2 yr 1 .
75 Spatial Variability of Organic Carbon to Total Nitrogen (C:N) and Organic Carbon to Total Phosphorus (C:P) Ratios across the Terrestrial and Aquatic Ecosystems Stoichiometric ratios of C:N soil and C:P soil were derived using the SOC, TN, and TP estimates in 2008/2009 (in kg m 2 ) derived from kriging. Note, that in Flo rida almost all soil carbon is in organic form, and thus, the assumption is made that SOC equals total soil carbon. The C:N water and C:P water were derived using the monthly harmonic means of TOC, TN, and TP loads (in kg m 2 ) in 2008/2009 in order to match the water load data to the same time period as soil data. Results and Discussion Historic Site specific Soil Organic Carbon Stocks The SOC stocks representing historic conditions are listed in Table 3 1. As can be seen, the historic SOC mean and median s tocks were lower than those of the current set. A wider range of SOC stocks and higher skewness coefficient were found in the historic set compared to the current data. The descriptive statistics of mean SOC stocks for DSh sa and DSh bf were similar with va lues of 3.7 and 3.6 kg C m 2 , respectively. The maximum and minimum SOC stocks ranged from 0.4 to 46.6 kg C m 2 for DSh sa and DSh bf (Table 3 1). Of the 290 DSh bf values greater th an 6.3 kg C m 2 . The large amounts of SOC were found in marshes and wetlands, which statistically can be considered as outliers, but from an ecological perspective are profoundly important to characterize carbon enriched soils. Soil suborders ranked from highest to lowest SOC stocks with Saprists > Aquolls > Aquults, showing mean values of 23.5, 12.4, and 6.4, kg C m 2 , respectively. The
76 relatively high SOC storage in these soils can be explained by the aquic moisture regime. The moisture condition indicat es that Saprists (wet Histosols), Aquolls (wet Mollisols), and Aquults (wet Ultisols) are saturated for extended periods of time by seasonal flooding. Saturated or flooded conditions limit the availability of oxygen for microbes to decompose organic matte r under anaerobic decomposition (Keller, 2011; Megonigal et al., 2003) . As a result of the slow decomposition rate, wetlands or peatland global soils have accumulated ap proximately one third of the soil carbon pool (Bridgham et al., 2006) . Vasques et al. (2010) and Ross (2013) had similar findings in subtropical Florida, where they found the highest amount of SOC stocks in organic soils (Histosols). In our study, the lowest SOC stocks were found in Udalfs < Orthods < Udults with mean values of 2.1, 2.3 and 2.5 kg C m 2 . Interestingly, the lowest SOC stocks (kg C m 2 ) were expected to be found in Psamments since they are considered quartz itic, sand rich deposits and are formed on well drained uplands. They ty pically lack capacity to hold organic matter in the topsoil. These findings were different in comparison to the results from Vaques et al. (2010) and Ross (2013), in which they concluded the lowest SOC stocks to be associated with Entisols. However, the relatively low SOC stock s in Psamments, with a mean value of 2.9 kg C m 2 , were also found in this study. In general, Udalfs, Orthods, and Udults are better drained than soils of aquic suborders and likely have more rapid carbon decomposition rates. In Spodosols, the presence of organic compounds can be found in a Bh horizon (Brady and Weil, 2008; Tan et al., 1999) . These Bh horizons usually occur below the topsoil (0 20 cm) investigated in this study.
77 Current S ite specific Soil Organic Carbon Stocks The mean SOC stock value for DSc sa was 5.2 kg C m 2 which was higher than DSc bf with a mean value of 4.6 kg C m 2 . In comparison to the historic dataset, the range of SOC stocks in the current dataset was narrower with 0.9 to 20.6 kg C m 2 (Table 3 1). Of the 234 soil profiles for DSc bf rd quantile) with mean values larger than 10.9 kg C m 2 . These high SOC stock samples were a lso found in swamps and wetlands. Like the historic dataset, Saprists under current soil conditions store the highest mean SOC stocks (10.8 kg C m 2 ). These results confirm the findings from Bernal and Mitsch (2012) , Chmura (2003) , and de Klein and van der Werf (2013) that the approp riate slow decomposition rate and long term carbon accumulation in wetlands are important for soil carbon storage. Aquolls and Aquepts also store high SOC content in the FL SRB with mean SOC stocks of 8.8 and 7.1 kg C m 2 . These soils are formed in aquic s oil moisture regime. The lowest SOC stocks in soil suborders were similar to the historic conditions, but with the inclusion of Psamments. The smallest SOC stock values were as follows: Udalfs < Psamments < Udults with means of 1.9, 2.4, and 3.4 kg C m 2 , respectively. Geospatial Patterns of DSh and DSc across the FL SRB As can be seen in Figure 3 1, sampling locations of DSh were impacted by a data gap, in which the spatial coverage was not as evenly distributed as in dataset DSc. The fitted semivariograms of DSh cal in Table 3 4 shows a larger range of 29,319 m than the range calculated for DSc cal with 29,169 m. A long range indicates regional spatial autocorrelation. These long range spatial dependence patterns found in our study was
78 also found in similar studies that investigated the soil carbon spatial variability in watersheds in Florida (Grunwald et al., 2010; Ross et al., 2013) . The nugget to sill ratios of the DSh cal and DSc cal datasets for OK and BK were simil ar with 50% suggesting a moderate range spatial structure. If the nugget to sill ratio is less than 25 percent, the variable is considered to have a strong spatial dependence, a ratio between 25 percent and 75 percent is considered to have a moderate spati al dependence, and otherwise the variable has a weak spatial dependence (Cambardella et al., 1994) . Table 3 5 provides descriptive statistics for historic and current SOC stock (0 20 cm) estimates as derived by OK and BK methods in two modes: (a) SOC estimates across the FL SRB and (b) SOC stock estimates within the FL SRB but with exclusion of high uncertainty areas. The maximums SOC kriged estimates from historic and current datasets in mode (a) and (b) wer e much lower than the field observed ones from DSh bf (46.6 kg C m 2 ) and DSc bf (20.6 kg C m 2 ), which suggests relatively high bias for large kriged estimates. The minimum SOC kriged estimates from historic and current datasets derived by OK and BK were hi gher than the observations from DSh bf (0.4 kg C m 2 ) and DSC bf (0.9 kg C m 2 ). According to Journel and Huijbregts (1978) estimates derived from are somewhat biased due to underestimation of high estimates and overestimation of low estimates. Interestingly, the mean SOC kriged estimates from historic (OK: 3.7 kg C m 2 , BK: 3.3 kg C m 2 ) and current datasets (OK: 4.9 kg C m 2 , BK: 4.4 kg C m 2 ) were similar to the means from the observed SOC stocks (DSh bf : 3.6 kg C m 2 , DSc bf : 4.6 kg C m 2 ), suggesting low bias in mean estimates. The kriged maps derived by OK and BK
79 showed similar spatial distributions of SOC stocks (Figure 3 4). This can be explained by the small block size used in the BK; thus, the differences between the point based OK and BK were relatively small. However, lower values of ME and RMSE from the validation set in BK indicated a lower bias in SOC stock estimate s when compared to OK. It should be noted that the negative and positive estimates in the ME are considered indicators of accuracy. The RMSE in BK was smaller than OK in both modes (a) and (b) (Table 3 4). These findings were not surprising since the weigh ted average of the SOC stocks over a block was calculated instead of a site (i.e., with point support) in order to minimize the estimation of error variance. The mean variance of the kriged estimates in the historic dataset was consistently higher than th e current ones for both, OK and BK. This is likely to be a result of the uneven sampling distribution in the historic dataset. The SOC stock estimates derived by OK and BK showed improvements in model predictions for current conditions (DSc) after eliminat ing the high variance portions, but not for historic conditions (Dsh) as indicated by ME and RMSE metrics (Table 3 4 and Figure 3 5). According to Tourassi et al. (2001) , there is a chance that the splitting of the full dataset into a calibrat ion set and a validation set may suffer from bias or variance in observed values. As shown in Table 3 5, the number of validation samples of the historic dataset in mode (a) remained at 87 samples, but the validation samples of the current datasets dropped from 70 to 62 samples during the high uncertainty elimination process which may have impacted the bias and error assessment. The FL SRB covers 19,665 km 2 , including 185 km 2 of water bodies. Soil organic carbon stocks were quantified for the land area, wh ich accounted for 19,480 km 2 . The
80 absolute historic and current SOC stocks are shown in Table 3 6. The historic SOC stocks ranged from 64.3 to 74.0 Tg and current SOC stocks ranged from 83.7 to 95.4 Tg across different approaches. In comparison with the n ortheast and east central regions studied by Ross et al. (2013) , soil conditions in the FL SRB (north central Florida) stored approximately 1.5 times and 3 times less than their findings for historic conditions and current conditions, respectively. Given that the total SOC stocks in Florida was 747 Tg (Xiong et al., 2014) , the FL S RB soils accounted for approximately 11 percent of the statewide SOC stocks. Quantification of Soil Organic Carbon Sequestration Rates Soil organic carbon sequestration rates for kriged estimates Gains and losses in SOC stocks (kg C m 2 ) were derived by s ubtracting the historic prediction maps from the current prediction maps and eliminating the high uncertainty areas (Figure 3 6). The variability of SOC changes ranged from 7.4 to 5.2 kg C m 2 (for center points of pixels with dimension 30 m x 30 m) deriv ed by OK, while the range of SOC gains and losses was narrower, from 6.4 to 4.6 kg C m 2 derived by BK (for blocks of size 30 mx 30 m). The mean SOC gains (after eliminating high uncertainties) of 1.0Â±1.3 kg C m 2 (OK) and 0.9Â±1.2 kg C m 2 (BK) averaged o ver the whole FL SRB region. The comparison of the kriged estimates yielded a net gain in SOC stocks of 30 g C m 2 yr 1 (BK) and 33 g C m 2 yr 1 (OK). These results indicate that this region has acted as a carbon sink over the past decades. Franzluebbers (2010) found that the SOC sequestration rates for the southeastern U.S. increased by 45 g C m 2 yr 1 , sampled to a 20 cm soil depth and increased by 84 g C m 2 yr 1 , sampled to a 25 cm soil depth. Minasny et al. (2011) reported an increase in SOC sequestration rate for tropical soils in Indonesia from 1970
81 to 2010 by approximately 10 g C m 2 yr 1 ienced a decrease in SOC many years before this time. This SOC increase was claimed as the result of increased crop production and biomass as a consequence of increased fertilizer use during the green revolution. In this study, however, there is no clear r elationship between fertilization as the main nutrient input in agricultural production systems and SOC sequestration rate. The statistic record of total fertilizer use by county from Florida Department of Agriculture and Consumer Services (FDACS) (2014) reported an unclear trend of TN and TP fertilizer use between 1945 and 2010. The highest TN fertilizer use occurred in 1976 1984 and in1999 and fluctuated over 2000 2010 within 14 counties i n the FL SRB (excluding Putnam county). Whereas, the largest TP fertilizer use emerged in 1965 1974, declined in 1975 and showed an inconsistent trend until 2010 (Figure 3 11). Soil organic carbon sequestration rates for collocated sites There were 68 soi l profiles that met the 200 m collocation criteria among the historic and current sites. As a result of a data hole in the middle of the study area, matched locations were found predominantly on the edge of the study area and inside the buffer zone. Althou gh the mean SOC change varied from small negative to positive values, overall the mean and median values were positive suggesting substantial SOC gains. When compared to the kriged estimates, the collocated SOC stock change values were lower with mean valu es of 0.5 kg C m 2 net gain and sequestration rate of 18 g C m 2 yr 1 across the study area. Figure 3 7 displays the magnitude and direction of the SOC change of each collocated site. About 62 percent of the collocated sites had experienced a net SOC gain, w hile 38 percent sites showed a SOC loss. The findings are very similar to those by Xiong (2013) , in which approximately 63 percent of the
82 collocated sites across the state showed a net gain and the remaining 37 percent have resulted in SOC loss. Our SOC sequestration rate was relatively large when compared to findings from Ross et al. (2013) who found an average increase of 0.8 g C m 2 yr 1 across four hydrologic basins in northeast and east central Florida. Other research in the U.S. showed approximate gains of 40 g C m 2 yr 1 for the central (Johnson et al., 2005) , 27 g C m 2 yr 1 for the northwestern (Liebig et al., 2005) , and 42 g C m 2 yr 1 for the southeastern regions (Franzluebbers and Follett, 2005) . Many internal soil formation processes and external environmental factors can manipulate carbon gains and losses in soils, such as climatic conditions, land use change, environmental conditions and decomposition, chemical and physical processes (Stockmann et al., 2013) . An overall net gain in the FL SRB is attributed perhaps due to condit ions of LC/LU and co benefits from water conservation program. Details of LC/LU change affecting SOC sequestration are discussed in the next section. Land cover/land use conditions and SOC change The SOC stocks increased by approximately 30 g m 2 yr 1 du ring a 30 year period that approximates the period where available LC/LU change was mapped (1988 and 2008). Land use change was pronounced in the basin (Chapter 2, Figures 2 10 and 2 11) and may explain some of the SOC change. Considering the high magnitu de of LC/LU change (> 1,000 km 2 change) (Table 3 7), the highest accumulation in SOC g m 2 yr 1 2 yr 1 2 yr 1 ) > 2 yr 1 (21.7 m 2 yr 1 2 yr 1 2 yr 1 ).
83 Other studies have shown that wetlands and conversion of other LC/LU types to wetland accumulated high amounts of carbon in soils. Craft et al. (2008) found net SOC gains in marsh soils with 24 g C m 2 yr 1 in the Okefenokee Swamp and 122 g C m 2 yr 1 under cypress in the Everglades with mean values of and 122 within 10 cm depth. Carbon sequestration rates of 99 to 190 g C m 2 yr 1 in the Everglades up to 60 cm deep were identified by Craft and Richardson (1993) . Ross (2013) found the largest SOC gains in Hardwood/Cypress Swamp and other wetland classes with 44 to 106.9 g C m 2 yr 1 among many other LC/LU classes covering agriculture, forest, and urban systems in a similar landscape in north east Florida. In our study, upland forests that remained in upland forests and all LC/LU classes that were converted to upland forests experienced net SOC gains. These findings were similar to a study presented by Ahn et al. (2009) , in which they concluded that the largest organic carbon storag es with 31.8 g C m 2 yr 1 from the uppermost 0 30 cm were associated with forested soils in a multi use watershed in northern Florida. On the other hand, Pregitzer and Palik (1997) found average SOC loss of 4.44 g C m 2 yr 1 in the shallow layers of soils undergoing conversion from cultivated land to pine plantation. The results from various studies suggest that the influence of plantation and natural forest on carbon pools vary across locations (Johnson, 1992; Morris et al., 2010, 2007; Paul et al., 2002; Powlson et al., 2011) . Xiong (2013) found that pine plantations in Florida showed an average SOC accumulation rate of 18.2 g m 2 yr 1 (0 20 cm) with insignificant difference of SOC stocks between natural pine forest and pine forest under management. Their results suggest ed that management of forest do not have major impact on SOC stock.
84 In our study the agricultural areas resulted in a slight i ncrease in SOC sequestration with small gains of 19.0 g C m 2 yr 1 when compared to other LC/LU classes. These findings contradict with scientific evidences that have determined depletion of SOC in agricultural soils, in particular by converting natural ec osystems to agricultural ecosystems or a plowing system (Davidson and Ackerman, 1993; Groenendijk et al., 2002; Kimble et al. , 1998; Reicosky, 2003) . Recently, some inconsistent results were found regarding agricultural practices and enhancement of carbon sequestration. Several studies found no significant difference in carbon sequestration between tillage and no tillage in the soil surface, but higher carbon accumulation in the soils were discovered in deeper profiles (Baker et al., 2007; Blanco Canqui and Lal, 2008; Poir ier et al., 2009) . In the FL SRB it is possible that increased SOC in agricultural soils was a consequence of the prominently implemented best management practices as well as extensive conservation programs. These programs aim to maximize agricultural pro ductivity while minimizing impact on water and retaining SOC in tilled/no tilled soil (Simonne et al., 2010) . Other factors such as crop types, crop density, cropping frequency, and fertilizat ion may also explain net SOC gains in agricultural systems (Luo et al., 2010; Sun et al., 2012) . In the FL SRB agriculture is composed mainly of specialty crops, such as improved pastures, hay fields, and row crops which indicates different LULC composition than in othe r studies focused on cash crops . Long term Trend Analysis of Total Organic Carbon, Total Nitrogen, and Total Phosphorus The 23 drainage areas ranged in size from 171 to 20,280 km 2 (Table 3 8). All flows were the average of the harmonic mean from water yea rs 2000 through 2010.
85 Trend analysis conducted for the monthly TOC loading showed an upward trend in eight drainage areas (out of 23 tested) which covered the Aucilla River (Aucl.00), upstream of the Santa Fe River (Sntf.01 and Sntf.03), the Upper Suwannee River (UpSuwn.01 to 04), and the Waccasassa River (Wacs.00) (Table 3 9). The increase in TN and TP loads were more pronounced than in TOC loads. An increasing trend of TN loads was found in the Coastal River (Cost.00), most of the Santa Fe River, the Upp er Suwannee River, and most of the Lower Suwannee River (Table 3 10). Similarly, TP loads (Table 3 11) had an upward trend in most of the drainage areas but slightly less than the TN trend (Table 3 10). The remaining drainage areas showed an insignificant trend which indicated little change in carbon and nutrient loading in the areas. Interestingly, while TOC, TN, and TP loads were strongly positive correlated with flows with R 2 of 0.9, 0.9, and 0.7, respectively, precipitation did not significantly change during the past decade. Total organic carbon , TN, TP loads may be attributed to nutrient and carbon influx via non point source pollution, local landscape settings, intricate aquifer, and spring hydrology of the system. An accent increase in TOC loads in the Upper Suwannee River has been documented by the North Florida Aquifer Replenishment Initiative (NFARI) (2013) . The upper part of the Suwannee River is known for its blackwater due to extensive peat deposits from the Okefenokee Swamp in South Georgia. Decomposition of vegetative bioma ss is claimed to provide large dissolved organic carbon to the river and is responsible for the dark water color (Meyer, 1990) . Low dissolved oxygen in blackwater conditions is common, particularly in summer when water flows are low and tempe ratures are high (Smock and Gilinsky, 1992; Todd et al., 2009) . It should be noted
86 that an increase in TOC loads occurred in the poorly drained areas. During high flows, the litter in the floodplains are picked up and transported to the streams (USEPA, 2002) . These operating processes occur in the blackwater system of the Suwannee River Basin that is dominated by wetlands and forests with combined coverage of 75 percent of the basin. High TOC loads may not affect human health as directly as nitrogen and phosphorus, but excessive organic carbon can disturb functions of an ecosystem via nutrient cycling and biogeochemical processes. High TOC indicates that much oxygen in the water is use d up during the natural process of decay of organic matter. Low oxygen affects aquatic animals to survive and impact human well being ultimately. The magnitude of annual TN load per area ( in kg km 2 yr 1 ) was associated with flows. Drainage areas that gen erated medium high and high flows transported TN loads between 200 and 1,500 kg km 2 yr 1 , whereas medium low and low flows transported TN loads between 17 and 200 kg km 2 yr 1 . Average monthly phosphorus loads showed a high magnitude in the Withlacoochee, Sntf.06, Sntf.07, UpSuwn.04, and the Lower Suwannee drainage areas (Table 3 11). This is mainly due to the river coming in contact with the parent material of the Hawthorne Group that contains high levels of phosphours rich carbonate fluorapatite (Maddox et al., 1992) , and partly due to discharges from phosphate mining (Hornsby and Mattson, 1998) . Average annual TN and TP loads were 244 and 20 kg km 2 yr 1 , respectively. T he Suwannee River has the highest nitrate nitrogen (NO 3 N) and TP load per area, compared to six other rivers in Georgia and Florida (Altamaha, St. Johns, Satilla, Ogeechee, Withlacoochee, and Ochlockonee Rivers) (Asbury et al., 1997) .
87 An upward trend of TN loads was observed pronouncedly in the lower Santa Fe River, the Upper Suwannee River, and the Lower Suwannee River. These findings were similar to the findings from Hornsby (2007) who indicated higher NO 3 N concentration trends in the Lower Suwannee River (comparable to the LwSuwn.04 in this study) during the period 1998 to 2006. The Suwannee River recharges and receives discharge from the Upper Floridan aq uifer (Pittman et al., 1997) . The inflow in the Suwannee River is directly due to ground water influx from the Upper Floridan aquifer. The poss ible nitrogen sources in streams are synthetic fertilizer, septic tanks, and animal waste (Andrews, 1994) . Bruland et al. (2008) in tracking nutrients that enter the Gulf of Mexico via the Suwannee Basin found a disproportionally high am ount of NO 3 N originating in the Santa Fe River. They attributed the high NO 3 N levels in soils to row crop agriculture and improved pasture sites based on seasonal spatial sampling in soils . The higher NO 3 N concentration was found in Ultisols and Spodoso ls that are not only prominent in the Santa Fe, but also dominate the whole Suwannee River Basin. An increase in TP loads was present in the Aucilla River, the Lower Santa Fe River, most of the Upper Suwannee River, and the Lower Suwannee River. Hornsby (2007) documented a decreasing trend of TP concentration from 1998 to 2006 at the Branford station (LwSuwn.04). Furthermore, GarcÃa et al. (2011) concluded that wastewater discharge contributed the largest amount of TP to the Suwannee River Estuary and streams, followed by background non point sources, agriculture, urban, and manure. It is also possible that low pH water runoff from the Okefenokee Swamp chemically reacts with the Hawthorne Group, resulting in phosphorus leach. The soi l phosphorus content ranged from < 2.5 to > 1000 mg kg 1 in
88 the Santa Fe River Watershed where elevated values were found in the subsoil due to the P rich geologic material these soils formed (Grunwald et al., 2010) . Phosphorus rich soils may ultimately contribute to the relatively high P l oads in surface waters found in the Santa Fe River Watershed, but also in the wider Suwannee Basin that consists of similar soils and geology. The relationships among nutrient loads, precipitation, and concentration are shown in Figure 3 12. Considering u pward trends in many drainage areas, there were a few that did not meet the numerical water quality standard proposed by the FDEP (2013) . In compliance with the Clean Water Act, the Environmental Protection Agency (EPA) completed an agreement in March 2013 for Florida to adopt its own rules to keep its aquatic ecosystem healthy by approving the numeric water quality standards (FDEP, 2013) The is a no more than one in three water monitoring stations comply with the TN and TP thresholds in the North Central region set at 1.87 mg L 1 and 0.30 mg L 1 , respectively. The remaining two stations (Aucl.00 and Cost.00) fall into the Panhandle East nutrient region adopting the TN and TP thresholds that cannot be exceeded with 1.03 mg L 1 and 0.18 mg L 1 , respectively (FDEP, 2013) . The Aucl.00, Cost.00, UpSuwn.03, and UpSuwn.04 exceeded the TN concentration threshold during the 1 0 year period (2000 2010), while the Aucl.00 and UpSuwn.03 showed surplus TP concentration. The primary concern with increases of TN and TP loads and concentrations is eutrophication of surface water and subsequent impacts on the ecological functioning of biotic systems. The input of nutrients will
89 stimulate microbial activities in water by increasing the rate of supply of organic matter to an ecosystem (Duarte, 2009) . Relationships between Soil Organic Carbon and Total Organic Carbon in Surface Water The average unit area TOC load i ncreased approximately 0.2 x 10 3 g C m 2 yr 1 for all drainage areas and increased 0.1 x 10 3 g C m 2 yr 1 at the 4 major drainage outlets of the Suwannee River Basin (2000 2010).The results coincided with organic carbon accretion rates in soils, with an overall average increase of 30 g m 2 yr 1 derived by multiplying the drainage area to retrieve SOC sequestration rates at each drainage basin. . Spatial cross correlations between unit area SOC stocks and TOC loads were confounded in the upstream of the San ta Fe River and Waccasassa River (i.e., the eastern portion of the study area). In general, these soils along the Sntf.01, Sntf.03, and Wacs.00 are poorly drained or somewhat poorly drained; thus, organic carbon from soils is transported horizontally via s urface runoff/erosion and lateral interflow. Khadka et al. (2014) pointed out that in the upper Santa Fe River (comparable to Sntf.01 to Sntf.05 in this study), soils and water release more carbon dioxide to the atmosphere when compare d to the lower Santa Fe River (comparable to Sntf.06 and Sntf.07 in this stud y). This is due to the two distinct hydrogeological settings of the region where much of the CO 2 is consumed during the carbonate dissolution reactions in the lower Santa Fe River. In contrast, other parts of the drainage areas along the Suwannee River ar e sandy, and are moderately well drained, or excessively drained. These soils present issues with irrigation and leaching with nutrients (U.S. Department of Agriculture (USDA), 1993) and possibly dissolved carbon. Though c arbon has been sequestered in
90 soils in the basin, the tendency of carbon efflux through the cascading lotic system into estuaries and the Gulf of Mexico were muted. Overall, Florida has relatively flat terrain with gentle slopes (0 5%), nonetheless, the s teeper slopes are found in the northern regions where the FL SRB is located. Topographic relief and elevation in the FL SRB become more muted with proximity to the Gulf of Mexico. In gentle sloping terrain surface soils materials, including organic carbon from the upper landscape positions, are prone to be transported to lower positions. This is a possible explanation for mismatched increases/decreases in SOC sequestration rate and TOC accumulation rates in soil and water systems. Variation in Carbon to Nu trient Ratios in Soils and Surface Water by Drainage Areas The carbon to nutrient ratios in soils are presented based on SOC, TN and TP stocks derived by BK only, because BK performed better than OK. The interpolation parameters and validation results for SOC stocks are described in Table 3 5 and for soil TN and TP are shown in Table 3 12. Figure 3 8 shows the patterns of SOC:TN and SOC:TP ratios across the FL SRB. Note that SOC and total soil carbon are assumed to be nearly equal in this basin because the amount of inorganic soil carbon is minuscule. The underlining criteria for stoichiometric ratios used for discussion are drawn from Unger et al. (1998) . They argue that a C:N less than 25 and the C:P less than 200 indicates net mineralization; if the C:N ratio is larger than 30 and the C:P is larger than 300, the process is net immobilizatio n, otherwise carbon and nutrients are naturally balanced. Based on mean values for SOC, TN, and TP stocks for each drainage area in soils and surface water, the following stoichiometric relationships were calculated: C:N =
91 17 and 9, respectively (Table 3 13). Spatial C:N ratios between the soil and lotic system were similar with the propensity of mineralization process (C:N < 25). For soils, net mineralization of N occurred across the 23 drainage areas with a range between 14 and 21 promoting the risk of N leaching into the aquifer and excess nitrogen being released to the atmosphere in the form of ammonia (NH 4 + ) or nitrous oxide (NO 3 ). Another study by Gundersen et al. (1998) suggested similar results, in which they found a C:N ratio less than 20 in the Lower Suwannee River and concluded that this area had the potential for greater N leaching. In our study, C:N ratios in surface water showed a wider range, from 2 to 39. Most drainage areas were low in C:N ratios, whereas the Aucl.00, Cost.00, Sntf.02, Sntf.03, and UpSuwn.04 with C:N ratios between 25 and 30 were identified; thus, indicating neither a gain nor loss of N. Only the UpSuwn .01 and UpSuwn.02 showed surface water net immobilization of N (C:N > 30) due to relatively low N. The mean soil C:N ratios in this study (C:N = 17) were lower than the findings from Grunwald et al. (2010) in which they reported mean ratios of C:N = 22.5 in the top soil (0 30 cm) within the Santa Fe Wat ershed. In contrast to C:N, the mean C:P ratios in soils were lower than those in surface water across all drainage areas. The C:P ratios in soils and the lotic system ranged from 46 to 288 and 3 to 866, respectively. The stoichiometric ratios in soils an d surface water were estimated at C:P = 108 and 84, respectively (Table 3 13). For soils, the net P mineralization took place in most areas (C:P < 200) except for the Cost.00, UpSuwn.01, UpSuwn.02, and UpSuwn.03, where neither net gain nor loss of P was fo und (200
92 can be explained by the naturally high carbon settings in the northeast area due to the bla ckwater environment. However, the UpSuwn.03 showed extremely low in C:P, with a mean value of 3 that might be described by the elevated P content from the P rich parent material derived from the Hawthorne geologic group. Overall, C:N and C:P were higher in the soils of the FL SRB than they were in the surface waters. Studies who compare C:N and C:P ratios in soils and surface waters across large basins, such as the FL SRB, are still rare. Others, such as Hecky et al. (1 993) reported that C:N and C:P ratios within freshwater systems were lower than lake sediments. The C:N:P ratios in this study were much lower than those in the Santa Fe River Watershed reported by Grunwald et al. (2010) with a soil C:N:P of 535:27:1 in 0 30 depth. McGroddy et al. (2004) suggested even a higher average C:N:P of 1,212:28:1 for foliage and 3,007:45:1 for litter. Note that the drainage areas used for calculation in this study did not cover most parts of the weste rn area where there was a high density of wetlands and forests. Moreover, the depth of soil profiles and types of land cover may force the differences. However, our C:N:P ratios of 108:6:1 were similar to Steven and Cole (1999) who proposed that the C:N:P should be similar across soil geographic regions in the world, with an averag e ratio of 108:8:1. In comparison to the global C:N:P of 106:16:1 in marine systems (Redfield, 1958) , our aquatic C:N:P results were lower, with mean values of 84:9:1 in rivers. Figure 3 8 shows the spatial distribution of soil N and soil P. Interestingly, the SOC and TN patterns of stocks mirrored each other, while TP stocks showed opposing patterns (Figure 3 9). Nutrient enriched and nutrient limited area separation using median values was also conducted to reflect the spatial SOC variability (Figure 3 10).
93 Within low TN areas (me dian 0.21 kg N m 2 ), the mean SOC (3.5 kg C m 2 ) tended to be lower, compared to high TN areas, where the mean SOC was 4.6 kg C m 2 . In contrast, within low TP areas (median 0.07 kg P m 2 ), SOC stocks showed opposing behavior with means of 4.8 kg C m 2 , whi le high TP areas contained lower means of 4.3 kg C m 2 (Figure 3 9). Conclusions Two major ecosystem services carbon sequestration and nutrient regulation in soils and surface waters were assessed in the FL SRB. In the context of global climate change , carbon sequestration plays a major role because it mutes the effects on climate warming. These beneficial regulating services at a global scale cannot be overstated because the threat that global climate change will intensify in the future is very real. Adaptation and mitigation efforts targeting to enhance the carbon sequestration in the ecosystem are viewed as mechanisms to enhance the carbon regulation ecosystem service. In addition, benefits of soil carbon sequestration are manifold including enhanced crop productivity, nutrient retention, increased soil cohesion between particles, reduced CO 2 emissions, improved water quality due to the reduction of nutrient run off, among many others The soil natural capital, here expressed as SOC and nutrient (N an d P) stocks, is critical for the long term sustainability and health of soils that subsequently secure food and fiber production. Agricultural and forest production is a major component of the economic system in the FL SRB enabling economic prosperity and livelihood of many residents in the basin. Our quantitative assessment of SOC stocks found that SOC ranged from 1.6 kg C m 2 to 13.8 kg C m 2 with a mean value of 3.4 kg C m 2 (based on BK), while current SOC stocks ranged from 2.0 to 8.3 kg C m 2 with a m ean value of 4.3
94 kg C m 2 (based on BK). The total absolute SOC stock was 66.2 Tg C in the historic dataset and 83.7 Tg C in the current dataset. These finding suggest that during the past decades, the FL SRB has acted as a soil carbon sink, which resulted in net SOC gains of 30 g C m 2 yr 1 derived by BK and 18 g C m 2 yr 1 derived by collocated historic current site observations. We assert that LC/LU and its change are closely interconnected with the carbon sequestration service. Although wetlands cover only 29% of the FL SRW it stores about 54 Tg of the total SOC stocks in the topsoil. Upland forests, w hich are extensive in the FL SRB cover about 46% of the area and store a total of 33.7 Tg SOC stocks in the topsoil. These contrast with carbon poor syst ems, such as rangelands and urban areas with much lower capacity to store carbon in soils. A confusion matrix provided ample evidence for major LC/LU shifts that have occurred within the basin including substantial increase in wetlands (927.4 km 2 ) and shif ts from upland forest to rangeland and agriculture that were interlinked with SOC sequestration. These findings suggest trends that will impact SOC storage under future LC/LU trajectories. Although there is no clear evidence, t he pro conservation strategi es implemented in Florida, e.g. Florida Forever Conservation Program, best management practices, and wetland restoration efforts, could have acted positive on the carbon storage in the basin; and thus, have profoundly enhanced the carbon regulation ecosyst em service. The monitoring program on these practices is needed to be addressed. Carbon sequestration and nutrient cycling were spatially correlated with soil nitrogen enriched areas with > 0.21 kg N m 2 covering 78% of the basin associated with SOC seque stration. The phosphorus enriched areas with > 0.007 kg P m 2 covering
95 76% of the basin have shown to enhance SOC sequestration. These findings that nitrogen and phosphorus in soils stimulate the accretion of carbon in soils have been confirmed in other st udies. But our study quantified the spatial patterns and confirmed only partial spatial cross correlation patters among carbon, nitrogen and phosphorus in soils and surface water that are complex. From the ecosystem service perspective the optimization of one service, e.g. soil carbon sequestration, may lead to the degradation of another service, e.g. nitrogen and phosphorus in soils and surface water. Nutrient enrichment from the ecological perspective may have negative effects, such as reduced biodiversit y and functioning of a system (e.g., N and P enriched wetlands) or impairment of drinking water quality. However, from an agricultural and forest production perspective nutrient enrichment is positive because it stimulates net ecosystem productivity, crop growth and enhances yield, up to a saturation level. This alludes to Instead, it has to be viewed in specific context considering space and time dimensions. We presented s everal maps that show the spatial distribution patterns of carbon and nutrients and their change across the basin that may serve to optimize carbon and nutrient regulation services in the future considering specific benefits, such as conservation of soil n atural capital, preservation of drinking water quality, soil carbon stocks promoting nutrient holding capacity in soils, or other. In the FL SRB 3 drainage basins, out of 23, showed impairment for TN and 2 drainage basins were impaired by TP considering th e Florida numeric water quality standards. Over a period of 10 years (2000 2010) 8, 16, and 13 drainage basins showed increasing trends for TOC, TN, and TP loads. Overall, the TOC loads in surface
96 waters within the whole drainage area slightly increased b y 0.2 x 10 3 g C m 2 yr 1 . The portions of the basin that showed the largest gains in SOC stocks did not spatially correlate with TOC load increases within drainage basins, suggesting that erosion and surface runoff were not the major process transporting soil carbon into the stream network. Rather, internal processes within the blackwater river system flowing through wetlands, floodplains, flatwoods and wet forests characterized by high net ecosystem productivity tend to retain carbon in the lotic system. The interaction effects between carbon and nutrients, and thus, the interaction between carbon sequestration and nutrient cycling ecosystem services, were assessed using C:N and C:P ratios which ranged from 14 to 21 (C:N) and 46 to 288 (C:P) in soils and ranged from 2 to 39 (C:N) and 3 to 866 (C:P) in surface water across the 23 drainage basins. Overall, net N and P mineralization dominated in soils and surface waters within the study area. These ratios pinpoint nutrient limitations in terrestrial and aqua tic ecosystems, as well as risks to water quality degradation, that might affect the functioning and integrity of natural human coupled ecosystems. Furthermore, the stoichiometric C:N:P ratio of 108:6:1 in soils and 84:9:1 in surface waters suggest that mi neralization, i.e., the net release of nutrients during the decomposition process, has dominated the Suwannee River Basin. This poses a potential threat given the fragile geology (karst topography and limestone bedrock) in the basin providing conduits for nutrient flushes into the aquifer, sand rich soils with high infiltration and percolation rates, coupled with high annual precipitation rates that promote cycling of nutrients through the soil water aquifer system and atmosphere. Human activities, includin g land use, management and tillage of soils, fertilization, conservation management,
97 prescribed fires, and others, have impacted both ecosystem services presented in this study. Those will be elaborated in more detail in C hapters 4 and 6.
98 Table 3 1 . Dat asets used and descriptive statistics of soil organic carbon (SOC) stocks at observation sites in the top soil (0 20 cm) (units in kg C m 2 ). Datasets Descriptions Number of s amples Mean Median Max Min Skewness STD 1 SEM 2 DSh sa Historic sites within the study area 169 3.7 2.7 46.6 0.4 6.4 4.5 0.3 DSh bf Historic sites within the study area and a 20 km buffer 290 3.6 2.6 46.6 0.4 6.1 3.9 0.2 DSh cal Historic calibration set randomly chosen: 70% of DSh bf 203 3.7 2.7 46.6 0.4 6.2 4.3 0.3 DSh val Historic val idation set randomly chosen: 30% of DSh bf 87 3.3 2.4 20.4 0.8 3.5 2.8 0.3 DSc sa Current sites within the study area 138 5.2 3.7 20.6 1.0 1.6 3.8 0.3 DSc bf Current sites within the study area and a 20 km buffer 234 4.6 3.3 20.6 0.9 1.8 3.4 0.2 DSc cal Cu rrent calibration set randomly chosen: 70% of DSc bf 164 4.6 3.4 17.8 0.9 1.5 3.2 0.3 DSc val Current validation set randomly chosen: 30% of DSc bf 70 4.6 3.1 20.6 1.1 2.2 3.8 0.5 1 STD refers to standard deviation. 2 SEM refers to standard error of mean.
99 T able 3 2 . Descriptive statistics of total nitrogen (TN) and total phosphorus (TP) stocks at observation sites in the top soil (0 20 cm) (units in kg m 2 ) in 2008/09. Variables Descriptions Number of s amples Mean Median Max Min Skewness STD 1 SEM 2 TN Obs ervations within the study area 138 0.31 0.21 1.06 0.00 1.49 0.23 0.02 Observations within the study area and a 20 km buffer 234 0.27 0.20 1.06 0.00 1.80 0.20 0.01 Calibration set randomly chosen: 70% of the total observations 164 0.28 0.21 1.06 0.00 1 .60 0.20 0.02 Validation set randomly chosen: 30% of the total observations 70 0.25 0.19 1.00 0.06 2.32 0.20 0.02 TP Observations within the study area 138 0.03 0.01 0.40 0.00 4.32 0.05 0.00 Observations within the study area and a 20 km buffer 234 0 .02 0.01 0.40 0.00 4.64 0.04 0.00 Calibration set randomly chosen: 70% of the total observations 164 0.03 0.01 0.40 0.00 4.66 0.05 0.00 Validation set randomly chosen: 30% of the total observations 70 0.02 0.01 0.17 0.00 2.59 0.03 0.00 1 STD refers to standard deviation. 2 SEM refers to standard error of mean.
100 Table 3 3 . Selected monthly parameters of river flow, total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) at the eight sub basins between water year 2000 and 2010. Sub ba sins Parameters No. of river gages Arithmetic m ean Geometric mean Harmonic mean Median Max Min STD 1 SEM 2 Alapaha Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 1 8,480.2 18.5 1.7 0.2 1,969.3 115.6 111.8 3,496.9 16.2 1.5 0.2 567.1 53.4 58.2 1,619.9 11.21 1.4 0.2 136.1 28.0 35.8 3,536.7 17.8 1.5 0.2 589.9 46.7 48.4 86,200.0 46.1 6.9 0.4 17,246.4 1,173.8 870.0 376.3 0.5 0.8 0.1 3.9 5.7 10.6 12,368.4 8.4 0.8 0.1 3,201.7 170.0 153. 6 1,084.8 0.7 0.1 0.0 280.8 14.9 13.5 Withlacoochee Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 2 13,584.1 10. 4 1.2 0.1 2,102.6 158.0 160.8 6,949.0 8.0 1.1 0.1 525.7 76.6 81.7 4,104.0 5.6 1.1 0.1 128.1 41.0 47.0 6,291.5 8. 5 1.3 0.1 595.4 67.0 73.6 122,000.0 32.3 2.4 0.3 33,105.2 1,387.2 1,774.5 813.5 1.5 0.5 0.1 15.0 7.9 11.0 18,579.2 6.7 0.3 0.1 4,244.4 214.3 239.8 2,018.5 0.7 0.0 0.0 443.6 23.2 25.8 Aucilla Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 1 2,863.4 27.3 1.2 0.1 769.1 38.4 16.6 479.8 22.4 1.0 0.1 105.9 4.9 2.4 0.2 15.8 0.8 0.0 0.0 0.0 0.0 523.0 25.9 1.1 0.1 143.7 7.3 3.2 28,200.0 56.1 3.2 0.2 10,813.0 333.5 193.1 0.0 2.0 0.2 0.0 0.0 0.0 0.0 5,031.6 13.9 0.6 0.0 1,668.0 66.0 31.4 444.7 1.2 0.1 0.0 147.4 5.8 2.8 Coastal Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 1 777.6 32.2 1.0 0.1 366.4 11.2 6.9 358.5 25.6 0.8 0.1 91.6 2.9 3.2 24.2 19.5 0.6 0.1 2.3 0.1 0.1 327.7 27.3 1.0 0.1 82.3 3.1 4.1 6,456.8 80.9 2.9 0.2 2,996.0 94.9 58.1 0.3 3.9 0.1 0.0 0.0 0.0 0.0 1,030.1 20.1 0.6 0.0 562.3 16.9 8.5 91.0 1.8 0.1 0.0 49.7 1.5 0.7 1 STD refers to standard deviation. 2 SEM refers to standard error of mean.
101 Table 3 3 . Continued. Sub basins Parameters No. of river gages Arithmetic m ean Geometric mean Harmonic m ean Median Max Min STD 1 SEM 2 Santa Fe Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 7 12,995.6 24.6 1.3 0.2 629.4 374.7 49.3 10,815.4 18.2 1.2 0.2 154.4 32.3 24.8 9,431.1 11.6 1.0 0.1 56.6 20.3 16.7 9,924.6 23.9 1.3 0.1 158.9 35.1 22.1 6 6,505.3 68.9 2.9 0.6 10,949.8 4,516.2 614.0 2,483.4 1.5 0.3 0.1 9.6 2.9 5.4 9,325.5 16.0 0.5 0.1 1,369.7 894.9 79.9 840.5 1.8 0.1 0.0 131.9 126.4 7.7 Upper Suwannee Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 4 6,621.2 40.9 1.7 1.5 3,757.1 110.3 192.9 2,292.9 38.3 1.6 1.1 1,013.0 32.6 97.6 861.4 30.2 1.5 0.9 214.3 9.7 53.9 2,206.8 40.5 1.7 0.9 1,215.7 34.3 100.4 68,981.2 79.0 3.9 9.6 42,587.6 1,295.4 2,977.1 149.9 7.8 0.4 0.4 15.9 1.1 7.3 11,258.8 12.7 0.6 1.5 6,675.2 196.8 329.5 985.7 1.1 0.1 0.1 584.2 17.2 29.1 Lower Suwannee Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 6 39,752.1 15.7 1.3 0.1 8,689.0 552.6 647.6 28,433.0 11.9 1.3 0.1 3,208.0 366.7 438.5 22,180.8 8.3 1.2 0.1 996.8 260.1 321.3 25,203.6 13.8 1.3 0.1 3,210.6 339.6 382.0 213,757.5 53.2 2.8 0.3 87,663.7 3,168.0 4,564.2 8,967.5 1.4 0.6 0.1 114.1 65.5 61.4 39,096.2 10.7 0.3 0.1 14,326.2 575.2 725.5 3,447.2 0.9 0.0 0.0 1,263.6 50.7 64.0 Waccasassa Flow (10 7 L mo 1 ) TOC conc. (mg L 1 ) TN conc. (mg L 1 ) TP conc. (mg L 1 ) TOC loads (10 3 kg mo 1 ) TN loads (10 3 kg mo 1 ) TP loads (10 2 kg mo 1 ) 1 292.9 11.34 0.6 0.1 57.4 2.3 2.0 215.2 8.2 0.5 0.1 17.7 1.1 1.5 175.4 6.4 0.5 0.1 9.2 0.7 1.2 169.2 6.6 0.5 0.1 11.2 0.7 1.3 1,056.6 42.3 1.4 0.1 417.9 12.3 10.1 79.3 2.2 0.1 0.0 2.8 0.2 0.4 277.1 10.3 0.3 0.0 104.9 3.3 2.1 43.3 1.6 0.1 0.0 16.4 0.5 0.3 1 STD refers to standard deviation. 2 SE M refers to standard error of mean.
102 Table 3 4 . Comparison of soil organic carbon (SOC) stocks (kg C m 2 ) in the top soil using calibration sets derived from ordinary kriging (OK) and block kriging (BK) in two modes: (a) SOC stock estimates derived from kriging across the whole FL SRB and (b) SOC stock estimates derived from kriging within the FL SRB but with exclusion of high uncertainty areas. Parameters (unit in kg C m 2 ) Mode (a): SOC stock estimates derived from kriging across the whole FL SRB Mode (b): SOC stock estimates derived from kriging with excluded high uncertainty areas Historic Current Historic Current OK BK OK BK OK BK OK BK Maximum 15.5 13.9 9.2 8.3 15.5 13.8 9.2 8.3 Minimum 1.8 1.6 2.2 2.0 1.8 1.6 2.2 2.0 Mean 3.7 3.3 4.9 4.4 3.8 3.4 4.8 4.3 Median 3.6 3.2 4.6 4.1 3.6 3.2 4.6 4.1 Standard deviation 1.3 1.2 1.3 1.2 1.3 1.2 1.3 1.2
103 Table 3 5 . Variogram parameters and validation prediction errors for SOC stocks(kg C m 2 ) using ordinary k riging (OK) and block kriging (BK) and a spherical model a in different modes: (a) soil organic carbon (SOC) stock estimates derived from kriging across the whole FL SRB and (b) SOC stock estimates derived from kriging within the FL SRB but with exclusion o f high uncertainty areas. Parameters Mode (a): SOC stock estimates derived from kriging across the whole FL SRB Mode (b): SOC stock estimates derived from kriging with excluded high uncertainty areas Historic Current Historic Current OK BK OK BK OK BK OK BK Nugget b (kg Cm 2 ) 2 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 Partial sill c (kg Cm 2 ) 2 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 Range (m) d 29319 29319 29169 29169 29319 29319 29169 29169 Lag value (m) 7400 7400 7292 7292 7400 7400 7292 7 292 Number of lags 7 7 10 10 7 7 10 10 Mean prediction error (kg Cm 2 ) 0.34 0.05 0.64 0.17 0.34 0.05 0.57 0.09 Root mean square error (kg Cm 2 ) 3.17 0.47 5.37 1.45 3.17 0.47 4.45 0.07 Number of validation samples 87 87 70 70 87 87 62 62 a Search neighborhood: minimum samples: 3, samples: 10. b Nugget and partial sill are reported on natural log transformed SOC stocks. c The partial sill is the sill minus the nugget. d Range: spatial autocorrelation derived from fitted semivariogram.
104 Table 3 6 . Soil organic carbon (SOC) estimates (0 20 cm soil profile) derived from ordinary kriging (OK) and block kriging (BK) in two modes: (a) SOC stock estimates derived from kriging across the whole FL SRB and (b) SOC stock estimates derived from kriging within the FL SRB but with exclusion of high uncertainty areas Mode Dataset Kriging technique Mean (kg C m 2 ) Standard deviation (kg C m 2 ) Area (km 2 ) SOC stocks (Tg C) (a) DSh cal a OK 3.7 1.3 19,480 72.1 BK 3.3 1.2 19,480 64.3 DSc cal b OK 4.9 1.3 19,480 95 .4 BK 4.4 1.2 19,480 85.7 (b) DSh cal OK 3.8 1.3 19,480 74.0 BK 3.4 1.2 19,480 66.2 DSc cal OK 4.8 1.3 19,480 93.5 BK 4.3 1.2 19,480 83.7 a DSh cal represents historic calibration set randomly chosen 70% of historic sites within the study area and a 20 km buffer. b DSc cal represents current calibration set randomly chosen 70% of current sites within the study area and a 20 km buffer.
105 Table 3 7 . Land cover/land use ( LC/LU ) confusion matrix from 1988/89 to 2006/08 in km 2 following the Florida Land Use and Cover Classification System (FLUCCS), Level I and median gains and losses of soil organic carbon (SOC) (g C m 2 yr 1 ) LC/LU change (km 2 ) 2006/08 UR AG RL UF WL BL TCU 1988/89 UR 205.3 (19.0) 13.6 (1.0) 6.7 (19.3) 38.4 (18.0) 10.7 (16.7) 5.1 (21 .7) 15.8 (13.0) AG 347.2 (16.7) 1,839.8 (19.0) 48.3 (19.0) 1,166.1 (17.3) 55.7 (19.0) 7.4 ( 6.0) 26.0 (13.7) RL 255.4 (19.0) 509.8 (21.0) 116.7 (19.0) 1,580.8 (21.7) 168.1 (27.0) 8.6 (14.7) 34.6 (20.3) UF 270.5 (25.0) 377.7 (22.7) 241.5 (20.7) 5,0 43.0 (31.3) 2,098.5 (43.7) 13.6 (23.0) 31.5 (26.3) WL 64.1 (19.3) 55.6 (19.3) 57.9 (18.0) 1,245.1 (27.7) 3,284.7 (34.0) 13.2 (78.0) 8.1 (21.3) BL 17.3 (38.7) 16.2 (42.0) 3.4 (44.0) 37.4 (49.7) 5.9 (67.0) 0.2 ( 8.0) 3.0 (46.0) TCU 2.5 (21.3) 0.7 ( 6.7) 0.4 (22.3) 3.5 (19.0) 1.2 (27.3) 0.03 (2.0) 25.1 (18.7) High magnitude of land cover/land use change with areas greater than 1,000 km 2 . Median values of soil organic carbon (SOC) stocks change are shown in parentheses. UR (1000 code in FLUCCS) den otes urban and built up (i.e., residential, commercial and services, industrial, extractive, institutional, recreational, open land). AG (2000 code in FLUCCS) denotes agriculture (i.e., cropland and pastureland, improved pastures, unimproved pastures, woo dland pastures, row crops, field crops, tree crops, feeding operations, nurseries and vineyards, specialty farms, other agriculture). RL (3000 code in FLUCCS) denotes rangeland (i.e., herbaceous, shrub and brushland, coastal scrub, mixed rangeland). UF ( 4000 code in FLUCCS) denotes upland forest (i.e., pineland, hardwood forest, xeric and mesic forest communities). WL (6000 code in FLUCCS) denotes wetland (i.e., wetland hardwood forest, wetland coniferous forest, wetland forested mixed, vegetated non for ested wetlands). BL (7000 code in FLUCCS) denotes barren land where the land has very little or no vegetation and limited potential to support vegetation communities (i.e., bare soil or rock). TCU (8000 code in FLUCCS) denotes transportation, communicati on, and utilities.
106 Table 3 8 . Drainage areas used in the hydrology analysis according the U.S. Geological Survey (USGS) Hydrologic Unit code and the Suwannee River Water Management District (SRWMD). Drainage Area ID used in the study USGS reference Sta tion ID Agency Sub basins Drainage area (km 2 ) Alph.00 With.01 With.02 Aucl.00 Cost.00 Sntf.01 Sntf.02 Sntf.03 Sntf.04 Sntf.05 Sntf.06 Sntf.07 UpSuwn.01 UpSuwn.02 UpSuwn.03 UpSuwn.04 LwSuwn.01 LwSuwn.02 LwSuwn.03 LwSuwn.04 LwSuwn.05 LwSuwn.06 Wacs.00 02317 620 02319000 02319394 02326500 02326000 02320700 02321000 02321500 02321898 02321975 02322500 02322800 02315000 02315500 02315520 02315550 02319500 02319800 02320000 02320500 02323000 02323500 02313530 ALA010C1 WIT020C1 WIT040C1 AUC050C1 ECN010C1 SFR010C1 NEW009C1 SFR030C1 SFR040C1 SFR050C1 SFR060C1 SFR070C1 SUW010C1 SUW040C1 SWF010C1 SUW070C1 SUW100C1 SUW120C1 SUW130C1 SUW140C1 SUW150C1 SUW160C1 WAC006C1 USGS USGS USGS SRWMD USGS SRWMD USGS USGS SRWMD SRWMD USGS USGS SRWMD USGS SRWMD SRWMD USGS USGS USGS U SGS USGS USGS SRWMD Alapaha Withlacoochee Withlacoochee Aucilla Coastal Santa Fe Santa Fe Santa Fe Santa Fe Santa Fe Santa Fe Santa Fe Upper Suwannee Upper Suwannee Upper Suwannee Upper Suwannee Lower Suwannee Lower Suwannee Lower Suwannee Lower Suwannee L ower Suwannee Lower Suwannee Waccasassa 4,201 1,729 2,421 1,872 473 171 465 1,384 1,973 2,339 2,738 3,794 4,714 5,562 5,667 6,015 13,418 14,046 14,466 15,553 19,652 20,280 467 USGS denotes the United States of Geological Survey. SRWMD denotes the Suwannee River Water Management District.
107 Table 3 9 . Mann Kendall trend analysis for monthly unit area loads of total organic carbon (TOC) (kg C km 2 ) by drainage area in harmonic mean between 2000 and 2010. Drainag e Area Area (km 2 ) TOC loads (kg C km 2 ) Months Kendall score Variance p value (two sided) Trend Seasonal Alph.00 With.01 With.02 Aucl.00 Cost.00 Sntf.01 Sntf.02 Sntf.03 Sntf.04 Sntf.05 Sntf.06 Sntf.07 UpSuwn.01 UpSuwn.02 UpSuwn.03 UpSuwn.04 L wSuwn.01 LwSuwn.02 LwSuwn.03 LwSuwn.04 LwSuwn.05 LwSuwn.06 Wacs.00 4,201 1,729 2,421 1,872 473 171 465 1,384 1,973 2,339 2,738 3,794 4,714 5,562 5,667 6,015 13,418 14,046 14,466 15,553 19,652 20,280 467 32.4 81 .2 47.9 0.0 4.9 1.6 5.7 7.8 0.2 0.2 38.3 73.0 34.3 55.1 4.5 60.4 87.6 87.2 74.0 87.6 52.4 5.8 19.6 130 65 118 127 128 48 63 127 112 130 132 95 132 131 120 129 129 132 132 130 119 131 41 9 18 14 143 614 4 74 237 1,375 486 671 374 317 1,204 1,263 1,070 1,124 70 470 530 453 111 553 254 1,907 950 1,434 1,787 235,712 12,659 28,427 230,251 158,163 246,892 258,419 96,742 258,419 252,612 194,367 241,25 9 1,860 258,419 258,419 246,892 189,567 252,612 7,927 0.84 0.56 0.71 0.0007 0.21 0.00003 0.16 0.004 0.22 0.18 0.46 0.31 0.02 0.01 0.02 0.02 0.1 0.36 0.3 0.36 0.8 0.27 0.004 Insignificant Insignificant Insignificant Increasing Insignificant Increasi ng Insignificant Increasing Insignificant Insignificant Insignificant Insignificant Increasing Increasing Increasing Increasing Insignificant Insignificant Insignificant Insignificant Insignificant Insignificant Increasing Yes Yes Yes Yes No No No No No No No No No No No No Yes Yes No No No No No
108 Table 3 10 . Mann Kendall trend analysis for monthly unit are a loads of total nitrogen (TN) (kg N km 2 ) by drainage area in harmonic mean between 2000 and 2010. Drainage Area Area (km 2 ) TN loads (kg N km 2 ) Months Kendall score Variance p value (two sided) Trend Seasonal Alph.00 With.01 With.02 Aucl.00 Cost.00 Sntf.01 Sntf.02 Sntf.03 Sntf.04 Sntf.05 Sntf.06 Sntf.07 UpSuwn.01 UpSuwn.02 UpSuwn.03 UpSuwn.04 LwSuwn.01 LwSuwn.02 LwSuwn.03 LwSuwn.04 LwSuwn .05 LwSuwn.06 Wacs.00 4,201 1,729 2,421 1,872 473 171 465 1,384 1,973 2,339 2,738 3,794 4,714 5,562 5,667 6,015 13,418 14,046 14,466 15,553 19,652 20,280 467 6.7 9.7 27.0 0.0 0.2 1.6 0.4 0.4 0.0 0.0 21.9 21.5 1.3 1.7 0.7 3.2 11.4 13.0 13.3 20.3 15.9 20.0 1.5 130 65 118 127 128 48 63 127 112 130 132 95 132 131 120 129 129 132 132 130 119 131 41 29 42 40 42 1,067 484 267 1,393 616 1,101 1,180 63 3 1,462 1,413 1,000 1,468 1,467 1,030 1,338 1,379 573 1,451 180 1,907 950 1,434 1,860 235,711 12,658 28,427 230,251 145,842 252,612 258,419 96,742 258,419 252,612 194,367 241,259 246,892 258,419 258,419 252,612 189,567 258,418 7,927 0.51 0.17 0.29 0.33 0.03 0.00002 0.11 0.004 0.11 0.03 0.02 0.04 0.004 0.005 0.02 0.003 0.003 0.04 0.01 0.01 0.19 0.004 0.04 Insignificant Insignificant Insignificant Insignificant Increasing Increasing Insignificant Increasing Insignificant Incre asing Increasing Increasing Increasing Increasing Increasing Increasing Increasing Increasing Increasing Increasing Insignificant Increasing Increasing Yes Yes Yes Yes No No No No No No No No No No No No No No No No No No No
109 Table 3 11 . Mann Kendall tre nd analysis for monthly unit area loads of total phosphorus (TP) (kg P km 2 ) by drainage area in harmonic mean between 2000 and 2010. Drainage Area Area (km 2 ) TP loads (kg P km 2 ) Months Kendall score Variance p value (two sided) Trend Seasona l Alph.00 With.01 With.02 Aucl.00 Cost.00 Sntf.01 Sntf.02 Sntf.03 Sntf.04 Sntf.05 Sntf.06 Sntf.07 UpSuwn.01 UpSuwn.02 UpSuwn.03 UpSuwn.04 LwSuwn.01 LwSuwn.02 LwSuwn.03 LwSuwn.04 LwSuwn.05 LwSuwn.06 Wacs.00 4,201 1,729 2,421 1,872 473 171 465 1,384 1,973 2,339 2,738 3,794 4,714 5,562 5,667 6,015 13,418 14,046 14,466 15,553 19,652 20,280 467 0.9 2.4 2.1 0.0 0.0 0.1 0.1 0.2 0.0 0.0 1.5 1.9 0.2 0.4 1.3 1.9 2.0 2.3 2.2 2.1 1.6 1.9 0.2 130 65 118 127 128 48 63 127 112 130 132 95 132 131 120 129 129 132 132 130 119 131 41 63 34 36 154 614 392 201 1,293 507 967 1,085 691 1,920 1,607 598 1,238 935 1,042 1,088 1,037 203 1,054 103 1,907 950 1,434 1, 820 235,712 12,659 28,427 230,251 145,841 246,892 258,418 96,742 258,419 252,612 194,367 241,259 246,892 258,419 258,419 252,612 189,567 258,419 7,926 0.15 0.27 0.34 0.0003 0.21 0.0005 0.24 0.01 0.19 0.05 0.03 0.03 0.0002 0.001 0.18 0.01 0.06 0.0 4 0.03 0.04 0.64 0.04 0.25 Insignificant Insignificant Insignificant Increasing Insignificant Increasing Insignificant Increasing Insignificant Increasing Increasing Increasing Increasing Increasing Insignificant Increasing Insignificant Increasing Increas ing Increasing Insignificant Increasing Insignificant Yes No Yes Yes No No No No No No No No No No No No No No No No No No No
110 Table 3 12 . Interpolation parameters and validation results derived by block kriging for soil nutrient properties lnN (log tran sformed total nitrogen in kg C m 2 ) and lnP (log transformed total phosphorus in kg C m 2 ) in the top soil (0 20 cm) for the current soil dataset in 2008/2009. Parameter lnN lnP Semivariogram model a Exponential Exponential Nugget b (kg C m 2 ) 2 0.20 0.79 Partial b sill (kg C m 2 ) 2 0.21 1.09 Range c (m) 33391 30000 Lag value (m) 7292 7292 Number of lags 10 10 Mean prediction error (kg C m 2 ) 0.01 0.02 Root mean square error (kg C m 2 ) 0.001 0.002 Number of validation samples 70 70 a Search neighborh ood: minimum samples: 3, samples: 10. b Nugget and partial sill are reported on natural log transformed TN and TP stocks. c Range: spatial autocorrelation derived from fitted semivariogram.
111 Table 3 1 3 . Spatial distribution o f mean carbon to nitrogen (C:N) and carbon to phosphorus (C:P) in soil and surface water for each of the drainage areas in the Suwannee River Basin . Drainage Area Soil (based on kg C m 2 ) Surface water (based on kg C m 2 ) C:N C:P C:N C:P Alph.00 With.01 With.02 Aucl.00 Cost.00 Sntf.01 Sntf.02 Sntf.03 Sntf.04 Sntf.05 Sntf.06 Sntf.07 UpSuwn.01 UpSuwn.02 UpSuwn.03 UpSuwn.04 LwSuwn.01 LwSuwn.02 LwSuwn.03 LwSuwn.04 LwSuwn.05 LwSuwn.06 Wacs.00 18 15 14 16 18 19 18 17 16 15 15 16 20 20 21 21 18 18 18 18 17 17 16 166 88 90 197 288 46 143 62 76 64 61 66 239 234 232 150 100 100 103 102 80 100 152 8 9 2 28 27 4 28 25 19 17 3 2 39 39 5 28 11 13 12 9 7 7 17 58 45 21 866 185 574 285 167 99 74 35 32 498 316 3 82 90 72 77 93 73 75 95 Average ratios 17 108 9 84 Average C:N:P 108:6:1 84:9:1 Drainage basins with incomplete soil data available.
112 Figure 3 1 . Spatial distribution of historic soil organic carbon (SOC) observations (n = 290) and soil organic carbon stocks (kg C m 2 ) in the top soil (0 20 cm). [Study area adapted from Suwannee River Water Management District (SRWMD), 1999. SRWMD Boundary. Map scale 1:24,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ; Historic SOC data adapted from a subset from the Florida Soil Characterization Database, 2012. Accessible through http://flsoils.ifas.ufl.edu or http://TerraC.ifas.ufl.edu].
113 Figure 3 2 . Spatial distribution of current soil organic carbon (SOC) obs ervations (n = 234) and soil organic carbon stocks (kg C m 2 ) in the top soil (0 20 cm). [Study area adapted from Suwannee River Water Management District (SRWMD), 1999. SRWMD Boundary. Map scale 1:24,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ; Current SOC data adapted from a subset from the Rapid Assessment and Modeling of Changes in Soil Carbon Storage and Turnover in a Southern Landscape. 2011 ] .
114 Figure 3 3 . Drainage area delineation from a digital elevation model and topographic attributes. Identification list corresponds to the 23 water monitoring stations within the Suwannee River Basin. [Elevation adapted from National Elevation Dataset (NED), 2009. Natio nal Elevation Dataset. Map scale 1:250,000. Accessible through http://ned.usgs.gov/ ; Water stations adapted from Suwannee River Water Management District (SRWMD), 1990s to 2013. Accessible through http://www.mysuwanneeriver.org/rivers.htm ; Streams computed from a digital elevation model and topographic attributes].
115 (a) (b) (c) (d) Figure 3 4 . Kriged estimated of historic soil organic carbon (SOC) stocks (kg C m 2 ) in the 20 cm depth derived from historic and current calibration datasets within th e core study area. Historic SOC estimates were derived by (a) ordinary kriging and (b) block kriging. Current SOC estimates were derived by (c) ordinary kriging and (d) block kriging.
116 (a) (b) (c) (d) Figure 3 5 . Variance maps derived from soil organic carbon (SOC) (kg C m 2 ) estimates classified by quantiles. Series of variance maps showing (a) ordinary kriging and (b) block kriging derived from historic SOC stock estimates, (c) ordinary kriging and (d) block kriging derived from current SOC st ock estimates.
11 7 (a) (b) Figure 3 6 . Prediction maps show soil organic carbon (SOC) gains and losses (kg C m 2 ) after eliminating high uncertainties. SOC change derived by subtracting historic prediction maps from current prediction maps using (a) ord inary kriging and (b) block kriging.
118 Figure 3 7 . Soil organic carbon (SOC) gains and losses (g C m 2 yr 1 ) derived from historic and current collocated sites within 200 m radius. .
119 (a) (b) Figure 3 8 . Spatial distribution patterns of (a) total ni trogen (TN) and (b) total phosphorus (TP) stocks in soil.
120 (a) (b) Figure 3 9 . Spatial distribution patterns of (a) soil organic carbon (SOC) to total nitrogen (C:N) and (b) soil organic carbon to total phosphorus (C:P) in the topsoil (0 20 cm) des cribing interrelationships between carbon and nutrient variability. The SOC and nutrient estimates to calculate ratios were derived from block kriging.
121 (a) (b) Figure 3 10 . Soil organic carbon (SOC) stock (kg C m 2 ) variability in (a) nitrogen limited and enriched areas and (b) phosphorus limited and enriched areas.
122 (a) (b) Figure 3 11 . Fertilizer use by county based on (a) total nitrogen and (b) total phosphorus between 1945 and 2010. [Adapted from Florida Department of Agriculture and Co nsumer Services (FDACS), 2014. Total Fertilizer and Nutrients by County (1945 to 2010). Accessible through http://www.freshfromflorida.com/Divisions Offices/Agricultural Environmental Services/Agriculture Industry/Fertilizer Manufacturers/Fertilizer Consum ption Tonnage Data ]
123 (a) (b) (c) Figure 3 12 . Time series plots of monthly mean flows, concentration and loads for (a) total organic carbon (TOC), (b) total nitrogen (TN), and (c) total phosphorus (TP) between 2000 and 2010 wateryears.
124 Figure 3 1 3 . Monthly mean precipitation and mean temperature with smoothing curves between 2000 and 2010 at LwSuwn.06 drainage area. The trend lines in the middle of the two graphs indicate no significant change in mean precipitation and mean temperature derived b y Mann Kendall test .
125 CHAPTER 4 ESTIMATION OF THE TE RRESTRIAL CARBON BUD GET Overview Carbon sequestration has become an important policy option to mitigate the increasing atmospheric greenhouse gases (GHG). The terrestrial biosphere can sequester signif icant amounts of anthropogenic carbon dioxide (CO 2 ) by the natural carbon uptake process through plant biomass and soils. Forest carbon sequestration can be enhanced by afforestation, reforestation, and agroforestry (Bellassen and Luyssaert, 2014; Gorte, 2009; Haim et al., 2014; Lin and Lin, 2013) . On agricultural lands, soil carbon sequestration can be enhanced by a variety of management practices, including crop residue management, erosion control, extended crop rotation, minimized tillage, and improved nutrient management (Govaerts et al., 2009; West and Marland, 2003, West and Marland, 2002) . Globally, terre strial carbon sequestration could accumulate more than 0.5 Pg C yr 1 (1 Pg = 10 15 kg) by 2040, mitigating about 6 to 23 percent of the emissions, and sequestering over 40 Pg C by the end of the 21st century to stabilize the atmospheric condition (Thomson et al., 2008) . Since feedback of the carbon cycle between the terrestrial system and climate change could impact the dynamics of ecosystem function and service, scientific and technical attempts to understand and develop progress toward options for creditable carbon accumulation have emerged (Hovi et al., 2003; Lecocq and Ambrosi, 2007) . This includes understandin g the factors controlling carbon allocation mechanisms in vegetation and soils. Hilgard (1906) , Dokuchaiev (Glinka, 1907) , and Jenny (1941) were some of the first researchers to describe the relationship between soil forming factors and soil
126 properties in so il formation. The CLORTP model (CL: climate, O: organism, R: relief, T: topography, P: parent material) is a well known conceptual framework that builds on the fundamental knowledge of soil formation. McBratney et al. (2003) formulated the empirica l SCORPAN approach by framing relationships between soils and factors (S: soil properties, C: climate, O: organism/vegetation, R: relief, P: parent material, A: age/time, N: space). Besides the environmental landscape factors, anthropogenic activities are considered as additional explicit feature influencing soil properties and climate regulation. The STEP AWBH model was developed in response to the modern world to support spatially and temporally explicit digital soil modeling (Grunwald et al., 2011b) . This conceptual model embraces soil (S), topography (T), ecology (E), parent material (P), atmosphere (A), water (W), biota (B), and human (H) factors together to account for the effects on s oil genesis. Both the SCORPAN and the STEP AWBH models are flexible enough to be implemented with various geostatistical techniques, and stochastic or ensemble learning methods. For example, the SCORPAN factors can map continuous soil depth (Malone et al., 2009) and can predict soi l organic carbon (SOC) at the regional scale using the scorpan kriging approach (Ungaro et a l., 2010) . As environmental covariates, the STEP AWBH factors can predict total carbon stocks using various data mining techniques, such as, regression trees, bagged trees, random forest, and support vector machines (Xiong et al., 2012). Here we focus o n the major carbon pools in the Suwannee River Basin in Florida (FL SRB) which are composed of below ground (soil) and above ground (biomass) carbon. This chapter adopts the STEP AWBH model and applies it to assess the
127 attainable capacity of a terrestrial ecosystem to store carbon. Theoretically, there are three levels of SOC sequestration: potential SOC (SOC potential ), attainable SOC (SOC attain ), and actual SOC (SOC actual ) (Baldock et al., 2007; Ingram and Fernandes, 2001) . In the terrestrial system, the potential carbon sequestration is defined by factors that set the physico chemical maximum li mit to storage (e.g., soil type, depth to bedrock, and mineralogy). The attainable carbon sequestration is set by factors that limit the inputs of carbon to the system (e.g., vegetation or land use, fertilization for biomass production, net primary product ion), which can be modified by humans (e.g., implementation of best management practices, reduction of burning of fossil fuels for energy consumption). The actual carbon storage is controlled by factors that modulate carbon storage or loss (e.g., drainage, tillage, management, respiration, or photosynthesis) and depends on a combination of internal environmental landscape factors, past and current anthropogenic forcings, and socio economic drivers. Carbon accumulation in the biosphere is finite (Johnston et al., 2009; Powlson et al., 2011) . Each soil has a unique carbon sequestration level dictated by soil properties (e.g., texture, mineralogy, bulk density, and depth) (Ingram and Fernandes, 2001; Six et al., 2002; Stewart et al., 2009, 2008) . Angers et al. (2011) and Hassink (1997) suggest ed the concept of SOC saturation that asserts that the quantity of stable SOC in soils is finite and determined by the amount of fine particles (clay and fine silt). The SOC potential is the difference between the theoretical SOC saturation value and measured amount of SOC. Despite the difficulties to determine a SOC saturation level , it has been used in various geographic regions and different situations (Carter et al., 2003; Liang et al., 2009; Six et al., 2002; Stewart et al., 2007; Zhao et al., 2006) . SOC
128 sequestration is limited by the content of cla y and fine silt fractions (Has sink, 1997) . Achieving carbon saturated conditions in sandy soils is easier than in fined textured soils; nevertheless, saturation may be observed in situations with very high carbon input or slow decomposition rates (Angers et al., 2011) . Soil depth also play an important role in explaining the SOC saturation. Hoyle et al. (2013) indicated that deeper soil (0 30 cm) had a larger attainable SOC sequestration when compared to the soil surface (0 10 cm), where they found that the upper 10 cm was largely saturated with carbon. The density of SOC is the concentration of SOC expressed per volume of soil in units of g C m 3 . It has been applied by Edmondson et al. (2012) and others to characterize soil carbon storage, similarly to SOC stocks expressed in units of g C m 2 per soil depth which expresses the sample support on a volume basis. In the FL SRB, dominant soils are sand rich which inherently does not promote SOC accretion. Thus, the actual and potential soil carbon content is expected to be similar. On the other hand, soils formed in aquic conditions are carbon rich and in the FL SRB such hydr ologic conditions are prominent. Importantly, the accretion of SOC in aquic and organic soils occurs mainly by increasing the levels of soil depth and is typically expressed in C cm yr 1 . O rganic soils , like mineral soils, may also further increase the den sity of SOC conjuring to the SOC potential level if conditions are favorable, e.g., if the amount of biotic residue that provides C input into soils is high and aquic conditions prevail leading to slow decomposition rates pr omulgating SOC accumulation. Vas ques et al. (2012) poin ted out that in Florida actual ecological landscape conditions (climate, hydrology, land use/land cover) are extremely complex forming a mosaic of carbon rich and carbon poor soils that are interspersed with each other.
129 The notion that numerous ecosystem functions, such as nutrient and water holding capacity, filtering of contaminant, mitigation of GHG (Post et al., 2009; Zvomuya et al., 2008) are optimized if both soil carbon and biomass production are at maximum levels . . During the second commitment period of the United Nations Framewo rk Convention on Climate Change (UNFCCC; 2012 2022), the consideration of reduced emissions from deforestation and forest degradation (REDD) by maximizing carbon storage and conserving ecosystems are included (UN REDD, 2014) . From an ecosystem perspective this provided by a region are optimized when the natural carbon capital is maximized across a landscape. However, this notion is constraint by the fact that site specific constraints, specifically human induced management and land use, impose limits to the attainable and actual carbon stocks. From socio economic and ecological perspectives , diversity in natural terrestrial carbon capital across a landscape is desirable to support bio , pedo , hydro diver sity encouraging the goal to maximize carbon accumulation in a landscape to reach the SOC potential . S ome concerns about maximizing rates of carbon sequestration have been addressed. Policies that aim to maximize carbon sequestration alone do not necessari ly lead to an agreement to maximize co benefits due to the complex relationship between carbon sequestration and biodiversity (Lin et al., 2013; Pichancourt et al., 2014; Putz and Redford, 2009) . The interdependence between biota and soil based carbon has been uptake documented in numerous studies (BÃ©langer and Pinno, 2008; Ravindran and Yang, n.d.; Wang et al., 2011; White II et al., 2009; Wu et al., 2009; Xiong et al., 2014) . Together they form the terrestrial carbon pool, which at the global scale accounts
130 approximately for 7% of the total carbon budget (Lal, 2008) . Like sequestration in soils, there are three levels that can be distinguished: potential (TerrC potential ), attainable (TerrC attain ), and actual terrestrial C (TerrC actual ). The potential terrestrial carbon is expressed as the maximum amount of car bon that can be stored in biomass independent of geographic location and environmental conditions, whereas the attainable terrestrial carbon describes the carbon storage that could be attained through land use management, vegetation species selection, clim ate adaptation, or other at a given geographic setting. And the actual terrestrial carbon storage is defined by the sum of below (soil) and above ground carbon stored in the system. We postulate that human induced management strategies combined with envir onmental fluctuations of climate, land use, and hydrology contribute to TerrC attain that is constraint by site specific soil landscape conditions. Importantly, which of these coupled human environmental combinations achieve the maximum attainable carbon st orage in terrestrial systems is usually not known. This will be investigated in this study. Two objectives are addressed: i) assess the spatially explicit relationships between SOC and environmental factors (soils, topography, ecology, parent material, at mosphere, water, biota, and human) and ii) assess the environmental value of actual and attainable terrestrial carbon capital (TerrC actual , TerrC attain ) across the FL SRB consisting of below ground (soil) and above ground (biomass) carbon. Materials and M ethods Above Ground and Below Ground Carbon Data A total of 234 soil samples in the topsoil (0 20 cm) were collected between 2008 and 2009 across the FL SRB and its buffer area (20 km around the FL SRB) based on the random design stratified by land cover/ land use and soil suborder classes. Each soil
131 sampling location was georeferenced with a differential global positioning system. Total carbon (TC) was measured by a combustion (Shimadzu TOC V/SSM 5000) gas analyzer, while inorganic carbon (IC) was analyze d by separate procedures. The CO 2 evolution was used to measure TC and IC. To measure TC, ball milled soil samples were combusted at 900 o C. To measure IC, ball milled soil samples were treated with 42.5 percent phosphoric acid (H 3 PO) and then combusted at 200 o C. The SOC concentration was calculated by subtracting the IC concentration from the obtained TC concentration (mg kg 1 ). The SOC concentration was converted to stock units (kg C m 2 ) using measured bulk density and soil depth. For above ground carbon assessment, we derived data from the Landfire project which provided a high resolution map of year 2000 baseline estimates of above ground biomass (National Biomass Carbon Data, NBCD 2000 Version 2) (Kellnorfer et al., 2013) . The above ground live and dry biomass in kg C m 2 was extracted on soil sampling loc ations. Environmental and Anthropogenic Covariates A comprehensive set of 172 environmental and human covariates representing the STEP AWBH factors was compiled from multiple data sources using ArcGIS software (Environmental Systems Research Institute, ES RI, Redlands, CA). Table 4 1 presents assembled spatial data layers for the characterization of landscape and socio economic properties used in this study. Predictors included 31 categorical and 141 continuous data types. Categorical data included soil tax onomic properties, drainage classes, runoff classes, parent materials, land cover/land use types, and best management practice implementation, whereas continuous data consisted of climate properties, organic matter content, topography characteristics, biot ic variables, and demographic and socio economic data.
132 Modeling the Relationships between SOC and STEP AWBH Factors This study is embedded in the STEP AWBH modeling concept which explicitly combines spatially and temporally explicit environmental and hum an variables that model the evolution of the soil ecosystem (Grunwald et al., 2011b) . The equation is outlined in Equation 4 1. ( 4 1 ) where, is the target soil realization, represents ancillary soil properties, represents topographic properties, represents ecological properties, represents the parent material and geologic properties, represe nts atmospheric properties, represents water properties, represents biotic properties, and represents human induced forcings, Is the number of predictors, is a pixel with size x (width = length = x) at a site specific on land, is the current time, is the time to with time steps is soil depth. The spatially explicit STEP factors capture the relative stable soil forming factors within a human time frame, while the AWBH fact ors account for time dependent variation (Thompson et al. , 2012) . A core concept of the STEP AWBH model is to data mine soil and environmental human covariates. Tree based modeling (one of the data machine learning methods) has the advantage of dealing with highly complex data, detecting non linear relationsh ips, handling missing data, and deriving high order interactions among data al., 2002) . In regression trees, the response variable is numerical, and predictors can be categorical and numer ic. In the late 1990s, ensemble learning methods that generate
133 multiple classifiers and aggregate their results, such as boosting and bagging classification trees, have drawn attention in various scientific areas (Breiman, 1996; Schapire et al., 1998) . Successive trees in the boosting method give extra weight to points that are wrongly predicted by previous predictors. A weighted vote is chosen for prediction in the last step (Liaw and Wiener, 2002) . In contrast to boosting, successive trees used in the bagging technique independently rely on earlier trees using the bootstrap sample. A simple majorit y vote is chosen from the prediction (Liaw and Wiener, 2002) . Breiman (2001) developed random forests (RF) by adding an additional randomness layer to the bagging method. In RF, each tree is built by randomly and repeatedly selecting predictors . When using splitting, the node in RF no longer represents the best split among all variables, but rather the best split among a subset of predictor variables. Random forests combine several individual trees and the outputs are the averaged result (Strobl et al., 2007) . In soil model development, the RF method has shown an increase in pre diction accuracy and robustness as compared to traditional interpolation methods (Kim et al., 2012) regression trees, bagged trees, boosted trees, and suppo rt vector machine (Xiong et al., 2012) . In this study, the RF regression method was used to identify the most powerful environmental predictive factor s to model SOC using the RandomForest package in R.14.2. (R development Core Team, 2011). The model was assessed using the validation method. The dataset (n = 234) was randomly divided into a calibration set (70%, n = 164) and a validation set (30%, n = 70 ). The coefficient of determination (R 2 ) was reported to assess the fit between observed and predicted values. The root mean square error (RMSE) in Equation 3 8
134 (Webster and Oliver, 2001b) and the residual prediction deviation (RP D) in Equation 4 2 (Islam et al., 2003) were reported for error assessment of the RF model. ( 4 2 ) where, is the root mean square error, is the standard deviation of the calibration or validation set, and is the number of observed or predicted values. It should be noted that the term RM SE is sometimes used interchangeably with the term root mean square deviation (RMSD). Assessing Terrestrial Carbon Stocks The TerrC actual stocks were derived by the summation of the measured SOC stocks in the top 20 cm of the soil and above ground biomass from the National Biomass Carbon Dataset (NBCD) database (kg C m 2 ) based on soil sampling locations (n = 234). A total of 43 predictors were used to predict the observed TerrC actual stocks using the RF model. To assess the TerrC attain stocks, we posit th at the STEP factor are not expected to substantially change within a human time frame (one generation), whereas AWBH factors are likely to be variable in time and may increase or decrease. The most powerful predictors in the first quantile of all STEP AWBH variables were selected (n = 43) and used to predict the TerrC attain stocks. In the model, the STEP factors were kept constant. The AWBH factors were varied by Â±10, Â±20, and Â±30 percent, respectively, by keeping AWBH factors constant, except for one of t hem that was increased/decreased one by one with the respective percentage value, until all factor combinations were assessed within upper and lower bounds. A simulated annealing was used to implement the approach at the 234 sites. The constraint was that AWBH factors were increased
135 and/or decreased up to the reasonable minimum and maximum amounts representing lower and upper bounds. The factor combination which amounts to the highest terrestrial carbon stocks simulated for the FL SRB at the 234 sites was p ostulated to equal the attainable terrestrial carbon stocks that could be obtained based on dynamic AWBH variables and relatively stable STEP conditions. Simulated annealing (SA) is one of the stochastic optimization methods of solving the global optimiza tion as introduced by Kirkpatrick et al. (1983) . This method is suitable for non parametric problems of a large variety of variables. The heart of the SA technique is an analogy with thermodynamics of the physical process of solid annealing. Starting from an initial state at high temperatures, the liquid molecules move freely with respect to one another. If the liquid is cooled slowly, it loses thermal mobility and forms a solid at the point of time it reaches a state of minimum energy for the system (Press et al., 1992) . The slow cooling procedure allows sufficient time for the redistribution of the atoms. From the initial situation with energy level , a perturbation in the state is introduced in the next level of energy (Aerts and Heuvelink, 2002) . The probability of accepting change is explained by the Boltzm ann distribution (Kirkpatrick et al., 1983) . (4 3) where, p is the probability of transition from lower energy level to higher energy level . Parameter is the Boltzman n constant (1.38 x 10 16 ergs/Kelvin) and is the temperature at the current thermal state (Press et al., 1992 ; Zomaya and Kazman, 2010) . In this case, in Equation 4 3 is computed and compared to , a random number from [0,1]. If is larger than , the new point is accepted and the algorithm
136 moves to the next level. If is smaller than , the new point is rejected (Goffe et al., 1994) . The gradual decrease of temperature is a crucial element of the SA procedure. The effective cooling rate should fall in between 0 and 1, typica lly 0.80 0.99 (Goffe et al., 1994; Wah et al., 2007) . For example, after times through the loops, the temperature reduces 80 percent of the temperature from the previous stage. The new temperature is given by (4 4) where, is the current temperature and 1 denotes the temperature at one step before the current state. Parameter r is a cooling rate (0.80). Figure 4 1 displays a workflow of the SOC stock predictions and quan tification approach for the TerrC attain stocks. The software used to operate the model was R 3.0.3. An initial temperature used in this study was set at 10 , 000 with the total of 60 iterations . The system time for execution of the code was about two minu tes. A minimum of 100 runs was test and calculated a mean value to set the seed of the model. To characterize the spatial distribution of terrestrial carbon stocks, the ordinary kriging technique was used to interpolate the 234 point observations of Terr C actual and TerrC attain predicted mean values to basin scale. The spatial autocorrelation of site specific TerrC attain obtained by different scenarios of change (i.e., 10, 20, and 30 percentage) and TerrC actual t in ArcGIS. Equation 4 (Anselin, 1995) . ( 4 5 )
137 where, 1 and +1, is the z score value for the terrestrial carbon stocks at location , and is the z score value for the terrestrial carbon stocks at neighboring observations . Spatial weights indicating the strength of connection between the paired terrestrial carbon locations of and are represented by . Results and Discussion Variable Importance and Spatial Variation in Soil Organic Carbon Descriptive statistics of SOC stocks are presented in Table 3 1. Figure 3 2 shows the spatial distribution of terrestrial carbon. The variables that emerged in the first quantile of the RF model showing predictive power were as foll ow: biota > soil > topography. Minor variables represented water, atmospheric properties, and parent material (Table 4 2). Xiong et al. (2014) had similar findings from a study in Florida, in which they found that the most controlling factor to be associated with SOC stocks are vegetation and soil water gradient. In this study, vegetation related variables such as vegetation type s, land cover/land use (LC/LU) classes, and normalized difference demonstrated strong interaction to SOC stocks. The relationship between the variation of SOC and different vegeta tion types or LC/LU change have been observed in many previous studies (Liao and Long, 2011; Oueslati et al., 2013; Ross et al., 2013; Yang et al., 2010) . Soil taxonomic variables, such as soil great group, suborder, and subgroup, were highly interrelated with SOC stocks as shown in the top ten explanatory variables (Table 4 2). Drainage classes were anticipated to have a control over the gains and losses of SOC across the study area (see Fi gure 2 7, and Figure 3 6). Areas with poorly
138 to very poorly drained soils tended to accumulate SOC content, whereas areas with well drained to excessively drained soils had a tendency to have net losses of SOC. These results suggest that topographic and wa ter related variables play a dominant role to infer on SOC storage in the basin. Soil slope showed up in the 11 th rank in the RF model and expressed negative correlation with SOC variation (Table 4 2). This indicated that the downslope positions were corre lated with high SOC, and vice versa, upslope positions showed the opposite behavior. Even though topographic terrain in Florida is relatively flat (0 5% slopes), the steeper slopes are found in northern regions of Florida where the FL SRB is located. One explanation of the impact of slopes on SOC is that the capacity to retain soil water in footslope positions supporting plant growth and promoting the slow decomposition of organic matter. Additionally, the surface soil materials from the upper landscape po sitions are transported to downslope resulting in higher SOC content in lower positions (Wilding et al., 1983) . Distance from stre ams or open water, available water capacity at 25 cm, hydrologic group, ponding frequency class, and soil runoff potential were water variables (W) in the model that demonstrated strong connectivity with SOC stocks. The effect of wetness in soil is conside red to be a major factor that controls vegetation growth and the decomposition process that are closely interlinked with SOC gain. The net primary production delivers input of carbon as above ground biomass into the soils and wetness in soil regulates dec omposition rates. These results were similar to those of Davison and Janssens (2006) , Jones et al. (2005) , Vasques et al. (2012) , and Xiong et al. (2014 ) .
139 In this study, parent material was found to have high influence on SOC, while the ecology variables (E factor) were not strongly associated with SOC. In comparison, Xiong et al. (2014) found that the ecoregion was one of the relevant variables that controlled SOC at regional scale in the State of Florida. Parent material factors, particularly surficial geology, consistently showed a strong relationship with SOC at statewide scale (Xiong et al., 2014) and at a regional scale (this study). Xion (2014) outcomes affirmed that parent material has an effect on soil formation and development. The influence of parent material on SOC stocks occurs through different sources, such as soil weathering, mineralogy, water permeability, nutrient sup ply, plant production and decomposition (Post et al., 2004) . Surprisingly , atmospheric variables seemed to have a muted effect on SOC across the state (Xiong et al., 2014) , even though numerous studies have reported nonlinear climate effects on carbon dynamics (Grunwald, 2009) . The long term trend analysis in Chapter 3 indicated insignificant changes in temperature and precipitation over a decade (Figure 3 13) . However, findings from this chapter showed that three variables of m onthly maximum temperature in summer (July, August, and September) were negatively correlated with SOC. The opposite was found for annual average precipitation that was positively correlated with SOC stocks. Ample research has been conducted to study the i nteractions between climate and SOC which are still debated fiercely. Climate is only one of various kinetic factors controlling decomposition and soil respiration rates (Davidson and Janssens, 2006) . Increases in CO 2 in the atmosphere and temperature ca n have a variety of different effects on SOC inputs through the
140 photosynthesis process, and SOC losses via respiration and decomposition (Ontl and Schulte, 2012) . Recent studies showed that temperature is strongly associated with soil respiration (Bond Lamberty and Thomson, 2010) . Bardgett (2011) argued that warmer temperature will elevate the decomposition of soil carbon shifting the soil carbon pool from a sink to a source of CO 2 . Alternatively, plants growing in elevated CO 2 concentrations and higher temperature fix more carbon through photosynthesis, producing greater biomass and contributing more carbon inputs (Drake et al., 1997) . However, greater root biomass may accelerate plant respiration, resulting in carbon losses (Hungate et al., 1997) . Considering the geographic locations, Wang et al. (2000) found that temperature alone explained only 19% of the variance in soil CO 2 flux in the Sierra Nevada in central California, while moisture alone explained 40%. Th ey also found that soil moisture, temperature, and site variables together improved the model significantly and explained 72% of the variation in soil CO 2 flux along the transects. Others found a positive correlation between mean annual temperature and SOC accumulation in temperate and warmer climatic zones (Poeplau et al., 2011) . An intermediate SOC levels can be found in tropic ecosystems due to high rates of both primary productivity and decomposition rate du e to concomitant warm temperature and abundant rainfall (Ontl and Schulte, 2012) . Future SOC gains will depend on the ability to sequester carbon offseting CO 2 emissions at a quicker rate than carbon losses due to soil respiration under projected global temperature. In this study we found a positive correlation between mean annual precipitation and SOC stocks. Precipitation is a major driver that promotes primary production and increases the roots and plant residues into soils in forms of SOC inputs (Gao et al., 2013; Ki rschbaum, 1995) . Some studies
141 discovered the opposite finding that the relationship of carbon sequestration to mean annual precipitation is negative (Derner and Schuman, 2007; Roa Fuentes et al., 2013) . A potential explanation for the negative relationship between SOC accumulation and mean annual precipitation is that higher amount of carbon input per unit area acc umulates in the dry area than in the moist (Roa Fuentes et al., 2013) . Soil texture is another factor explaining a negative relationship between precipitation and carbon storage. According to JobbÃ¡gy and Jackson (2000) organic matter translocate vertically faster in coarse textured soils with high permeability than fine textured ones. Human variables were not considered powerful predictors in the model as they ranked in the middle and bottom of all variables. Fertilizer consumption (74 th ) had the most influence among the H factors, followed by best management practice (BMP) implementation (100 th ) of all variables, while populatio n growth ranked near the bottom of all predictors. This indicates that human factors may have an indirect effect or little impact on SOC. For example, the amount of fertilizer used increases crop productivity, resulting in SOC increases under BMP implement ation (Snyder et al., 2009) . Rapid population growth rates and high population density have led to food scarcity and deforesta tion in many areas around the world, creating stressors on LC/LU conversion and affecting above ground and below ground SOC storage (Walker and Desanker, 2004) . Most of the counties in the FL SRB have population growth rates that were as small as 2.7 percent and as high as 26.4 percent from 2000 to 2010 (U.S. Census Bureau, 2010c) . These results suggest that anthropogenic factors indirectly influence SOC accumulation, mainly due to choice in LU/LC, intensity of agricultural and forest
142 pr oduction, irrigation management, and other management decisions that influence carbon dynamics. Estimates of Terrestrial Carbon Budget Descriptive statistics and the spatial distribution of TerrC actual stocks are illustrated in Table 4 4 and Figure 4 2. Th e amount of carbon increased almost twofold in the terrestrial ecosystem (mean of 9.2 kg C m 2 ) in comparison to shallow soils with 0 20 cm depth (mean of 5.2 kg C m 2 ) at a site specific basis. The results imply that about 44 percent of carbon storage is found above ground and 56 percent of carbon is stored in soils. Han et al. (2007) reported the total terrestrial carbon storage in the southeast and south central United States and found that Florida has the second highest terrestrial carbon pool in this region with 3,505 Tg C in soils, 252 Tg C in forest, 0.3 Tg C in crop, and 0.7 Tg C in pasture. Note that the SOC pools in the Han et al. (2007) study accounted for the total soil profile (to the depth of measurement) based on detailed work of Bliss et al. (1995) . SOC stocks contribute approx imately 93 percent of terrestrial carbon in Florida (Bliss et al., 1995; Han et al., 2007) . In our study relatively high carbon stocks were observed in swamps and forests. The highest mean TerrC actual stock values were found in mixed wetland forests (16.6 kg C m 2 ), followed by swamps (14.6 kg C m 2 ), and pinelands (11.0 kg C m 2 ). Trees are considered as a carbon sink in the terrestrial system (Bracho et al., 2011; Han et al., 2007; Montagnini and Nair, 2004) . The fundamental process of carbon sequestration revolves around photosynthesis, respi ration, and decomposition (Nair and Nair, 2003). Carbon enters the ecosystem via photosynthesis through leaves, assimilates carbon in above ground biomass, and transports below ground via roots, and remains in the litter deposition in soils (Montagnini and Nair, 2004) . Other studies in Florida suggested that
143 high NPP occurs in deciduous forest (36.69 Tg C) and evergreen forest (35.09 Tg C), while crops and urban contained relati vely low NPP with total NPP of 20.22 and 11.05 Tg C (Milesi et al., 2003) . Deyong et al. (2009) concluded that the annual loss in NPP of about 45.93 Gg C between 1999 and 2005 in Shenzhen China was the effect from urban sprawl, in which a natural landscape was converted to urban land use. According to a study by Bracho et al. (2011) , above ground tree net primary productivity was a major carbon sink after four years of plantation. They further recommended that tree plantations in the southeastern United States, such as pines, have the potential to contribute to the carbon offset of carbon emissions (Bracho et al., 2011) . In our study, the lowest TerrC actual stocks were present in row/field crops with a mean value of 1.2 Tg C (2.7 kg C m 2 ). Low amounts of carbon stocks probably were lost during harvest and crop rotation when most of the carbon is stored in crop residues. Lal (2008) and Montagnini and Nair (2004) suggested numerous strategies that help increase carbon stocks in the agricultural syst ems such as agroforestry, no till farming, diverse crop rotation, and integrated nutrient management. The modeled TerrC attain estimates under different environmental forcings (i.e., AWBH forcings) clearly showed that factor combinations (i.e., climatic pr operties, water, and biota) together had strong effects on carbon storage. This makes sense because these ecological components function complexly in the ecosystem. Assuming that predictors from Table 4 2 are changed by 10 percent, our model simulated a m inimal increase in carbon stocks with mean values of 9.3 kg C m 2 and median values of 8.8 kg C m 2 . The model predicted the average TerrC attain values of 9.4 kg C m 2 and median values of 8.8 and 9.0 kg m 2 when the environmental factors changed by 20 an d 30
144 percent, respectively (Table 4 5). The TerrC actual stocks almost reached 100 percent of the modeled TerrC attain values, indicating this terrestrial system was close to the saturation condition. The absolute increase from predicted TerrC actual to Terr C attain ( ) in terrestrial carbon stocks amounted to 4.3 Tg C (mean) and 13.2 Tg C (median) (Table 4 5). The findings from this study were similar to various other studies. Soils with low SOC levels tend to show a linear response to increasing carbon i nputs (Kong et al., 2005) , whereas soils with initial high levels of SOC amount show little or no response corresponding to additional C inputs in some long term experiments (Campbell et al., 1991; Huggins et al., 1998) . In comparison to the dryland ecosystems studied by Holye et al. (2013) , the attainable SOC stocks assessed by the Roth C model suggested an additional storage capacity, ranging from 3.3 12.8 to 7.0 27.0 kg C m 2 , or incr ease by 5 45 percent, depending on the types of land use and soil type combinations. A recent study conducted by Feng et al. (2013) concluded that fine silt and cl ay fractions did not exhibit C saturation behavior under long term manure amendments. The possibility to accrete carbon stocks in organic soils should be considered beyond carbon saturation coupled to soil texture as outlined by Hassink (1997) . Organic soils, in particular Histos ols, ha ve a slower decomposition rate than its carbon accumulation rate to the degree that the organic materials stockpile to a significant depth (Huang et al., 2011) . To our knowledge, research related to carbon saturation in highly organic soil s is still limiting. In the FL SRB organic soils covered about 6 percent and aquic soils about 49 percent where both conjecture to carbon saturation and depth bas ed accretion of carbon operate.
145 In our study, LC/LU variables were the most powerful predictor for both SOC and TerrC actual stock. Thus, LC/LU types are to target as a management strategy to increase terrestrial carbon sequestration in the FL SRB. For exam ple, the mean TerrC actual for row/field crops was 2.7 kg C m 2 which differed substantially from mean TerrC attain ( 30 percent scenario) with 5.8 kg C m 2 . These findings suggest that about 50 percent carbon in soils could be sequestered under row/field cr ops to achieve TerrC attain . Similarly, about 17.5 percent of soil carbon could be sequestered under bare soils to reach the maximum attainable carbon content represented by the 30 percent scenario. The predicted TerrC attain (30%) for wetlands, pineland, and hardwood forests were 13.7 , 9.1, and 10.5 kg C m 2 , while the Terrc actual were 11.9, 9.4, and 10.3 kg C m 2 . The TerrC actual and TerrC attain stocks in pineland and forests were quite similar in values with differences in TerrC actual and TerrC attain of 0.3 kg C m 2 (pineland) and 0.3 kg C m 2 ( hardwood forests). This is confounded by the fact that forests are undergoing cycles from regeneration/planting to mature stands and harvest with approximate rotational cycles of 20 25 years in Pinus palustris (slas h pine) and Pinus taeda ( loblolly pine ) prominently found in the FL SRB. Thus, the carbon values sequestered in the biomass of forests and underground vary widely over rotational cycles. The TerrC stock values over rotational cycles in forest ranged from 0 .4 to 0.8 kg C m 2 yr 1 (Bracho et al., 2011) . The biomass trend to the age of 25 year rotation cycle of the loblolly pine indicated that root biomass at this typical rotation was about 55 percent higher than in t he mature mixed bottomland hardwood (Noormets et al., 2012) . spatial autocorrelation at a site specific basis of TerrC actual and TerrC attain e for
146 both TerrC actual and TerrC attain (Table 4 5), serves as a proxy to express the heterogeneity of terrestrial carbon within the basin. Although from an ecosystem services perspective the aim is to enhance and maximize carbon storage in an ecosystem due to its many positive benefits, such as enhanced nutrient holding capacity and soil health, another important criteria is the heterogeneity of terrestrial carbon in an ecosystem. To sustain and promote pedo and biodiversity and the functioning and resilien ce of an ecosystem carbon rich and carbon poor are as are beneficial. The decline attain compared to TerrC actual suggests that the heterogeneity of terrestrial carbon in the basin decrease d as the tot al carbon stored in the basin increase d (Table 4 5 and Fig. 4 3). The spatial autocorrelation s were weak and nearly pure nugget effect of TerrC actual and TerrC attain residuals ; weak and thus , were not modeled (e.g., using regression kriging ) . T he spatial variation of terrestrial carbon stocks in different situations showcasing the reference set (TerraC actual ) and TerrC attain under the three scenarios ( 10, 20, and 30 percent variations) derived by the ordinary kriging are shown in Figure 4 3 . Interestingly , overall the spatial distribution in TerrC stocks in the four figures expressed similar patterns, although the local expressions differed. Under the scenario trajectory from 10% to 30% the areas low in carbon declined and the carbon rich areas became mor e extensive when compared to the TerrC actual values. Model Calibration and Validation Results We reported three statistical measurements of model performance. R 2 indicates goodness of fit in the model, forming the overall measure of the accuracy of a regr ession. The RMSE measures the average magnitude of the error; thus a lower value shows better predictions.
147 Our prediction model of SOC stocks using 172 STEP AWBH variables was able to account for 93 percent in calibration mode (n = 164) and 44 percent in validation mode (n = 70) of the variation across the study area (Figure 4 4). Root mean square error values measured the difference between the values predicted by the model and the values actually observed from the field. The RMSE value of 2.65 kg C m 2 in the validation set was higher than 1.33 kg C m 2 in the calibration set, and RPD values of 1.28 kg C m 2 in the validation set were lower than 2.58 kg C m 2 in the calibration set. A better model performance in the calibration set was expected because t he model was constructed based on the same data set. A separate validation set was used to test the performance of the calibration models. In general, for the quantitative soil spatial prediction model, Beckett and Webster (1971) suggested that an R 2 of 0.50 or less is common while an R 2 over 0.70 is unusual. Islam et al. (2003) indicated that a calibration of R 2 < 0.50 was unac ceptable. The validation R 2 for SOC stocks in our study was slightly lower than the one presented by Xiong et al. (2014), who investigated 1,080 soil samples (0 20 cm depth) for the prediction of SOC stocks using cross validation in the calibration set. In their study using the simulated annealing technique and RF model, the R 2 value obtained was 0.62 with an RMSE of 2.65 kg C m 2 . As recommended by Batten (1998) , the range and standard deviation of the population should not be overlooked. In this study, the calculated RPD demonstrated excellent prediction reliability in the calibration set but weak rel iability in the validation set. A major disadvantage of RF is that it is like a structure and makes it difficult to interpret the complexity of relationships among predictors (Prasad et al., 2006) . The purpose of this study did not aim to compare the
148 prediction performance from different techniques, but apply the same method (i.e., Random Forest combin ed with simulated annealing) to assess the sensitivity of model responses (i.e., SOC and TerrC stocks). The prediction of TerrC actual in the RF model improved in comparison to the SOC prediction (Figure 4 5). The calibration R 2 was 0.95 with a small RMSE of 1.78 kg C m 2 and excellent RPD of 3.64 kg C m 2 . Validation R 2 was 0.68 with an RMSE of 3.66 kg C m 2 and a moderate RPD of 1.69 kg C m 2 . High values of RMSE in the validation set indicated that there were some discrepancies between estimations and actual values. In comparison to McCarty et al. (2002), their study obtained validation R 2 values of 0.82 but contained larger prediction errors with an RMSE of 5.5 kg C m 2 . Islam et al. (2003) were able to obtain validation R 2 values of 0.72 for organic c arbon prediction and an RPD of 1.7 kg C m 2 , which were very similar to our findings. Overall, the RF model performed quite well for the TerrC actual prediction. In ecological applications, where samples are much more variable and interactions are complex, it is difficult to compare the model performance from different studies side by side. Dunn et al. (2002) recommended that acceptable values of measurements should depend on the intended application of the predicted values. Conclusions We assessed the relationships between SOC stocks and STEP AWBH factors in the FL SRB and found that biotic, soil, topographic, and water related factors played crucial roles in determining SOC storage, whereas human factors seemed to be fading from being strong predictors. Among the strong predictors, maximum temperature in summer time and me an annual precipitation also stood out as controlling factors for SOC storage. The whole basin stores tremendous amounts of carbon, multiple times
149 larger than atmospheric carbon, with about 75 Tg of C in topsoils and 92 Tg of C in biomass. The SOC stocks were only assessed in the topsoil; and thus, our soil carbon assessment is conservative because we expect to find substantial more carbon stored in the subsoil in a landscape where carbon rich soils, such as Spodosols and Histosols, are prominent. There wa s no single environmental factor that imparted most control on SOC storage, but instead intricate combinations of STP AWB variables, that contributed 43 strong predictors out of a set of 172 potential predictors, explained SOC stocks. Ecosystem services, specifically carbon sequestration services, provide ample benefits including enhancement of crop productivity, offset of atmospheric CO 2 , restoration of habitat and biodiversity, and improvement of soil water and nutrient retention. Carbon storage in soils and bound in biomass is significant because it helps to counteract global climate change. Our study demonstrated a novel approach to assess terrestrial carbon through management, adaptation, and mitigation that is attainable in a subtropical basin consist ing of a mosaic of different land uses embedded in a complex soil landscape. This approach is generalizable and transferable to any other landscape setting. Rather than one dominant factor, a combination of factors (i.e., LC/LU , soil, topography, climate, and water related properties) were used to control variations of the actual terrestrial carbon stocks. The model predicted that the TerrC actual stock values were close to the TerrC attain stock values in some instances (e.g., under wetlands) suggesting the presence of saturation in this area, while in other regions (e.g., under row/field crops) ample opportunities exist to enhance carbon sequestration. Thus, the major goal is to preserve the carbon stored in the terrestrial system of the FL SRB and enhance them through optimized carbon management. Land use adaptations have
150 much potential to reach TerrC attain , specifically land use conversions from cropland to systems with larger NPP. Bare soils, which represent marginal soils, also bear potential to elevate carbon storage if improved through management. This aligns with suggestions to utilize marginal soils for biofuel production (e.g. Arundo donax or other grasses with high NPP) because they would not compete with other land uses, such as food production, na tural conservation area, or urban uses. Such management adaptations do not only enhance above ground but also below ground carbon storage. Nevertheless, rather than offsetting CO 2 emissions, there are other co benefits of increased levels of carbon seques tration to the ecosystem functions (e.g., improvements in crop productivity, food security, soil aggregation that enhances nutrient storage, and water holding capacity). The spatially explicit modeling of actual and attainable terrestrial carbon stocks all ows identifying carbon poor areas and implementing conservation and carbon management strategies. Hence, our approach has much value to outline pathways into a carbon rich future.
151 Table 4 1 . Assembled environmental and human covariates representing STEP AWBH factors. Variable 1 Factor Data type 2 # of variable Source 3 Original scale/ resolution (m) Date Soil order S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil suborder S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil great group S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil subgroup S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil family particle size class S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil family CEC activity class S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil family reaction class S Cat. 1 USDA/NRCS/SSU RGO 1:24,000 2009 Soil family temperature class S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil moisture subclass S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil muck S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil sand content (0 20 cm) S Cont. 1 USDA/NRCS/SSU RGO 1:24,000 2009 Soil silt content (0 20 cm) S Cont. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil clay content (0 20 cm) S Cont. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil organic matter content (0 20 cm) S Cont. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil leaching pot ential S Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Flooding frequency class S/W Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Ponding frequency class S/W Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Annual minimum water table S/W Cont. 2 USDA/NRCS/SSURGO 1:24,000 2009 So il runoff potential S/W Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Drainage class S/W Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Soil available water capacity (0 25 cm) S/W Cont. 1 USDA/NRCS/SSURGO 1:24,000 2009 Hydrologic group S/W Cat. 1 USDA/NRCS/SSURGO 1:24,0 00 2009 Hydration expansion S/W Cat. 1 USDA/NRCS/SSURGO 1:24,000 2009 Surface soil moisture S/W Cont. 1 SMOS 15,000 2010 Soil slope T Cont. 1 USDA/NRCS/SSURGO 1:24,000 2009 DEM elevation (30m, 90m, 1km) T Cont. 3 USGS/NED 30/90/1000 1999 DEM slope (3 0m, 90m, 1km) T Cont. 3 USGS/NED 30/90/1000 1999 DEM flow accumulation (30m, 90m, 1km) T Cont. 3 USGS/NED 30/90/1000 1999 DEM compound topographic index (30m, 90m, 1km) T Cont. 3 USGS/NED 30/90/1000 1999 Distance from sinkhole T Cont. 1 FGS N/A 2009 Di stance from coast T Cont. 1 FMRI 1:40,000 1993 Distance from stream T Cont. 1 USGS 1:100,000 2002 Distance from open water T Cont. 1 USGS 1:2,000,000 2006 S: soil, T: topography, E: ecology, P: parent material, A: atmosphere, W: water, B: bi ota, H: human
152 Table 4 1 . Continued. Variable 1 Factor Data type 2 No. of variable Source 3 Original scale/ resolution (m) Date Ecoregion E Cat. 1 USGS 1:250,000 1995 P hysiographic province E Cat. 1 USGS 1:2,500,000 2000 Environmental geology P Cat. 1 USGS 1:250,000 2001 Surficial geology P Cat. 1 USGS 1:100,000 1998 Surficial geology epoch and period P Cat. 3 USGS 1:100,000 1998 Precipitation 4 A Cont. 13 PRISM Clima te Group 4,000 2000 2008 Temperature 5 A Cont. 26 PRISM Climate Group 4,000 2000 2008 Solar radiation 6 A Cont. 14 NOAA/NCDC/NARR 32,000 1979 2009 Forest canopy B Cont. 4 LANDFIRE 30 2009 Above ground biomass B Cont. 2 LANDFIRE/NBCD 30 1999 2002 Canopy coverage and imperviousness B Cont. 1 MRLC/NLCD 30 2001 Land cover class B Cat. 1 MRLC/NLCD 30 2001 Monthly average MODIS EVI B Cont. 12 MODIS4NACP 500 2005 Monthly average MODIS NDVI B Cont. 12 MODIS4NACP 500 2005 Monthly average MODIS LAI B Cont. 12 MODIS4NACP 500 2005 Monthly average MODIS FPAR B Cont. 12 MODIS4NACP 500 2005 Annual primary productivity (GPP, NPP) B Cont. 2 WHRC/NBCD 30 2005 Vegetation type B Cat. 1 LANDFIRE 30 2009 Biophysical settings B Cat. 1 LANDFIRE 30 2009 Vegetation height B Cont. 1 LANDFIRE 30 2009 Vegetation cover B Cont. 1 LANDFIRE 30 2009 Land cover/land use cover B Cat. 1 FFWCC 30 2003 Land cover/land use cover B Cat. 1 MRLC/NLCD 30 2006 Cropland data layer B Cat. 1 USDA/NASS 30 2004 Population growth H Cont. 1 U .S. Census Bureau County level 2010 Population density H Cont. 1 U.S. Census Bureau County level 2000 2010 Annual household median income H Cont. 1 U.S. Census Bureau County level 2006 2010 Fertilizer consumption H Cont. 1 FDACS County level 2000 2009 Best management practice implementation H Cat. 1 FDACS County level 2010 1 Variable abbreviation: CEC, Cation Exchange Capacity; DEM, Digital Elevation Model; EVI, Enhanced Vegetation Index; FPAR, Fraction of Photosynthetically Active Radiation; LAI, Leaf Area Index; MODIS, Moderate resolution Imaging Spectroradiometer; NDVI, Normalized Difference Vegetation. 2 Data type abbreviation: Cat., categorical data; Cont., continuous data. 3 Data source abbreviation: FDACS, Florida Department of Agriculture and Consu mer Services; FFWCC, Florida Fish and Wildlife Conservation Commission; LANDFIRE, LANDscape FIRE and resource management tools project;
153 MODIS4NACP, MODIS for North American Carbon Project; MRLC, Multi resolution Land Characteristics Consortium; NARR, North American Regional Reanalysis; NASS, National Agricultural Statistics Service; NBCD, National Biomass and Carbon Dataset; NED, National Elevation Dataset; NLCD, National Land Cover Data; NOAA, National Oceanic and Atmospheric Administration; NCDC, National Climatic Data Center; NRCS, Natural Resources Conservation Service; PRISM, Parameter elevation Regressions on Independent Slopes Model; SMOS, Soil Moisture and Ocean Salinity; SSURGO, Soil Survey Geographic Database; USDA, United States Department of Agri culture; WHRC, Woods Hole Research Center. 4 Precipitation variables comprise 12 monthly averages over 2000 2008 and 1 yearly average. 5 Temperature variables comprise 12 monthly minimum temperature average over 2000 2008, 1 yearly minimum temperature avera ge, 12 monthly maximum temperature average over 2000 2008, 1 maximum temperature average. 6 Solar radiation variables comprise 12 monthly averages over 1979 2009, 1 yearly average, and 1 monthly average.
154 Table 4 2 . Top forty three most important variab les (the first quantile) derived by random forest model using STEP AWBH factors to predict soil organic carbon (SOC) stocks at 0 20 cm within the Suwannee River Basin, Florida and descriptive statistics of continuous variables Variable 1 Unit/ scale Factor Var. imp 2 Mean SD 1 Max Min Skew ness CV 1 Correlation with SOC 3 Vegeta tion type factor B 103.36 NA NA NA NA NA NA NA Soil great group factor S 83.94 NA NA NA NA NA NA NA Soil organic matter content (0 20 cm) % weight S 65.91 4.08 8.85 70.00 0.00 4.80 2.17 0.44* Land cover/land use cover (2003) factor B 65.73 NA NA NA NA NA NA NA Soil suborder factor S 62.61 NA NA NA NA NA NA NA Biophysical settings factor B 60.68 NA NA NA NA NA NA NA Surficial geology factor P 52.17 NA NA NA NA NA NA NA Vegetation cover % B 49.27 NA NA NA NA NA NA NA Soil subgroup facto r S 48.53 NA NA NA NA NA NA NA Land cover class factor B 43.96 NA NA NA NA NA NA NA Soil slope % T 35.06 1.97 1.58 10.00 0.00 1.91 0.81 0.43* Drainage class factor S/W 33.41 NA NA NA NA NA NA NA MODIS NDVI of the 105 th day of year days B 30.62 0.72 0.09 0.87 0.17 1.51 0.13 0.33* Land cover/land use cover (2006) factor B 26.28 NA NA N A NA NA NA NA Vegetation height m B 25.07 4.48 3.72 10.00 0.00 0.07 0.83 0.24* MODIS EVI of the 257 th day of year days B 24.38 4.98 0.64 8.07 2.20 0.18 0.13 0.01 MODIS NDVI of the 137 th day of year days B 20.73 7.67 0.85 9.17 1.78 2.03 0.11 0.34* MODIS NDVI of the 257 th day of year days B 20.05 0.80 0.07 0.91 0.46 1.76 0.09 0.26* MODIS LAI of the 353 th day of year days B 19.58 29.28 23.33 250.00 6.00 7.36 0.80 0.17* DEM flow accumulation m 2 T 19.12 550.76 6234.37 94660.00 0.00 1 4.90 11.32 0.00 Monthly maximum temperature (Sep) o C A 19.11 31.46 0.23 32.35 31.01 0.72 0.01 0.20* Above ground live dry biomass (FIA) kg m 2 B 19.10 6.72 4.16 16.90 1.20 0.54 0.62 0.14 Annual primary productivity (NPP) kg m 2 B 18.15 8.79 4.16 32.77 4.20 3.55 0.47 0.15* DEM compound topographic index 30m T 15.47 12.72 5.31 25.76 5.55 0.35 0.42 0.10 Surficial geology epoch and period factor P 14.38 NA NA NA NA NA NA NA Above ground live dry biomass (NCE) kg m 2 B 14.27 7.20 4.49 20.00 2.00 1.04 0.62 0.19* MODIS EVI of the 225 th day of year days B 14.03 5.33 0.67 8.24 2.43 0.16 0.12 0.04 Distance from stream km T 13.59 2.76 3.07 14.67 0.03 1.81 1.11 0.08 Canopy coverage and imperviousness % B 13.07 47.61 38.52 99.00 0.00 0.21 0.81 0.25* Distance from open water km T 12.17 0.92 1.07 7.64 0.03 3.67 1.16 0.13* Distance from sinkhole km T 11.93 10.02 8.25 36.78 0.03 1.24 0.82 0.19* Soil available water capacity (0 25 cm) cm S/W 11.75 2.22 1.16 6.97 0.00 2.0 4 0.52 0.34* S: soil, T: topography, E: ecology, P: parent material, A: atmospher e, W: water, B: biota, H: human
155 Table 4 2 . Continued. Variable 1 Unit/ scale Factor Var. imp 2 Mean SD 1 Max Min Skew ness CV 1 Correlation with SOC 3 Monthly maximum temper ature (Aug) o C A 11.74 32.86 0.25 33.69 32.26 0.08 0.01 0.13* MODIS FPAR of the 353 th day of year days B 11.55 0.82 0.19 2.50 0.42 5.32 0.24 0.12 Hydrologic group factor W 11.46 NA NA NA NA NA NA NA MODIS NDVI of the 289 th day of year days B 11.28 0.77 0.07 0.91 0.47 1.30 0.09 0.23* DEM compound topographic index 90m T 10.93 13.67 4.77 23.46 3.51 1.02 0.35 0.01 Annual mean precipitation cm A 10.92 13.93 0.67 15.45 12.92 0.58 0.05 0.10 Ponding frequency class factor S/W 10.62 NA NA NA NA NA N A NA MODIS NDVI of the 73 th day of year days B 10.60 0.67 0.09 0.82 0.18 1.31 0.14 0.26* Monthly maximum temperature (Jul) o C A 10.46 33.08 0.25 33.83 32.32 0.02 0.01 0.14* Soil runoff potential factor S/W 10.15 NA NA NA NA NA NA NA DEM compound t opographic index 1km T 10.06 737.71 184.67 1507.00 430.00 0.92 0.25 0.09* 1 Variable abbreviation: DEM, Digital Elevation Model; MODIS, Moderate resolution Imaging Spectroradiometer, EVI, Enhanced Vegetation Index; NDVI, Normalized Difference Vegetation; LAI, Leaf Area Index; FPAR, Fraction of Photosynthetically Active Radiation; FIA, Forest Service Inventory and Analysis; NPP, Net Primary Productivity; NCE, Nationally Consistent Allometric Equations; SD, Standard Deviation; CV, Coefficient of Variation. 2 Variable importance is defined as the total decrease in node impurities variable splitting averaged over all trees in the forest. It is comparable to the residual sum of squares in the regression. 3 Spearman correlation coefficient Continuous data *Significant spearman correlation at 0.05 significance level.
156 Table 4 3 . Strong relevant variables derived by random forest model using STEP AWBH factors to predict terrestrial carbon stocks within the Suwannee River Basin, Florida an d descriptive statistics of continuous variables Variable 1 Unit/ scale Factor Var. imp 2 Mean SD 1 Max Min Skew ness CV 1 Correlation with SOC 3 Canopy coverage and imperviousness % B 1105.35 47.61 38.52 99.00 0.00 0.21 0.81 0.72* Above ground live dry biom ass (NCE) kg m 2 B 963.25 7.20 4.49 20.00 2.00 1.04 0.62 0.60* Above ground live dry biomass (FIA) kg m 2 B 604.94 6.72 4.16 16.90 1.20 0.54 0.62 0.58* Land cover/land use cover (2003) factor B 407.64 NA NA NA NA NA NA NA Land cover class factor B 407.46 NA NA NA NA NA NA NA Soil subgroup factor S 282.73 NA NA NA NA NA NA NA Vegetation cover % B 282.62 NA NA NA NA NA NA NA Soil great group factor S 218.33 NA NA NA NA NA NA NA Vegetation type factor B 214.05 NA NA NA NA NA NA NA Surfi cial geology factor P 177.65 NA NA NA NA NA NA NA Biophysical settings factor B 150.40 NA NA NA NA NA NA NA Soil suborder factor S 147.02 NA NA NA NA NA NA NA Soil organic matter content (0 20 cm) % weight S 133.77 4.08 8.85 70.00 0.00 4.80 2.17 0.42* Land cover/land use cover (2006) factor B 133.32 NA NA NA NA NA NA NA Distance from open water km T 72.77 0.92 1.07 7.64 0.03 3.67 1.16 0.03 MODIS NDVI of the 137th day of year days B 70.07 7.67 0.85 9.17 1.78 2.03 0.11 0.40* Hydrologic group factor W 6 7.74 NA NA NA NA NA NA MODIS NDVI of the 105th day of year days B 67.68 0.72 0.09 0.87 0.17 1.51 0.13 0.37* MODIS EVI of the 257th day of year days B 66.53 4.98 0.6 4 8.07 2.20 0.18 0.13 0.05 Annual primary productivity (NPP) kg m 2 B 66.02 8.79 4.16 32.77 4.20 3.55 0.47 0.23* DEM flow accumulation m 2 T 65.52 550.76 6234.37 94660.00 0.00 14.90 11.32 0.08 MODIS NDVI of the 257th day of year days B 6 0.45 0.80 0.07 0.91 0.46 1.76 0.09 0.34* Monthly maximum temperature (Sep) o C A 60.27 31.46 0.23 32.35 31.01 0.72 0.01 0.14* DEM compound topographic index 30m T 60.16 12.72 5.31 25.76 5.55 0.35 0.42 0.18* Distance from stream km T 58.7 2 2.76 3.07 14.67 0.03 1.81 1.11 0.12 Drainage class factor S/W 57.63 NA NA NA NA NA NA NA Soil runoff potential factor S/W 55.28 NA NA NA NA NA NA NA Monthly maximum temperature (Jul) o C A 54.61 33.08 0.25 33.83 32.32 0.02 0.01 0.18* MO DIS NDVI of the 73th day of year days B 54.44 0.67 0.09 0.82 0.18 1.31 0.14 0.35* Surficial geology epoch and period factor P 53.82 NA NA NA NA NA NA NA Monthly maximum temperature (Aug) o C A 47.33 32.86 0.25 33.69 32.26 0.08 0.01 0.20* D EM compound topographic index 90m T 47.29 13.67 4.77 23.46 3.51 1.02 0.35 0.12* S: soil, T: topography, E: ecology, P: parent material, A: atmospher e, W: water, B: biota, H: human
157 Table 4 3. Contined. Variable 1 Unit/ scale Factor Var. imp 2 Mean SD 1 Max Min Skew ness CV 1 Correlation with SOC 3 Annual mean precipitation cm A 44.28 13.93 0.67 15.45 12.92 0.58 0.05 0.20* Vegetation height m B 43.43 4.48 3.72 10.00 0.00 0.07 0.83 0.43* Distance from sinkhole km T 39.84 10.02 8.25 36.78 0.03 1.24 0.82 0.14* Soil available water capacity (0 25 cm) cm S/W 37.47 2.22 1.16 6.97 0.00 2.04 0.52 0.30* MODIS LAI of the 353th day of year days B 34.21 29.28 23.33 250.00 6.00 7.36 0.80 0.05 MODIS NDVI of the 289th day of year days B 34.10 0.77 0.07 0.9 1 0.47 1.30 0.09 0.34* MODIS FPAR of the 353th day of year days B 33.11 0.82 0.19 2.50 0.42 5.32 0.24 0.13 DEM compound topographic index 1km T 32.83 737.71 184.67 1507.00 430.00 0.92 0.25 0.10 MODIS EVI of the 225th day of year days B 32.73 5.33 0. 67 8.24 2.43 0.16 0.12 0.01 Soil slope % T 18.21 1.97 1.58 10.00 0.00 1.91 0.81 0.38* Ponding frequency class factor S/W 11.80 NA NA NA NA NA NA NA 1 Variable abbreviation: DEM, Digital Elevation Model; MODIS, Moderate resolution Imaging Spectroradio meter, EVI, Enhanced Vegetation Index; NDVI, Normalized Difference Vegetation; LAI, Leaf Area Index; FPAR, Fraction of Photosynthetically Active Radiation; FIA, Forest Service Inventory and Analysis; NPP, Net Primary Productivity; NCE, Nationally Consisten t Allometric Equations; SD, Standard Deviation; CV, Coefficient of Variation. 2 Variable importance is defined as the total decrease in node impurities variable splitting averaged over all trees in the forest. It is comparable to the residual sum of squares in the regression. 3 Spearman correlation coefficient Continuous data *Significant spearman correlation at 0.05 significance level.
158 Table 4 4 . Descriptive statistics of terrestrial carbon stocks at observation sites derived by above ground carbon (biom ass) and below ground carbon (top soil at 20 cm depth) (units in kg C m 2 ). Descriptions Number of samples Mean Median Max Min Skewness STD 1 SEM 2 Observations within the study area 138 9.8 8.5 28.8 1.1 0.8 6.8 0.6 Observations within the study area and a 20 km buffer 234 9.2 8.4 28.8 1.1 0.8 6.4 0.1 Calibration set randomly chosen: 70% of the total observations 164 9.6 8.9 28.8 1.1 0.6 6.5 0.5 Validation set randomly chosen: 30% of the total observations 70 8.3 7.3 28.8 1.6 1.2 6.2 0.7 1 STD refers to standard deviation. 2 SEM refers to standard error of mean. Table 4 5 . Descriptive statistics of terrestrial carbon stocks based on observations (n = 234 sites) and predictions under different scenarios derived by simulated annealing an d random forest modeling. Terrestrial carbon stocks ( kg C m 2 ) Mean Median Max Min Variance z score Observed actual terrestrial carbon (based on site observations) 9.21 (179.41) 8.35 (162.66) 28.81 (561.22) 1.11 (21.62) 41.35 (805.50) 0.34 1.75* Predicted actual terrestrial carbon (based on simulated annealing and random forest) 9.08 (176.88) 8.16 (158.96) 24.26 (472.58) 2.00 (38.96) 28.93 (563.56) 0.44 2.18* Predicted attainable terrestrial carbon ( 10% environme ntal variation) 9.31 (181.36) 8.84 (172.20) 21.61 (420.96) 3.73 (72.66) 15.34 (298.82) 0.31 1.62 Predicted attainable terrestrial carbon ( 20% environmental variation) 9.39 (182.92) 8.84 (172.20) 21.61 (420.96) 3.66 (71.30) 13.17 (256.55) 0.28 1.45 Predicted attainable terrestrial carbon ( 30% environmental variation) 9.43 (183.70) 9.03 (175.90) 21.17 (412.39) 4.33 (84.35) 13.28 (258.69) 0.29 1.51 *Significant at the 90% significance level (p value < 0.10). Total terrestrial carbon stocks in Tg C are shown in parentheses. Terrestrial carbon stocks are reported standardize by area (kg C m 2 ) and as absolute values in Tg C in parenthesis. d from simulated annealing and random forest.
159 Figure 4 1 . A workflow for soil organic carbon (SOC) stocks prediction and the attainable terrestrial (TerrC attain ) carbon prediction from the comprehensive set of environmental human predictor variables.
160 Figure 4 2 . Spatial distribution of terrestrial carbon stocks (kg C m 2 ) based on current soil organic carbon stocks (0 20 cm) and above ground biomass stocks (n = 234). [Soil carbon data adapted from a subset from the Rapid Assessment and Modeling of Changes in Soil Carbon Storage and Turnover in a Southern Landscape, 2011; Above ground Biomass adapted from Kellnorfer et al., 2013. North American Carbon Program (NACP), Above ground Biomass and Carbon Baseline Data (NBCD2000.V2). Accessible through htt p://www.whrc.org/mapping/nbcd/ ].
161 (a) (b) (c) (d) 1 Search neighborhood: 3 minimum samples, 4 angular sector, 5 samples per sector; 2 Mean error (kg C m 2 ); 3 Root mean square error (kg C m 2 ). Figure 4 3 . Terrestrial carbon stock estimates de rived by ordinary kriging: (a) observed actual terrestrial carbon stocks; (b), (c), and (d) attainable terrestrial carbon stocks after environmental variables change by 10 percent, 20 percent, and 30 percent. 1 Model: Stable Range (m): 48,337.23 Nugget: 0.28 Partial s ill: 0.40 2 ME: 0.03 3 RMSE: 6.1 1 Model: Spherical Range (m): 40,999.47 Nugget: 0.12 Partial s ill: 0.07 2 ME: 0.01 3 RMSE: 3.64 1 Model: Stable Range (m): 80,909.15 Nugget: 0.03 Partial s ill: 0.13 2 ME: 0.05 3 RMSE: 3.43 1 Model: Stable Range (m): 118,571.08 Nugget: 0.03 Partial s ill: 0.13 2 ME: 0.04 3 RMSE: 3.43
162 (a) (b) Figure 4 4 . Independent validat ion of predicted soil organic carbon (SOC) stocks (kg C m 2 ) at the 20 cm soil depth using random forest model for (a) the calibration set (70%) and (b) validation set (30%) . R 2 denotes the coefficient of determination, RMSD is the root mean square of devi ation (kg C m 2 ), and RPD is the residual prediction deviation (kg C m 2 ). There are 172 explanatory variables used in the model.
163 (a) (b) Figure 4 5 . Independent validation of predicted actual terrestrial carbon stocks (kg C m 2 ) using random fo rest model for (a) the calibration set (70%) and validation set (30%) . R 2 denotes the coefficient of determination, RMSD is the root mean square of deviation (kg C m 2 ), and RPD is the residual prediction deviation (kg C m 2 ). There are 43 explanatory vari ables used in the model.
164 CHAPTER 5 SOCIO ECONOMIC PERSPECTIVE S OF ECOSYSTEM SERVI CES Overview While most people realize the importance of ecosystem services, many case studies show how previously overlooked valuation of ecosystem services resulted in ra pid degradation and depletion of ecosystems (Chee, 2004; Daily et al., 2000) . Ecosystem services do not regularly fall within the sp here of markets; rather they tend (Costanza et al., 1997) . As such, this leads to the idea of re framing decisions and prompting improved management of natural capital by valuing ecosystem services as part of the decision making process. Some argue that valuation of ecological systems is either impossible or unwise; for example, intangibles such as human life, aesthetics, or long term ecological benefits cannot be assigned a monetary value. Costanza et al. (1997) stated that even though ecosystem valuation is certainly difficult and bound by uncertainties, nevertheless it should be undertaken. In the twenty first centu ry, the valuation of ecosystem services has become a significant evolving research area (Turner et al., 2003) . Valuation studies have contributed co mprehensive information for the policy making process, in particular they support preference based approaches that are compatible with a monetary system (Turner et al., 2003) . The principle of ecosystems is not about placing dollar values ($) on the environment, but rather stressing the effect of marginal change in ecosystem services under the fundamental concept of trade offs against general common goods and services (Hanley and Shogren, 2002; Randall, 2008) . A stud y linking ecology and economy is therefore critical for assessing the trade offs inherent in managing human societies within ecological systems (Farber et al., 2006) .
165 People may value ecosystem services depending upon their scale perception (AraÃ±a and LeÃ³n, 2012) , immediate direct effect (Boissiere et al., 2013; Howe et al., 2013) , gov ernance management (Costanza and Liu, 2014) , and/or demographic and socio economic background (Andersen et al., 2012; Peixer et al., 2011) . Many other confounding factors may lead to valuations of climate, carbon, and nutrient regulation that differ from their biophysical values. However, it is unclear how people in the Suwannee River Basin of Florida (FL SRB) value ecosystem services. The main goal of this study was to assess ecosystem services from a socio economic perspective. The three objectives were to (i) investigate the perceptions of households in the FL SRB in regard to ecosy stem services, (ii) explore socio economic made by people to protect ecosystem services and their willingness to pay to protect these ecosystem services. Valuation Appr oaches for Non market Goods and Services Decision makers are increasingly accepting the task of formulizing environmental policies to address the issue of valuation of ecosystem services. It is important for decision makers to consider stakeholders (the pu blic) when formulating environmental (Hicks, 2002) . Valuation of non market goods has been a key component of environme ntal economics for many years (Haipeng and Xuxuan, 2012) . Two fundamental approaches for val uing non market goods are revealed preference methods and stated preference methods (Bennett and Blamey, 2001; Martinsson, 2002) . Both techniques have advantages and disadvantages.
166 Revealed preference methods estimate values of non market goods by relying on evidence of how people behave through their actions in markets. Methods such as random utility and travel cost, averting behavior, hedonics, and production function are consider ed revealed preferences (National Research Council (NRC), 2004) . One main advantage of this technique is the reliance on actual choices to avoid the potential i ssues allied with hypothetical responses, and thus, are less susceptible to response relying on observable trips (travel cost method) or health expense (averting behavior) , analyses are restricted to the attribute that cannot be observed (Hicks, 2002) , for example, climate regulation, carbon sequestration, and nutrient cycling in this study. More details of the limitations are clearly described in Abley (2000) , Kroes (1988) , and NRC (2004) . Unlike the revealed preference methods, the stated preference methods are commonly used to estimate preferences of the welfare effects of non market impacts through hypothetical choice scenarios. While several methods of stated preferences are available, the best known stated preference method s include contingent valuation method (CVM) and choice experiments (CE) (Haipeng and Xuxuan, 2012) . The concept of CVM is rooted in individual utility maximization. Respondents are asked to express their maximum willingness to pay (WTP) for or minimum willingness to accept (WTA) a specific change in an environmental condition through the open e nded question, bidding game, or payment card (Fujiwara and Campbell, 2011; Mitchell and Carson, 1989) . Despite being one of the most commonly used methods for valuation of non mark et goods, its use faces problems with dissatisfaction about verbal protocol
167 analysis, variation in WTP across surveys, and bias incidence of protest bids (Diamond and Hausman, 1994; Hanley et al., 1998) . Some studies have addressed this issue by comparing hypothetical to actual values, and generally found the ex istence of bias (Champ et al., 1997; Cummings et al., 1995) . While the controversial debates of CVM are on going, the CE approach, developed by Louviere and Hensher (1982) and Louviere and Woodworth (1983) , has gained popularity in a variety of research fields (Ãlvarez Farizo et al., 2009; Boxall et al., 1996; Broadbent, 2013; Hainmueller et al., 2014; Hoehn et al., 2010; Taylor and Longo, 2009; Vollm er et al., 2013) . The CE is sometimes known as multi attribute utility choice analysis (Milon et al., 2000) or conjoint analysis (Farber and Griner, 2000) . Hanley et al. (1998) and Stevens et al. (2000) viewed the CE methods as a generalization close ended CVM involving two or more goods and services. CE methods allo w researchers to focus on valuing marginal changes as multi dimensional attributes rather than discrete changes (Hanley et al., 2001) . Choosing between choices encourages respondents to explore their preferences and trade offs in more detail (Stevens et al., 2000) and allows researchers anagement plans (Nalle et al., 2004) . Multiple components across choices can be estimated in terms of the relative explanatory power of a single behavioral outcome (Wallander, 2009) ; t herefore, when a choice set includes a price or cost factor as an attribute, economic values such as WTP can be estimated (Boxall et al., 1996) . Since attribute levels of choice situation, including price/cost, are designed in a systematic fashion, the performance in measuring the marginal value of changes and multiple characteristics of environmental
168 programs as positive is expect ed to be meaningful (Boxall et al., 1996; Hanley et al., 2001) . Both CVM and CE methods rely on hypothetical scenarios, and according to the neo classical economic theory, they should provide si milar results (Stevens et al., 2000) . In fact, mixed results of the comparison between hypothetical CE values have been reported (Broadbent, 2012, 2013; Broadbent et al., 2010; Cameron et al., 2002; Carlsson and Martinsson, 2001; Hensher, 2010; List et al., 2006; Loomis et al., 2009; Lusk and Schroeder, 2004; Olof and Henrik, 2008; Ready et al., 2010; Stevens et al., 2000) , thus a more critical discussion of deriving WTP from different packages is needed. Choice Experiments This technique requires a careful choice design that helps reveal the factors influencing choic es. A structured CE method consists of six main stages: 1) selection of attributes, 2) assignment of levels, 3) choice of experimental design, 4) construction of choice sets, 5) measurement of preferences, and 6) estimation (Hanley et al., 2001) . Identification of the attribute space such as levels and ranges must be relevant, realistic, and feasible for the environmental program questions being asked. One of these attributes is usually a monetary cost that allows estimating WTP (Hanley et al., 2001, 1998) . A bundle of interdependent services such as climate regulation and carbon sequestration can provide more meaningful values than summing the values of independent service levels w hen using CE (Gloulder and Kennedy, 1997) . In addition, a baseline status quo is typically included in the assignment of levels (Boxall et al., 1996) . Various statistical informed methods can be used to combine attribute levels. Two widely used approaches to combine attribute levels are the complete factorial
169 design and the fractional factorial design that are available through specialized software. The former method provides the estimation of all the main effects and interactions of the attributes that are suitable for a small set of att ribute levels. The latter method allows the reduction of scenario combinations to avoid too many choices for respondents; however, some interactions are inestimable (Louviere, 1988) . The experimental designs group profiles into choice sets in the form of individual, pair, or group options. Individual preferences can be measured by ratings, rankings, or choices (Hanley et al., 2001) . Then, an estimation procedure such as ordinary least squares (OLS) regression or logistic regression can be used to estimate ranking of preferences and WTP. When the dependent variables are categorical data, logistic regression such as the conditional logit model is often used (Alldredge and Griswold, 2006) . Many studies relevant to a variety of ecosystem services have been followed in attributes. For exa mple, Adamowicz et al. (1 998) implemented both the CE and contingent valuation to measure passive use values of habitat enhancement. Bullock et al. (1998) used the CE approach to measure preferences of respondents for deer hunting and landscape change in the Scottish Highlands. In a study by Milon et al. (2000) , individuals were asked to identify the importance of ecosystem restoration plans based on five multiattibute choices to value the Everglades ecosystem. Birol et al. (2006) scale farms. Brouwer et al. (2010) assessed spatial preference heterogeneity related to water quality improvements that included hydrogeographical units and levels of water quality improvements in the experimental designs.
170 A main interest of this chapter is the evaluation of broad ecosystem services associated with climate change, water quality, and food productivity. Ecosystem services identified in this study include climate/carbo n regulation, nutrient control, and agricultural/forestry production. Administration agency, location, and cost are also included in choice sets in this study. Framework and Choice Experiment Modeling This section outlines the theoretical framework for th e study of discrete choice to assess objectives (ii) and (iii). Discrete choice modeling engages the prediction of an generally take a random utility view of the choices in which decision makers are assumed to reach their maximum utilities (Train, 2009) . Logistic regression is widely used for modeli ng binary or multiple discrete response variables. Thr ee models related to this study multinomial logit model, conditional logi t model, and nested logit model are discussed in the materials and methods section. Random Utility Maximization (RUM) t heory Both multinomial and conditional logit models are used to investigate a behavior, known as random utility maximization (RUM) models valuation (Manski, 1977) . It is assumed that an individual would select an option with a higher level of utility than that associated with all other options. U ij = V ij ( X) + (4 1) The indirect utility ( U ) for each respondent ( i ) is made up of two components. The first part, a deterministic element ( V ij ), also known as a systematic component, which determines a linear index of the attributes X of differen t alternatives j in a choice set.
171 Another part of the utility function, a stochastic element ( ), also known as a random (Hanley et al., 2001) (Equation 4 1). P[(Um i > Un i ), n m] = P[(Vm i Vn i ) > ( m i n i )] (4 2) The probability ( P ) that an individual prefers option m to other options n is based on the utility function theory and for all n ( n ) not equal to m (Equation 4 2). When the probability of ( Um i > Un i ) equa ls the probability of (Vm i Vn i ), the condition is greater than the random error component. While the deterministic element that is related to the alternatives can be measured, the error terms ( ) cannot be measured. The random component is there fore assumed to be part of the Weibull probability distribution with an extreme value. This implies that the probability of choosing alternative i can be described in the logistic distribution (Greene, 1997; McFadden, 1974) . P ij = (4 3) where C is a choice set of the individual, e is an exponential function. A scale parameter often cannot be separated and typically is assumed to be 1. The logit model for Equation 4 3 can be estimated using maximum likelihood methods with the log likelihood function given by Hole (2006) . (4 4) where is an indicator variable that equals 1 if alternative j is chosen by respondent i , and zero otherwise. N is the number of respondents ( i N ) and C indica tes the j alternative in a choice set of the individual ( j C ).
172 In general, the multinomial and conditional logit models for utility function can be written in a simple linear regression form. V i = (4 5) where is a constant that captures the effects of utility on any attribute excluded from the choice attributes. The constant value is a meaningful interpretation only when all attributes can reasonably be zero. The vector of coeffic ients to is attached to a vector of attributes to and is the coefficient for the cost of an alternative . Even though the multinomial and conditional logit models share the same conceptual framework, the distin ction between the two models was pointed out by Hoffman and Duncan (1988) . They explained that the multinomial logit uses the the conditional logit uses properties of choice attributes as explanatory variables to make choice probabilities. In addition, the conditional logit needs t o specify which individuals belong to which matched set or stratum (McFadden, 1974) . The multinomial and conditional logit models impose the restriction that the alternatives are independent of one another. Violation of the independence of irrelevant alternatives (IIA) restriction lead s to the cross elasticities between all pairs of alternatives being identical (Wen and Koppelman, 2001) . The most common technique to relax the IIA is the nested logit (Williams, 1977) . The nested logit is often used when there is a set of alternatives experiencing vague decision. It allows the alternatives to be correla ted and accommodates degrees of interdependence among pairs of alternatives in a choice set. Details of the logit implementation are discussed in the methodology section.
173 Choice Experiments (CE) and Willingness to Pay (WTP) Since CE is embedded in the uti lity of maximization and demand theory (Bateman et al., 2003) , the coefficients in CE provide parameter estimates of the utility of e ach attribute (Ginsburgh and Throsby, 2013) . The cost/price coefficient can be interpreted as an estimate of the negative marginal utility of income. The measure of marginal change in welfare marginal willingness to pay (MWTP) in this case can be derived from the parameter estimates by the following form ulae (Hanemann, 1984; Hanley et al., 2001; Parsons and Kealy, 1992) : (4 6) (4 7) where, , the coefficient on the monetary attribute, provides the marginal uti lity of income. and represent indirect utility functions before and after the ecosystem management policy change. In the conditional logit model or the multinomial logit model, the marginal value of an attribute is the ratio of an attribute to the coefficient of cost/price: (4 8) where, is the coefficient of any of the non monetary attributes and is the coefficient for cost/price. The simplified formula of the ratio of coefficients given in Equation 4 quantity of an attribute (Ginsburgh and Throsby, 2013) . Although the MWTP can be assumed to have positive or negative values in Equation 4 8, it is possible that individuals would maintain the status quo situation. T he possibility of a negative MWTP
174 is appropriately present (Hanemann, 1984) . The mean and the med ian of the MWTP distribution cannot be calculated separately when the utility function is assumed to be linear. The mean and the median values, therefore, coincide (Hanemann, 1984) . Materials and Methods Survey Design ecosystem services in the FL SRB. The outcome of the survey was aimed to provide a better understanding from the social side and eventually to be integrated with the biophysical and modeling chapters to complete the ecosystem service valuation goal. The mail survey consisted of three major parts. The first part (sect ions 1 to 4) was designed to gauge opinions and experiences of respondents related to ecosystem issues. This part of the questionnaire included general questions about ecosystem services, climate regulation, soil carbon sequestration (storage), and nutrien t cycling. Responses were obtained using three point and five point Likert rating scales, dichotomous choice, and nominal questions. The second part (section 5) contained the the next economic information, gender, age, education, Florida residency, and income. The questionnaire and choice attributes used in the study were developed based on evidence from the lite rature. The complete questionnaire is shown in Appendix A. Choice Experiments Design protecting ecosystem services and improving environmental quality. To address the goal of the stud y, a set of limited choices in the form of programs was provided where
175 each program included four attributes. The first attribute was based on the type of ecosystem services, including climate/carbon storage, nutrient control, and agricultural and forestry production. The second attribute represented the administrative agency that would be supported through voluntary donations, annual taxes, or utility fees to manage the ecosystem services. Three agencies were included: the Suwannee River Water Management D istrict (SRWMD), county governments, and non government organizations (NGO). The third attribute targeted the size of the area to be managed for ecosystem services implementation, ranging from anywhere within the FL SRB to 20 miles (32 km) or 5 miles (8 km attribute was added to investigate the economic values of the program being chosen. Respondents were asked to choose $5, $25, or $50 as the per annum cost of WTP per household for the preferred program o ver a five year period. These numbers were set based on findings from Kreye et al. (2011) , which compiled 17 studies that examined WTP for water protection and applied a benefit transfer method to estimate an annual household WTP within four regions in Florida. The attributes and the levels of the attributes are presented in Table 5 1. Four attributes with three levels each are also known as four three level factors in a factorial design. A fractional factorial design was utilized to construct the optimal paired comparison choices as described by Street et al. (2005) . The paired comparis on method combined choice and rating with two specific choice sets and one opt out option. Respondents were asked to select the most preferred alternative of the two specific options or none of these (neither) option. The choice sets were created using t he fractional factorial designs which allowed reducing the number of scenario
176 combinations to the optimal sets (Street et al., 2005) . Thus, the minimum possible choice sets that remained for this study included nine questions. The nine questions were then divided into two sets of surveys, A and B, to avoid excessive time c onsuming circumstances from too many questions. Thus five choice sets were designed in Set A and four choice sets were included in Set B. The fractional factor ial designs are shown in Table 5 2. The MWTP values were estimated using a nested logit model with quantitative cost input. The coefficient of the WTP was specified to be held constant. This restriction allowed the distribution of WTP to be estimated directly from the distribution of non monetary coefficients because two distributions took the sam e form in the model (Goett and Hudson, 2000) . Survey Sampling and Implementation All settlements within fifteen counties were sorted based on census geographic and demographic data. In 2010, the number of households within the study area was 122,424 (U.S. Census Bureau, 2010c) . A total of 60 postal zip codes were identified using ArcGIS Â® software. According to the zip co des, the sampling area was slightly larger than the entire FL SRB boundary du e to edge conciliation (Figure 5 1). An initial random sample of 4,000 household mailing addresses was created by a professional marketing firm ( MSG , In c.). A comparison of the survey sample to the population using data from the American Community Survey five year estimates for 2007 2011 is shown in Table 5 3. Of the 4,000 households surveyed 2,000 households were assigned to receive survey set A and the other 2,000 were assigned to receive survey set B. The questionnaire and survey protocol were reviewed and approved by the Institutional Review Boards (IRB) at the University of Florida to ensure the rights of the
177 survey participants under the federal reg ulations. The standard mail survey approach, known as the Tailored Design Method (Dillman e t al., 2009) , was followed. This approach involves three principles to encourage response: reducing survey errors, developing a good set of survey procedures, and building positive social exchange. Each household was first contacted via an introductory postcard (September 12, 2012), then by a survey mailing with envelopes (September 19, 2012), followed by a thank you/reminder postcard (September 26, 2012). Two weeks later (October 17, 2012), a second mailing was sent, followed by a reminder postcard (Oct ober 24, 2012). By the middle of December, a total of 764 useable surveys had been returned (Figure 5 1). The overall response rate was about 19 percent. Analysis Demography, perceptions, and attitudes Demographic data were analyzed using descriptive stat istics. The frequency, percentage, mean, and standard deviation for each question were reported. The multinomial logit model was used in predicting the perception and attitude each respondent on the importance of ecosystem services. The explanatory variables for the multinomial logit model included gender, age, educational level, residency period, distance of geocoded household location from the Suwannee River, and Rur al Urban Commuting Area (Table 5 4). The gender, age, educational level, and residency period Suwannee River was computed using ArcGIS Â® software (Figure 5 1). Urban and non urban areas in this study were defined according to the United States Department of (Economic Research Services
178 (ERS), 2013) . The Rural Urban Commuting Area Codes (RUCAs) are based on the 2000 census and 2004 zip codes. This study consisted of 24 urban areas and 36 rural areas (Table 5 5 and Figure 5 1). The likelihood ratio test and Wald chi square statistics were reported for logistic regressions (Bewick et al., 2005; Engle, 1 984) . The likelihood ratio test evaluated the overall relationship between the independent variables and the dependent variables. It measured the difference between the maximum likelihood estimate (MLE) and the null value when the parameter is zero. The t est statistic is approximated by a chi squared distribution (Banerjee and Wellner, 2005; Bewick et al., 2005) . The Wald statistic test evaluated the significance of the individual coefficients in the logit model, a nd is distributed as a chi square with one degree of freedom. According to Banerjee and Wellner (2005) , Bewick et al. (2005) , and Tabachnick and Fidell (2001) , the likelihood ratio test is more powerful than the Wald test and is considered to be the superior method. T his study reports the results from both tests to determine the overall impacts of the individual coefficients between the multiple independent variables and categorical dependent variables. Choice experiment analysis This analysis focused on section 5 of the survey, the economic valuation part. These responses provided 2,690 observations, or 8,070 cases. Our CE approach assumed that an individual selected one of the specific program (A or B) options or the neither option. The nested logit model was the mos t appropriate method to describe this behavior. This way it put the opt out choice in a different nest from the ecosystem protection program choices. The nested logit was conducted with the SAS Â® software
179 using the Proc MDC (Multinomial Discrete Choice) f unction. A simple sche matic is illustrated in Figure 5 4. Two models were described. Model 1 included the opt out option (not interested in either program), while Model 2 excluded the opt out option. The inclusion of the opt out option in Model 1 would ca pture the real behavior of individuals, which indicated the likelihood of choices for all respondents and estimated the overall marginal willingness to pay (MWTP) for the population. Model 2 set the opt out option as a reference level, indicating the like lihood and MWTP only for respondents who chose program option A or B. Both models used effects coding for all variables except WTP that coded quantitatively (Bech and Gyrd Hansen, 2005; Jaccard, 2001) . The likelihood of preference for each ecosystem protection program was calculated relative to the alternat ives within each stratum. The model indicated how the attributes of a program affect the likelihood that an average respondent chose the program. The odds ratios and standard errors with a significant level were reported. The results should be interpreted as a preference ranking from most preferred attribute level to least preferred attribute level. To calculate the marginal WTP of each non monetary variable, a simple linear utility function provided the estimated coefficient of the non monetary variable d ivided by the estimated coefficient of the monetary variable as explained in Equation 4 8 (Krinsky and Robb, 1986) . Results Descriptive Statistics Out of the 4,000 households sampled, there was a total of 764 responses, or about a 19 percent response rate. All responses were used even though the number of
180 responden ts varied by each question. The percentage response rate and mean values corresponding to individua l questions are shown in Table 5 7. Demographics In general, respondents were more likely to be women (52%), who were late/middle aged to elderly (40% ages 4 5 64 years old and 32% ages 65 84 years old) and highly educated (29% with graduate or professional degrees, 21% with some areas and 22 percent lived in rural areas. T he primary land use reported was housing (73.5%), followed by other (13.5%), agriculture (6%), recreation (4%), and timber production (3%). Of all the households sampled, 84 percent participated in at least one environmental group or conservation organizat ion. Length of residency in this area was Of the respondents who answered the question, the majority of their annual household income fell into the first three categorie s: $25,000 $49,999 (20%), less than $25,000 (18%), and $50,000 $74,999 (16%). Because about 20 percent of the respondents preferred not to provide income information, the income variable was excluded from the explanatory variables in the multinomial logit models since the high number of non responses affected the performance of the logistic model predictions. Perceptions and attitudes on ecosystem s ervices Table 5 about ecosystem Respondents were familiar with water quality protection, carbon storage, climate regulation, and nutrient cycling, with mean levels of 2.23, 2.00, 2.00, and 1.79,
181 (2.92) the highest, f f 2.14). Follow caus ed by human activities (Figure 5 respondents completely believed that anthropogenic activities result in global climate change, followed by 25 percent somewhat completely believed this to be the case, 22 percent somewhat believed, 10 percent somewhat do not believe at all, and about 11 pe rcent did not believe at all. Participants were then divided into three groups based on how aggressive they believed government should be involved in global change policies (Fiture 5 4). In the survey, about 94 percent of the group wanting more aggressive government policies on climate change completely believe in government involvement, as opposed to only 2 percent of the group wanting less aggressive government policies and 4 percent of the group wanting no change. In contrast, only 3 percent of the grou p wanting more aggressive government policies on climate change do not believe in any government
182 policies, as opposed to 80 percent for the group wanting less aggressive government policies and 17 percent of the group wanting no change. When asked about e ffects due to climate change that respondents have observed in the community during the past 10 years, 20 percent of respondents reported not observing any changes. The remaining 80 percent experienced more drought than they did floods or hurricanes. Othe r climate effects included extreme temperatures in winter and summer, lower aquifer and lake levels, and dry wells. It should be noted that some believed that these effects were part of natural climate cycles and not climate change. The survey question a bout soil carbon sequestration/storage revealed that 52 stocks have increased, 22 percent believed they have decreased, and 14 percent answered no significant change. Wh en respondents were asked to respond about trade offs between productivity and water quality, most respondents did not agree with with a mean of 1.7 (using a scale of 1 Socio economic factors influencing p erceptio ns and a ttitudes The likelihood ratio tests from the multinomial logit modeling were calculated to evaluate the overall relationship between the independent varia bles and the dependent variable s (Table 5 9). The model suggested that among six socio economic variables, three factors, including gender (GEN), age (AGE), and educational levels (EDU), were es. The variable for length of residency (RESD), rural/urban commuting areas (RUCs), and distance from the main river to home (DIST) were not significantly different for all ecosystem services.
183 The results of the Wald statistics for the model indicated in dividual effects (Tables 5 10 t o 5 12). Respondents with more than a high school education were significantly more likely to be familiar with ecosystem service terms, more aware of the importance of ecosystem services, and more concerned about environmenta l issues as compared to respondents with less than a college education. In particular, respondents with less than a college degree were significantly less likely to be concerned about global climate change and showed less awareness of the importance of nat ure recreation and tourism as compared to respondents with a college or professional degree. However, different educational levels were not significant determinants to differentiate perceptions of the importance of clean air. In terms of gender, males (as compared to females) were more likely to be familiar with carbon storage terms, the importance of clean air, soil fertility renewal for better agricultural productivity, nature based recreation. Respondents over the age of 65 years old were significantly less likely to recognize specific terms, such as climate regulation and carbon storage, and were also less likely to be concerned about global climate change and poor water quality. Preferences Table 5 13 displays parameter estimates from the nested logi t models on how individuals choose the A, B, or opt out option. The underlining assumption was that every time respondents made a choice, they would select the alternative that showed the higher level of utility. The results from the two models (i.e., Mo dels 1 and 2) were reported. Model 1 included the opt out option, indicating the likelihood of choices for all respondents and overall MWTP for the population; Model 2 excluded the opt out option, indicating the likelihood and MWTP only for respondents wh o chose option A or B. Both
184 models estimate how the attributes of the ecosystem service program affect the probability that an individual would select that choice. Because the models were in the form of a logit transformation, the coefficients in the tab le are the log of odds. The positive and negative signs of the coefficients indicate the rank of preferences. To interpret these coefficients, they need to be anti logged to the odds ratio. The odds ratio shows how strongly the presence or absence of one v ariable is interpretable relative to the presence or absence of another variable being fixed in a given population (e.g., how many times option A or B is stronger than the opt out option). Ecosystem service preferences within the improvement g roup Models 1 and 2 showed similar preference rankings, although there were no statistically significant differences in preferences with respect to types of ecosystem service. In comparison within a choice block I (Table 5 13), in examining the coefficients, Model 1 and 0.045 for Model 2), followed by agricultural/ forestry production (0.005 for Model 1 and 0.024 for Model 2). Climate/carbon regulation was the least preferred am ong ecosystem service types to be managed ( 0.004 for Model 1 and 0.015 for Model 2). All of the services (i.e., nutrient control, agricultural/forestry production, and climate/carbon regulation) were ranked higher than the status quo. The odds ratios were calculated by comparing coefficients of each attribute level to the opt out option. Thus, the odds ratios from both Models 1 and 2 generated the same values. For example, the odds ratio of nutrient control to opt out option in Model 1 was exp 0.026 / exp 0 .019 = 1.05 and Model 2 was exp 0.045 / exp 0 = 1.05 as well. This suggests that the odds of preferring nutrient control management were 1.05 times the odds of preferring
185 not to have any programs, when all else being equal. The odds ratios were 1.02 for the were 1.02 times the odds of favoring not to choose any ecosystem protection program. the odds of favoring not to implement any programs when other variables were held constant. ervice Ecosystem service preferences within the p rogram administration g roup The findings offer some interesting insights into whi ch organizational approach to government) to manage ecosystems among the other three options of administrative management (SRWMD, government, NGO). In comparison within a choic e block II (Table 5 while the coefficients of the SRWMD were 0.048 for Model 1 and 0.067 for Model 2. on. 0.103 for Model 1 and 0.085 for Model 2. Respondents out option. Although no statistical significance was out ( exp 0.0 82 / exp 0.019 for Model 1 and exp 0.101 / exp 0 for Model 2). The respondents preferred preferred to opt out by the odds of 0.92 times.
186 From the models, the administration attr ocation of c ategory Respondents expressed strong pre ference for ecosystem protection programs to be implemented anywhere within the basin. This can be explained by the high positive estimated coefficients of 0.222 (Model 1 with statistical significance) and 0.241 (Model 0.272 for Model 1 and 0.253 for Model 2) or the opt out option. The odds ratio s indicated that the odds of preferring to have the preferring not to have any ecosystem programs. This odds ratio was calculated by exp 0.222 / exp 0.019 in Model 1 and exp 0.241 / exp 0 in Model 2. The odds of favoring to have to the opt out option were 1.10 times. The odds of preferring to implement ecosystem protection programs within 5 miles of preferring not to have ecosystem protection programs, with all else being equal (choice block III, Table 5 13). ecosystem protection
187 Economic Estimates Ta ble 5 13 presents the annual household MWTP results for each attribute to protect the ecosystems through donations, taxes, or utility fees over a period of five years. Overall, the MWTP in Model 2 was higher because households were willing to accept the pr oposition of an ecosystem protection program with associated costs. High and low MWTP reflect the order of preferences for each attribute. For example, respondents showed a greater preference for the ecosystem protection program to be implemented througho ut the whole river basin. To demonstrate the procedure for calculating MWTP, consider the value $1.55 in Model 1. The average WTP cost coefficient was 0.017 and the coefficient of nutrient control was 0.026. Therefore, the MWTP was derived by ( 0.026/ 0.017) = $1.55 in Model 1. This indicated that the inclusion of nutrient control improvement would support an increase of $1.55 per household, per year. Positive values of MWTP indicate the willingness to pay in $, whereas negative MWTP values express th e opposite (i.e., the unwillingness to pay/engage in that specific service). The largest MWTP was $13.22 es for ecosystem services (climate/carbon regulation; nutrient control; and agricultural and forestry production) were very small with less than $1.55 (Model 1). Even smaller MWTP values were found for Model 2. system service protection program was estimated to be $0.12 for the model that included the opt out option
188 (Model 1) and $1.24 for the model that excluded the opt out option (Model 2). The most highly preferred combination across attributes of Model 1 had an overall average MWTP of $6.55 (derived through: (1.55+4.88+13.22)/3), while Model 2 showed $7.67 (derived through: (2.67+5.99+14.35 ) /3). In 2010, the number of households in fifteen counties was 122,424, which gave an average MWTP approximately of 122, 424*0.12 = $14,691 per year (Model 1) and 122,424*1.24 = $151,806 per year (Model 2) to implement ecosystem protection programs. Discussion Socio demographic, Geographic Factors and Ecosystem Services Socio demographic factors influencing ecosystem servic es perceptions and attitudes are shown to be fairly similar to previous studies. Educational level and age, for example, are positively associated with knowledge, awareness, and concerns; that is, respondents who are younger and have a higher educational level are the most concerned about the environment (Franzen and Meyer, 2010; Gelissen, 2007; Marquart Pyatt, 2008; Marquart Pyatt 2007) . This study found that the level of education highly influenced the level of environmental concern among the respondents (Engel and PÃ¶tschke, 1998; Liere and Dunlap, 1980; Olofsson and Ã–hman, 2006; Scott and Willit s, 1994) . However, regardless of educational level, social consciousness about environmental issues ideally should be developed at an early age. It was pointed out by numerous researchers that raising environmental awareness is more important than seekin g higher education to protect ecosystem services (Arslan, 2012; Pooley and . The relationship between respondent age and ecosystem services perspectives and attitudes are less clear in the FL SRB. Th e late middle aged group (45 64 years)
189 showed a higher degree of environmentally friendly attitudes than the elderly group (greater than 65 years) in 5 of the 14 areas analyzed. The young to middle aged respondents (18 44 years) showed higher perception a nd attitude levels toward ecosystem services, although the results were insignificant. Some previous studies indicated that younger people tended to be more concerned about environmental issues than older people (Engel and PlÃ¶tschke, 1998 ; van Liere and Dunlap, 1980 ). Olofsson and Ã–hman (2006) found the same pattern although it was a weak relationship. A possible explanation in regard to the relationship between age and ec osystem service perceptions and attitudes is that younger people are more open to change in environmental attitudes (United Nation Environment Programme (UNEP) and United Nation Educational, Scientific, and Cultural Organization (UNESCO), 2001) . In the FL SRB, i be more concerned about a sustainable future for their children. The results concerning gender are more ambiguous. While a higher number of responses came from females, males were more likely to recognize the importance of ecosystem services (i.e., familiarity with carbon storage, water quality protection, clean air, soil fertility, global climate change concerns, poor soil fertility, and agricultural productivity) than were f emales. These findings were similar to findings from other studies (e.g. Ofei Manu, 2009; Steel, 1996; Stern and Dietz, 1994; Tindall et al., 2003; Zelezny et al., 2000) . McEvo y (1972) argued that because males were more active in politics and more educated than females in general, they had more concerns over environmental problems. A comparative study between the United States and twelve European countries in relation to gender and env ironmental views undertaken by
190 Somma and Tolleson Rinehart (1997) found that there was no difference between American men and women regarding perspectives on the environm ent, but that there was a significant difference between European men and women, with European men being more concerned about the environment than their female counterparts. The gender based differences in environmental perceptions and attitudes between the American and European groups probably was probably due to differences in the two (Kashima et al., 2014) as well as social and ethical concerns (Oughton, 2013) . The length of residency, residency areas, and distance from the main stream of els. Long time residents preferred balancing ecosystem protection and river resource use, whereas newcomers preferred ecosystem preservation (McCool and Martin, 1994; Vaske et al., 2001) . Dissimilar results were reported by Cantrill (1998) in his study of people living in the Lake Superior basin (LSB). He found that residents who had lived less than fifteen years in the LSB referenced the ecosystems in their communication, while long time residents strongly recognized the natural environment in their sense of place and were more sensitive to huma n impacts than short time residents (Sheldon et al., 1984) . In the case of residency zones (i.e., urban, large rural town, and small/isolated town), the relationship between residency areas and attitude towards ecosystem services was insignificant. However, urban residents were slightly more like ly to be environmentally concerned than rural residents in most areas. Similar findings were also reported in Trembley and Dunlap (1978) and Yu (2014) , in which they concluded that urban residents were generally exposed to higher pollution levels and received more
191 information when compared to rural residents. On the other hand, rural residents rely on nat ural resources for their living, thus focus more on issues directly related to agriculture production. Despite the levels of perceptions and attitudes, when it came to willingness to location to be managed dominated other factors. Interestingly, people were more willing to pay for an ecosystem management service farther away from their homes rather than closer to their homes. It is possible that although residents would like to live in a well managed ecosystem, they are reluctant to be bothered by some activities in their neighborhood areas. To summarize this section, individual level factors such as education, age, and gender affected ecosystem service perception and attitude, while place related factors such as duration of residency, residency area, and distance between the main river to homes showed a muted effect. It should be noted that levels of perception and attitude are not necessarily followed by environmentally friendly beh avior. Rather, the results from this study provide an understanding of the socio demographic characteristics of people in the FL SRB considering the context of ecosystem services. Other factors such as general beliefs, political affiliation, and behavior s should be further analyzed in future studies. Scale Sensibility One of the ways of analyzing ecosystem service preferences and levels of willingness to pay is scale sensibility. In this study three types of preference attribute scales were used: knowled ge (type of ecosystem services preferences), political (administrative agency preferences), and spatial (managed location preferences).
192 When considering ecosystem management with regards to type of services, a knowledge scale is necessary. Among the inve stigated ecosystem services in this study, the most significant need of households is better water quality. This implies that residents in the FL SRB are concerned about water quality impairment in the basin because it directly impacts their drinking water (e.g., wells). An increasing trend of nutrient loads (total nitrogen and total phosphorus) downstream in the Santa Fe River and the Upper and Lower Suwannee River, and excessive nutrient concentration in four drainage areas (see Chapter 3) are somehow ass water quality in the FL SRB. Residents receive more information about clean water protection issues from agencies, such as the federal Environmental Protection Agency (EPA), SRWMD, the Suwannee River Partnership (SRP) , and environmental groups through targeted water quality campaigns and educational programs than they do ent, the funds will target water restoration, rather than soils or climate change (Florida Department of Environmental Protection (FDEP), 2014) . One of t he reasons for climate regulation and carbon sequestration are ranked as less important because there is often great disparity between perceptions concerning long term global climate change and current local/regional perception of ecological problems (Hulme, 2010) . Because climate change and carbon sequestration occur over a long time scale, they are not perceived as being of immediate local concern (Garvey, 2008) . In addition, as Gl obalScan (2009) pointed out, people have become more pessimistic about the environm ent, with climate change confidence slipping from first to fourth in the major issues that worry people the
193 science (Kahan, 2012) . As a result, the anthropogenic climate ch ange controversy continues. Cook et al. (2013) analyzed 11,944 climate abstracts and found that 66.4 percent of all the abstracts showed no position of anthropogenic global warming, 32.6 percent endorsed human causing global warming, and 0.7 percent de nied any anthropogenic global warming. Weber (2013) discussed the idea that personal experience and beliefs of individuals strongly affect their preferences. Results from the survey show that political scale is an impo rtant management factor because the farther the governing/administrative agency is removed from the community, the less trust and engagement occurs in the community. As explained in (2004) , the interaction between local (county) government agencies and stakeholders (residents) is crucial since local agencies act as mediators between stakeholders and state/federal governing agencies (Lubell, 2004) . For exa mple, the SRWMD is the state agency that regulates water issues in the FL SRB, and is accountable to the state government, not the local government. Local agencies often allow more flexibility in customizing regulations to local situations. This explains why the FL SRB, rather than state (SRWMD) government agencies or federal (EPA) and non governmental agencies. One of the reasons for the least favor for NGO may be du e to the long time for trust development in these agencies. For example, the Okeechobee Basin project took five years to build trust in the NGO involved (M. Clark, personal communication, September 12, 2013. In the bigger picture, NGOs are not as
194 financi ally independent as the government upon whom they are often financially dependent (Bebbington et al., 2005; Shah, 2005) . Spatial scale as it relates to location management preferences in this study is used to determine whether the effect of an ecosystem service improvement or disturbance takes place locally or at a distant location (RodrÃguez et al., 2006) . Most families have lived in the FL SRB for several generations. They are conservative, religious, and community oriented (T. Obreza, personal communication, September 12, 2013). As such, they do not want to be regulated and they believe that they know best how to manage the land. This explains why implementing any programs within 5 to 20 miles from these households was not favored. As stated by Obreza (personal communication, September 12, 2013), people in the FL SRB have a mindset where mentioned in studies by Bengtsson et al. (2003) and van Jaarsveld et al. (2005) , adverse impacts may occur as a result of the mismatc h between the intent of the watershed decision and the area of implementation. An attempt to maintain or increase one service may cause substantial declines in other services on a broader (basin) scale (Tilman et al., 2002; Vidal Legaz et al., 2013) . For example, nitrogen and phosphorus loads per unit area in the Upper Suwannee and Lower Suwannee River Basins increased significantly between 2000 and 2010. I ncreases in nutrient loading coincided with increases in soil carbon sequestration rates (Chapter 3) and crop and livestock sales along these sub basins. Economic Valuation in Choice Experiments (CE) In the context of ecosystem services conservation and p lanning, the CEs used in this survey can provide the bottom line for decision making. The average household
195 MWTPs per year in Model 1 ($0.12) and Model 2 ($1.24) were much lower than the findings from a study by Kreye et al. (2011) in the same area. They estimat ed an annual household WTP of $4.29 for water quality programs that use land acquisition and $70.72 for programs that do not use easement type strategies. Also, our results indicated low values in particular when compared to the total value of ecosystem s ervices in the Lake Okeechobee watershed, Florida provided by Shrestha and Alavalapati (2004) . In their study, the pub lic willingness to pay for improvements in water quality, carbon sequestration, and wildlife habitat through silvopasture practice was $ 30.24 $71.17 per household per year for five years derived by a random parameter logit model . A carbon sequestration imp rovement program alone was $58.05 per household per year at the moderate environmental improvement level . In the Everglades, Florida, willingness to pay for the full restoration plan was measured approximately $60 $70 per household per year over a ten year period (Milon et al., 1999) . Hulme (2010) argued that there are risks associated with monetary estimations for climate change. He cautioned that when long term environmental problems are converted into a single monetary metric, it provides biased interpretation, particularly at a global scale. Ecosystem services are considered externalities and are associated here is no direct ownership. The payment for the watershed ecosystem services program, for example, reported low WTP to improve downstream water resources (~1% of annual income), which is also considered local commons (Neef and Thomas, 2009) . However, de Groot et al. (2012) argued that when the data are collected systematically at coarser scale and converted to a universal common set of units ($/unit area/year), the monetary
196 expressions of ecosystem services are meaningful. Global Scan (2009) reported that sustainability experts in 76 countries broadly viewed eco nomic instruments as the most effective means to combat the climate change issue. While economic incentives are not the panacea to solve environmental issues, the valuation method can be used as an additional tool to other integrative instruments. Input f rom natural and social scientists are recommended to improve the CE method. Above all, the most important point of the ecosystem service valuation of this study is not about expressing values in monetary units, but giving guidance in understanding public preferences and the relative current values. These preferences and values aim to prioritize managing or allocating resources between competing demands and sustainable uses. Conclusions The preservation of water, soil, and clean air has profound implication s for the sustainability of goods and benefits provided to residents in the FL SRB that are embedded within larger economies, global systems (e.g., global climate system and global carbon and nutrient cycles), socio cultural and political contexts (e.g., f ederal regulations by the United States Environmental Protection Agency). The beliefs and perspectives of local residents of the basin identified nutrient control (water quality) as the most important service, agricultural and forestry production as somew hat important, and climate regulation and carbon sequestration were ranked as least important ecosystem services. This stands in sharp contrast to scientific based perspectives derived from empirical observations, monitoring of water quality, climatic tren ds, and watershed modeling that attest that the global climate change phenomena exists and drinking water quality in the basin is not imminently threatened due to long term efforts
197 focused on implementation of best management practices and environmental re gulations. The sequestration of carbon in soils and biomass in the basin is substantial and offers opportunities to ad apt to climate change impacts (C hapter 4). These findings suggest that subjective individual perspectives and objective science based ecos ystem knowledge is profoundly divergent. Politics and decisions in regard to land and water being and system based perspectives. The latter viewing the natural capital of e cosystems as resources to sustain regional communities and societies at larger scale (i.e., across the whole FL SRB) that are in competition with economic, human capital, and other interests. This scale sensitivity coupled with low value attribution to key ecosystem services and environmental consciousness by residents has numerous implications. In such case, tension is likely to arise that mutes the willingness for voluntary (e.g. action groups), mandatory (through regulation), or monetary actions to maint ain and/or enhance ecosystem services. In addition, stakeholder engagement and embracing of political instruments to optimize ecosystem services is substantially hampered. For example, stringent greenhouse gas emission regulations, conservation programs ta rgeting soil carbon sequestration, carbon tax, carbon trust funds or emission trading banks to enhance the climate and carbon regulation ecosystem services are less likely to be embraced by local residents if they do not value these services, as indicated in this study. Some limitations in understanding the complexity of climate, relatively low household income (about 40% of households earn less than $50,000/yr), educational level, low participation in environmental groups (< 13%) may have
198 contributed to the low valuation of the climate and carbon regulation services, and higher benefits associated with agricultural and forestry production, and nutrient control services. The d issociation between local needs of residents (e.g., drinking water from the regional Floridan aquifer) and global phenomena and issues (e.g., global climate change, clean air) that are abstract and challenging to experience directly (e.g., rising mean glob al temperatures can be measured but not sensed individually) may explain the overall low valuation of climate and carbon regulation services. Respondents welcome the implementation of ecosystem services anywhere in the basin, but not close to their home ( within 5 miles). They trust most county and regional agencies (SRWMD) than NGOs (e.g., Suwannee River Partnership) in efforts to enhance services provided by ecosystems. Ecosystem services focus on the benefits that people derive from ecosystems, which are composed of soil, water, air, and vegetation. All of them are part of the Global Commons, that are essential for the survival and lifelihood of people (e.g., clean drinking water resources, food production, etc.), but are too often taken for granted; and thus, devalued over profit, economic growth, or materialistic individualistic goals. Over 65 percent of the residents in the FL SRB live in urban centers that may be somewhat dissociated from soil, water, and food capital geographically and mentally associ ated with rural settings. According to the choice experiment survey the willingness of residents to pay for ecosystem services was extremely low with annual MWTP of $0.12 (Model 1) and $1.24 (model 2), much less than a meal at a fast food restaurant. The se findings suggest that residents in the FL SRB have other preferences than services provided by the very basin they live in. Interestingly, this is somewhat confounded by the fact that the level of
199 outdoor activities (e.g., hiking, tubing in rivers, fish ing) and appreciation of beautiful natural areas (e.g., State Parks) in the FL SRB is quite high. This study provided evidence that the perception and valuation of ecosystem services is based on the beliefs, needs, and preferences people associate with e cosystems. The spectrum of diverse responses to the ecosystem service survey was vast and provides an important foundation for politics and decision making. We believe that the coupling of socio cultural valuation, economic assessment, and political instru ments is saliently important to sustain and enhance ecosystem services.
200 Table 5 1 . Attributes and attribute levels in choice experiments . Scheme attribute Attribute levels Type of ecosystem services to be managed Climate/carbon regulation Nutrient contr ol Agricultural and forestry production Status quo Program administration Suwannee River Water Management District County (government) Non government organizations (NGO) Status quo Location of area to be managed Anywhere within the district Within 20 miles from your home Within 5 miles from your home Status quo Willingness to pay $5 $25 $50 Status quo
201 Table 5 2 . Optimal fractional factorial design survey of four 3 level attributes of ecosystem services survey in the Suwannee River Basin, Florida . Choice set Program A Program B Ecosystem services Program administration Location WTP 1 Ecosystem services Program administration Location WTP 1 Climate/carbon regulation SRWMD 2 Anywhere 4 $5 Nutrient control NGO Within 20 mi $50 2 Climate/carbon reg ulation County Within 20 mi $50 Nutrient control SRWMD Within 5 mi $25 3 Climate/carbon regulation NGO 3 Within 5 mi $25 Nutrient control County Anywhere $5 4 Nutrient control SRWMD Within 20 mi $25 Agricultural/fore st production NGO Within 5 mi $5 5 Nutrient control County Within 5 mi $5 Agricultural/fore st production SRWMD Anywhere $50 6 Nutrient control NGO Anywhere $50 Agricultural/fore st production County Within 20 mi $25 7 Agricultural/forest production SRWMD Within 5 mi $50 Climate/carbon regulation NGO Anywhere $25 8 Agricultural/forest production County Anywhere $25 Climate/carbon regulation SRWMD Within 20 mi $5 9 Agricultural/forest production NGO Within 20 mi $5 Climate/carbon regulation County Within 5 mi $50 1 Willingness to pa y 2 Suwannee River Water Management District (SRWMD) 3 Non government organization (NGO) 4 Anywhere within the Suwannee River Water Management District
202 Table 5 3 . Number of households sampled and number in population in the study area . County HH 1 % HH HH sampled % HH sampled HH response % HH response Alachua 2 97,542 40.32 1,837 45.93 422 55.38 Baker 2 8,333 3.44 67 1.68 7 0.92 Bradford 2 9,188 3.80 163 4.08 18 2.36 Columbia 24,127 9.97 491 12.28 67 8.79 Dixie 5,380 2.22 98 2.45 16 2.10 Gilchrist 6,009 2.48 112 2.80 23 3.02 Hamilton 4,441 1.84 90 2.25 11 1.44 Jefferson 2 5,313 2.20 106 2.65 17 2 .23 Lafayette 2,474 1.02 33 0.83 12 1.57 Levy 2 16,034 6.63 272 6.80 52 6.82 Madison 6,939 2.87 132 3.30 15 1.97 Putnam 2 29,061 12.01 44 1.10 13 1.71 Suwannee 15,810 6.53 321 8.03 59 7.74 Taylor 7,632 3.15 148 3.70 25 3.28 Union 3,665 1.51 86 2.15 5 0.66 Total 241,948 100.00 4,000 100.00 762 100.00 1 HH denotes households 2 Only portions of Alachua, Baker, Bradford, Jefferson, Levy, and Putnam Counties were included in the study area; however, Table 4 3 shows the total number of households in each county. Source: The American Community Survey five year county estimates for 2007 2011.
203 Table 5 4 . Explanat ory variables used in multinomial logit models . Variables Description Gender 1 = Female 2 = Male (base group) Age 1= 18 to 44 years 2 = 45 to 64 years 3 = 65 years and greater (base group) Educational level 1 = Elementary and high school 2 = Some coll Residency period 1 = Less than 10 years 2 = 10 to 19 years 3 = More than 19 years (base group) Rural Urban Commuting Area Codes (RUCs) 1 = Isolated small rural town 2 = Large rura l town 3 = Urban (base group) Distance in miles (kilometers)
204 Table 5 5 . Zip codes and Rural Urban Commuting Area Codes (RUCAs) within the Suwannee River Basin, Florida . FID ZIP County RUCAs FID ZIP Count y RUCAs 1 32008 Suwannee 10.5 31 32350 Madison 10.4 2 32013 Lafayette 10 32 32356 Taylor 8 3 32024 Columbia 5 33 32359 Taylor 8 4 32025 Columbia 4 34 32605 Alachua 1 5 32038 Columbia 5.2 35 32606 Alachua 1 6 32040 Baker 7.1 36 32607 Alachua 1 7 3204 4 Bradford 2 37 32608 Alachua 1 8 32052 Hamilton 7 38 32609 Alachua 1 9 32053 Hamilton 10.4 39 32615 Alachua 2 10 32054 Union 7.3 40 32618 Alachua 2 11 32055 Columbia 4 41 32619 Gilchrist 10.4 12 32058 Bradford 8.3 42 32621 Levy 2 13 32059 Madison 10 .6 43 32622 Bradford 2 14 32060 Suwannee 7.4 44 32625 Levy 10 15 32061 Columbia 4 45 32626 Levy 10 16 32062 Suwannee 6 46 32628 Dixie 7 17 32064 Suwannee 7 47 32631 Alachua 2 18 32066 Lafayette 10 48 32640 Alachua 2 19 32071 Suwannee 10.5 49 32641 Al achua 1 20 32083 Union 9 50 32643 Alachua 2 21 32087 Baker 8.1 51 32648 Dixie 10.6 22 32091 Bradford 7.3 52 32653 Alachua 1 23 32094 Suwannee 6 53 32666 Putnam 2 24 32096 Hamilton 5 54 32668 Levy 3 25 32331 Madison 10 55 32669 Alachua 2 26 32336 Jef ferson 2 56 32680 Dixie 7 27 32340 Madison 7 57 32693 Gilchrist 10.4 28 32344 Jefferson 10.1 58 32694 Alachua 2 29 32347 Taylor 7 59 32696 Levy 2 30 32348 Taylor 7 60 34449 Levy 10 RUCAs scheme used: Urban: 1.0, 1.1, 2.0, 2.1, 3.0, 4.1, 5.1, 7.1, 8.1 , and 10.1 Large rural city/town: 4.0, 4.2, 5.0, 5.2, 6.0, and 6.1 Small and isolated small rural town: 7.0, 7.2, 7.3, 7.4, 8.0, 8.2, 8.3, 8.4, 9.0, 9.1, 9.2, 10.0, 10.2, 10.3, 10.4, 10.5, and 10.6 Source: United States Department of Agriculture, Economic Research Service, 2004
205 Table 5 6 . Variables for socio economic analysis . Variables Description Survey section Scale range ID Survey codes FAM_CLI Familiarity of climate regulation term 1 1 3 FAM_CAR Familiarity of carbon storage term 1 1 3 FAM_WA T Familiarity of water quality protection term 1 1 3 FAM_NUT Familiarity of nutrient cycling term 1 1 3 IMP_AGR Importance of agricultural production 1 1 3 IMP_FOR Importance of forest production 1 1 3 IMP_WATQ Importance of good water quality 1 1 3 I MP_AIRQ Importance of clean air 1 1 3 IMP_SOIL Importance of renewing soil fertility 1 1 3 IMP_TOUR Importance of nature recreation and tourism 1 1 3 CONC_CLICH Concerns about global climate change 1 1 3 CONC_WATQ Concerns about poor water quality 1 1 3 CONC_SOIL Concerns about poor soil fertility 1 1 3 CONC_AGRI Concerns about low agricultural productivity 1 1 3 BELIEF Degree of belief in global climate change which is caused by human activities 2 1 5 DROUGHTS More droughts in the past 10 years 2 1 =yes, 0 =no FLOODS More floods in the past 10 years 2 1=yes, 0 =no FIRES More natural fires in the past 10 years 2 1=yes, 0 =no STORMS Increase in the intensity of storms or hurricanes 2 1=yes, 0 =no OTHER_EFF Other effects due to climate change 2 NA NOCHANGE Have not observed any effects caused by climate change in the past10 years 2 1=yes, 0 =no POLICIES Degree of government policies enforcement 2 1 5 SOC_SEQ Soil carbon sequestration/storage 3 1=yes, 0 =no PROvsWAT Agreement between farm produc tivity and water quality trade offs 4 1 5 RNOF_FARM Runoff from farms degrades water quality 4 1=yes, 0 =no RNOF_URB Runoff from urban areas degrades water quality 4 1=yes, 0 =no INDTRY Industry discharges degrade water quality 4 1=yes, 0 =no SEPTIC L eakage from septic systems degrades water quality 4 1=yes, 0 =no OTHER_POL Other pollution sources 4 NA Do not know what causes water quality degradation 4 1=yes, 0 =no LU CONTROL Improve water quality by having stricter land use controls 4 1 =yes, 0 =no REGULATION Improve water quality by stating regulations for surface water runoff 4 1=yes, 0 =no EDUCATE Improve water quality by educating residents about water conditions 4 1=yes, 0 =no WAT_TREAT Improve water quality by building water trea tment facilities 4 1=yes, 0 =no
206 Table 5 6. Continued. Variables Description Survey section Scale range OTHER_IMPROVE Other measures to improve water quality 4 NA DON'T KNOW Do not know how to improve water quality 4 1=yes, 0 =no RESPDNT Respondents 5 1 641 SET Questionnaire set A or B 5 A/B QUEST Choice sets 5 A = 1 5, B = 1 4 ALT Alternatives; Program A, Program B, not interested 5 0,1,2 RES Response to alternatives, give 1 if the alternative is selected and 0 when it is not 5 1=yes, 0 =no ASC Alternative specific constant for each program A and B 5 1=yes, 0 =no CAR Dummy variables for the climate/carbon storage 5 1=yes, 0 =no NUT Dummy variables for the nutrient control 5 1=yes, 0 =no AGR Dummy variables for the agricultural and fore stry production 5 1=yes, 0 =no WMD Dummy variables for the Suwannee Water River Management District 5 1=yes, 0 =no CNT Dummy variables for the county 5 1=yes, 0 =no NGO Dummy variables for the non government organization 5 1=yes, 0 =no ANY Dummy va riables for managing ecosystems anywhere within the Suwannee River Basin area 5 1=yes, 0 =no IN20 Dummy variables for managing ecosystems within 20 miles of their homes 5 1=yes, 0 =no IN5 Dummy variables for managing ecosystems within 5 miles of their homes 5 1=yes, 0 =no WTP Dummy variables for the annual amount per household that willing to pay over a period of 5 years 5 $5, $25, $50 STR Stratification variable indentifying each combination of respondents and questions 5 NA GEN Gender of respond ent 5 1 2 AGE Age of respondent 5 1 5 EDU Education level of respondent 5 1 6 RESD Residency period of respondent 5 1 6 INCM Annual household income of respondent 5 1 6 Variables used in choice experiment analysis
207 Table 5 7 . Demographic statistics of survey respondents Variables Number of responses Percentage (%) Gender (n = 734) Male Female Prefer not to answer 330 380 24 44.96 51.77 3.27 Age (n = 740) 18 24 years 25 44 years 45 64 years 65 84 years 8 5 years and greater Prefer not to answer 30 151 294 235 14 16 4.05 20.41 39.73 31.76 1.89 2.16 Educational levels (n = 735) Elementary school (through 9 th grade) High school or GED Some college Bach Graduate or professional degree Prefer not to answer 7 108 154 94 135 215 22 0.95 14.69 20.95 12.79 18.37 29.25 3.00 Agricultural land ownership (n = 734) Yes No 158 576 21.53 78.47 Primary land use (n = 720) Agriculture Timber production Recreation Only housing Others 44 24 26 529 97 6.11 3.34 3.61 73.47 13.47 Residency period in the area (n = 743) Less than 10 years 10 14 years 15 19 years 20 24 year s 25 30 years More than 30 years 211 94 69 65 68 233 3 28.40 12.65 9.29 8.75 9.15 31.36 0.40 Environmental group participation (n = 729) Yes No 96 615 18 13.17 84.36 2.47 Annua l household income (n =725) Less than $25,000 $25,000 $49,999 $50,000 $74,999 $75,000 &99,000 $100,000 $149,000 $150,000 or more Prefer not to answer 132 142 115 85 70 37 144 18.21 19.59 15.86 11.72 9.66 5.10 19.86 Rural Urban Commuting Area Rural Urban 264 498 34.65 65.35
208 Table 5 8 . Descriptive statistics of respondent attitudes about ecosystem services (on scale of 1 3) Variables Number of responses Percentage (%) Mean Standard deviation Part 1 Familiarity of ecosystem services Familiarity of climate regulation (n=746) Very familiar (scale 3) Somewhat familiar (scale 2) Not familiar (scale 1) 147 353 209 37 19.70 47.32 28.02 4.96 2.00 0.80 Familiarity of carbon storage (n=737) Very familiar (scale 3) Somewhat familiar (scale 2) Not familiar (scale 1) 116 250 330 41 15.74 33.92 44.78 5.56 2.00 0.85 Familiarity of water quality protection (n=744) Very familiar (scale 3) Somewhat familiar (scale 2) Not familiar (scale 1) 310 313 96 25 41.67 42.07 12.90 3.36 2.23 0.82 Familiarity of nutrient cycling (n=740) Very familiar (scale 3) Somewhat familiar (scale 2) Not familiar (scale 1 ) 176 264 263 37 23.78 35.68 35.54 5.00 1.79 0.89 Part 2 Importance of ecosystem services Importance of agricultural production (n=745) Very important (scale 3) Somewhat important (scale 2) Not important (scale 1) 454 229 37 25 60.94 30.74 4.96 3.36 2.49 0.74 Importance of forest production (n=740) Very important (scale 3) Somewhat important (scale 2) Not important (scale 1) 442 221 52 25 59.73 29.86 7.03 3.38 2.46 0.77 Importance of good water quality (n=752) Very important (scale 3) Somewhat important (scale 2) Not important (scale 1) 711 26 4 11 94.55 3.46 0.53 1.46 2.92 0.46 Importance of clean air (n=747) Very important (scale 3) Somewhat important (scale 2) Not important (scale 1) 685 46 4 12 91.70 6.16 0.54 1.60 2.88 0.46 Importance of renewing soil fertility (n=741) Very important (scale 3) Somewhat imp ortant (scale 2) Not important (scale 1) 500 196 22 23 67.48 26.45 2.97 3.10 2.58 0.70
209 Table 5 8. Continued. Variables Number of responses Percentage (%) Mean Standard deviation Importance of nature recreation and tourism ( n=741) Very important (scale 3) Somewhat important (scale 2) Not important (scale 1) 390 278 56 17 52.63 37.52 7.56 2.29 2.40 0.73 Part 3 Concerns about ecosystem services Concerns about global climate change (n=754) Very concerned (scale 3) Somewhat concerned (scale 2) Not concerned (scale 1) 301 269 169 15 39.92 35.68 22.41 1.99 2.14 0.83 Concerns about poor water quality (n=749) Very concerned (scale 3) Somewhat con cerned (scale 2) Not concerned (scale 1) 509 184 46 10 67.96 24.57 6.14 1.33 2.62 0.97 Concerns about poor soil fertility (n=746) Very concerned (scale 3) Somewhat concerned (scale 2) Not concerned (scale 1) Do 296 339 86 25 39.68 45.44 11.53 3.35 2.21 0.78 Concerns about agricultural productivity (n=751) Very concerned (scale 3) Somewhat concerned (scale 2) Not concerned (scale 1) 312 317 92 30 41.54 42.21 12. 25 4.00 2.22 0.81
210 Table 5 9 . Multinomial logit analysis of socio the likelihood ratio test . Statement Chi Square GEN AGE EDU RESD RUCs DIST Familiarity of climate regulat ion 9.31* 9.46* 33.68* 6.01 2.65 1.78 Familiarity of carbon storage 15.25* 13.70* 31.97* 2.17 2.27 1.63 Familiarity of water quality protection 5.99* 1.56 40.31* 5.95 6.47 1.83 Familiarity of nutrient cycling 3.08 6.14 23.06* 3.45 2.19 1.56 Importance of agricultural production 1.34 0.90 20.95* 3.56 4.62 1.12 Importance of forest production 5.88 6.19 16.49* 2.78 9.19 2.35 Importance of good water quality 3.61 4.03 11.40* 4.08 2.31 0.99 Importance of clean air 10.36* 8.10 5 .77 1.05 2.46 1.18 Importance of renewing soil fertility 28.87* 5.83 10.66* 2.55 1.66 0.37 Importance of nature recreation and tourism 2.71 9.96* 19.29* 1.64 0.43 5.46 Concerns about global climate change 34.72* 9.80* 16.79* 5.10 8.31 3.20 Conc erns about poor water quality 4.76 13.96* 10.84* 5.81 1.23 0.06 Concerns about poor soil fertility 26.41* 5.23 16.24* 1.06 1.41 0.00 Concerns about agricultural productivity 14.57* 2.18 23.55* 3.58 0.80 0.48 Degree of freedom 2 4 4 4 4 2 *Signifi cance at the 95% confidence level (p < 0.05).
211 Table 5 10 . Multinomial logit estimates of respondent familiarity about ecosystem services using Wald statistic tests . Variables Not/Somewhat familiar know Odds ratio Std error Odds ratio Std error Climate regulation Gender a Female 1.23 0.33 0.66 0.37 Age b 18 44 years 45 64 years 1.88 2.84* 0.48 0.39 1.77 3.35* 0.54 0.43 Educational levels c Elementary school & high school 0.28* 0.69 0.44 0.44 0.05* 0.40 0.61 0.48 Residency period d Less than 10 years 10 19 years 1.26 0.95 0.42 0.40 0.79 1.19 0.48 0.44 RUCs Isolated small rural town Large rural town 0.45 0.53 0.61 0.61 0. 60 0.77 0.60 0.72 Distance Distance from Suwannee River 1.01 0.01 1.02 0.02 Carbon storage Gender a Female 1.13 0.30 0.78* 0.36 Age b 18 44 years 45 64 years 2.04 2.47* 0.45 0.35 2.38 4.19* 0.52 0.41 Educational le vels c Elementary school & high school 0.33* 0.71 0.40 0.39 0.05* 0.44 0.65 0.44 Residency period d Less than 10 years 10 19 years 1.28 0.86 0.40 0.36 1.23 1.15 0.46 0.42 RUCs Isolated small rural town La rge rural town 0.54 0.48 0.79 0.58 0.65 0.73 0.58 0.71 Distance Distance from Suwannee River 1.01 0.01 1.00 0.01 a Reference category for gender is male. b Reference category for age is the age of greater than 65 years. c Reference category for ed d Reference category for residency period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
212 Table 5 10. Continued. Variables Not/Somewhat familiar Very know Odds ratio Std error Odds ratio Std error Water quality protection Gender a Female 0.90 0.40 0.61 0.41 Age b 18 44 years 45 64 years 1.19 1.49 0.57 0.45 0.97 1.40 0.58 0.46 Educational levels c Elementary school & high school 0.32* 1.41 0.51 0.58 0.09* 1.02 0.54 0.58 Residency period d Less than 10 years 10 19 years 1.02 0.53 0.52 0.45 0.70 0.54 0.54 0.47 RUCs Isolated small rural town Large rural town 0.51 1.12 0.66 0.79 0.52 0.57 0.67 0.82 Distance Distance from Suwannee River 1.02 0.02 1.01 0.01 Nutrient cycling Gender a Female 1.00 0.32 0.72 0.35 Age b 18 44 years 45 64 years 2.04 2.08* 0.48 0.36 2.11 2.47* 0.52 0.40 Educational levels c Elementary school & high school 0.31* 0.96 0.41 0.43 0.12* 0.80 0.50 0.45 Residency period d Less than 10 years 10 19 years 1.04 0.78 0.41 0.38 0.70 0.71 0.45 0.42 RUCs Isolated small rural town Large rural town 0.58 0.83 0.52 0.62 0.47 0.85 0.58 0.69 Distance Distance from Suwannee River 1.01 0.01 7.01 0.01 a Reference category for gender is male. b Reference category for age is the age of greater than 6 5 years. c d Reference category for residency period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
213 Table 5 11 . Multinomial logit estimates of re tests . Variables Not/Somewhat familiar know Odds ratio Std error Odds ratio Std error Agricultural production Gen der a Female 0.70 0.38 0.83 0.01 Age b 18 44 years 45 64 years 1.35 1.17 0.54 0.43 1.19 1.22 0.53 0.41 Educational levels c Elementary school & high school 0.12* 0.48 0.51 0.48 0.23* 0.72 0.48 0.48 Residency period d Less than 10 years 10 19 years 1.11 1.18 0.46 0.49 0.81 1.27 0.45 0.47 RUCs Isolated small rural town Large rural town 1.57 0.84 0.69 0.73 2.48 1.30 0.67 0.69 Distance Distance from Suwannee River 1.01 0.01 1.01 0. 01 Forest production Gender a Female 0.56 0.37 0.83 0.36 Age b 18 44 years 45 64 years 1.64 0.94 0.55 0.40 1.58 1.37 0.54 0.39 Educational levels c Elementary school & high school 0.18* 0 .44 0.49 0.45 0.27* 0.69 0.47 0.44 Residency period d Less than 10 years 10 19 years 1.33 1.56 0.46 0.46 1.09 1.14 0.45 0.44 RUCs Isolated small rural town Large rural town 0.93 0.40 0.64 0.70 1.81 1.06 0.61 0.65 Distance Distance from Suwannee River 1.00 0.01 1.01 0.01 a Reference category for gender is male. b Reference category for age is the age of greater than 65 years. c d Reference category for residen cy period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
214 Table 5 11. Continued. Variables Not/Somewhat familiar know Odds ratio Std error Odds ratio Std error Good w ater quality Gender a Female 0.32 0.68 0.63 0.55 Age b 18 44 years 45 64 years 1.81 0.57 1.01 0.76 2.09 1.35 0.87 0.59 Educational levels c Elementary school & high school 0.75 0.54 0.92 0. 88 0.19* 0.45 0.77 0.74 Residency period d Less than 10 years 10 19 years 1.44 1.96 0.77 0.97 0.69 1.99 0.62 0.81 RUCs Isolated small rural town Large rural town 0.75 0.80 1.23 1.23 1.28 0.62 1.01 0.98 Distance Distance from Suwann ee River 1.02 0.03 1.02 0.02 Clean air Gender a Female 0.32* 0.58 0.92 0.48 Age b 18 44 years 45 64 years 1.78 0.57 0.82 0.66 1.75 1.52 0.73 0.53 Educational levels c Elementary school & high school Some college & associ 0.57 0.69 0.77 0.71 0.29 0.63 0.65 0.62 Residency period d Less than 10 years 10 19 years 0.74 1.65 0.68 0.77 0.80 1.62 0.56 0.68 RUCs Isolated small rural town Large rural town 0.37 0.41 1.01 1.07 0.85 0.61 0.83 0.88 Distance Distance from Suwannee River 1.00 0.02 1.01 0.02 a Reference category for gender is male b Reference category for age is the age of greater than 65 years. c d Reference c ategory for residency period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
215 Table 5 11. Continued. Variables Not/Somewhat familiar know Odds ratio Std error Odds rat io Std error Renewing soil fertility Gender a Female 0.41* 0.39 1.07 0.36 Age b 18 44 years 45 64 years 1.40 0.85 0.57 0.42 1.39 1.31 0.54 0.40 Educational levels c Elementary school & high school degree 0.21* 0.65 0.51 0.48 0.35* 0.86 047 0.46 Residency period d Less than 10 years 10 19 years 1.31 1.87 0.47 0.51 1.03 1.53 0.45 0.49 RUCs Isolated small rural town Large rural town 0.68 0.45 0.62 0.71 0.65 0.43 0.58 0.66 Dis tance Distance from Suwannee River 1.00 0.01 1.00 0.01 Nature recreation and tourism Gender a Female 0.50 0.43 0.53 0.43 Age b 18 44 years 45 64 years 1.30 1.23 0.64 0.47 2.22 2.04 0.64 0.47 Educational levels c Elementar y school & high school 0.07* 0.21* 0.70 0.68 0.10* 0.22* 0.70 0.68 Residency period d Less than 10 years 10 19 years 1.15 1.64 0.52 0.56 0.96 1.55 0.52 0.56 RUCs Isolated small rural town Large rural town 0.89 0.67 0.71 0.76 0.91 0.65 0.70 0.75 Distance Distance from Suwannee River 1.00 0.02 1.02 0.02 a Reference category for gender is mal. b Reference category for age is the age of greater than 65 years. c Reference category for educational leve d Reference category for residency period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
216 Table 5 12 Wald statistic tests . Variables Not/Somewhat familiar know Odds ratio Std error Odds ratio Std error Global climate change Gender a Female 0.33* 0.52 0.86 0.52 Age b 18 44 years 45 64 years 4.00 5.76* 0.83 0.67 4.13 5.32* 0.83 0.68 Educational levels c Elementary school & high school 0.05* 0.11* 1.10 1.08 0.05 0.08 1.10 1.08 Residency period d Less than 10 years 10 19 years 0.86 1.98 0.61 0.70 0.97 2.88 0.61 0.70 RUCs Isolated small rural town Large rural town 3.46 2.02 0.99 0.90 2.23 0.78 0.61 0.70 Distance Distance from Suwannee River 1.04 0.02 1.04 0.02 Poor water quality Gender a Female 0.93 0. 55 1.36 0.54 Age b 18 44 years 45 64 years 3.37 2.84 0.87 0.69 2.76 4.33* 0.86 0.68 Educational levels c Elementary school & high school 0.15* 1.00 0.70 0.81 0.17* 1.10 0.69 0.80 Residency period d Less than 10 years 10 19 years 0.75 1.05 0.68 0.71 0.82 1.74 0.67 0.70 RUCs Isolated small rural town Large rural town 1.10 0.86 0.87 0.99 0.87 0.60 0.86 0.97 Distance Distance from Suwannee River 1.00 0.02 1.00 0.02 a Reference cate gory for gender is male. b Reference category for age is the age of greater than 65 years. c d Reference category for residency period is more than 19 years. *Significance at the 95% confidence level (p < 0.05).
217 Table 5 12. Continued. Variables Not/Somewhat familiar know Odds ratio Std error Odds ratio Std error Poor soil fertility Gender a Female 0.44* 0.39 1. 03 0.40 Age b 18 44 years 45 64 years 2.70 2.26 0.58 0.44 2.26 2.30 0.59 0.45 Educational levels c Elementary school & high school 0.19* 0.92 0.49 0.51 0.34* 1.19 0.50 0.52 Residency period d Less than 10 years 10 19 years 0.71 0.93 0.48 0.48 0.63 0.86 0.49 0.48 RUCs Isolated small rural town Large rural town 0.98 0.54 0.65 0.72 0.94 0.66 0.66 0.72 Distance Distance from Suwannee River 1.00 0.01 1.03 0.01 Low agricultural prod uctivity Gender a Female 0.43* 0.39 0.77 0.01 Age b 18 44 years 45 64 years 1.52 1.71 0.54 0.44 1.64 1.87 0.54 0.44 Educational levels c Elementary school & high school 0.17* 0.87 0.49 0.50 0.38* 1.40 0.49 0.50 Residency period d Less than 10 years 10 19 years 0.64 0.64 0.47 0.47 0.66 0.88 0.48 0.46 RUCs Isolated small rural town Large rural town 0.97 0.73 0.65 0.74 1.11 0.97 0.65 0.74 Distance Distance from Suwannee River 1.01 0.01 1.01 0.01 a Reference category for gender is male. b Reference category for age is the age of greater than 65 years. c d Reference category for residency period is m ore than 19 years. *Significance at the 95% confidence level (p < 0.05).
218 Table 5 13 . Nested logit model results for choice experiments . Attribute Model 1 a Model 2 b Odds ratio Preferences Coefficient MWTP ($) c Coefficient MWTP ($) c Choice b lock I Type of ecosystem services to be managed Climate/carbon regulation 0.004 0.24 0.015 0.88 1.01 3 Nutrient control 0.026 1.55 0.045 2.67 1.05 1 Agricultural and forestry production 0.005 0.32 0.024 1.43 1.02 2 Status quo for ecosystem management 0.019 1.11 0 0 1 4 Choice block II Program administration Suwannee River Water Management district 0.048 2.88 0.067 3.99 1.07 2 County 0.082 4.88 0.101 5.99 1.11 1 Non gover nment organizations 0.103 6.15 0.085 5.05 0.92 4 Status quo for program administration 0.019 1.12 0 0 1 3 Choice block III Location of area to be managed Anywhere within the basin 0.222* 13.22 0.241 14.35 1.27 1 Within 20 miles of your home 0.076 4.55 0.095 5.67 1.10 2 Within 5 miles of your home 0.272 16.17 0.253 15.04 0.78 4 Status quo for area to be managed 0.019 1.12 0 0 1 3 Willingness to pay per year 0.017* 0.12 0.017* 1.24 Number of responses 2690 Number of cases 8070 Log likelihood 2832 *Significance at the 95% confidence level (p < 0.05). a Logit model includes opt out option. b Logit model excludes opt out option. c Annual household MWTP over a period of five years.
219 Figure 5 1 . The Suwannee River and location of responses returned within study area . [Study boundary adapted from Suwannee River Water Management District (SRWMD), 1999. SRWMD Boundary. Map scale 1:24,000. Accessible through http://www.srwmd.state.fl.us/index.aspx?NID=319 ; Streams computed from a digital elevation model and topographic attributes]
220 Figure 5 2 . Degree of belief in global climate change caused by human activities (n=749) . Figure 5 3 . Opinion of r espondents on local government policies based on their beliefs in global climate change .
221 Figure 5 4 . Decision tree structure of the nested logit model with two levels. Level 1: selecting program or not; Level 2: which program to choose.
222 CHAPTER 6 SYNTHESIS OF ECOSYST EM SERVICE VALUES BA SED ON BIOPHYSICAL, ECOLOGICAL, AND SOCI O ECONOMIC MEASUREMENT S Overview The integral analysis between nature and humans is challenging because it involves a wide range of disciplines. These challenges arise from c omplex links of ecosystem service delivery processes that often exceed the specific boundary domains (i.e., atmospheric, biotic, pedogenic, hydrologic, and anthropogenic) . The dynamics of ecological and social development across time and spatial heterogene ity contribute to the complexities (Landuyt et al., 2013) . Th e ecological process and its relationship with social interactions comprise the cause and effect chains that exhibit nonlinear and transient behaviors (Green et al., 2005) . To address the ecosystem service dynamics incorporating people and nature, various approaches ha ve been implemented such as discourse based valuation (Wilson and Howarth, 2002) , participatory ma pping (Lynam, 1999) , alternative scenarios (Wolle nberg et al., 2000) , vision/pathway scenario (Wollenberg et al., 2000) , global unified metamodel of the biosphere (GUMBO) (Boumans et al., 2002) , and Bayesian belief network (BBN) (Cain, 2001; Clark, 2005) . Among these techniques and other options, there is an increasing use of BBN modeling because of the numerous distinct advantages. Applications of the BBN model in environment al modeling include spatially explicit land use change decisions, habitat importantly, ecosystem service modeling (Aretano et al., 2013; Celio et al., 20 14; Chen and Pollino, 2012; Haines Young, 2011; Poppenborg and Koellner, 2014; Smith et al., 2007; StelzenmÃ¼ller et al., 2010; Tattari et al., 2003) .
223 This chapter brings together the knowledge from previous chapters that links selected ecosystem service s (i.e., climate regulation, carbon sequestration, and nutrient cycling) and human well being at the regional scale, with the aim of clarifying and valuing the importance of the ecosystem service from a social ecological perspective under different feasibl e scenarios. The BBN model may not be suitable when accurate predictions are required , but it is appropriate for more uncertain situations . In our view importance in a lignment with Lynam et al. (2007) . The prediction results are useful for comparing alternative scenarios aiding decision making from a holistic viewpoint. Our objective was to develop a framework to bundle and harmonize biophysical and human perce ived benefits, specifically to synthesize interactions between ecosystem services. Bayesian Belief Networks (BBN) Bayesian belief networks are multivariate statistical models linked by probabilities for model transparency (Aretan o et al., 2013) . They are somewhat similar to decision trees, and as alternatives of decision pathways provide expected utilities (McCann et al., 2006) . Two important structural model components of BBNs consist of (1) a directed acyclic graph (DAG) that encodes interdependencies between the variables connected by arrows, and (2) a conditional probability table (CPT) that denotes the influence of the causal variables of the links in the graph (Aguilera et al., 2011; Jensen and Nielsen, 2007) . In DAG, arrows represent directional cause effect relations between the system variables. Each arrow starts in a parent node and ends in a child node. Each node (variable) has a CPT that contains a set of stated values, which are often discrete, mutually exclusive, and collectively exhaustive (Sun and MÃ¼ller, 2013) . The DAG and
224 CPT can be developed by empirical observation, literature and/ o r can be advanced by expert judgment. Mathematically, the interactions in the network are defined by the conditional dependencie (Bayes and Price, 1763) . For example, a random variable with the parents has a CPT of the form . The conditional probability yields (6 1) Decomposing Equati on 6 1 with the chain rule of joint probability, this becomes ) (6 2) where arcs represents causal relations among these variables. The nodes are defined by their CPTs tha t interact with the states of any parent nodes. The conditional probability of a variable does not depend on all other variables, but specifies only from its parents in the network (Pearl, 2009) . Advantages and Shortcomings of the BBN Model in Ecosystem Service Mode ling There are distinct advantages of the BBN model over the conventional statistical methods. First, the BBN system dynamic model is well suited t o problems involving complexity and uncertainty, especially in the social ecological models (McCann et al., 2006) . Seco nd, combinations of qualitative and quantitative inputs from empirical data and expert judgment are allowed for the model analysis (Haines Young, 2011; Kuikka et al., 1999) . Third, the BBN model can simplify complex systems th rough key variables and their relationships that can be viewed as graphical tools. It thus helps in the communication among ecologists, model users, policy makers, and stakeholders (Cain,
225 2001; Smith et al., 2011) . Fourth, the model structure and pro bability tables can be refined when data are updated, and are accounted for prior knowledge and missing data, which other traditional decision tree analyses or modeling approaches cannot do (McCann et al., 2006) . Last but not least, the BBN model is useful for the testing of future scenarios. It provides a guidance framework of the likelihood consequences of future events (Wooldridge, 2003) . However, the use of BBN model remains challenging. In reality, parameters and variables in environmental research are mixed with discrete and continuous data, and the BBN model can only handle continuous values in a limited man ner. Therefore, the states of variables are discretized over the discrete domain. This way, only rough characteristics of the original distribution are captured (Friedman and Goldszmidt, 1996) . Although automatic discretization techniques have been developed, no sa tisfactory methods have been found yet (Uusitalo, 2007) . A fully speci fied probabilistic model is required to have empirically observable, quantifiable, or defensible characteristics (McCann et al., 2006) . The elicitation of expert judgment is often needed, yet it is difficult to obtain knowledge that can be converted into probabilities. Uusitalo (2007) pointed out the reasons for this caveat. Many researchers in ecology related fields are familiar with field sampling or experiments. It is challenging for th em to provide numbers without depending on data. They are used to classical statistical means that have point estimates and confidence intervals, and often feel uncertain in generating inputs in terms of distribution. The use of expert judgment should be c arried out with care in order to deliver the model in a defendable fashion (Marcot et al., 2006) . Clemen
226 and Winkler (1999) recommended that experts should be familiar with the definitions and connection of variables and should agree on the model structure. In addition, the feedback loops and inclusion of temporal scale d imensions are limited in a DAG (McCann et al., 2006; Uusitalo, 2007) . Materials and Methods We developed the conceptual BBN model by identifying the important system variables and forming the connectors between variables. A combination of empirical data from field observations, statistical representations, theoretical data from the literature sources, findings from chapters of this dissertation, and probabilist ic quantities from expert knowledge was extensively utilized along with spatial, temporal, and survey data from previous studies in order to represent the most meaningful outcomes. Scenarios in the model represent imagined possible future outcomes that are highly influenced by the perception and valuation of people, public awareness, decision making, and politics. In this study, a higher score is better in terms of benefits people can obtain from the ecosystem services. Computing the probabilities is base d on a stochastic simulation method by measuring the frequency/probability of events that occur during simulation runs (Pearl, 1988) . Bayesian Belief Network (BBN) Construction The diagram derived from the concept models and available datasets from various sources was developed into a BBN model using the Netica software 4.16 ( Norsys Software Corporation (NORSYS), 2014 , http://www.norsys.com/netica.html ). In ure nodes are variables representing either empirical, calculated parameters, or
227 probabilities that can be regulated by actions (i.e., all nodes in Tiers 1, 2, 3, and 4, Figure 6 1). The input nodes without arrows pointing to them are the parent nodes (i.e ., scenario, annual household income, soil carbon to nitrogen ratio, soil carbon to phosphorus ratio, importance of ecosystem services, and concerns of the ecosystem services). Input nodes with incoming and outgoing arrows are called child nodes. A decisio n node denotes control events or variables that can be interpreted or implemented directly by a decision maker (i.e., overall scores). Theoretically, a decision node should be accompanied by utility nodes (i.e., carbon sequestration scores, climate regulat ion scores, and nutrient cycling scores), which express preferences over outcomes. More details about implementation of the BBN model are provided in numerous studies (Cain, 2001; Kragt, 2009; Marcot et al., 2006; Nyberg et al., 2006; Spiegelhalter et al., 1993) . The number of discretization states in nature nodes was limite d to a maximum of three states, except scenarios and land cover/land use coverage, in order to keep CPTs in a manageable fashion given a large number of explanatory variables as recommended by Spiegelhalter et al. (1993) . From Figure 6 1, Tier 1 of the model consisted of scenarios that were developed qualitatively to examine the effects of direct drivers (landscape, riverine, and climatic properties) and indirect drivers (regulations, income, and awareness of ecosystem services) on eco system services in the future. The scenarios were developed based on a review from the Intergovernmental Panel on Climate Change (IPCC) report (Forster et al., 2007) . National England Research Report (Creedy et al., 2009) , UK National Ecosystem Assessment (NEA) (2011) , and a Malta case study (Morris et al., 2011) . These reports were based on international and national scales. We therefore adjusted
228 the scenarios to fit a regional basin scale with high and low ecosystem functioning and social impact variants. Altogether, four story lines for the mid of 21 st century were created: Grow with Awareness (GA), Gain Eco nomic value (GE), Go toward the Projection (GP), and Business as Usual (BU). Tier 2 showed the interactions among variables influenced directly or indirectly by the scenarios. For example, land cover/land use (LC/LU) coverage, temperature, precipitation, r egulations, and energy consumption acquire direct probabilities of occurrence from the scenario selection. Whereas carbon to nutrient ratios, soil organic carbon (SOC) sequestration rate, soil carbon stocks, above ground carbon stocks, net primary producti vity (NPP), carbon dioxide emission (CO 2 ), implementations of best management practices (BMPs), average annual water flows, and nutrient loading in water were indirectly influenced by the scenarios, and they were directly influenced by the variables above them. Tier 3 included the biophysical quality of carbon sequestration, climate regulation, and nutrient cycling variables. Tier 4 combined human perspectives towards awareness of ecosystem services. The data in Tier 4 were collected from the Ecosystem Serv ice Survey in the Suwannee River Basin during the fourth quarter of 2012. Details of the variables used and the data sources are shown in Table 6 1. Scenario Features Grow with Awareness (GA) describes the future situation in the Suwannee River Basin in Fl orida (FL SRB) where people are optimistic about the means to mitigate CO 2 emissions. In this scenario, they are aware of the importance of ecosystem services and have concerns about environmental issues. Local residents are eager to change their environme ntal footprint to keep a good balance between the quality of life and the quality of natural resources. Conservation lands expand in the future. Agricultural and
229 urban areas have minimal changes in spite of higher population density. Existing environmenta l regulations are strictly enforced. A majority of the median household income falls in medium level. Temperature and precipitation are not anticipated to change quickly in the next several decades. Gain Economic Value (GE) describes a rapid economic grow th as it becomes a greater priority in a competitive, global environment. In this scenario, people are worried more about their income than they are about environmental consequences. Most people would not perceive climate change as an immediate threat in t heir lifetime or the lifetime of their children. Neither political nor public involvement is strong enough to drive a positive change. Current environmental regulations remain but are less effective. Household income is expected to increase. Mean temperat ure and precipitation may reach or exceed the projected estimation. Go toward the Projection (GP) describes the trend of model projection from various sources such as LC/LU change, and temperature and precipitation prediction. In this scenario, the number of people engaging or not engaging in environmental programs is almost the same. Some are forced to follow the regulatory standards. It would take longer than a generation to rectify all the environmental problems. While there is not much change in local a spects and behaviors such as regulations and household income, external forces are the more dominant controlling factors. Business as Usual (BU) represents the projection based on the current socio ecological trends. This scenario is most likely to lie b etween two extremes GA and GE but is not as extreme as the GP.
230 Parameterizing the Model After identifying the variables and links between them, we assigned states and probabilities to each variable. The states of each node signified the conditions or possible values that node can assume. These values can be expressed as numerical numbers, interval values, or probability distributions (MartÃn de Santa Olalla et al., 2005) . The state values varied based on the available data. We discretized data in most cases and used a probability distribution to represent valu es. The mean standard deviation values were the initial starting values that were assigned for a middle class. Values larger than a middle range were assigned a higher state, while values below a middle range were assigned a lower state (Table 6 1). Some n odes with limited data were elicited from the literature and from discussions with experts as suggested by Kragt (2009) . BNN GIS application The geospatial analyses within the ArcGIS software was extensively used to populate the probabilities of spatio temporal data. For example, spatial annual mean precipitation data collecte d during 1971 to 2010 (PRISM Climate Group, 2012) were calculated using average and standard deviation values and then were discretized, with the total area divided into three groups: higher than average (>1,444 mm yr 1 ), long term average (1,383 to 1,444 mm yr 1 ), and lower than average (<1,383 mm yr 1 ). The percentage of LC/LU coverage was derived by estimating a specific LC/LU type by areal coverage a (e.g., urban area (6%), agricultural land (14%), upland forest (47%), wetland (29%), and others (4%). So me nodes may have various parents which added complexity to the analysis. In such cases, we created a conditional statement using the raster calculator tool in ArcGIS on multiple layers to populate the chances of
231 occurrence. The BBN GIS application was app lied in several nodes in this study (i.e., Annual_temp, Annual_precip, CtoN_ratio_soil, CtoP_ratio_soil, SOC_seqst_rate, Soil_C, Above_ground_C, Implement_BMPs defined in Table 6 1). Survey incorporation Results from the Ecosystem Services Survey in the Suwannee River Basin (Chapter 5) were exploited to populate the probability distribution, particularly in the ecosystem services and regulation nodes. For example, respondents were clustered into four main groups representing the scenario features. Then th e percentage of responses for each group toward the aggressiveness of local government policies was calculated. The awareness of each ecosystem function category (i.e., awareness of soil quality, air quality) was interpreted based on understanding the impo rtance of the services. Expert involvement The BBN model also relied on empirical and literature data; thus, a small portion of the probability input was required by experts. A group of five senior experts were sent conditional probability tables along w ith related information about the system variables. They were asked to give their opinions about the structure of the BBN model and to provide their belief for chance node. Scores of ecosystem services The outcomes of the different scenarios were compared with each other using the overall score. The overall score represents the qualitative information about the selected ecosystem services within the study area that benefit human well being. The scores contained no units since they were derived from various types of data across social and ecological contexts. The overall score was the sum of utility values from
232 three selected ecosystem services, ranging from positive 90 to negative 90. In a study by Haines Young (2011) , a similar scorin g system was applied for bridging ecosystem ecosystem services are in better conditions when compared to lower negative values. The utility scores used in calculating the impact of the different scenarios on the ecosystem service situation are shown in Table 6 2. Although these numbers could be arbitrary, they were useful for comparing policy options from a multi criteria assessment or trade off analysis to achieve a pr eferable outcome. A percentile of the reference scores were then used to define overall ecosystem service conditions: very good (54 to 90), good (18 to 54), fair (0 to18), poor ( 18 to 0), bad ( 54 to 18), very bad ( 90 to 54). These scores are rooted in the idea of measuring to what extent humans can biogeochemical processes. Model Assessment Wherever possible, data should undergo model evaluation. The quantitative model can be eval uated by predictive accuracy assessment, in which a train and test approach is used to quantitatively evaluate the model ( Dlamini, 2010; Pollino et al., 2007) . Jakeman et al. (2006) validated, the subjective criteria and transparency of the modeling procedure shoul d be included. In cases of scenario analysis, predictive accuracy evaluation is limited. Using approaches in this study. By integrating various combinations of inputs and examining behavior of the model is consistent with the current understanding of the real system
233 (Chen and Pollino, 2012) . Sensitivity analysis of node s is helpful to identify variations in model parameters of whether a variable behaves sensitively or insensitively to other variables (Marcot, 2012; Pollino et al., 2007) . The entropy reduction is a key method used in the Netica software that informs the mutual information between the query variable (Q) and the varying variab le (F). The expected reduction in entropy of Q is due to a finding at F (NORSYS, 2014) . (6 3) (6 4) where is the entropy reduction, is the entropy of before any new findings, is the entropy of after new findings. The refers to the sum over all states of , and is the sum of all states of . and are the probabilities of variables and , respectively. is the joint probability of variables and . Results and Dis cussion The BNN diagrams for different scenarios are shown in Figures 6 2 to 6 6 should be viewed from the top downwards and from the bottom upwards to the hexagonal utility nodes (Carbon_seq_ESscores, Climate_reg_ESscores, Nutrient_cyc_ESscores). These th ree utility nodes measured the different outcomes for the ecosystem service stressors and were divided into six ecosystem service condition classes. Results of the 6 2 where the ecosystem service condition was considered good (47.70 overall score). a value of 48.73 (Figure 6 3). The score fell in the good range of the ecosystem service
234 condition. The GP scenario performed similar with overall score value of 48.52 (Figure 6 4). One explanation of the slightly lower score of the GP scenario is that some drivers are global ones, such as rise in temperature and increase in precipitation, t hat occur beyond the basin scale; however, the changes could impact the ecosystem condition in the FL SRB. In addition, even if full enforcement of existing and new regulations is anticipated to be higher in the GP scenario, bottom up governance strategies and high awareness by residents are required. These reasons are similar to the findings by Ainsworth et al. (2012) and Saarman and Carr (2013) , in which the most effective ecosystem based management in marine and fisheries (provisioning services) was a ssociated with a balance in the top down and bottom up schemes. However, the benefits that people could gain from the ecosystem services from both BU and GP are in the good category. Figures 6 5 and 6 6 demonstrate the outcomes for the two most contrasting scenarios, GE and GA. The GE scenario was expected to perform the worst among all the scenarios because people were more concerned about money than about the environment. Increases in energy demand and land use change occur as a result of economic growth. The overall score was 59.08, which was considered the lowest 5). In contrast, when the GA scenario was chosen, the model clearly showed a very positive value of 62.74, or a very g ood ecosystem service condition. This implies that complete awareness of the sustainable use of natural resources enhances ecosystem services. Using the same scoring scheme ( 90 to 90) but with different inputs, Haines Young (20 11) scenario in his study had the highest value (24.16) because large numbers of people
235 frequently engaged with the environment and a well functioning biodiversity was in ab scenario had the lowest value ( tre nd of business as usual and economic growth (Haines Young, 2011) , which were described separately in our study. Results from the sensitivity analysis were reported based on types of services. Tables 6 3 to 6 5 show the entropy reduction and variance of beliefs . The entropy indicates the level of abstraction adopted to construct the BBN model . The probability distribution that has the highest entropy under the given constraints is the most rational choice (Jaynes, 1957) . In our case, the entropy reduction is related to the extent to which nodes and higher level constructs are sensitively used. The sensitivity analysis not only helps verify the model validity, but also explains the most influential and informative variables with respect to the target variable (Sun and MÃ¼ller, 2013) . Entropy reduction and variance of beliefs determine the degree and ranking order of the influence of parent nodes (stressors) on the outcome of the selected services. Results for the carbon sequestration node (Carbon_seq_ESscores) revealed that awareness of soil quality was the most sensitive variable to the carbon sequestration and had the greatest influence on the carbon sequestration score, followed by the levels of terrestrial carbon (Table 6 3). These similar patterns were present in the sensitivity a nalysis for climate regulation (Climate_reg_ESscores) and nutrient cycling services (Nutrient_cyc_ESscores). Awareness of air quality and the condition of air quality were
236 ranked the most influential v ariables for climate regulation and reveal ed the highes t sensitiv ity level (Table 6 4 ) , while water quality and awareness of water quality were the most powerful in impacting nutrient cycling services (Table 6 5). It is not surprising that the benefits ecosystem services provide and awareness ranked top in the sensitivity analysis. One explanation is that the BBN model from the biophysical view was built using ecosystem structure, function, and flow. Through the processes, these compositions were directly linked to the delivery of the final ecosystem services. The quantitative studies conducted by Balvanera et al. (2006) and Cardinale et al. (2012) indicated that biodiversity (provisioning services) in a specific ecosystem positively correlated with the regulating and supporting services . The scores (values) of carbon sequestration, climate regulating, and nutrient cycling, were impacted by services people belief that they receive. Hence, the influence of awareness on ecosystem service value is profoundly meaningful. Personal awareness a nd appreciation for the environment lead to pro environmental behavior and are more likely to be embraced in the long run (Barnes, 2005) . Direct impacts from deteriorated ecosystem health on human well being amplify public awareness and a ction toward environmental issues. Thus, people are more likely to engage in enhancing ecosystem services. For example, the impact from tree loss along urban streets stimulated higher awareness, concerns, and involvement in environmental issues (Hunter, 2011) . In our study, the influential stration ecosystem service (Tables 6 3 to 6 5). Interestingly, in the socio economic ecosystem service survey ( C hapter 5) the opposite trend was
237 found where residents were most concerned about water quality in context of clean drinking water due to the exp erienced threat to impairment of surface water and ground water in the FL SRB than soil or air quality. In contrast, the BBN is more comprehensive and includes multiple perspectives socio economic, ecological and biophysical and thus provides a more wh 2 regulation nodes. NPP represents the flux of carbon into the ecosystem (gross photosynthesis minus plant and animal respir ation). A debate whether NPP and CO 2 relate positively or negatively to each other is ongoing. DeLucia et al. (1999) concluded that elev ated CO 2 resulted in a consistent increase in NPP but at the limited capacity of the system , in particular the rapid growing forest that represent the upper limit for the carbon sequestration . Beedlow et al. (2004) argued that CO 2 fertili zation did not always raise NPP and carbon sequestration because there are a number of factors that diminished carbon sequestration such as nutrient s , temperature, and precipitation . The scenario node and regulations mutually played a dominant role in i nfluencing the climate regulation and nutrient cycling in the sensitivity analysis. This implies that when regulations and scenarios are changed, the scores or values of climate regulation and water quality will be altered. The results from the BBN model f urther suggested that other factors rather than awareness showed a high degree of influence in each temperature impacting water quality.
238 Conclusions The goal of the BBN model in this study was to show the consequences of the choices that people make grounded in biophysical environmental settings of the FL SRB. With the BU (Business as Usual) scen ario, the chances of good ecosystem services are roughly equivalent to the GP (Go toward the Projection) scenario, although tighter regulatory measures are expected and increases in temperature and precipitation are anticipated. This implies that within t he FL SRB, the ecosystem services in general provide satisfactory benefits to the community. Focusing on awareness enhancement rather than economic growth allows more room for improving the environment. Economic growth is commonly associated with increase d energy use, which leads to more greenhouse gas emissions and diminished air quality. Intense activities in response to the economic growth scenario could have an inverse impact on water quality through increased nutrient loading. Furthermore, lack of env ironmental awareness drives the GE (Gain Economic value) scenario to the level with the lowest values that reduce benefits for residents. In the GE scenario despite degrading ecosystem services evidenced by low water, air, and soil quality people value oth er things more, even if their health suffers. On the other hand, the GA (Grow with Awareness) scenario had the highest probability of gaining maximum ecosystem service benefits. When ecosystems provide benefits to human well being (e.g., save drinking wat er and healthy soils) and people recognize, appreciate, and value environmental resources, ecosystem services clearly increase to the highest level. With GA the morally induced environmental ethics of people stimulates residents, stakeholders, and decision makers to act on behalf of the environment (Schmidtz and Willott, 2012) through conservation and protection activities and programs.
239 Although the GA scenario seems to be the ideal scenario, it is possible to combine other strategies to improve the ecosystem services based on economic incentives (e.g., payment for ecosystem services and the existing cost share program) rather than environmental concerns. Both public awareness and natural benefits derived from ecosystem services are sensitive to the model. Policies and legislation that address soil, water, and a ir quality should be planned parallel with raising public awareness. This BBN model may not capture every aspect of the complex processes of climate regulation, carbon sequestration, and nutrient cycling services, which is why this study focuses on the vis ible socio ecological relationships with outcomes obtained by different alternatives. The outcomes can be changed if new information or beliefs are updated relative to new situations. The structure of the BBN model from this study is useful for more compre hensive and integrated assessment on a larger, smaller, or similar scale. Importantly, the BBN model combined socio economic, ecological, and biophysical aspects related to three interconnected ecosystem services. It is the synthesis of different realms an d perspectives that made this BBN study more holistic when compared to more one sided ecosystem service investigations (e.g., willingness to pay or sole water quality analysis). Summary and Synthesis This dissertation contributes to the ecosystem service field by integrating concepts and developing methods for understanding and analyzing the interactions between soil, water, and climate in their regulating and supporting roles. The fundamental framework is rooted in the relationship between ecosystem con dition and management and human well being. Empirical and secondary data are utilized
240 extensively across the sciences making it an interdisciplinary analysis. Altogether, this work provides a better understanding of the patterns and dynamics within a spec ific socio ecological context in the Suwannee River Basin in Florida. Summary of Findings Chapter 3 characterized carbon sequestration, climate and nutrient regulation ecosystem services from a biophysical perspective. Findings from interpolation (krigin g) and paired comparison (collocated sites specific) identified the top soils acting as a carbon sink over the past 40 years in the Suwannee River Basin, southeastern United States. The land use and other biotic effects as well as soil landscape conditions on soil organic carbon (SOC) accumulation were dominant over the muted effects imposed by climate. Climate, as characterized by mean annual precipitation and temperature, did not show any significant trends over the past decades in this region. The larges t accumulation in SOC stocks occurred with the conversion of upland forests to wetlands, followed by wetlands wetlands and upland forests upland forests. The trend of total organic carbon (TOC) loading in surface water coincided with an increase in SOC st ocks on a unit area basis. Correlations between TOC loads in surface water and SOC stocks in soils only partially coincided spatially possibly due the transport processes that translocate carbon within the basin. TOC loading variation varied among the upst ream blackwater river system and the downstream stream system with a high density of springs. An upward trend of total nitrogen (TN) and total phosphorus (TP) loads were most pronounced, especially downstreams in the Santa Fe River, Upper Suwannee River, and Lower Suwannee River. The increases in nutrient loading in surface water were observed in the areas most prone to nutrient leaching due to sandy soil texture and well drained to excessively drained soils. The long term trend analysis of TN and
241 TP conce ntrations showed that more than 87 percent of 23 drainage areas were safe under the Florida numerical water quality standard. The stoichiometric ratios of C:N and C:P varied across the drainage areas. Overall, net N and P mineralization dominated in soils and surface water (low ratios), except in a few drainage areas that showed neither gain nor loss in P in surface water. Chapter 3 also provides quantitative relationships of C:N:P ratios for a better understanding of the biogeochemistry of soils and surfac e waters. Chapter 4 employed the random forest (RF) technique and the STEP AWBH conceptual model to explore the spatially explicit relationships between SOC stocks and environmental human driving factors. Results indicated that the highest predictors expla ining SOC stocks were biota, followed by soil/topography and water. The first quantile of STEP AWBH variables in the RF model were selected to estimate terrestrial carbon stocks using the simulated annealing optimization method. The combination of environ mental factors clearly showed the impacts on the different levels of terrestrial carbon stocks. In comparison to the actual carbon stocks, there is some potential to enhance carbon storage to reach the attainable carbon levels when perturbation in envir onmental factors occurred by 10, 20, and 30 percent. A novel contribution of this chapter provides a step towards modern machine learning and optimization methods under the umbrella concept of the STEP AWBH model. The simulations of attainable terrestrial carbon stocks provide valuable insight into possible futures that can guide carbon management. Chapter 5 summarized the perception, preferences, and valuation of ecosystem services provided by residents in the study area. Fundamental assessment of social ,
242 demographic, and economic context revealed that gender, age, and education were services. In the preference assessment, the nutrient control in relation to water quality was the most important service, followed by agricultural and forestry production, while climate regulation and carbon sequestration were ranked the least important among the ecosystem services provided in the survey. Respondents preferred county government and the Suwannee River Water Management District (SRWMD) to manage ecosystems, rather than non government organization. In terms of location of area to be loser to their homes. The disparities based on human and biophysical perspectives were explained by scale sensibility based on both a political and spatial basis. The monetary valuation using the willingness to pay scheme was low for all aspects of ecosyst em service, management, and locations partially due to the predominant Global Commons view. Chapter 6 integrated results from Chapter 3,4, and 5 along with knowledge from t of ecosystem service value was developed in order to compare outcomes between alternative scenarios. As expected, the GA scenario provided the largest ecosystem benefits to the system, while the GE scenario derived only minimal ecosystem benefits. The BU scenario provided satisfactory benefits that leave ample room to achieve higher levels of services. Sensitivity analysis showed the degree of variable importance for climate regulation, carbon sequestration, and nutrient cycling. The values were less like
243 Synthesis We assessed the role of ecosystems in regulating and supporting climate, soil, and water, and in delivering services for the well being of societies. Soil is accounted for as a host medium for other regulating services (i.e., climate regulation and nutrient cycling). Quantification of soil carbon helps identify and evaluate regulating and supporting services explicitly. Global carbon estimates are rough and often based on legacy data , therefore regional carbon assessment with current data is profoundly important for our understanding of the carbon cycle. For example, Guo and Gifford (2002) and JobbÃ¡gy and Jackson (2000) , cited in the review paper by Stockmann et al. (2013) , indicated that the total SOC stocks amounted to 2,344 Gt C, while 1,500 Gt of organic carbon stored 1 m deep and about 615 Gt C stored in the top 20 cm. Furthermore, it has been estimated that the total global soil carbon (C) pool including wetl ands and permafrost ( 3 ,250 Pg C) is about five times the biotic pool (650 Pg C) and about four times the atmospheric pool (780 Pg C) (Field et al., 2007) . In the southeast and south central United States, soil C pool estimates amount to about 16,535 Tg C to the depth of soil profile and biotic pool stores about 4,567 Tg C (LC/LU includes forest, crop, pasture) (Han et al., 2007) . In the Suwannee River Basin, Florida soils are carbon rich with wetlands covering about 29% of the basin and veg etation characterized by high net primary productivity with upland forests covering about 46% of the basin. According to findings from this research the carbon budget in the Suwannee River Basin in Florida entails approximately 168 Tg C in the top meter so ils (including 75 Tg C from the top 20 cm soils), 92 Tg C in live and dry biomass, and 0.05 Tg C in surface water (Table 6 6). Carbon fluxes include 0.9 kg m 2 yr 1 from NPP (a total of 175 Tg C) , 0.1 kg C m 2 yr 1 in
244 the atmosphere (a total of 2 Tg C) , a nd 1 kg C CO 2 day 1 from human respiration (a total of 0.03 Tg C) (Table 6 7) . Thus, in the regional ecosystem with high density of vegetation, soils store about 84 times as much carbon as the atmosphere and about 1. 8 times as much as the biomass. This sy nthesis suggests that ecosystems in slow development areas capture and regulate much more carbon in the terrestrial system as compared to the atmosphere. This implies that people living in a similar type of environment are more likely to obtain the same be nefits from soil quality and air quality. In the ecosystem process, there are interactions among services. It is important to consider trade offs and synergies among ecosystem services. Synergies describe the situation that when one ecosystem service is i mproved, co benefits to other ecosystem services are also improved. For example, implementations of the Total Maximum Daily Loads (TMDLs) attempt to reduce pollutants and nutrient loads into water bodies and thereby improve water quality and water conserva tion on a site specific, regional, and watershed basis. The co benefits from nutrient reduction (nutrient cycling services) are effective at offsetting greenhouse gas emissions (GHGs) to the atmosphere (climate regulation services). Another example is Best Management Practices (BMPs), which are designed to benefit water quality and water conservation (nutrient cycling services) while maintaining or enhancing agricultural production (provisioning services). To achieve this goal, the improvement of soil ferti lity (carbon sequestration) is necessary through conservation and soil carbon storage management. Applications to improve soil carbon via fertilization and irrigation are important but trade offs need to be addressed. Adding more nutrients might not enhanc e soil quality when soil carbon accumulation reaches the saturation limit. Excessive nutrient application has also reverse impact on
245 other services, such as water quality. Moving forward, renewable energy sounds promising to reduce GHGs, but trade offs bet ween CO 2 reduction and energy use associated with those activities must be considered in form of complete live cycle analysis. Carbon market opportunities in forestry and agriculture in Florida have been valued collectively $340 million per year based on a carbon market value of $20 per ton CO 2eq (Mulkey et al., 2008) . The Flo rida Farm Bureau Federation and AgraGate Climate Credits Corp agreed to partner in providing carbon credit services to farmers, ranchers, and private forest owners. The carbon credits are expected to sell to the Chicago Climate Exchange (CCX) and return be n e fits to the agricultural producers. The intention of the program is virtuous , but carbon management practices typically include many challenges . These challenges can severely constrain the implementation of carbon credit policies, which include political willpower, opposition of industrial sectors , efforts to collect carbon offsets data , economic efficiency , and low willingness to pay (Jenkins, 2014) . This is also the case in the FL SRB where the willingness to pay for ecosystem services, in particular carbon sequestration is very low. The most important point we have learned from our s tudy is that the majority of the general public is likely to make a decision on their experience , values and beliefs about specific environmental issues. The public perceptions of the climate change and carbon related problems are not high enough yet to a llow carbon trading programs. F rom an implementation point of view, the challenges remain unless local people directly experience the impacts, successful stories of the program in the same geographic area are prove n , superior political
246 feasibility and inst rument are established , and economic environmental harmony is reached . This dissertation considers only three ecosystem services within a wide range of full services. Still it is impossible to represent every single aspect of the benefits from each select ed service. The functioning ecosystem processes and socio economic interactions are complex so no single approach can be used to assess all the ecosystem services. This work expands on the integration of disciplinary knowledge in a common conceptual framew ork leading to a truly inter disciplinary approach (Eigenbrode et al., 2007) tha t views the total environment consisting of human ecosystem coupled systems. Basic interdisciplinary knowledge and implementation of ecosystem service tools should include GIS, statistical analysis (e.g., descriptive statistics, trend analysis, and probabi lity), geostatistics/geospatial analysis, machine learning, social economic demographic background of the study area, soil water climate sciences, uncertainty assessment, optimization techniques, and economic modeling. A comprehensive ecosystem service pro ject to form interdisciplinary teams is essential in any study on the subject.
247 Table 6 1 . Overview of variables used in the Bayesian Belief Network (BBN) model , states, data sources , and justification technique . Node Description States Source/Litera ture Justification method Scenario Feasible scenarios that could happen based on the choices people make. GA (go with awareness) GE (gain economic value) GP (go toward the projection) BU (business as usual) After Creedy et al., 2009 Literature review LC/LU_coverage The percentages of land use/land cover in five categories. Other LC/LU classes include rangeland, barren land, and utilities (30m resolu tion). UR (urban area) AG (agricultural land) UF (upland forest) WL (wetland) Other (other land cover types) Florida Department of Environmnet Protection (FDEP), 2009; GeoPlan, 2006 Literature review; GIS application Annual_temp Average annual temperature from 1971 through 2010 (rescaled to 30m resolution). Data were divided into three categories using mean an d standard deviation values as a threshold. Higher than average (>1,444 mm) Long term average (1,328 1,444 mm) Lower than average (<1,328 mm) PRISM Climate Group, 2012; Song et al., 2013 Literature review; GIS application Annual_precip Average annua l precipitation from 1971 through 2010 (rescaled to 30m resolution). Data were divided into three categories using mean and standard deviation values as a threshold. Higher than average (>20.5 o C) Long term average (19.9 20.5 o C) Lower than average (<19.9 o C) PRISM Climate Group, 2012; Song et al., 2013 Literature review; GIS application Regulations The percentage of responses people had toward the local government policies be more or less aggressive in dealing with climate chan More aggressive Less aggressive No change Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation
248 Table 6 1. Continued Node Description States Source/Literature Justification method Energy_consump Hou sehold energy consumption compared with income. High energy use Medium energy use Low energy use American Coalition for Clean Coal Electricity (ACCCE), 2012 ; KEMA, 2010 ; Saunders, 2011 Literature review Annual_hh_income The median household income within the Suwannee River Basin in Florida (FL SRB) between 2000 and 2006. Abo ve median Median Below median U.S. Census Bureau, 2010 Survey incorporation CtoN_ratio_soil The percent coverage of soil organic carbon (SOC) sto cks (kg C m 2 ) to total nitrogen (TN) stocks (kg N m 2 ) in soils within the FL SRB. <25 25 30 Unger et al., 1998 ; Spatial distribution of SOC and TN stocks (2008/09) derived by block kriging from Chapter 3. Original data the FLSCP (Grunwald et al., 2011a) GIS application CtoP_ratio_soi l The coverage of soil organic carbon stocks (kg C m 2 ) to total phosphorus (TP) stocks (kg C m 2 ) in soils within the FL SRB. <200 200 300 >300 Unger et al., 1998 ; Spatial distribution of SOC and TP stocks (2008/09) derived by block kriging from Chapter 3. Original data the FLSCP (Grunwald et al., 2011a) GIS application SOC_seqst_rate The soil organic carbon sequestration rate. The rates were derived by subtraction of historic SOC from current SOC estimates divided by number of year of measurement during approximately 40 years. High (>50 g m 2 yr 1 ) Median (16 50 g m 2 yr 1 ) Low (<16 g m 2 yr 1 ) Spatial distribution of historic SOC stocks (196 1996) and current SOC stocks (2008/09) derived by block kriging from Chapter 3. Original data Assessment and Modeling of Changes in So il Carbon Storage and Turnover in a Southern Landscape (FLSCP) (Grunwald et al., 2011a) GIS application NPP An average net primary productivity (NPP) 2000 2006 from MODIS1743 data. High (>12 kg C m 2) Medium (5 12 kg C m 2) Low (<5 kg C m 2) MODIS subsetted land products, Collection 5 (MODIS 17 team, 2009) . GIS application
249 Table 6 1. Continued Node Description States Source/Literature Justification method Soil_C An average of amount of soil org anic carbon stocks per unit area. High (>5.6 kg C m 2 ) Medium (3.2 5.6 kg C m 2 ) Low (<3.2 kg C m 2 ) Unger et al., 1998 ; Spatial distribution of SOC stocks (2008/09) derived by block kriging from Chapter 3. Original data the FLSCP (Grunwald et al., 2011a) G IS application Above_ground_C Th e amount of live and dry above ground biomass derived from the National Biomass and Carbon Dataset (NBCD) data (version 2). High (>9.9 kg C m 2 ) Medium (0.3 9.9 kg C m 2 ) Low (<0.3 kg C m 2 ) Kellnorfer et al., 2013 GIS application CO2_emissions Carbon dioxide (CO 2 ) emissions to the atmosphere Increase Decrea se No change Gurney et al., 2009; KEMA, 2010 Literature review; Expert involvement Avg_annual_flows The average annual discharge at the LwSuwn.06 (USGS reference: 02323500) from 2000 to 2010. H igh flows (>8 x 10 11 L mo 1 ) Medium flows (2 8 x 10 11 L mo 1 ) Low flows (<2 x 10 11 L mo 1 ) U.S. Geological Survey (USGS), 2010 GIS application Implement_BMPs The percent coverage of best management practice (BMPs) implementation i n agricultural lands within the study area from 2010 through 2012. Yes No Office of Agricultural Water Policy (OAWP), 2012 GIS application Surplus_nutrnt_water The probability of excessive nutrient concentration above numerical standard with or without implementation of BMPs. Increase Decrease No change Florida Department of Environmnet Protection (FDEP), 2013 ; Long term nutrient concentration analysis (2000 2010) from Chapter 3 Statistical analysis; Expert involvement Nutrnt_loads _water The probability of having amount of nutrient loads. Increase Decrease No change Long term nutrient loading analysis (2000 2010) from Chapter 3 Statistical analysis; Expert involvement
250 Table 6 1. Continued Node Description States Source/Literature Justification method Terrestrial_C The amount of terrestrial carbon stocks from above ground and below ground biomass. High (>14.6 kg C m 2 ) Medium (4.4 14.6 kg C m 2 ) Low (<4.4 kg C m 2 ) Grunwald et al., 2011; Kellnorfer et al., 2013 GIS application Air_quality The probability of air quality condition und er three possible outcomes thatCO 2 emissions have (i) increased, (II) decreased, or (iii) remained insignificant change under environmental pressure. Good Moderate Unhealthy FDEP, 2007 Literature review Water_quality The probability of water quality condition considering the long te rm trend analysis from 2000 to 2013 and the impaired water body according to the Florida numeric water quality standard. Good Acceptable Bad FDEP, 2013 Expert involvement Imp_soil_quality The percent response to the importance of soil quality question. Important Not important Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) GIS application Soil_quality_concer n The percent response to the soil quality concern question. Concerned Not concerned Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Aware_soil_quality Awareness of soil quality based on importance and concerns towards soil quality. Aware Unaware Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Imp_air_quality The percent response to the importance of air quality question. Important Not important Ecosy stem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation
251 Table 6 1. Continued. Node Description States Source/Literature Justification method Air_quality_concern The percent response to the air quality concern qu estion. Concerned Not concerned Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Aware_air_quality Awareness of air quality based on importance and concerns towards air quality. Aware Unaware Ecosystem S ervices Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Imp_water_quality The percent response to the importance of water quality question. Important Not important Ecosystem Services Survey in the Suwannee River Basin, 20 12 (this dissertation) Survey incorporation Water_quality_concern The percent response to the water quality concern question. Concerned Not concerned Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Awa re_water_quality Awareness of water quality based on importance and concerns towards water quality. Aware Unaware Ecosystem Services Survey in the Suwannee River Basin, 2012 (this dissertation) Survey incorporation Carbon_seq_ESscores Assigned utility sco res for carbon sequestration ecosystem services with a range from 30 to +30. +30, +20, +10, 10, 20, 30 Haines Young, 2011 Literature review Climate_reg_ESscores Assigned utility scores for climate regulation ecosyst em services with a range from 30 to +30. +30, +20, +10, 10, 20, 30 Haines Young, 2011 Literature review Nutrient_cyc_ESscores Assigned utility scores for nutrient cycling ecosystem services with a range from 30 to +3 0. +30, +20, +10, 10, 20, 30 Haines Young, 2011 Literature review Overall_scores Sum of carbon sequestration, climate regulation, and nutrient cycling ecosystem service scores representing overall ecosystem service val ues of different scenarios. Very good Good Fair Poor Bad Very bad Haines Young, 2011 Percentile calculation
252 Table 6 2 . Utility function structure used for calculating impacts of socio ecological scenar ios on the ecosystem service values. States of awareness input States of ecological input Utility value for ecosystem services Awareness of soil quality Terrestrial carbon stocks Carbon sequestration ecosystem service scores Aware High 30 Aware Medium 20 Aware Low 10 Unaware High 10 Unaware Medium 20 Unaware Low 30 Awareness of air quality Air quality Climate regulation ecosystem service scores Aware Good 30 Aw are Moderate 20 Aware Unhealthy 10 Unaware Good 10 Unaware Moderate 20 Unaware Unhealthy 30 Awareness of water quality Water quality Nutrient cycling ecosystem service scores Aware Good 30 Aware Acceptable 20 Aware Bad 10 Unaware Good 10 Unaware Acceptable 20 Unaware Bad 30
253 Table 6 3 . Sensitivity analysis results performed for the carbon sequestration ecosystem ser vice node (Carbon_seq_ESscores). Node (variable) Entropy reduction (%) Variance of beliefs Justification method Aware_soil_quality 9.67 0.0286398 Survey incorporation Terrestrial_C 6.86 0.0162089 GIS application Soil_quality_concern 1.27 0.0037353 Surv ey incorporation Imp_soil_quality 0.54 0.0016217 Survey incorporation Above_ground_C 0.37 0.0009971 GIS application CO2_emissions 0.05 0.0001254 Expert involvement Soil_C 0.04 0.0001201 GIS application LC/LU_coverage 0.02 0.0000606 Literature review; GIS application SOC_seqst_rate 0.003 0.0000087 GIS application NPP 0.003 0.0000072 GIS application Table 6 4 . Sensitivity analysis results performed for the climate regulation ecosystem service node (Climate_reg_ESscores). Node (variable) Entropy red uction (%) Variance of beliefs Justification method Aware_air_quality 18.9 0.0489859 Survey incorporation Air_quality 10.3 0.0230205 Literature review Air_quality_concern 1.74 0.0042750 Survey incorporation CO2_emissions 0.64 0.0014835 Expert invo lvement Imp_air_quality 0.15 0.0042750 Survey incorporation NPP 0.06 0.0014835 GIS application Annual_temp 0.03 0.0004019 Literature review; GIS application Scenario 0.01 0.0001498 Literature review Regulations 0.005 0.0000601 Survey incorpo ration Annual_precip 0.004 0.0000335 Literature review; GIS application Table 6 5 . Sensitivity analysis results performed for the nutrient cycling ecosystem services node (Nutrient_cyc_ESscores). Node (variable) Entropy reduction (%) Variance of bel iefs Justification method Water_quality 15.7 0.0249650 Statistical analysis; Expert involvement Aware_water_quality 6.06 0.0147331 Survey incorporation Nutrnt_loads_water 5.81 0.0098815 Statistical analysis; Expert involvement Avg_annual_flows 1. 64 0.0027927 GIS application Surplus_nutrnt_water 0.40 0.0008026 Expert involvement Implement_BMPs 0.20 0.0003995 GIS application Imp_water_quality 0.19 0.0004470 Survey incorporation Water_quality_concern 0.18 0.0003870 Survey incorporation R egulations 0.05 0.0000962 Survey incorporation Scenario 0.01 0.0000276 Literature review
254 Table 6 6 . Carbon content in the Suwannee River Basin, Florida. Carbon pools Carbon value (kg C m 2 ) Area Carbon budget (Tg C yr 1 ) Period Sources Top soil (0 20 cm) 3.8 19,480 km 2 75 2008/09 Rapid Assessment and Modeling of Changes in Soil Carbon Storage and Turnover in a Southern Landscape (Grunwald et al., 2011a) Deep soil (0 100 cm) 8.6 19,480 km 2 168 1925 2011 gSURRGO database (Natural Resources Conservation Service (NRCS), 2013) Biomass 4.7 19,480 km 2 92 1999 2002 National Biomass and Carbon Dataset (NBCD2000 Version 2.0) (Kellnorfer et al., 2013) Surface water a 0.002 23,092 km 2 0.05 2000 2010 Suwannee River Water Management District (SRWMD) and U.S. Geological Survey; this dissertation a Calculated from drainage areas with outlets to the Gulf of Mexic o. Table 6 7. Carbon fluxes in the Suwannee River Basin, Florida. Carbon fluxes Units Fluxes Carbon budget in the area Period Sources Net primary productivity 19,480 km 2 9 kg C m 2 yr 1 175 Tg C 2000 200 9 MODIS17 for North American Carbon Program (NACP ) (2009) Atmosphere (CO 2 ) 19,665 km 2 0. 1 kg m 2 yr 1 2 Tg C 1999 2008 Vulcan Project (Gurney et al., 2009) Human a 337,812 p eople in the basin 1 kg CO 2 day 1 0.03 Tg C 2010 Carbon Dioxide Information Analysis Center (CDIAC), (2013) ; U.S. Census, (2010b) a The rate of carbon dioxide exhalation is approximately 1 kg/day/person. An amount of kg CO 2 was converted to kg C in the table.
255 Figure 6 1 . Conceptual model illustrating the key variables used to predict the overall ecosystem service scores as a function of carbon sequestration, climate regulation, and nutrient cycling under multiple socio ecological drivers.
256 Fi gure 6 2 . Results of the Bayesian Belief N etwork when all scenarios (BU, Business as U sual; GP, Go toward Projection; GE, Gain E conomic value; GA, Go with A wareness) are equally likely.
257 Figure 6 3 . Results of the Bayesian B elief Network for scenari o BU (Business as U sual).
258 Figure 6 4 . Results of the Bayesian Belief Network for scenario GP ( G o toward the P rojection).
259 Figure 6 5 . Results of the Bayesian Belief Network for scenario GE ( Gain E conomic value).
260 Figure 6 6 . Results of the Bayesian Belief Network for scenario GA ( Go with A wareness).
261 APPENDIX SURVEY INSTRUMENT Ecosystem Services Survey in the Suwannee River Basin This survey is being conducted by the Soil and Water Science Department at the University of Florida to evalu in the Suwannee River Basin. The survey is being given to a random sample of households throughout 15 counties in the Suwannee River Water Management District (SRWMD) . Your participation on this sur vey is voluntary, and you do not have to answer any question that you do not wish to, however your full participation would be greatly appreciated. There are no direct benefits or risks to you for participating in the study. Please have an adult (age 18 o r older) in your household complete this survey. The survey will require about 10 minutes to complete. Your responses will be anonymous and results will be published only in summary form. If you have any questions about this survey, you may contact the inv estigator (see below) or the faculty advisor Alan W. Hodges (telephone 352 392 1881 ext. 312, email email@example.com). For questions about your rights as a research participant, contact the University of Florida Institutional Review Board (telephone 352 39 2 0433). Thank you for your cooperation! Sincerely, Pasicha Chaikaew Ph.D. student, Principal Investigator University of Florida Soil and Water Science Department PO Box 110290, Gainesville, FL 32611 Telephone 352 392 1812, email firstname.lastname@example.org
262 Secti on 1: Ecosystem services in your community Ecosystem services refer to the tangible and intangible benefits that people obtain from ecosystems, such as providing agricultural products, purifying air and water, cycling soil nutrients, stabilizing climate, p roviding wildlife habitats and human recreation. 1. How familiar are you with the following terms about ecosystem services? Please circle the number . Ecosystem services terms Not familiar Somewhat familiar Very familiar know Climate regulation 1 2 3 X Carbon storage 1 2 3 X Water quality protection 1 2 3 X Nutrient cycling 1 2 3 X 2. What types of ecosystem services are important to you in the Suwannee River Basin? Please circle Ecosystem service Not important Somewhat important Very important know Agricultur al production 1 2 3 X Forest production 1 2 3 X Good water quality 1 2 3 X Clean air 1 2 3 X Renewing soil fertility 1 2 3 X Nature recreation and tourism 1 2 3 X 3. What are your concerns regarding possible impacts of current environmental condition s that affect your quality of life? Please circle the number representing your degree of concern about each Environmental condition Not worried Somewhat worried Very worried know Global climate change 1 2 3 X Poor water quality 1 2 3 X Poor soil fertility 1 2 3 X Low agricultural productivity 1 2 3 X
263 Section 2: Climate regulation ed by Do not believe at all Somewhat believe Completely believe know 1 2 3 4 5 X 5. Based on your experience, what e ffects attributed to climate change have you observed in your community during the past 10 years or more (check any that apply). roughts or low rainfall or hurricanes 6. In your opinion, should state and local government policies be more or less aggressive in dealing with climate change? Please circle the numb Much less aggressive No change Much more aggressive know 1 2 3 4 5 X Section 3: Soil carbon sequestration (storage) 7. Consider the figure below showin g that one of the important ways to reduce carbon dioxide (CO 2 ) emissions in the atmosphere is to capture carbon into the biomass of plants or soils. Within the Suwannee River Water Management District, forestry and agriculture are dominant land uses that can play an important role in moderating climate change by increasing the storage of carbon in plants and the soil.
264 How do you think soil carbon stocks have been changed in your area in the past 30 years? nificant change Section 4: Nutrient cycling 8. Theoretically soils become more fertile and plant production increases when nitrogen and phosphorus are added. However, an excess of nutrients applied can affect water quality in nearby surface water bodies and groundwater. To what degree do you agree with the statement: ? Not agree at all Somewhat agree Very much agree know 1 2 3 4 5 X
265 9. What type of activitie s do you think most degrade the water quality in your area. Please check all that apply. Rainfall runoff from urban areas (i.e. parking lots, streets, yards) Industry discharges Leakage from septic systems Please specify__________ _________________________ 10. What are the best ways, in your opinion, to improve water quality in rivers or streams in your area? Please rank the following options in order of preference (1 = most preferred, 4 = least preferred) Stricter la nd use controls _____ State regulations for surface water runoff _____ Educate residents about water conditions _____ Building water treatment facilities _____ Other measures _____ Please specify_________________________________ ____ _ Section 5: Economic valuation Suppose that a program was implemented in your area to protect ecosystem services and improve environmental quality. The program would be financed through either voluntary donations, annual taxes or utility fees. The progra m would be organized in terms of four attributes: Type of ecosystem services to be managed 1) climate/carbon storage, 2) nutrient control, 3) agricultural and forestry production. Program administration 1) taxes levies by the Suwanee River Water Managem ent District (SRWMD), 2) taxes or utility fees levied by county government, or 3) donations to non government organizations (NGO). Location of area to be managed The area within the Suwannee River Water Management District that should be managed: 1)
266 an ywhere within the district, 2) within 20 miles of your home, or 3) within 5 miles of your home. Willingness to pay Amount per household annually you would be willing to spend over a period of 5 years to pay for the program: 1) $5, 2) $25, or 3) $50. 11. Please indicate your preference for how the environmental protection/improvement program should be designed in the following series of five choices between two different programs by checking the box below (A vs. B), or check if neither program is preferre d or acceptable. Assume that you can choose only ONE program for each choice. CHOICE 1 Program A Program B Type of ecosystem services to be managed Climate/carbon regulation Nutrient control Program administration SRWMD taxes Donation to Non Govern ment Organization Not interested in either program Location of area to be managed Anywhere within Water Management District Within 20 miles of my home Willingness to pay per year $5 $50 Please mark the program you would prefer
267 CHOICE 2 Pro gram A Program B Type of ecosystem services to be managed Climate/carbon regulation Nutrient control Program administration Donation to Non Government Organization County taxes Not interested in either program Location of area to be managed Within 5 miles of my home Anywhere within Water Management District Willingness to pay per year $25 $5 Please mark the program you would prefer CHOICE 3 Program A Program B Type of ecosystem services to be managed Nutrient control Agricultural /for estry production Program administration County taxes SRWMD taxes Not interested in either program Location of area to be managed Within 5 miles of my home Anywhere within Water Management District Willingness to pay per year $5 $50 Please mark the program you would prefer
268 CHOICE 4 Program A Program B Type of ecosystem services to be managed Agricultural/ forestry production Climate/carbon regulation Program administration SRWMD taxes Donation to Non Government Organization Not inter ested in either program Location of area to be managed Within 5 miles of my home Anywhere within Water Management District Willingness to pay per year $50 $25 Please mark the program you would prefer CHOICE 5 Program A Program B Type of ec osystem services to be managed Agricultural /forestry production Climate/carbon regulation Program administration Donation to Non Government Organization County taxes Not interested in either program Location of area to be managed Within 20 miles of m y home Within 5 miles of my home Willingness to pay per year $5 $50 Please mark the program you would prefer
269 LIST OF REFERENCES Abley, J., 2000. Stated preference techniques and consumer dec challenges to old assumptions (Working Paper). Adamowicz, W., Boxall, P., Williams, M., Louviere, J., 1998. Stated Preference Approaches for Measuring Passive Use Values: Choice Experiments and Contingent Valuation. Am. J. Agric. Econ. 8 0, 64 75. doi:10.2307/3180269 Aerts, J.C.J.H., Heuvelink, G.B.M., 2002. Using simulated annealing for resource allocation. Int. J. Geogr. Inf. Sci. 16, 571 587. doi:10.1080/13658810210138751 Aguilera, P.A., FernÃ¡ndez, A., FernÃ¡ndez, R., RumÃ, R., SalmerÃ³n, A., 2011. Bayesian networks in environmental modelling. Environ. Model. Softw. 26, 1376 1388. doi:10.1016/j.envsoft.2011.06.004 Ahn, M. Y., Zimmerman, A.R., Comerford, N.B., Sickman, J.O., Grunwald, S., 2009. Carbon Mineralization and Labile Organic Carbo n Pools in the Sandy Soils of a North Florida Watershed. Ecosystems 12, 672 685. doi:10.1007/s10021 009 9250 8 Ainsworth, C.H., Morzaria Luna, H., Kaplan, I.C., Levin, P.S., Fulton, E.A., Cudney Bueno, R., Turk Boyer, P., Torre, J., Danemann, G.D., Pfister , T., 2012. Effective ecosystem based management must encourage regulatory compliance: A Gulf of California case study. Mar. Policy 36, 1275 1283. doi:10.1016/j.marpol.2012.03.016 Alan Pounds, J., Bustamante, M.R., Coloma, L.A., Consuegra, J.A., Fogden, M. P.L., Foster, P.N., La Marca, E., Masters, K.L., Merino Viteri, A., Puschendorf, R., Ron, S.R., SÃ¡nchez Azofeifa, G.A., Still, C.J., Young, B.E., 2006. Widespread amphibian extinctions from epidemic disease driven by global warming. Nature 439, 161 167. do i:10.1038/nature04246 Alldredge, J.R., Griswold, J., 2006. Design and Analysis of Resource Selection Studies for Categorical Resource Variables. J. Wildl. Manag. 70, 337 346. doi:10.2193/0022 541X(2006)70[337:DAAORS]2.0.CO;2 Ãlvarez Farizo, B., Gil, J.M., Howard, B.J., 2009. Impacts from restoration strategies: Assessment through valuation workshops. Ecol. Econ. 68, 787 797. doi:10.1016/j.ecolecon.2008.06.012 American Coalition for Clean Coal Electricity (ACCCE), 2012. Energy Cost Impacts on American Famili es, 2001 2012. Andersen, M.D., Kerr, G.N., Lambert, S.J., 2012. Cultural differences in environmental valuation (2012 Conference, August 31, 2012, Nelson, New Zealand No. 136040). New Zealand Agricultural and Resource Economics Society.
270 Andrews, W.J., 1994 . Nitrate in ground water and spring water near four dairy farms in North Florida, 1990 93 (No. WRI 94 4162). United States Geological Survey. Angers, D.A., Arrouays, D., Saby, N.P.A., Walter, C., 2011. Estimating and mapping the carbon saturation defici t of French agricultural topsoils. Soil Use Manag. 27, 448 452. doi:10.1111/j.1475 2743.2011.00366.x Anselin, L., 1995. Local indicators of spatial association LISA. Geogr. Anal. 27, 93 115. doi:10.1111/j.1538 4632.1995.tb00338.x AraÃ±a, J.E., LeÃ³n, C.J., 2 012. Scale perception bias in the valuation of environmental risks. Appl. Econ. 44, 2607 2617. doi:10.1080/00036846.2011.566188 Aretano, R., Petrosillo, I., Zaccarelli, N., Semeraro, T., Zurlini, G., 2013. People perception of landscape change effects on e cosystem services in small Mediterranean islands: A combination of subjective and objective assessments. Landsc. Urban Plan. 112, 63 73. doi:10.1016/j.landurbplan.2012.12.010 Arslan, S., 2012. The Influence of Environment Education on Critical Thinking and Environmental Attitude. Procedia Soc. Behav. Sci. 55, 902 909. doi:10.1016/j.sbspro.2012.09.579 Asbury, C.E., Oaksford, E.T., National Water Quality Assessment Program (U.S.), Geological Survey (U.S.), 1997. A comparison of drainage basin nutrient input s with instream nutrient loads for seven rivers in Georgia and Florida, 1986 90, Wat er Ayanu, Y.Z., Conrad, C., Nauss, T., Wegmann, M., Koellner, T., 2012. Quantifying and Mapping Ecosystem Service s Supplies and Demands: A Review of Remote Sensing Applications. Environ. Sci. Technol. 46, 8529 8541. doi:10.1021/es300157u Bagstad, K.J., Semmens, D.J., Waage, S., Winthrop, R., 2013. A comparative assessment of decision support tools for ecosystem servi ces quantification and valuation. Ecosyst. Serv. 5, 27 39. doi:10.1016/j.ecoser.2013.07.004 Bagstad, K.J., Villa, F., Johnson, G., Voigt, B., 2011. ARIES Artificial Intellidence for Ecosystem Services: A Guide to Models and Data, Version 1.0 Beta, The AR IES Consortium. Bilbao, Spain. Available at http://www.ariesonline.org/toolkit.html. Baker, J.M., Ochsner, T.E., Venterea, R.T., Griffis, T.J., 2007. Tillage and soil carbon sequestration What do we really know? Agric. Ecosyst. Environ. 118, 1 5. doi:10.10 16/j.agee.2006.05.014
271 Balvanera, P., Pfisterer, A.B., Buchmann, N., He, J. S., Nakashizuka, T., Raffaelli, D., Schmid, B., 2006. Quantifying the evidence for biodiversity effects on ecosystem functioning and services. Ecol. Lett. 9, 1146 1156. doi:10.1111 /j.1461 0248.2006.00963.x Banerjee, M., Wellner, J.A., 2005. Score Statistics for Current Status Data: Comparisons with Likelihood Ratio and Wald Statistics. Int. J. Biostat. 1, 1001 1001. Bardgett, R.D., 2011. Plant soil interactions in a changing world. F1000 Biol. Rep. 3. doi:10.3410/B3 16 Barnes, J., 2005. Awareness to Action: the journey toward a deeper ecological literacy 5. Bateman, I.J., Carson, R.T., Day, B., Hanemann, W.M., Hanley, N., Hett, T., Jones Lee, M., Loomes, G., M Swanson, S., 2003. Guidelines for the Use of Stated Preference Techniques for the Valuation of Preferences for Non Market Goods. Edward Elgar Publishing, Cheltenham, UK. Batten, G.D., 1998. Plant analys is using near infrared reflectance spectroscopy: the potential and the limitations. Aust. J. Exp. Agric. 38, 697 706. Bayes, M., Price, M., 1763. An Essay towards Solving a Problem in the Doctrine of Chances. By the Late Rev. Mr. Bayes, F. R. S. Communicat ed by Mr. Price, in a Letter to John Canton, A. M. F. R. S. Philos. Trans. 53, 370 418. doi:10.1098/rstl.1763.0053 Bebbington, A., Farrington, J., Lewis, D.J., Wellard, K., 2005. Reluctant Partners? Non Governmental Organizations, the State and Sustainable Agricultural Development. Routledge. Bech, M., Gyrd Hansen, D., 2005. Effects coding in discrete choice experiments. Health Econ. 14, 1079 1083. doi:10.1002/hec.984 Beckett, P.H.T., Webster, R., 1971. Soil variability: a review. Soils Fertil. 34, 1 15. BÃ© langer, N., Pinno, B.D., 2008. Carbon sequestration, vegetation dynamics and soil development in the Boreal Transition ecoregion of Saskatchewan during the Holocene. CATENA 74, 65 72. doi:10.1016/j.catena.2008.03.005 Bellassen, V., Luyssaert, S., 2014. Car bon sequestration: Managing forests in uncertain times. Nature 506, 153 155. doi:10.1038/506153a Bengtsson, J., Angelstam, P., Elmqvist, T., Emanuelsson, U., Folke, C., Ihse, M., Moberg, F., NystrÃ¶m, M., 2003. Reserves, Resilience and Dynamic Landscapes. A mbio 32, 389 396.
272 Bennett, J., Blamey, R., 2001. The Choice Modelling Approach to Environmental Valuation. Edward Elgar Publishing. Bernal, B., Mitsch, W.J., 2012. Comparing carbon sequestration in temperate freshwater wetland communities. Glob. Change Bio l. 18, 1636 1647. doi:10.1111/j.1365 2486.2011.02619.x Bewick, V., Cheek, L., Ball, J., 2005. Statistics review 14: Logistic regression. Crit. Care 9, 112 118. doi:10.1186/cc3045 Birol, E., Smale, M., Gyovai, Ã., 2006. Using a choice experiment to estimate valuation of agrobiodiversity on Hungarian small farms. Environ. Resour. Econ. 34, 439 469. Bishop, R.C., Welsh, M.P., 1992. Existence values in benefit cost analysis and damage assessment. Land Econ. 68, 405. Blanco Canqui, H., Lal, R., 2008. No Tillage and Soil Profile Carbon Sequestration: An On Farm Assessment. Soil Sci. Soc. Am. J. 72, 693. doi:10.2136/sssaj2007.0233 Bliss, N.B., Waltman, S.W., Petersen, G.W., 1995. Preparing a soil carbon inventory for the United States using geographic info rmation systems, eds. CRC Press, Boca Raton, Fla. Boissiere, M., Locatelli, B., Sheil, D., Padmanaba, M., Sadjudin, E., 2013. Local perceptions of climate variability and change in tropical forests of Papua, Indonesia. Ecol. Soc. art. 13. doi:http://dx.doi .org/10.5751/ES 05822 180413 Bond Lamberty, B., Thomson, A., 2010. Temperature associated increases in the global soil respiration record. Nature 464, 579 582. doi:10.1038/nature08930 Boumans, R., Costanza, R., Farley, J., Wilson, M.A., Portela, R., Rotman s, J., Villa, F., Grasso, M., 2002. Modeling the dynamics of the integrated earth system and the value of global ecosystem services using the GUMBO model. Ecol. Econ. 41, 529 560. doi:10.1016/S0921 8009(02)00098 8 Boxall, P.C., Adamowicz, W.L., Swait, J., Williams, M., Louviere, J., 1996. A comparison of stated preference methods for environmental valuation. Ecol. Econ. 18, 243 253. doi:10.1016/0921 8009(96)00039 0 Boyd, J., Banzhaf, S., 2007. What are ecosystem services? The need for standardized environme ntal accounting units. Ecol. Econ. 63, 616 626. Bracho, R., Starr, G., Gholz, H.L., Martin, T.A., Cropper, W.P., Loescher, H.W., 2011. Controls on carbon dynamics by ecosystem structure and climate for southeastern U.S. slash pine plantations. Ecol. Monogr . 82, 101 128. doi:10.1890/11 0587.1
273 Brady, N.C., Weil, R.R., 2008. The Nature and Properties of Soils, 14th ed. Prentice Hall, Upper Saddle River, NJ. Braund, M., 2008. The Structures of Perception: An Ecological Perspective. Kritike 2, . Breiman, L., 19 84. Classification and regression trees. Chapman & Hall. Breiman, L., 1996. Bagging Predictors. Mach. Learn. 24, 123 140. doi:10.1023/A:1018054314350 Breiman, L., 2001. Random Forests. Mach. Learn. 45, 5 32. doi:10.1023/A:1010933404324 Bridgham, S.D., Mego nigal, J.P., Keller, J.K., Bliss, N.B., Trettin, C., 2006. The carbon balance of North American. Wetlands 26, 889 916. doi:10.1672/0277 5212(2006)26[889:TCBONA]2.0.CO;2 Broadbent, C., 2012. Hypothetical Bias, Consequentiality and Choice Experiments. Econ. Bull. 32, 2490 2499. Broadbent, C.D., 2013. Evaluating mitigation and calibration techniques for hypothetical bias in choice experiments. J. Environ. Plan. Manag. 0, 1 19. doi:10.1080/09640568.2013.839447 Broadbent, C.D., Grandy, J.B., Berrens, R.P., 2010. Testing for Hypothetical Bias in a Choice Experiment Using a Local Public Good: Riparian Forest Restoration. Int. J. Ecol. Econ. Stat. 19, 1 19. Brouwer, R., Martin Ortega, J., Berbel, J., 2010. Spatial Preference Heterogeneity: A Choice Experiment. Land Econ. 86, 552 568. Bruland, G.L., Bliss, C.M., Grunwald, S., Comerford, N.B., Graetz, D.A., 2008. Soil nitrate nitrogen in forested versus non forested ecosystems in a mixed use watershed. Geoderma 148, 220 231. doi:10.1016/j.geoderma.2008.10.005 Bullock, C.H., Elston, D.A., Chalmers, N.A., 1998. An application of economic choice experiments to a traditional land use deer hunting and landscape change in the Scottish Highlands. J. Environ. Manage. 52, 335 351. doi:10.1006/jema.1997.0179 Burkett, V., Ritschar Murray, B., Jagtap, S., Cruise, J., 2001. Potential Consequences of Climate Variability and Change for the Southeastern United States, in the Potential Consequences of Climate Variability an d Change: Foundation Report, National Assessment Synthesis Team for the U.S. Global Change Research Program. Cambridge University Press, UK.
274 Burkett, V.R., Wilcox, D.A., Stottlemyer, R., Barrow, W., Fagre, D., Baron, J., Price, J., Nielsen, J.L., Allen, C. D., Peterson, D.L., Ruggerone, G., Doyle, T., 2005. Nonlinear dynamics in ecosystem response to climatic change: Case studies and policy implications. Ecol. Complex. 2, 357 394. doi:10.1016/j.ecocom.2005.04.010 Cain, J., 2001. Planning improvements in natu ral resource management. Guidelines for using Bayesian networks to support the planning and management of development programmes in the water sector and beyond. Centre for Ecology and Hydrology, Wallingford. Cambardella, C.A., Moorman, T.B., Parkin, T.B., Karlen, D.L., Novak, J.M., Turco, R.F., Konopka, A.E., 1994. Field Scale Variability of Soil Properties in Central Iowa Soils. Soil Sci. Soc. Am. J. 58, 1501. doi:10.2136/sssaj1994.03615995005800050033x Cameron, T.A., Poe, G.L., Ethier, R.G., Schulze, W.D. , 2002. Alternative Non market Value Elicitation Methods: Are the Underlying Preferences the Same? J. Environ. Econ. Manag. 44, 391 425. doi:10.1006/jeem.2001.1210 Campbell, C.A., Zentner, R.P., Bowren, K.E., Townley Smith, L., Schnitzer, M., 1991. Effect of crop rotations and fertilization on soil organic matter and some biochemical properties of a thick Black Chernozem. Can. J. Soil Sci. 71, 377 387. doi:10.4141/cjss91 036 Cantrill, J.G., 1998. The environmental self and a sense of place: Communication fo undations for regional ecosystem management. J. Appl. Commun. Res. 26, 301 318. doi:10.1080/00909889809365509 Carbon Dioxide information Analysis Center (CDIAC), 2013. Frequently Asked Global Change Questions. Cardinale, B.J., Duffy, J.E., Gonzalez, A., Ho oper, D.U., Perrings, C., Venail, P., Narwani, A., Mace, G.M., Tilman, D., Wardle, D.A., Kinzig, A.P., Daily, G.C., Loreau, M., Grace, J.B., Larigauderie, A., Srivastava, D.S., Naeem, S., 2012. Biodiversity loss and its impact on humanity. Nature 486, 59 6 7. doi:10.1038/nature11148 Carlsson, F., Martinsson, P., 2001. Do Hypothetical and Actual Marginal Willingness to Pay Differ in Choice Experiments?: Application to the Valuation of the Environment. J. Environ. Econ. Manag. 41, 179 192. doi:10.1006/jeem.200 0.1138 Carlton, S.J., Jacobson, S.K., 2013. Climate change and coastal environmental risk perceptions in Florida. J. Environ. Manage. 130, 32 39. doi:10.1016/j.jenvman.2013.08.038
275 Carpenter, S.R., Mooney, H.A., Agard, J., Capistrano, D., Defries, R.S., DÃa z, S., Dietz, T., Duraiappah, A.K., Oteng Yeboah, A., Pereira, H.M., Perrings, C., Reid, W.V., Sarukhan, J., Scholes, R.J., Whyte, A., 2009. Science for managing ecosystem services: Beyond the Millennium Ecosystem Assessment. Proc. Natl. Acad. Sci. U. S. A . 106, 1305 1312. doi:10.1073/pnas.0808772106 Carson, R.T., Flores, N.E., Meade, N.F., 2001. Contingent Valuation: Controversies and Evidence. Environ. Resour. Econ. 19, 173 210. doi:10.1023/A:1011128332243 Carter, M.R., Angers, D.A., Gregorich, E.G., Boli nder, M.A., 2003. Characterizing organic matter retention for surface soils in eastern Canada using density and particle size fractions. Can. J. Soil Sci. 83, 11 24. Celio, E., Koellner, T., GrÃªt Regamey, A., 2014. Modeling land use decisions with Bayesian networks: Spatially explicit analysis of driving forces on land use change. Environ. Model. Softw. 52, 222 233. doi:10.1016/j.envsoft.2013.10.014 Champ, P.A., Bishop, R.C., Brown, T.C., McCollum, D.W., 1997. Using Donation Mechanisms to Value Nonuse Benef its from Public Goods. J. Environ. Econ. Manag. 33, 151 162. doi:10.1006/jeem.1997.0988 Chee, Y.E., 2004. An ecological perspective on the valuation of ecosystem services. Biol. Conserv. 120, 549 565. Chen, S.H., Pollino, C.A., 2012. Good practice in Bayes ian network modelling. Environ. Model. Softw. 37, 134 145. doi:10.1016/j.envsoft.2012.03.012 Chmura, G.L., Anisfeld, S.C., Cahoon, D.R., Lynch, J.C., 2003. Global carbon sequestration in tidal, saline wetland soils. Glob. Biogeochem. Cycles 17, 1111. doi:1 0.1029/2002GB001917 R., Franklin, J.F., MacMahon, J.A., Noss, R.F., Parsons, D.J., Peterson, C.H., Turner, M.G., Woodmansee, R.G., 1996. The Report of the Ecological Soc iety of America Committee on the Scientific Basis for Ecosystem Management. Ecol. Appl. 6, 665 691. doi:10.2307/2269460 Christy, J.R., Norris, W.B., Redmond, K., Gallo, K.P., 2006. Methodology and Results of Calculating Central California Surface Temperatu re Trends: Evidence of Human Induced Climate Change? J. Clim. 19, 548 563. doi:10.1175/JCLI3627.1 Clark, J.S., 2005. Why environmental scientists are becoming Bayesians. Ecol. Lett. 8, 2 14. doi:10.1111/j.1461 0248.2004.00702.x Clemen, R.T., Winkler, R.L., 1999. Combining Probability Distributions From Experts in. Risk Anal. 19, 187 203. doi:10.1023/A:1006917509560
276 Cook, J., Nuccitelli, D., Green, S.A., Richardson, M., Winkler, B., Painting, R., Way, R., Jacobs, P., Skuce, A., 2013. Quantifying the consensu s on anthropogenic global warming in the scientific literature. Environ. Res. Lett. 8, 024024. doi:10.1088/1748 9326/8/2/024024 Raskin, R.G., Sutton, P., Belt, M. van Online 15 May 1997 Doi101038387253a0 387, 253 260. doi:10.1038/387253a0 Costanza, R., Liu, S., 2014. Ecosystem Services and Environmen tal Governance: Comparing China and the U.S. Asia Pac. Policy Stud. n/a n/a. doi:10.1002/app5.16 Council on Environmental Quality (CEQ), 2013. Principles and Requirements for federal Investments in Water Resources. Council on Environmental Quality, Washing ton, DC. Craft, C., Washburn, C., Parker, A., 2008. Latitudinal Trends in Organic Carbon Accumulation in Temperate Freshwater Peatlands, in: Vymazal, D.J. (Ed.), Wastewater Treatment, Plant Dynamics and Management in Constructed and Natural Wetlands. Sprin ger Netherlands, pp. 23 31. Craft, C.B., Richardson, C.J., 1993. Peat Accretion and N, P, and Organic C Accumulation in Nutrient Enriched and Unenriched Everglades Peatlands. Ecol. Appl. 3, 446 458. doi:10.2307/1941914 Creedy, J., Doran, H., Duffield, S., Environment in 2060 Issues, Implications and Scenarios (National England Research Report No. NERR031). East Parade, Sheffield, UK. Cummings, R.G., Harrison, G.W., RutstrÃ¶m, E.E., 1995. Homegrown Values and H ypothetical Surveys: Is the Dichotomous Choice Approach Incentive Compatible? Am. Econ. Rev. 85, 260 266. Services: Societal Dependence on Natural Ecosystems. Island Press, Was hington, DC. Daily, G.C., SÃ¶derqvist, T., Aniyar, S., Arrow, K., Dasgupta, P., Ehrlich, P.R., Folke, C., Jansson, A., Jansson, B. O., Kautsky, N., Levin, S., Lubchenco, J., MÃ¤ler, K. G., Simpson, D., Starrett, D., Tilman, D., Walker, B., 2000. The value of nature and the nature of value. Science 289, 395 396. doi:10.1126/science.289.5478.395 Dale, V.H., Polasky, S., 2007. Measures of the effects of agricultural practices on ecosystem services. Ecol. Econ. 64, 286 296. doi:10.1016/j.ecolecon.2007.05.009
277 Davi dson, E.A., Ackerman, I.L., 1993. Changes in soil carbon inventories following cultivation of previously untilled soils. Biogeochemistry 20, 161 193. doi:10.1007/BF00000786 Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomp osition and feedbacks to climate change. Nature 440, 165 173. doi:10.1038/nature04514 Davies, R.B., 1987. Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika 74, 33 43. doi:10.1093/biomet/74.1.33 De Groot, R., Bra nder, L., van der Ploeg, S., Costanza, R., Bernard, F., Braat, L., Christie, M., Crossman, N., Ghermandi, A., Hein, L., Hussain, S., Kumar, P., McVittie, A., Portela, R., Rodriguez, L.C., ten Brink, P., van Beukering, P., 2012. Global estimates of the valu e of ecosystems and their services in monetary units. Ecosyst. Serv. 1, 50 61. doi:10.1016/j.ecoser.2012.07.005 De Groot, R.S., Wilson, M.A., Boumans, R.M.J., 2002. A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecol. Econ. 41, 393 408. doi:10.1016/S0921 8009(02)00089 7 De Klein, J.J.M., van der Werf, A.K., 2013. Balancing carbon sequestration and GHG emissions in a constructed wetland. Ecol. Eng. doi:10.1016/j.ecoleng.2013.04.060 us, K.E., 2000. Classification and regresison trees: A powerful yet simple techinique for ecological data analysis. Ecology 81, 3178 3192. doi:10.1890/0012 9658(2000)081[3178:CARTAP]2.0.CO;2 DeLucia, E.H., Hamilton, J.G., Naidu, S.L., Thomas, R.B., Andrews , J.A., Finzi, A., Lavine, M., Matamala, R., Mohan, J.E., Hendrey, G.R., Schlesinger, W.H., 1999. Net Primary Production of a Forest Ecosystem with Experimental CO2 Enrichment. Science 284, 1177 1179. doi:10.1126/science.284.5417.1177 Denizman, C., Randazz o, A.F., 2000. Post miocene subtropical karst evolution, lower Suwannee River Basin, Florida. Geol. Soc. Am. Bull. 112, 1804 1813. Derner, J.D., Schuman, G.E., 2007. Carbon sequestration and rangelands: A synthesis of land management and precipitation effe cts. J. Soil Water Conserv. 62, 77 85. Deyong, Y., Hongbo, S., Peijun, S., Wenquan, Z., Yaozhong, P., 2009. How does the conversion of land cover to urban use affect net primary productivity? A case study in Shenzhen city, China. Agric. For. Meteorol., Spe cial Section on Water and Carbon Dynamics in Selected Ecosystems in China 149, 2054 2060. doi:10.1016/j.agrformet.2009.07.012 Diamond, P.A., Hausman, J.A., 1994. Contingent Valuation: Is Some Number better than No Number? J. Econ. Perspect. 8, 45 64.
278 Dillm an, D.A., Smyth, J.D., Christian, L.M., 2009. Internet, Mail, and Mixed Mode Surveys: The Tailored Design Method, 3rd ed. John Wiley & Sons, NY. Dlamini, W.M., 2010. A Bayesian belief network analysis of factors influencing wildfire occurrence in Swaziland . Environ. Model. Softw. 25, 199 208. doi:10.1016/j.envsoft.2009.08.002 Drake, B.G., GonzÃ lez Meler, M.A., Long, S.P., 1997. More efficient plants: A consequence of rising atmospheric CO2? Annu. Rev. Plant Physiol. Plant Mol. Biol. 48, 609 639. doi:10.1146 /annurev.arplant.48.1.609 Duarte, C.M., 2009. Global Loss of Coastal Habitats: Rates, Causes and Consequences. FundaciÃ³n BBVA. Dunn, B.W., Batten, G.D., Beecher, H.G., Ciavarella, S., 2002. The potential of near infrared reflectance spectroscopy for soil a nalysis a case study from the Riverine Plain of south eastern Australia. Aust. J. Exp. Agric. 42, 607 614. Dutton, A., Edwards Jones, G., Macdonald, D.W., 2010. Estimating the value of non use benefits from small changes in the provision of ecosystem ser vices. Conserv. Biol. J. Soc. Conserv. Biol. 24, 1479 1487. doi:10.1111/j.1523 1739.2010.01536.x Economic Research Services (ERS), 2013. Rural Urban Continuum Codes, Available at http://www.ers.usda.gov/data products/rural urban continuum codes/.aspx#.U3Sp pvldV5J. Department of Agriculture. Edmondson, J.L., Davies, Z.G., McHugh, N., Gaston, K.J., Leake, J.R., 2012. Organic carbon hidden in urban ecosystems. Sci. Rep. 2. doi:10.1038/srep00963 Egoh, B., Reyers, B., Rouget, M., Richardson, D.M., Le Maitre, D.C ., van Jaarsveld, A.S., 2008. Mapping ecosystem services for planning and management. Agric. Ecosyst. Environ. 127, 135 140. doi:10.1016/j.agee.2008.03.013 K., Morse, W., Nielsen Pincus, M., Stephens, J., Winowiecki, L., Bosque Perez, N.A., 2007. Employing philosophical dialogue in collaborative science. BioScience 57, 55 64. doi:10.1641/B570109 Endter Wada, J., Blahna, D., Krannich, R., Brunson, M., 1998. A Framework f or Understanding Social Science Contributions to Ecosystem Management. Ecol. Appl. 8, 891 904. doi:10.1890/1051 0761(1998)008[0891:AFFUSS]2.0.CO;2 Engel, U., PÃ¶tschke, M., 1998. Willingness to pay for the environment: Social structure, value orientations a nd environmental behaviour in a multilevel perspective. Innov. Eur. J. Soc. Sci. Res. 11, 315 332. doi:10.1080/13511610.1998.9968571
279 Engle, R.F., 1984. Wald, likelihood ratio, and Lagrange multiplier tests in econometrics (Handbook of Econometrics). Elsevi er. ERDAS Inc., 1989. Land Cover Land Use Classification for the Suwannee River Water Management District. FL. EsbjÃ¶rn Hargens, S., Zimmerman, M.E., 2009. Integral Ecology: Uniting Multiple Perspectives on the Natural World. Shambhala Publications. Farber, S., Costanza, R., Childers, D.L., Erickson, J., Gross, K., Grove, M., Hopkinson, C.S., Kahn, J., Pincetl, S., Troy, A., Warren, P., Wilson, M., 2006. Linking ecology and economics for ecosystem management. BioScience 56, 121 133. doi:10.1641/0006 3568(200 6)056[0121:LEAEFE]2.0.CO;2 Environ. Sci. Technol. 34, 1407 1412. doi:10.1021/es990727r Farrell, M.D., Good, J., Hornsby, D., Janicki, A., Mattson, R., Upchurch, S., Champion, K., Chen, J., Grabe, S., Malloy, K., Nijbroek, R., Schneider, J., Wade, D., Wessel, M., 2005. Technical Report: MFL establishment for the Lower Suwannee River and estuary Little Fanning, Fanning and Manatee Springs. Suwannee River Water Management Distric t in association with Water Resource Associates, Inc. and SDII Global Corporation. Feng, M., Liu, S., Euliss Jr., N.H., Young, C., Mushet, D.M., 2011. Prototyping an online wetland ecosystem services model using open model sharing standards. Environ. Model . Softw. 26, 458 468. doi:10.1016/j.envsoft.2010.10.008 Feng, W., Xu, M., Fan, M., Malhi, S.S., Schoenau, J.J., Six, J., Plante, A.F., 2013. Testing for soil carbon saturation behavior in agricultural soils receiving long term manure amendments. Can. J. So il Sci. 1 14. doi:10.4141/cjss2013 012 Fernald, E., Purdum, E.D., 1998. Water Resources Atlas of Florida. Institute of Science and Public Affairs, Florida State University, Tallahassee, FL. Field, C.B., Sarmiento, J., Hales, B., 2007. The carbon cycle of N orth America in a global context, in: A.W. King, L. Dilling, G.P. Zimmerman, D.M. Fairman, R.A. Houghton, G. Marland, A.Z. Rose, T.J. Wilbanks (Eds.), The First State of the Carbon Cycle Report (SOCCR), Synthesis and Assessment Product 2.2, Report by the U .S. Climate Change Science Program and the Subcommittee on Global Change Research. pp. 21 28. Fisher, B., Turner, R.K., 2008. Ecosystem services: Classification for valuation. Biol. Conserv. 141, 1167 1169. doi:10.1016/j.biocon.2008.02.019 Fisher, B., Turner, R.K., Morling, P., 2009. Defining and classifying ecosystem services for decision making. Ecol. Econ. 68, 643 653. doi:10.1016/j.ecolecon.2008.09.014
280 Florida Department of Agriculture and Consumer Services (FDACS), 2014. Total Fertilize r and Nutrients by County (1945 to 2010), Available at http://www.freshfromflorida.com/Divisions Offices/Agricultural Environmental Services/Agriculture Industry/Fertilizer Manufacturers/Fertilizer Consumption Tonnage Data. Florida Department of Environmen tal, P., 2001. Basin Status Report: Suwannee (Including Aucilla, Coastal, Suwannee and Waccasassa Basins in Florida). Florida Department of Environmental Protection (FDEP), 2001. Basin Status Report: Suwannee (Including Aucilla, Coastal, Suwannee and Wacca sassa Basins in Florida). Florida Department of Environmental Protection (FDEP), 2004a. Integrated Water Quality Assessment for Florida: 2004 305(b) Report and 303(d) List Update. Bureau of Water Resource Management, Tallahassee, FL. Florida Department of Environmental Protection (FDEP), 2004b. SRWMD 2004 Land Use. Available at http://www.dep.state.fl.us/gis/datadir.asp. Florida Department of Environmental Protection (FDEP), 2009. SRWMD 2009 Land Use. Available at http://www.dep.state.fl.us/gis/datadir.asp. Available at http://content.govdelivery.com/accounts/FLDEP/bulletins/a227ff. DEP Press Office. Flor ida Department of Environmnet Protection (FDEP), 2007. County Air Quality Indes, http://www.dep.state.fl.us/air/air_quality/countyaqi.htm. Florida Department of Agriculture and Consumer Services (FDACS). Florida Department of Environmnet Protection (FDEP), 2009. Land cover/land use. Florida Department of Environmnet Protection (FDEP), 2013. Implementation of or streams, spring vents, lakes, and selected estuaries, http://www.dep.state.fl.us/secretary/news/2013/03/NNC_Implementation_3 11 13.pdf. Florida Fish and Wildlife Commission (FFWC), 2003. Florida Vegetation and Land Cover Data Derived from 2003 Landsat E TM+Imagery by B.Styes et al. Office of Environmental Services, Tallahassee, FL.
281 Forster, P. et al., 2007. Changes in atmospheric constituents and in radiative forcing, in: Solomon, S., Qin, D., M. Manning, Z. Chen (Eds.), Climate Change 2004: The Physical Science Basis, Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, UK. Franzen, A., Meyer, R., 2010. Environmental Attitudes in Cross National Perspective: A Multile vel Analysis of the ISSP 1993 and 2000. Eur. Sociol. Rev. 26, 219 234. doi:10.1093/esr/jcp018 Franzluebbers, A.J., 2010. Achieving soil organic Ccrbon sequestration with conservation agricultural systems in the Southeastern United States. doi:10.2136/sssaj 2009.0079 Franzluebbers, A.J., Follett, R.F., 2005. Greenhouse gas contributions and mitigation potential in agricultural regions of North America: Introduction. Soil Tillage Res., Greenhouse Gas Contributions and Mitigation Potential in Agricultural Regio ns of North America 83, 1 8. doi:10.1016/j.still.2005.02.020 Freeman, A.M.I., 1993. The Measurement of Environmental and Resource Values: Theory and Methods. Washington, DC. Friedman, N., Goldszmidt, M., 1996. Discretizing continuous attributes while learn ing Bayesian Networks, in: In Proc. ICML. Morgan Kaufmann, pp. 157 165. Fujiwara, D., Campbell, R., 2011. Valuation Techniques for Social Cost Benefit Analysis: Stated Preference, Revealed Preference, and Subjective Well Being Approaches, A Discussion of t he Current Issues. Department for Work and Pensions, London. Gao, Y., Zhou, X., Wang, Q., Wang, C., Zhan, Z., Chen, L., Yan, J., Qu, R., 2013. Vegetation net primary productivity and its response to climate change during 2001 2008 in the Tibetan Plateau. S ci. Total Environ. 444, 356 362. doi:10.1016/j.scitotenv.2012.12.014 GarcÃa, A.M., Hoos, A.B., Terziotti, S., 2011. A Regional Modeling Framework of Phosphorus Sources and Transport in Streams of the Southeastern United States. J. Am. Water Resour. Assoc. AWRA 47, 991 1010. doi:10.1111/j.1752 1688.2010.00517.x Garvey, J., 2008. The Ethics of Climate Change: Right and Wrong in a Warming World. Continuum, London. Gelissen, J., 2007. Explaining Popular Support for Environmental Protection A Multilevel Analysis of 50 Nations. Environ. Behav. 39, 392 415. doi:10.1177/0013916506292014
282 GeoPlan, 2006. 2005 and 2060 Comparison of Landuse Allocation (1000 Friends of Florida Project), https://www.geoplan.ufl.edu/lucis/1000friends_results.html. Ginsburgh, V., Throsby, C .D., 2013. Handbook of the economics of art and culture. Vol. 2 Vol. 2. Sciences, Russian Pedological Investigations. Office of the Academy of Science U.S.S.R., Leningrad. Glo beScan, 2009. Global Climate Change Survey: Climate Confidence Monitor 2009, Available at http://www.globescan.com/component/edocman/?view=category&id=4&Itemid=5 91. Gloulder, L.H., Kennedy, D., 1997. Valuing ecosystem services: Philosophical bases and empi Dependence on Natural Ecosystems. Island Press, Washington, DC. Willingness to pay for Service Attributes. En ergy J. 21, 1. Goffe, W.L., Ferrier, G.D., Rogers, J., 1994. Global optimization of statistical functions with simulated annealing. J. Econom. 60, 65 99. Goovaerts, P., 1997. Geostatistics for Natural Resources Evaluation, Applied Geostatistics Series. Oxf ord University Press, NY. Govaerts, B., Verhulst, N., Castellanos Navarrete, A., Sayre, K.D., Dixon, J., Dendooven, L., 2009. Conservation Agriculture and Soil Carbon Sequestration: Between Myth and Farmer Reality. Crit. Rev. Plant Sci. 28, 97 122. doi:10. 1080/07352680902776358 Green, J.L., Hastings, A., Arzberger, P., Ayala, F.J., Cottingham, K.L., Cuddington, K., Davis, F., Dunne, J.A., Fortin, M. J., Gerber, L., Neubert, M., 2005. Complexity in Ecology and Conservation: Mathematical, Statistical, and Com putational Challenges. BioScience 55, 501 510. doi:10.1641/0006 3568(2005)055[0501:CIEACM]2.0.CO;2 Greene, W.H., 1997. Econometric Analysis, 3rd ed. Prentice Hall, NJ. Groenendijk, F.M., Condron, L.M., Rijkse, W.C., 2002. Effects of afforestation on organi c carbon, nitrogen and sulfur concentrations in New Zealand hill country soils. Geoderma 108, 91 100. doi:10.1016/S0016 7061(02)00125 8 Grubbs, J.W., 1998. Recharge Rates to the Upper Floridan Aquifer in the Suwannee River Water Management District, Florid a (No. WRI 97 4283). United States Geological Survey.
283 Grumbine, R.E., 1994. QuÃ© es manejo del ecosistema? Conserv. Biol. 8, 27 38. doi:10.1046/j.1523 1739.1994.08010027.x Grunwald, S. (Ed.), 2006. Environmental Soil Landscape Modeling Geographic Inform ation Technologies and Pedometrics. CRC Press. Grunwald, S., 2008. Role of soils to sequester carbon, in: S. Mulkey, J. Alavalapati, A. Hodges, A.C. Wilkie, S. Grunwald (Eds.), Opportunities for Greenhouse Gas Reduction by Agriculture and Forestry in Flori da. University of Florida, School of Natural Resources and Environment Department of Environmental Defense, Washington, DC. Grunwald, S., 2009. Multi criteria characterization of recent digital soil mapping and modeling approaches. Geoderma 152, 195 207. doi:10.1016/j.geoderma.2009.06.003 Assessment and Modeling of Changes in Soil Carbon Storage and Turnover in a NIFA AFRI grant No. 2007 35107 18368). Grunwald, S., Thompson, J.A., Boettinger, J.L., 2011b. Digital Soil Mapping and Modeling at Continental Scales: Finding Solutions for Global Issues. Soil Sci. Soc. Am. J. 75, 1201. doi:10.2136/sssaj2011.0025 Grunwald, S., Vasques, G.M., Comerford, N.B., Bruland, G.L., Bliss, C.M., 2010. Regional modeling of carbon, nitrogen, and phosphorus geospatial patterns, in: G. Hanrahan (Ed.), Modeling of Pollutants in Complex Environmental Systems. ILM Publ, Herfordshire, UK. Gundersen, P., Emmett, B.A., Kjon ass, O.J., Koopmans, C.J., Tietema, A., 1998. Impacts of nitrogen deposition on nitrogen cycling: a synthesis. Nutr. Cycl. Agroecosystems NUTR CYCL AGROECOSYST 37 55. Guo, L.B., Gifford, R.M., 2002. Soil carbon stocks and land use change: a meta analysis . Glob. Change Biol. 8, 345 360. doi:10.1046/j.1354 1013.2002.00486.x Guo, Y., Amundson, R., Gong, P., Yu, Q., 2006. Quantity and Spatial Variability of Soil Carbon in the Conterminous United States. Soil Sci. Soc. Am. J. SSSAJ 70. doi:10.2136/sssaj2005. 0162 Gurney, K.R., Mendoza, D., Zhou, Y., Fischer, M., Miller, C., Geethakumar, S., de la Rue du, S., 2009. The Vulcan Project: High resolution fossil fuel combustion CO2 emissions fluxes for the United States. Environ. Sci. Technol. 43. doi:10.1021/es9008 06c Haim, D., White, E.M., Alig, R.J., 2014. Permanence of agricultural afforestation for carbon sequestration under stylized carbon markets in the U.S. For. Policy Econ. 41, 12 21. doi:10.1016/j.forpol.2013.12.008
284 Haines Young, R., 2011. Exploring ecosyst em service issues across diverse knowledge domains using Bayesian Belief Networks. Prog. Phys. Geogr. 35, 681 699. doi:10.1177/0309133311422977 Hainmueller, J., Hopkins, D.J., Yamamoto, T., 2014. Causal Inference in Conjoint Analysis: Understanding Multidi mensional Choices via Stated Preference Experiments. Polit. Anal. 22, 1 30. doi:10.1093/pan/mpt024 Haipeng, Z., Xuxuan, X., 2012. Combining Stated Preference and Revealed Preference Methods for the Valuation of Non market Goods. Chin. J. Popul. Resour. Env iron. 10, 121 126. doi:10.1080/10042857.2012.10685119 Han, F.X., Plodinec, M.J., Su, Y., Monts, D.L., Li, Z., 2007. Terrestrial carbon pools in southeast and south central United States. Clim. Change 84, 191 202. doi:10.1007/s10584 007 9244 5 Hanemann, W.M ., 1984. Welfare Evaluations in Contingent Valuation Experiments with Discrete Responses. Am. J. Agric. Econ. 66, 332 341. doi:10.2307/1240800 Hanley, N., Mourato, S., Wright, R.E., 2001a. Choice Modelling Approaches: A Superior Alternative for Environment al Valuatioin? J. Econ. Surv. 15, 435 462. doi:10.1111/1467 6419.00145 Hanley, N., Mourato, S., Wright, R.E., 2001b. Choice modelling approaches: A superior alternative for environmental valuatioin? J. Econ. Surv. 15, 435 462. doi:10.1111/1467 6419.00145 H anley, N., Shogren, J.F., 2002. Awkward choices: Economics and nature conservation, in: D.W. Bromley, J. Paavola (Eds.), Economics, Ethics and Environmental Policy: Contested Choices. Blackwell Publishers Ltd, Oxford. Hanley, N., Wright, R.E., Adamowicz, V ., 1998. Using choice experiments to value the environment. Environ. Resour. Econ. 11, 413 428. doi:10.1023/A:1008287310583 Harwell, M.A., Long, J.F., Bartuska, A.M., Gentile, J.H., Harwell, C.C., Myers, V., Ogden, J.C., 1996. Ecosystem management to achie ve ecological sustainability: The case of South Florida. Environ. Manage. 20, 497 521. doi:10.1007/BF01474652 Hassink, J., 1997. The capacity of soils to preserve organic C and N by their association with clay and silt particles. Plant Soil 191, 77 87. doi :10.1023/A:1004213929699 Hawkins, E., Sutton, R., 2011. The potential to narrow uncertainty in projections of regional precipitation change. Clim. Dyn. 37, 407 418. doi:10.1007/s00382 010 0810 6
285 Hecky, R.E., Campbell, P., Hendzel, L.L., 1993. The stoichiom etry of carbon, nitrogen, and phosphorus in particulate matter of lakes and oceans. Limnol. Oceanogr. 38, 709 724. Hensher, D.A., 2010. Hypothetical bias, choice experiments and willingness to pay. Transp. Res. Part B Methodol. 44, 735 752. Hicks, R.L., 20 02. A Comparison Of Stated And Revealed Preference Methods For Fisheries Management (2002 Annual meeting, July 28 31, Long Beach, CA No. 19853). American Agricultural Economics Association (New Name 2008: Agricultural and Applied Economics Association). Hi lgard, E.W., 1906. Soils: Their Formation, Properties, Composition, and Relations to Climate and Plant Growth in the Humid and Arid Regions. Macmillan. Hirsch, R.M., Slack, J.R., 1984. A Nonparametric Trend Test for Seasonal Data With Serial Dependence. Wa ter Resour. Res. 20, 727 732. doi:10.1029/WR020i006p00727 Hirsch, R.M., Slack, J.R., Smith, R.A., 1982. Techniques of trend analysis for monthly water quality data. Water Resour. Res. 18, 107 121. doi:10.1029/WR018i001p00107 Hoehn, J.P., Lupi, F., Kaplowit z, M.D., 2010. Stated Choice Experiments with Complex Ecosystem Changes: The Effect of Information Formats on Estimated Variances and Choice Parameters. J. Agric. Resour. Econ. 35. Hoffman, S.D., Duncan, G.J., 1988. Multinomial and Conditional Logit Discre te Choice Models in Demography. Demography 25, 415 427. doi:10.2307/2061541 Hole, A.R., 2006. Small sample properties of tests for heteroscedasticity in the conditional logit model (Health, Econometrics and Data Group (HEDG) Working Paper No. 06/04). HEDG, c/o Department of Economics, University of York. Hornsby, D., Ceryak, R., 1999. Groundwater Quality Report 1998. (No. WR99 03). Suwannee River Water Management District, Live Oak, FL. Hornsby, H. d., Mattson, R., 1998. Surfacewater Quality and Biological Monitoring Annual Report 1997 (No. WR98 03). Suwannee River Water Management District, Live Oak, FL. Hornsby, H.D., 2007. Influences on the distribution and occurrence of nitrate nitrogen and total phosphorus in the water resources of the Suwannee River Wa ter Management District (Ph.D. Dissertation). University of Florida, Gainesville, FL. Hovi, J., Skodvin, T., Andresen, S., 2003. The persistence of the Kyoto Protocol: Why other Annex I countries move on without the United States. Glob. Environ. Polit. 3, 1 23.
286 Howe, P.D., Markowitz, E.M., Lee, T.M., Ko, C. Y., Leiserowitz, A., 2013. Global perceptions of local temperature change. Nat. Clim. Change 3, 352 356. doi:10.1038/nclimate1768 increasing soil organic carbon stocks in dryland agricultural systems. Soil Res. 51, 657 667. Hrabanski, M., Bidaud, C., Le Coq, J. F., MÃ©ral, P., 2013. Environmental NGOs, policy entrepreneurs of market based instruments for ecosystem services? A compari son of Costa Rica, Madagascar and France. For. Policy Econ. 37, 124 132. doi:10.1016/j.forpol.2013.09.001 Huang, P.M., Li, Y., Sumner, M.E., 2011. Handbook of Soil Sciences: Properties and Processes, Second Edition. CRC Press. Huggins, D.R., Buyanovsky, G. A., Wagner, G.H., Brown, J.R., Darmody, R.G., Peck, T.R., Lesoing, G.W., Vanotti, M.B., Bundy, L.G., 1998. Soil organic C in the tallgrass prairie derived region of the corn belt: Effects of long term crop managmeent. Soil Tillage Res. 47, 219 234. Hulme, M., 2010. Problems with making and governing global kinds of knowledge. Glob. Environ. Change 20, 558 564. doi:10.1016/j.gloenvcha.2010.07.005 Hungate, B.A., Holland, E.A., Jackson, R.B., Chapin, F.S., Mooney, H.A., Field, C.B., 1997. The fate of carbon in grasslands under carbon dioxide enrichment. Nature 388, 576 579. doi:10.1038/41550 Hunter, M.R., 2011. Impact of ecological disturbance on awareness of urban nature and sense of environmental stewardship in residential neighborhoods. Landsc. Urban Plan. 1 01, 131 138. doi:10.1016/j.landurbplan.2011.02.005 Ingram, J.S.I., Fernandes, E.C.M., 2001. Managing carbon sequestration in soils: concepts and terminology. Agric. Ecosyst. Environ. 87, 111 117. doi:10.1016/S0167 8809(01)00145 1 Intergovernmental Panel on Climate Change (IPCC), 2007. Climate Change 2013: The Physical Science Basis, Working Group I Contribution to the 4th Assessment Report Changes to the Underlying Scientific/Technical Assessment. IPCC, Cambridge University, NY and UK. Intergovernmental P anel on Climate Change (IPCC), 2013. Climate Change 2013: The Physical Science Basis, Working Group I Contribution to the 5th Assessment Report Changes to the Underlying Scientific/Technical Assessment. IPCC, Geneva, Switzerland. Islam, K., Singh, B., Mc Bratney, A., 2003. Simultaneous estimation of several soil properties by ultra violet, visible, and near infrared reflectance spectroscopy. Soil Res. 41, 1101 1114.
287 Jaarsveld, A.S. van, Biggs, R., Scholes, R.J., Bohensky, E., Reyers, B., Lynam, T., Musvoto , C., Fabricius, C., 2005. Measuring conditions and trends in ecosystem services at multiple scales: the Southern African Millennium Ecosystem Assessment (SAfMA) experience. Philos. Trans. R. Soc. B Biol. Sci. 360, 425 441. doi:10.1098/rstb.2004.1594 Jacca rd, J., 2001. Interaction Effects in Logistic Regression. SAGE. Jackson, B., Pagella, T., Sinclair, F., Orellana, B., Henshaw, A., Reynolds, B., Mcintyre, N., Wheater, H., Eycott, A., 2013. Polyscape: A GIS mapping framework providing efficient and spatial ly explicit landscape scale valuation of multiple ecosystem services. Landsc. Urban Plan. 112, 74 88. doi:10.1016/j.landurbplan.2012.12.014 Jakeman, A.J., Letcher, R.A., Norton, J.P., 2006. Ten iterative steps in development and evaluation of environmental models. Environ. Model. Softw. 21, 602 614. doi:10.1016/j.envsoft.2006.01.004 Jaynes, E.T., 1957. Information Theory and Statistical Mechanics. Phys. Rev. 106, 620 630. doi:10.1103/PhysRev.106.620 Jenkins, J.D., 2014. Political economy constraints on carb on pricing policies: What are the implications for economic efficiency, environmental efficacy, and climate policy design? Energy Policy 69, 467 477. doi:10.1016/j.enpol.2014.02.003 Jenny, H., 1941. Factors of Soil Formation. McGraw Hill, NY. Jensen, F.V., Nielsen, T.D., 2007. Bayesian Networks and Decision Graphs, 2nd ed. Springer Science + Business Media, LLC, New York, N.Y. JobbÃ¡gy, E.G., Jackson, R.B., 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Eco l. Appl. 10, 423 436. doi:10.1890/1051 0761(2000)010[0423:TVDOSO]2.0.CO;2 Johnson, D.W., 1992. Effects of forest management on soil carbon storage. Water. Air. Soil Pollut. 64, 83 120. doi:10.1007/BF00477097 Johnson, J.M.F., Reicosky, D.C., Allmaras, R.R., Sauer, T.J., Venterea, R.T., Dell, C.J., 2005. Greenhouse gas contributions and mitigation potential of agriculture in the central USA. Soil Tillage Res., Greenhouse Gas Contributions and Mitigation Potential in Agricultural Regions of North America 83, 7 3 94. doi:10.1016/j.still.2005.02.010 Johnston, A.E., Poulton, P.R., Coleman, K., 2009. Chapter 1 Soil Organic Matter: Its Importance in Sustainable Agriculture and Carbon Dioxide Fluxes, in: Donald L. Sparks (Ed.), Advances in Agronomy. Academic Press, pp . 1 57.
2 88 Jones, C., McConnell, C., Coleman, K., Cox, P., Falloon, P., Jenkinson, D., Powlson, D., 2005. Global climate change and soil carbon stocks; predictions from two contrasting models for the turnover of organic carbon in soil. Glob. Change Biol. 11, 154 166. doi:10.1111/j.1365 2486.2004.00885.x Journel, A.G., Huijbregts, C.J., 1978. Mining Geostatistics. Academic Press, NY. Kahan, D., 2012. Why we are poles apart on climate change. Nature 488, 255 255. doi:10.1038/488255a Kareiva, P., Tallis, H., Rick etts, T.H., Daily, G.C., Polasky, S., 2011. Natural Capital: Theory and Practice of Mapping Ecosystem Services (OUP Catalogue). Oxford University Press. Kashima, Y., Paladino, A., Margetts, E.A., 2014. Environmentalist identity and environmental striving. J. Environ. Psychol. 38, 64 75. doi:10.1016/j.jenvp.2013.12.014 Katz, B.G., Catches, J.S., Bullen, T.D., Michel, R.L., 1998. Changes in the isotopic and chemical composition of ground water resulting from a recharge pulse from a sinking stream. J. Hydrol. 211, 178 207. doi:10.1016/S0022 1694(98)00236 4 Kautz, R., Stys, B., Kawula, R., 2007. Florida vegetation and land use change between 1985 89 and 2003. Agric. Nat. Resour. Sci. 70, 12 23. Keim, B.D., Fontenot, R., Tebaldi, C., Shankman, D., 2011. Hydroclim atology of the U.S. Gulf Coast Under Global Climate Change Scenarios. Phys. Geogr. 32, 561 582. doi:10.2747/0272 3622.214.171.1241 Keller, J.K., 2011. Wetlands and the global carbon cycle: what might the simulated past tell us about the future? New Phytol. 192 , 789 792. doi:10.1111/j.1469 8137.2011.03954.x Kellnorfer, J., Walker, W., Kirsch, K., Fiske, G., Bishop, J., Lapoint, L., Hoppus, M., Westfall, J., 2013. NACP Abovegournd Biomass and Carbon Baseline Data, V.2 (NBCD 2000), U.S.A., 2000 dataset. Oak Ridge, TN. KEMA, 2010. 2009 California Residential Appliance Saturation Study (No. CEC 200 2010 004 ES). California Energy Commission. Khadka, M.B., Martin, J.B., Jin, J., 2014. Transport of dissolved carbon and CO2 degassing from a river system in a mixed silic ate and carbonate catchment. J. Hydrol. 513, 391 402. doi:10.1016/j.jhydrol.2014.03.070 Kidd, C.V., 1992. The evolution of sustainability. J. Agric. Environ. Ethics 5, 1 26. doi:10.1007/BF01965413
289 Kim, J., Grunwald, S., Osborne, T.Z., Revero, R.G., Robbins , R., Yamataki, H., 2012. Spatial resolution effects of remote sensing images on digital soil models in aquatic ecosystems, in: Minasny, B., McBratney, A.B., Malone, B.P. (Eds.), Digital Soil Assessments and beyond: Proceedings of the Fifth Global Workshop on Digital Soil Mapping 2012, Sydney, Australia. CRC Press, London. Kimble, J.M., Follett, R.F., Cole, C.V., 1998. The Potential of U.S. Cropland to Sequester Carbon and Mitigate the Greenhouse Effect. CRC Press. Kirkpatrick, S., Gelatt, C.D., Vecchi, M.P ., 1983. Optimization by Simulated Annealing. Science 220, 671 680. doi:10.1126/science.220.4598.671 Kirschbaum, M.U.F., 1995. The temperature dependence of soil organic matter decomposition, and the effect of global warming on soil organic C storage. Soil Biol. Biochem. 27, 753 760. doi:10.1016/0038 0717(94)00242 S Kong, A.Y.Y., Six, J., Bryant, D.C., Denison, R.F., van Kessel, C., 2005. The relationship between carbon input, aggregation, and soil organic carbon stabilization in sustainable cropping system s. Soil Sci. Soc. Am. J. 69, 1078. doi:10.2136/sssaj2004.0215 Konrad, C.E., Fuhrmann, C.M., Billiot, A., Keim, B.D., Kruk, M.C., Kunkel, K.E., Needham, H., Shafer, M., Stevens, L., 2013. Climate of the southeast USA: Past, present, and future, in: T.I. Kei th, K. Dow, L. Carter, J. Anderson (Eds.), Climate of the Southeast United States: Variability, Change, Impacts, and Vulnerability, Sourtheast Climate Consortium. Island Press, Washington, DC. Kragt, M.E., 2009. A Beginners Guide to Bayesian Network Modell ing for Integrated Catchment Management. Landscape Logic. Kremen, C., 2005. Managing ecosystem services: what do we need to know about their ecology? Ecol. Lett. 8, 468 479. doi:10.1111/j.1461 0248.2005.00751.x Kreye, M.M., Adams, D.C., Borisova, T., Escob edo, F.J., 2011. The value of water quality protection programs: A meta analysis. J. For. 109, 580. Krinsky, I., Robb, A.L., 1986. On Approximating the Statistical Properties of Elasticities. Rev. Econ. Stat. 68, 715 719. doi:10.2307/1924536 Kroes, E.P., S heldon, R.J., 1988. Stated Preference Methods: An Introduction. J. Transp. Econ. Policy 22, 11 25. Kruskal, W.H., Wallis, W.A., 1952. Use of Ranks in One Criterion Variance Analysis. J. Am. Stat. Assoc. 47, 583 621. doi:10.2307/2280779
290 Kuikka, S., HildÃ©n, M., Gislason, H., Hansson, S., Sparholt, H., Varis, O., Kuikka, S., HildÃ©n, M., Gislason, H., Hansson, S., Sparholt, H., Varis, O., 1999. Modeling environmentally driven uncertainties in Baltic cod (Gadus morhua) by Bayesian influence diagrams. Can. J. Fi sh. Aquat. Sci. 56, 629 641. Labiosa, W.B., Forney, W.M., Esnard, A. M., Mitsova Boneva, D., Bernknopf, R., Hearn, P., Hogan, D., Pearlstine, L., Strong, D., Gladwin, H., Swain, E., 2013. An integrated multi criteria scenario evaluation web tool for partic ipatory land use planning in urbanized areas: The Ecosystem Portfolio Model. Environ. Model. Softw. 41, 210 222. doi:10.1016/j.envsoft.2012.10.012 Lal, R., 2008. Carbon sequestration. Philos. Trans. R. Soc. B Biol. Sci. 363, 815 830. doi:10.1098/rstb.2007. 2185 A review of Bayesian belief networks in ecosystem service modelling. Environ. Model. Softw. 46, 1 11. doi:10.1016/j.envsoft.2013.03.011 Layke, C., 2009. Measuring Ecosystem Service Indicators, WRI Working Paper. World Resources Institute, Washington, DC. Lecocq, F., Ambrosi, P., 2007. The Clean Development Mechanism: history, status, and prospects, in: M. Cropp er (Ed.), Association of Environmental and Resource Economists. Oxford University Press, UK. Liang, A., Yang, X., Zhang, X., McLaughlin, N., Shen, Y., Li, W., 2009. Soil organic carbon changes in particle size fractions following cultivation of Black soils in China. Soil Tillage Res. 105, 21 26. doi:10.1016/j.still.2009.05.002 Liao, H., Long, J., 2011. [Variation of soil organic carbon under different vegetation types in Karst Mountain areas of Guizhou Province, southwest China]. Ying Yong Sheng Tai Xue Bao J. Appl. Ecol. Zhongguo Sheng Tai Xue Xue Hui Zhongguo Ke Xue Yuan Shenyang Ying Yong Sheng Tai Yan Jiu Suo Zhu Ban 22, 2253 2258. Liaw, A., Wiener, M., 2002. Classification and Regression by randomForest. R News 2, 18 22. Liebig, M.A., Morgan, J.A., Reed er, J.D., Ellert, B.H., Gollany, H.T., Schuman, G.E., 2005. Greenhouse gas contributions and mitigation potential of agricultural practices in northwestern USA and western Canada. Soil Tillage Res., Greenhouse Gas Contributions and Mitigation Potential in Agricultural Regions of North America 83, 25 52. doi:10.1016/j.still.2005.02.008 Liere, K.D.V., Dunlap, R.E., 1980. The Social Bases of Environmental Concern: A Review of Hypotheses, Explanations and Empirical Evidence. Public Opin. Q. 44, 181 197.
291 Light, H.M., Darst, M.R., Lewis, L.J., Howell, D.A., 2002. Hydrology, Vegetation, and Soils of Riverine and Tidal Floodplain Forests of the Lower Suwannee River, Florida, and Potential Impacts of Flow Reductions (No. 1656A), U.S. Geological Survey Professional Pa per. Prepared in cooperation with the Suwannee River Water Management District, Denver, CO. Lin, B.B., Macfadyen, S., Renwick, A.R., Cunningham, S.A., Schellhorn, N.A., 2013. Maximizing the Environmental Benefits of Carbon Farming through Ecosystem Service Delivery. BioScience 63, 793 803. doi:10.1525/bio.2013.63.10.6 Lin, C., Lin, C. H., 2013. Comparison of carbon sequestration potential in agricultural and afforestation farming systems. Sci. Agric. 70, 93 101. doi:10.1590/S0103 90162013000200006 List, J., Sinha, P., Taylor, M., 2006. Using choice experiments to value non market goods and services: Evidence from field experiments (Natural Field Experiments No. 00278). The Field Experiments Website. Loomis, J.B., Asmus, C., Bell, P., 2009. A Comparison of Ac tual and Hypothetical Willingness to Pay of Parents and Non The Case of Nitrates in Drinking Water (2007 Annual Meeting, July 29 August 1, 2007, Portland, Oregon TN No. 9358). American Agricultural Economics Associat ion (New Name 2008: Agricultural and Applied Economics Association). Louviere, J.J., 1988. Analyzing Decision Making: Metric Conjoint Analysis. SAGE. Louviere, J.J., Hensher, D.A., 1982. Design and analysis of simulated choice or allocation experiments in travel choice modeling. Transp. Res. Rec. Louviere, J.J., Woodworth, G., 1983. Design and Analysis of Simulated Consumer Choice or Allocation Experiments: An Approach Based on Aggregate Data. J. Mark. Res. 20, 350 367. doi:10.2307/3151440 Lubell, M., 2004. Collaborative watershed management: a view from the grassroots. Policy Stud. J. 32, 341 361. doi:10.1111/j.1541 0072.2004.00069.x Luo, Z., Wang, E., Sun, O.J., 2010. Can no tillage stimulate carbon sequestration in agricultural soils? A meta analysis of p aired experiments. Agric. Ecosyst. Environ. 139, 224 231. doi:10.1016/j.agee.2010.08.006 Lusk, J.L., Schroeder, T.C., 2004. Are Choice Experiments Incentive Compatible? A Test with Quality Differentiated Beef Steaks. Am. J. Agric. Econ. 86, 467 482. Lynam, T., 1999. Adaptive analysis of locally complex systems in a globally complex world. Ecol. Soc. 3, 13.
292 Lynam, T., de Jong, W., Sheil, D., Kusumanto, T., Evans, K., 2007. A review of tools for incorporating community knowledge, preferences, and values into decision making in natural resources management. monitoring program: background hydrogeochemistry [WWW Document]. URL http://aquaticcommons.org/1307/ (accessed 3.27.14). M aes, J., Paracchini, M.L., Zulian, G., 2011. A European Assessment of the Provision of Ecosystem Services: Towards and Atlas of Ecosystem Services, Joint Research Centre. Institute for Environment and Sustainability, Ispra, Italy. Malcolm, J.R., Liu, C., N eilson, R.P., Hansen, L., Hannah, L., 2006. Calentamiento global y extinciones de especies endÃ©micas en sitios de importancia para la biodiversidad. Conserv. Biol. 20, 538 548. doi:10.1111/j.1523 1739.2006.00364.x Malone, B.P., McBratney, A.B., Minasny, B. , 2009. Mapping continuous soil depth functions in the Edgeroi district, NSW, Australia, using terrain attributes and other environmental factors. Mangel, M., Talbot, L.M., Meffe, G.K., Agardy, M.T., Alverson, D.L., Barlow, J., Botkin, D.B., budowski, G., Clark, T., Cooke, J., Crozier, R.H., Dayton, P.K., Elder, D.L., Fowler, C.W., Funtowicz, S., Giske, J., Hofman, R.J., Holt, S.J., Kellert, S.R., Kimball, L.A., Ludgwig, D., Magnusson, K., III, B.S.M., Mann, C., Norse, E.A., Northridge, S.P., Perrin, W.F., Perrings, C., Norse, E. a., Northridge, S.P., Perrin, W.F., Perrings, C., Peterman, R.M., Rabb, G.B., Regier, H.A., III, J.E.R., Sherman, K., Sissenwine, M.P., Smith, T.D., Starfield, A., Taylor, R.J., Tillman, M.F., Toft, C., Twiss, J.R., Jr., Wilen, J., Young, T.P., 1996. Principles for the conservation of wild living resources. Ecol. Appl. 6, 338 362. doi:10.2307/2269369 Mann, H.B., 1945. Nonparametric Tests Against Trend. Econometrica 13, 245 259. doi:10.2307/1907187 Manski, C.F., 1977. The structure of random utility models. Theory Decis. 8, 229 254. doi:10.1007/BF00133443 Marcot, B.G., 2012. Metrics for evaluating performance and uncertainty of Bayesian network models. Ecol. Model. 230, 50 62. doi:10.1016/j.ecolmodel.2012.01.013 Marcot, B.G., Steventon , J.D., Sutherland, G.D., McCann, R.K., 2006. Guidelines for developing and updating Bayesian belief networks applied to ecological modeling and conservation. Can. J. For. Res. 36, 3063 3074. doi:10.1139/x06 135 Marquart Pyatt, S.T., 2007. Concern for the Environment Among General Publics: A Cross National Study. Soc. Nat. Resour. 20, 883 898. doi:10.1080/08941920701460341
293 Marquart Pyatt, S.T., 2008. Are There Similar Sources of Environmental Concern? Comparing Industrialized Countries. Soc. Sci. Q. 89, 131 2 1335. doi:10.1111/j.1540 6237.2008.00567.x MartÃn de Santa Olalla, F.J., DomÃnguez, A., Artigao, A., Fabeiro, C., Ortega, J.F., 2005. . Water Manag., Special Issue on Land and Water Use: Environmental Management Tools and Practices 77, 21 36. doi:10.1016/j.agwat.2004.09.029 Martinez foundation for ecological econo mics. Ecol. Econ. 26, 277 286. doi:10.1016/S0921 8009(97)00120 1 Martinsson, P., 2002. Using Choice Experiments for Non Market Valuation (EEPSEA Special and Technical Paper No. sp200205t2). Economy and Environment Program for Southeast Asia (EEPSEA). McBra tney, A.B., MendonÃ§a Santos, M.L., Minasny, B., 2003. On digital soil mapping. Geoderma 117, 3 52. doi:10.1016/S0016 7061(03)00223 4 McCann, R.K., Marcot, B.G., Ellis, R., 2006. Bayesian belief networks: applications in ecology and natural resource managem ent. Can. J. For. Res. 36, 3053 3062. McCool, S.F., Martin, S.R., 1994. Community Attachment and Attitudes Toward Tourism Development. J. Travel Res. 32, 29 34. doi:10.1177/004728759403200305 McEvoy, J., 1972. The American concern with the environment, in: Burch, Jr. et al, W.R. (Ed.), Social Behavior, Natural Resources and the Environment. Harper and Row, NY. McFadden, D., 1974. Conditional logit analysis of qualitative choice behavior, in: Azrembka, P. (Ed.), Frontiers in Econometrics. Academic Press, NY. McGroddy, M.E., Daufresne, T., Hedin, L.O., 2004. Scaling of C:N:P Stoichiometry in Forests Worldwide: Implications of Terrestrial Redfield Type Ratios. Ecology 85, 2390 2401. Megonigal, J.P., Hines, M.E., Visscher, P.T., 2003. Anaerobic Metabolism: Linka ges to Trace Gases and Aerobic Processes, in: Treatise on Geochemistry. Elsevier, pp. 317 424. Meyer, J.L., 1990. A Blackwater Perspective on Riverine Ecosystems. BioScience 40, 643 651. doi:10.2307/1311431
294 Milesi, C., Elvidge, C.D., Nemani, R.R., Running , S.W., 2003. Assessing the impact of urban land development on net primary productivity in the southeastern United States. Remote Sens. Environ., Urban Remote Sensing 86, 401 410. doi:10.1016/S0034 4257(03)00081 6 Millennium Ecosystem Assessment (MA), 200 3. Ecosystem and Human Well Being, A Framework for Assessment. Island Press, Washington, DC. Millennium Ecosystem Assessment (MA), 2005a. Ecosystem and Human Well Being, Volume 1: Current State and Trends. Island Press, Washington, DC. Millennium Ecosystem Assessment (MA), 2005b. Ecosystem and Human Well Being, Synthesis. Island Press, Washington, DC. Millennium Ecosystem Assessment (MA), 2005c. Ecosystem and Human Well Being, Volume 4: Multiscale Assessments. Island Press, Washington, DC. Miller, J.A., 198 6. Hydrogeological Framework of the Floridan Aquifer System in Florida and in Parts of Georgia, Alabama, and South Carolina (Professional Paper No. 1403 B). U.S. Geological Survey (USGS). Milon, J.W., Hodges, A.W., Rimal, A., 2000. Multiattribute choice an alysis in ecosystem restoration planning. Environ. Pract. 2, 176 187. doi:10.1017/S1466046600001393 Milon, J.W., Hodges, A.W., Rimal, A., Kiker, C.F., Casey, F., 1999. Public Preferences and Economic Values for Restoration of the Everglades/South Florida E cosystem. Food and Resource Economics Department, University of Florida, Gainesville, FL. Minasny, B., Sulaeman, Y., Mcbratney, A.B., 2011. Is soil carbon disappearing? The dynamics of soil organic carbon in Java. Glob. Change Biol. 17, 1917 1924. doi:10.1 111/j.1365 2486.2010.02324.x The influence of the Atlantic Warm Pool on the Florida panhandle sea breeze. J. Geophys. Res. Atmospheres 116, n/a n/a. doi:10.1029/2 010JD015367 Mitchell, R., Carson, R., 1989. Using Surveys to Value Public Goods: The Contingent Valuation Method. John Hopkins Press, Baltimore. MODIS 17 team (Steve Running, lead), 2009. MODIS Land Subsets for North American Carbon Program (NACP), Collect ion 5. Oak Ridge National Laboratory Distributed Active Archive Center (ORNL DAAC). Montagnini, F., Nair, P.K.R., 2004. Carbon sequestration: An underexploited environmental benefit of. Agrofor. Syst. 61 62, 281 295. doi:10.1023/B:AGFO.0000029005.92691.79
295 Morris, J.B., Tassone, V., de Groot, R., Camilleri, M., Moncada, S., 2011. A framework for participatory impact assessment: involving stakeholders in European policy making, a case study of land use change in Malta. Ecol. Soc. 16, 1 12. Morris, S.J., Bohm, S., Haile Mariam, S., Paul, E.A., 2007. Evaluation of carbon accrual in afforested agricultural soils. Glob. Change Biol. 13, 1145 1156. doi:10.1111/j.1365 2486.2007.01359.x Morris, S.J., Conant, R., Mellor, N., Brewer, E., Paul, E.A., 2010. Controls on S oil Carbon Sequestration and Dynamics: Lessons from Land use Change. J. Nematol. 42, 78 83. Mulkey, S., Alavalapati, J., Hodges, A.W., Wilkie, A.C., Grunwald, S., 2008. Opportunities for Greenhouse Gas Reduction through Forestry and Agriculture in Florida, Executive Summary. School of Natural Resources and Environment, University of Florida, Gainesville, FL. Myers, D.B., Grunwald, S., Vasques, G.M., Harris, W.G., 2011. Harmonizing legacy and Nal le, D.J., Montgomery, C.A., Arthur, J.L., Polasky, S., Schumaker, N.H., 2004. Modeling joint production of wildlife and timber. J. Environ. Econ. Manag. 48, 997 1017. doi:10.1016/j.jeem.2004.01.001 National Research Council (NRC), 2004. Valuing Ecosystem S ervices: Toward Better Environmental Decision Making. The National Academies Press, Washington, DC. Natural Resources Conservation Service (NRCS), 2006a. Soil Survey Geographic (SSURGO) Database, Available at http://soildatamart.nrcs.usda.gov. Natural Reso urces Conservation Service (NRCS), 2006b. Land Resource Regions and Major Land Resource Areas of the United States, the Caribbean, and the Pacific Basin, U.S. Department of Agriculture Handbook 296. Natural Resources Conservation Service (NRCS), 2013. Nati onal Soil Survey Database. Neef, A., Thomas, D., 2009. Rewarding the upland poor for saving the commons? Evidence from Southeast Asia. Int. J. Commons 3, 1 15. Nelson, E., Mendoza, G., Regetz, J., Polasky, S., Tallis, H., Cameron, Dr., Chan, K.M., Daily, G .C., Goldstein, J., Kareiva, P.M., Lonsdorf, E., Naidoo, R., Ricketts, T.H., Shaw, Mr., 2009. Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape scales. Front. Ecol. Environ. 7, 4 11. doi:10.18 90/080023
296 Noormets, A., McNulty, S.G., Domec, J. C., Gavazzi, M., Sun, G., King, J.S., 2012. The role of harvest residue in rotation cycle carbon balance in loblolly pine plantations. Respiration partitioning approach. Glob. Change Biol. 18, 3186 3201. doi :10.1111/j.1365 2486.2012.02776.x Norsys Software Corporation (NORSYS), 2014. Netica. Available at www.norsys.com. North Florida Aquifer Replenishment Initiative (NFARI), 2013. Upper Floridan Aquifer Regional Recharge Concepts and Feasibility Study, Availa ble at http://floridaswater.com/keystoneheights/pdfs/UFA_Recharge_Study.pdf. Nyberg, J.B., Marcot, B.G., Sulyma, R., 2006. Using Bayesian belief networks in adaptive management. Can. J. For. Res. 36, 3104 3116. doi:10.1139/x06 108 Ofei Manu, P., 2009. Gend er and environment in the Japanese workplace. Int. J. Innov. Sustain. Dev. 4, 150 164. doi:10.1504/IJISD.2009.028069 Office of Agricultural Water Policy (OAWP), 2012. Enrollment in Best Managment Practices, http://www.freshfromflorida.com/Divisions Offices /Agricultural Water Policy/Enroll in BMPs. Florida Department of Agriculture and Consumer Services (FDACS). Olof, J. S., Henrik, S., 2008. Measuring Hypothetical Bias in Choice Experiments: The Importance of Cognitive Consistency. BE J. Econ. Anal. Policy 8, 1 10. Olofsson, A., Ã–hman, S., 2006. General Beliefs and Environmental Concern Transatlantic Comparisons. Environ. Behav. 38, 768 790. doi:10.1177/0013916506287388 Ontl, T.A., Schulte, L.A., 2012. Soil Carbon Storage. Nat. Educ. Knowl. 3, 35 35. Oueslat i, I., Allamano, P., Bonifacio, E., Claps, P., 2013. Vegetation and Topographic Control on Spatial Variability of Soil Organic Carbon. Pedosphere 23, 48 58. doi:10.1016/S1002 0160(12)60079 4 Oughton, D.H., 2013. Social and ethical issues in environmental r emediation projects. J. Environ. Radioact. 119, 21 25. doi:10.1016/j.jenvrad.2011.08.019 Parsons, G.R., Kealy, M.J., 1992. Randomly drawn opportunity sets in a Random Utility model of lake recreation. Land Econ. 68, 93 106. doi:10.2307/3146746 Pascual, U., Muradian, R., Brander, L.M., Gomez Baggethun, E., Martin Lopez, B., Verma, M., Armsworth, P., Christie, M., Cornelissen, H., Eppink, F., Farley, J., Loomis, J.B., Person, L., Perrings, C., Polasky, S., 2010. The economics of valuing ecosystem services and biodiversity, in: P. Kumar (Ed.), The Economics of Valuting Ecosystem Services and Biodiversity. Earthscan, London.
297 Paul, K.I., Polglase, P.J., Nyakuengama, J.G., Khanna, P.K., 2002. Change in soil carbon following afforestation. For. Ecol. Manag. 168, 24 1 257. doi:10.1016/S0378 1127(01)00740 X Pearl, J., 1988. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann. Pearl, J., 2009. Causality: Models, Reasoning, and Inference, 2nd ed. Cambridge University Press. Pe ixer, J., Giacomini, H.C., Petrere Jr, M., 2011. Economic valuation of the Emas waterfall, Mogi GuaÃ§u River, SP, Brazil. An. Acad. Bras. CiÃªnc. 83, 1287 1302. Peter A. Beedlow, D.T.T., 2004. Rising atmospheric CO 2 and carbon sequestration in forests. Fron t. Ecol. Environ. FRONT ECOL Env. 2, 315 322. doi:10.1890/1540 9295(2004)002[0315:RACACS]2.0.CO;2 Pichancourt, J. B., Firn, J., ChadÃ¨s, I., Martin, T.G., 2014. Growing biodiverse carbon rich forests. Glob. Change Biol. 20, 382 393. doi:10.1111/gcb.12345 Pittman, J.R., Hatzell, H.H., Oaksford, E.T., 1997. Spring contributions to water quantity and nitrate loads in the Suwannee River during base flow in July 1995 (No. 97 4252), Water Resource Investigations Report. U.S. Geological Survey, Reston,VA. Planert , M., n.d. Simulation of Regional Ground Water Flow in the Suwannee River Basin, Northern Florida and Southern Georgia. Plummer, M.L., 2009. Assessing benefit transfer for the valuation of ecosystem services. Front. Ecol. Environ. 7, 38 45. doi:10.1890/080 091 Poeplau, C., Don, A., Vesterdal, L., Leifeld, J., Van Wesemael, B., Schumacher, J., Gensior, A., 2011. Temporal dynamics of soil organic carbon after land use change in the temperate zone carbon response functions as a model approach. Glob. Change Bi ol. 17, 2415 2427. doi:10.1111/j.1365 2486.2011.02408.x Poirier, V., Angers, D.A., Rochette, P., Chantigny, M.H., Ziadi, N., Tremblay, G., Fortin, J., 2009. Interactive Effects of Tillage and Mineral Fertilization on Soil Carbon Profiles. Soil Sci. Soc. Am . J. 73, 255. doi:10.2136/sssaj2008.0006 Pollino, C.A., Woodberry, O., Nicholson, A., Korb, K., Hart, B.T., 2007. Parameterisation and evaluation of a Bayesian network for use in an ecological risk assessment. Environ. Model. Softw., Bayesian networks in w ater resource modelling and management 22, 1140 1152. doi:10.1016/j.envsoft.2006.03.006 Beliefs are What is Needed. Environ. Behav. 32, 711 723. doi:10.1177/00139165003250 07
298 crop choice using socio psychological measurements of expected benefits of ecosystem services. Environ. Model. Softw. 1 8. doi:10.1016/j.envsoft.2014.03.006 Post, W.M., Am onette, J.E., Birdsey, R., Garten, C.T., Izaurralde, R.C., Jardine, P.M., Jastrow, J., Lal, R., Marland, G., Mccarl, B.A., Thomson, A.M., West, T.O., Wullschleger, S.D., Metting, F.B., 2009. Terrestrial Biological Carbon Sequestration: Science for Enhancem ent and Implementation, in: Mcpherson, B.J., Sundquist, E.T. (Eds.), Carbon Sequestration and Its Role in the Global Carbon Cycle. American Geophysical Union, pp. 73 88. Post, W.M., Izaurralde, R.C., Jastrow, J.D., McCarl, B.A., Amonette, J.E., Bailey, V.L ., Jardine, P.M., West, T.O., Zhou, J., 2004. Enhancement of Carbon Sequestration in US Soils. BioScience 54, 895 908. doi:10.1641/0006 3568(2004)054[0895:EOCSIU]2.0.CO;2 Powlson, D.S., Whitmore, A.P., Goulding, K.W.T., 2011. Soil carbon sequestration to m itigate climate change: a critical re examination to identify the true and the false. Eur. J. Soil Sci. 62, 42 55. doi:10.1111/j.1365 2389.2010.01342.x Prasad, A.M., Iverson, L.R., Liaw, A., 2006. Newer Classification and Regression Tree Techniques: Baggin g and Random Forests for Ecological Prediction. Ecosystems 9, 181 199. doi:10.1007/s10021 005 0054 1 Pregitzer, K.S., Palik, B.J., 1997. Changes in ecosystem carbon 46 years after establishing red pine (Pinus resinosa Ait.) on abandoned agricultural land i n the Great Lakes region, in: E.A. Paul, E.T. Elliott, K. Paustian, C.V. Cole (Eds.), Soil Organic Matter in Temperate Agroecosystems: Long Term Experiments in North America. CRC Press, Boca Raton. (PCAST), 2011. Report to the President: Sustaining Environmental Capital: Protecting Society and the Economy. Executive Office of the President of the United States. Press, W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P., 1992. Numerical recipes in C: the art of scientific computing. Chapter 10: minimization or maximization of functions, 2nd ed. Cambridge University Press, New York. PRISM Climate Group, 2012. PRISM spatial climate layers, http://www.prism.oregonstate.edu/. Puri, H.S., Vernon, R.O., 1964. Summary of the Geology of Florida and a Guidebook to the Classic Exposures (Florida Geological Survey: Special publication, #5), Revised edition. ed. Florida Geological Survey. Putz, F.E., Redford, K.H., 2009. Dangers of carbon based conservation. Gl ob. Environ. Change 19, 400 401. doi:10.1016/j.gloenvcha.2009.07.005
299 Randall, A., 1991. Total and nonuse values, in: J.B. Braden, C.D. Kolstad (Eds.), Measuring the Demand for Environmental Quality. Amsterdam, North Holland. Randall, A., 2008. Benefit cost considerations should be decisive when there is nothing more important at stake, in: Bromley, D.W., Paavola, J. (Eds.), Economics, Ethics, and Environmental Policy. Blackwell Publishers Ltd, pp. 53 68. Raudsepp Hearne, C., Peterson, G.D., Bennett, E.M., 2 010. Ecosystem service bundles for analyzing tradeoffs in diverse landscapes. Proc. Natl. Acad. Sci. U. S. A. 107, 5242 5247. doi:10.1073/pnas.0907284107 Ravindran, A., Yang, S. S., n.d. Effects of vegetation type on microbial biomass carbon and nitrogen i n subalpine mountain forest soils. J. Microbiol. Immunol. Infect. doi:10.1016/j.jmii.2014.02.003 Ready, R.C., Champ, P.A., Lawton, J.L., 2010. Using Respondent Uncertainty to Mitigate Hypothetical Bias in a Stated Choice Experiment. Land Econ. 86, 363 381. Redfield, A.C., 1958. The biological control of chemical factors in the environment. Am. Sci. 46, 205 221. Reicosky, D.C., 2003. Tillage Induced CO2 Emissions and Carbon Sequestration: Effect of Secondary Tillage and Compaction, in: GarcÃa Torres, L., Ben ites, J., MartÃnez Vilela, A., Holgado Cabrera, A. (Eds.), Conservation Agriculture. Springer Netherlands, pp. 291 300. Roa Fuentes, L.L., Hidalgo, C., Etchevers, J.D., Campo, J., 2013. The effects of precipitation regime on soil carbon pools on the Yucata n Peninsula. J. Trop. Ecol. 29, 463 466. doi:10.1017/S0266467413000552 RodrÃguez, J.P., Beard, T.D., Bennett, E.M., Cumming, G.S., Cork, S.J., Agard, J., Dobson, A.P., Peterson, G.D., 2006. Trade offs across space, time, and ecosystem services. Ecol. Soc. 11, 28. Ross, C.W., Grunwald, S., Myers, D.B., 2013. Spatiotemporal modeling of soil organic carbon stocks across a subtropical region. Sci. Total Environ. 461 462, 149 157. doi:10.1016/j.scitotenv.2013.04.070 Saarman, E.T., Carr, M.H., 2013. The Californi a Marine Life Protection Act: A balance of top down and bottom up governance in MPA planning. Mar. Policy, Governing marine protected areas: towards social ecological resilience through institutional diversity 41, 41 49. doi:10.1016/j.marpol.2013.01.004 Sa unders, H.D., 2011. An Income Based Analysis of Historical U.S. Energy Consumption.
300 Schapire, R.E., Freund, Y., Bartlett, P., Lee, W.S., 1998. Boosting the margin: a new explanation for the effectiveness of voting methods. Ann. Stat. 26, 1651 1686. doi:10. 1214/aos/1024691352 Schmidtz, D., Willott, E., 2012. Environmental Ethics what Really matter, what really works. Oxford University Press, UK. Schneider, P.D., Upchurch, P.P., Chen, P.J., Cain, C., 2008. Simulation of Groundwater Flow in North Florida and South Central Georgia. Suwannee River Water Management District, Live Oak, FL. Scott, D., Willits, F.K., 1994. Environmental Attitudes and Behavior A Pennsylvania Survey. Environ. Behav. 26, 239 260. doi:10.1177/001391659402600206 Scott, T.M., 1988. The L ithostratigraphy of the Hawthorn Group (Miocene) of Florida, Florida Geological Survey Bullentin 59. Tallahassee, FL. Scott, T.M., Campbell, K.M., Rupert, F.R., Arthur, J.A., Missimer, T.M., Lloyd, J.M., Yon, J.W., Duncan, J.G., 2001. Geologic Map of the S tate of Florida, Florida Geological Survey Map Series 146. Seidl, A.F., Moraes, A.S., 2000. Global valuation of ecosystem services: application to the Pantanal da Nhecolandia, Brazil. Ecol. Econ. 33, 1 6. doi:10.1016/S0921 8009(99)00146 9 Shah, A., 2005. N on Governmental Organizations of Development Issues. Global Issues, Available at http://www.globalissues.org/article/25/non governmental organizations on development issues. Sheldon, P.J., Var, T., Var, T., 1984. Resident attitudes to tourism in North Wale s. Tour. Manag. 5, 40 47. doi:10.1016/0261 5177(84)90006 2 Sherrouse, B.C., Clement, J.M., Semmens, D.J., 2011. A GIS application for assessing, mapping, and quantifying the social values of ecosystem services. Appl. Geogr. 31, 748 760. doi:10.1016/j.apgeo g.2010.08.002 Shrestha, R.K., Alavalapati, J.R.R., 2004. Valuing environmental benefits of silvopasture practice: a case study of the Lake Okeechobee watershed in Florida. Ecol. Econ. 49, 349 359. doi:10.1016/j.ecolecon.2004.01.015 Simonne, E., Hutchinson, C., DeValerio, J., Hochmuth, R., Treadwell, D., Wright, A., Santos, B., Whidden, A., McAvoy, G., Zhao, X., Olczyk, T., Gazula, A., Ozores Hampton, M., 2010. Current Knowledge, Gaps, and Future Needs for Keeping Water and Nutrients in the Root Zone of Vege tables Grown in Florida. HortTechnology 20, 143 152.
301 Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil organic matter: Implications for C saturation of soils. Plant Soil 241, 155 176. doi:10.1023/A:1016125726789 Smith, C.S., Howes, A.L., Price, B., McAlpine, C.A., 2007. Using a Bayesian belief network to predict suitable habitat of an endangered mammal The Julia Creek dunnart (Sminthopsis douglasi). Biol. Conserv. 139, 333 347. doi:10.1016/j.biocon.2007.06.025 Smith, P., Ashmore, M.R., Black, H.I.J., Burgess, P.J., Evans, C.D., Quine, T.A., Thomson, A.M., Hicks, K., Orr, H.G., 2013. REVIEW: The role of ecosystems and their management in regulating climate, and soil, water and air quality. J. Appl. Ecol. 50, 812 829. do i:10.1111/1365 2664.12016 Smith, R.I., Dick, J.M., Scott, E.M., 2011. The role of statistics in the analysis of ecosystem services. Environmetrics 22, 608 617. doi:10.1002/env.1107 Smith, V.K., 1987. Nonuse Values in Benefit Cost Analysis. South. Econ. J. 54, 19 26. doi:10.2307/1058800 Snyder, C.S., Bruulsema, T.W., Jensen, T.L., Fixen, P.E., 2009. Review of greenhouse gas emissions from crop production systems and fertilizer management effects. Agric. Ecosyst. Environ., Reactive nitrogen in agroecosystems: Integration with greenhouse gas interactions 133, 247 266. doi:10.1016/j.agee.2009.04.021 Soil Survey Staff, 2010. Keys to Soil Taxonomy, 11th ed. United States Department of Agriculture, Natural Resources Conservation Service, Washington, DC. Somma, M., Tolleson Rinehart, S., 1997. Tracking the Elusive Green Women: Sex, Environmentalism, and Feminism in the United States and Europe. Polit. Res. Q. 50, 153 169. doi:10.2307/449033 Song, X., Tian, H., Xu, X., Hui, D., Chen, G., Sommers, G., Marzen, L., Liu, M., 2013. Projecting terrestrial carbon sequestration of the southeastern United States in the 21st century. Ecosphere 4, art88. doi:10.1890/ES12 00398.1 Spiegelhalter, D.J., Dawid, A.P., Lauritzen, S.L., Cowell, R.G., 1993. Bayesian Analysis in Expert Sys tems. Stat. Sci. 8, 219 247. doi:10.1214/ss/1177010888 Spruill, T.B., Showers, W.J., Howe, S.S., 2002. Application of classification tree methods to identify nitrate sources in ground water. J. Environ. Qual. 31, 1538 1549. Steel, B.S., 1996. Thinking Glob ally and Acting Locally?: Environmental Attitudes, Behaviour and Activism. J. Environ. Manage. 47, 27 36. doi:10.1006/jema.1996.0033
302 StelzenmÃ¼ller, V., Lee, J., Garnacho, E., Rogers, S.I., 2010. Assessment of a Bayesian Belief Network GIS framework as a pr actical tool to support marine planning. Mar. Pollut. Bull. 60, 1743 1754. doi:10.1016/j.marpolbul.2010.06.024 Stern, P.C., Dietz, T., 1994. The Value Basis of Environmental Concern. J. Soc. Issues 50, 65 84. doi:10.1111/j.1540 4560.1994.tb02420.x Steven, F.J., Cole, M.A., 1999. Cycles of Soil Carbon, Nitrogen, Phosphorus, Sulfur, Micronutrients. John Wiley & Sons, NY. Stevens, T.H., Belkner, R., Dennis, D., Kittredge, D., Willis, C., 2000. Comparison of contingent valuation and conjoint analysis in ecosy stem management. Ecol. Econ. 32, 63 74. doi:10.1016/S0921 8009(99)00071 3 Stevenson, R.B., 2007. Schooling and environmental education: contradictions in purpose and practice. Environ. Educ. Res. 13, 139 153. doi:10.1080/13504620701295726 Stewart, C.E., Pa ustian, K., Conant, R.T., Plante, A.F., Six, J., 2007. Soil carbon saturation: concept, evidence and evaluation. Biogeochemistry 86, 19 31. doi:10.1007/s10533 007 9140 0 Stewart, C.E., Paustian, K., Conant, R.T., Plante, A.F., Six, J., 2009. Soil carbon sa turation: Implications for measurable carbon pool dynamics in long term incubations. Soil Biol. Biochem. 41, 357 366. doi:10.1016/j.soilbio.2008.11.011 Stewart, C.E., Plante, A.F., Paustian, K., Conant, R.T., Six, J., 2008. Soil carbon saturation: Linking concept and measurable carbon pools. Soil Sci. Soc. Am. J. 72, 379 392. Stockmann, U., Adams, M.A., Crawford, J.W., Field, D.J., Henakaarchchi, N., Jenkins, M., Minasny, B., McBratney, A.B., Courcelles, V. de R. de, Singh, K., Wheeler, I., Abbott, L., Ange rs, D.A., Baldock, J., Bird, M., Brookes, P.C., Chenu, C., D., Zimmermann, M., 2013. The knowns, known unknowns and unknowns of sequestration of soil organic carbon. Agric. Ecos yst. Environ. 164, 80 99. doi:10.1016/j.agee.2012.10.001 Stone, E.L., Harris, W.G., Brown, R.B., Kuehl, R.J., 1993. Carbon storage in Florida spodosols. Soil Sci. Soc. Am. J. USA. Street, D.J., Burgess, L., Louviere, J.J., 2005. Quick and easy choice sets: Constructing optimal and nearly optimal stated choice experiments. Int. J. Res. Mark. 22, 459 470. doi:10.1016/j.ijresmar.2005.09.003 Strobl, C., Boulesteix, A. L., Zeileis, A., Hothorn, T., 2007. Bias in random forest variable importance measures: Illust rations, sources and a solution. BMC Bioinformatics 8, 1 21. doi:10.1186/1471 2105 8 25
303 Sun, Z., MÃ¼ller, D., 2013. A framework for modeling payments for ecosystem services with agent based models, Bayesian belief networks and opinion dynamics models. Envir on. Model. Softw., Thematic Issue on Spatial Agent Based Models for Socio Ecological Systems 45, 15 28. doi:10.1016/j.envsoft.2012.06.007 Suwannee River Partnership (SRP), 2002. Reasonable Assurance Documentation for the Suwannee River, Santa Fe River and Suwannee River Estuary, Available at http://www.dep.state.fl.us/water/watersheds/assessment/docs/2002Update/suwra final.pdf. Suwannee River Water Management District (SRWMD), 1995. Land Cover Land Use Classification for the Suwannee River Water Management D istrict, Available at http://www.dep.state.fl.us/gis/datadir.asp. FL. Swallow, B.M., Sang, J.K., Nyabenge, M., Bundotich, D.K., Duraiappah, A.K., Yatich, T.B., 2009. Tradeoffs, synergies and traps among ecosystem services in the Lake Victoria basin of East Africa. Environ. Sci. Policy 12, 504 519. doi:10.1016/j.envsci.2008.11.003 Tabachnick, B.G., Fidell, L.S., 2001. Using multivariate statistics. Pearson Education, Boston. Takatsuka, Y., Cullen, R., Wilson, M., Wratten, S., 2009. Using stated preference te chniques to value four key ecosystem services on New Zealand arable land. Int. J. Agric. Sustain. 7, 279 291. doi:10.3763/ijas.2009.0334 Tallis, H., Polasky, S., 2009. Mapping and valuing ecosystem services as an approach for conservation and natural resou rce management. Ann. N. Y. Acad. Sci. 1162, 265 283. doi:10.1111/j.1749 6632.2009.04152.x Tallis, H.T., Ricketts, T., Guerry, A., Wood, S.A., Sharp, R., Nelson, E., Ennaanay, D., Wolny, S., Olwero, N., Vigerstol, K., Pennington, D., Mendoza, G., Aukema, J. , Foster, J., Forrest, J., Cameron, D., Arkema, K., Lonsdorf, E., Kennedy, C., Verutes, G., Kim, C., Guannel, G., Papenfus, M., Toft, J., Marsik, M., Bernhartdt, Standford, CA. Tan, Z., Harris, W.G., Mansell, R.S., 1999. Water table dynamics across an aquod udult transition in Florida flatwoods. Soil Sci. 164, 10 17. Tattari, S., Schultz, T., Kuussaari, M., 2003. Use of belief network modelling to assess the impact of buffer zone s on water protection and biodiversity. Agric. Ecosyst. Environ. 96, 119 132. doi:10.1016/S0167 8809(02)00233 5 Taylor, T., Longo, A., 2009. Valuation of Marine Ecosyst Application of Choice Experiments to Value Algal Bloom in the Black Sea Coast of Bulgaria (Working Paper). Department of Economics, University of Bath, Bath, UK.
304 Thompson, J.A., Roecker, S.M., Grunwald, S., Owens, P.R., 2012. Digital soil mapping: Interactions with and applications for hydropedology, in: Lin, H.S. (Ed.), Hydropedology Synergistic Intergration of Pedology and Hydrology. Academic Press, Elsevier B.V., pp. 665 709. Thomson, A.M., CÃ©sar Izaurralde, R., Smith, S.J., Clar ke, L.E., 2008. Integrated estimates of global terrestrial carbon sequestration. Glob. Environ. Change 18, 192 203. doi:10.1016/j.gloenvcha.2007.10.002 Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agricultural sustainability and intensive production practices. Nature 418, 671 677. doi:10.1038/nature01014 Tindall, D.B., SCOTT Davies, Celine, M., 2003. Activism and Conservation Behavior in an Environmental Movement: The Contradictory Effects of Gender. Soc. Nat. Resour. 16, 909 932. doi:10.1080/716100620 Todd, M.J., Vellidis, G., Lowrance, R.R., Pringle, C.M., 2009. High Sediment Oxygen Demand Within an Instream Swamp in Southern Georgia: Implications for Low Dissolved Oxygen Levels in Coastal Blackwater Streams1. JAWRA J. Am. Water Resour. Assoc. 45, 1493 1507. doi:10.1111/j.1752 1688.2009.00380.x Tourassi, G.D., Frederick, E.D., Markey, M.K., Jr, C.E.F., 2001. Application of the mutual information criterion for feature selection in computer aided diagnosis. Med. Phys. 28, 2394 2402. doi:10.1118/1.1418724 Train, K., 2009. Discrete Choice Models with Simulation, 2nd ed. Cambridge University Press, NY. Tremblay, K.R., Dunlap, R.E., 1978. Rural Urban Residence and Concern with Environmental Quality: A Replication and Extension. Rural Soc iol. Turner, R.K., Paavola, J., Cooper, P., Farber, S., Jessamy, V., Georgiou, S., 2003. Valuing nature: lessons learned and future research directions. Ecol. Econ. 46, 493 510. doi:10.1016/S0921 8009(03)00189 7 U.S. Census Bureau, 2010a. Censuses of Popul ation and 2010 County Redistricting Files, Available at http://quickfacts.census.gov/qfd/index.html. U.S. Census Bureau, 2010b. Median Household Income by County: 2006 2010 5 year Estimates. U.S. Census Bureau, 2010c. Population by County (2000 2010), Avai lable at http://quickfacts.census.gov/qfd/states/12000.html. U.S. Census Bureau, 2014. State and County QuickFacts, Available at http://quickfacts.census.gov/qfd/states/12000.html.
305 U.S. Department of Agriculture (USDA), 1993. South East Idle Suwannee River Area Watershed Protection Plan and Environmental Assessment. Soil Conservation Services, Gainesville, FL. U.S. Department of Agriculture (USDA), 2007. Florida: County Summary Highlights 2007, Available at http://www.agcensus.usda.gov/Publications/2007/F ull_Report/Volume_1,_Chapt er_2_County_Level/Florida/. U.S. Environmental Protection Agency (USEPA), 1983. Methods for Chemical Analysis of Water and Wastes. Environmental Monitoring and Support Laboratory, Cinncinnati, OH. U.S. Environmental Protection Age ncy (USEPA), 2002. Turkey Branch dissolved oxygen TMDL: Alapaha River Basin (HUC 03110202) in the Suwannee River Basin, Georgia. Department of Natural Resources, Atlanta, GA. U.S. Environmental Protection Agency (USEPA), 2014. Nutrient Pollution Policy and Data: Nitrogen and Phosphorus Pollution Data Access Tool, Available at http://water.usgs.gov/nwc/explain_data.html. U.S. Geological Survey (USGS), 2009. National Elevation Dataset (NED). Available at http://nationalmap.gov/. U.S. Geological Survey (USGS), 2010. Daily Mean Flow Data at the Suwannee River near Wilcox (Station reference 02323500) between 2000 and 2010 water year. Original unit cubic feet per second, http://www.mysuwanneeriver.org/rivers.htm. U.S. Geological Survey (USGS), 2014. Explanations f or the National Water Conditions, Available at http://water.usgs.gov/nwc/explain_data.html. Uchida, E., Anderson, C.M., Swallow, S.K., 2007. Marketing Ecosystem Services from Agricultural Land: Stated Preferences over Payment Mechanisms and Actual Sales of Farm Wildlife Contracts (2007 Annual Meeting, July 29 August 1, 2007, Portland, Oregon TN No. 9955). American Agricultural Economics Association (New Name 2008: Agricultural and Applied Economics Association). UK National Ecosystem Assessment (NEA), 2011. The UK National Ecosystem Assessment: Synthesis of the Key Findings, UNEP WCMC. Cambridge, UK. Ungaro, F., Staffilani, F., Tarocco, P., 2010. Assessing and mapping topsoil organic carbon stock at regional scale: A scorpan kriging approach conditional on s oil map delineations and land use. Land Degrad. Dev. 21, 565 581. doi:10.1002/ldr.998 Unger, P.W., Sharpley, A.N., Steiner, J.L., Papendick, R.I., Edwards, W.M., 1998. Soil management research for water conservation and quality, in: F.J. Pierce, W.W. Frye (Eds.), Advanced in Soil and Water Conservation. Ann Arbor Press, MI.
306 United Nation Environment Programme (UNEP) and United Nation Educational, Scientific, and Cultural Organization (UNESCO), 2001. Is the Future Yours?, Research Projecton Youth and Sustain able Consumption. UNEP/UNESCO, Paris. United Nations (UN), 2013. World Population Prospects: The 2012 Revision (ST/ESA/SER.A/336), Volume I: Comprehensive Table. Department of Economic and Social Affairs. UN REDD, 2014. Beyond Carbon: Ecosystem based benef its of REDD+, Available at http://www.un redd.org/. The United Nations Collaborative Programme on Reducing Emissions from Deforestation and Forest Degradation in Developing Countries. Uusitalo, L., 2007. Advantages and challenges of Bayesian networks in en vironmental modelling. Ecol. Model. 203, 312 318. doi:10.1016/j.ecolmodel.2006.11.033 Van Jaarsveld, A.S., Biggs, R., Scholes, R.J., Bohensky, E., Reyers, B., Lynam, T., Musvoto, C., Fabricius, C., 2005. Measuring conditions and trends in ecosystem service s at multiple scales: the Southern African Millennium Ecosystem Assessment (SAfMA) experience. Philos. Trans. R. Soc. B Biol. Sci. 360, 425 441. Vaske, J.J., Donnelly, M.P., Williams, D.R., Jonker, S., 2001. Demographic influences on environmental value or ientations and normative beliefs about national forest management. Soc. Nat. Resour. 14, 761 776. Vasques, G.M., Grunwald, S., Comerford, N.B., Sickman, J.O., 2010. Regional modelling of soil carbon at multiple depths within a subtropical watershed. Geoder ma 156, 326 336. doi:10.1016/j.geoderma.2010.03.002 Vasques, G.M., Grunwald, S., Myers, D.B., 2012. Associations between soil carbon and ecological landscape variables at escalating spatial scales in Florida, USA. Landsc. Ecol. 27, 355 367. doi:10.1007/s10 980 011 9702 3 Vasques, G.M., Grunwald, S., Myers, D.B., 2012. Multi scale behavior of soil carbon at nested locations in Florida, USA. Landsc. Ecol. J. 27, 355 367. Vasques, G.M., Grunwald, S., Sickman, J.O., Comerford, N.B., 2010. Upscaling of Dynamic So il Organic Carbon Pools in a North Central Florida Watershed. Soil Sci. Soc. Am. J. 74, 870. doi:10.2136/sssaj2009.0242 Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S., 4th ed. Spring Street, NY. Verstraete, M.M., Pinty, B., Myneni, R .B., 1996. Potential and limitations of information extraction on the terrestrial biosphere from satellite remote sensing. Remote Sens. Environ. 58, 201 214. doi:10.1016/S0034 4257(96)00069 7
307 Vidal Legaz, B., MartÃnez FernÃ¡ndez, J., PicÃ³n, A.S., Pugnaire, F.I., 2013. Trade offs between maintenance of ecosystem services and socio economic development in rural mountainous communities in southern Spain: A dynamic simulation approach. J. Environ. Manage. 131, 280 297. doi:10.1016/j.jenvman.2013.09.036 Villa, F. , Bagstad, K.J., Johnson, G.W., Voigt, B., 2011. Scientific instruments for climate change adaptation: estimating and optimizing the efficiency of ecosystem service provision. Econ. Agrar. Recur. Nat. 11. Vira, B., Adams, W.M., 2009. Ecosystem services and conservation strategy: beware the silver bullet. Conserv. Lett. 2, 158 162. doi:10.1111/j.1755 263X.2009.00063.x Vollmer, D., Ryffel, A.N., Djaja, K., GrÃªt Regamey, A., 2013. Determining Demand for Riparian Ecosystem Services: A Spatially Explicit Discret e Choice Experiment in Jakarta, Indonesia (SSRN Scholarly Paper No. ID 2373389). Social Science Research Network, Rochester, NY. Wah, B.W., Chen, Y., Wang, T., 2007. Simulated annealing with asymptotic convergence for nonlinear constrained optimization. J. Glob. Optim. 39, 1 37. doi:10.1007/s10898 006 9107 z Walker, S.M., Desanker, P.V., 2004. The impact of land use on soil carbon in Miombo Woodlands of Malawi. For. Ecol. Manag. 203, 345 360. doi:10.1016/j.foreco.2004.08.004 Walkley, A., Black, I., 1934. An examination of Degtjareff method for determining soil organic matter and a proposed modification of the chromic acid titration method. Soil Sci 37, 29 38. Wallace, K.J., 2007. Classification of ecosystem services: Problems and solutions. Biol. Conserv. 13 9, 235 246. doi:10.1016/j.biocon.2007.07.015 Wallander, L., 2009. 25 years of factorial surveys in sociology: A review. Soc. Sci. Res. 38, 505 520. doi:10.1016/j.ssresearch.2009.03.004 Wang, Y., Amundson, R., Niu, X. F., 2000. Seasonal and altitudinal vari ation in decomposition of soil organic matter inferred from radiocarbon measurements of soil CO2 flux. Glob. Biogeochem. Cycles 14, 199 211. doi:10.1029/1999GB900074 Wang, Y., Fu, B., LÃ¼, Y., Chen, L., 2011. Effects of vegetation restoration on soil organi c carbon sequestration at multiple scales in semi arid Loess Plateau, China. CATENA 85, 58 66. doi:10.1016/j.catena.2010.12.003 Water Resources Institute (WRI), 2008. Ecosystem Services: A Guide for Decision Makers. World Resources Institute, Washington, D C.
308 Water Resources Institute (WRI), 2012. The Corporate Ecosystem Services Review: Guidelines for Identifying Business Risks and Opportunities Arising from Ecosystem Change, Version 2.0. World Resources Institute, Washington, DC. Weber, E.U., 2013. Psychol ogy: Seeing is believing. Nat. Clim. Change 3, 312 313. doi:10.1038/nclimate1859 Webster, R., Oliver, M.A., 2001a. Geostatistics for Environmental Scientists. John Wiley & Sons Ltd., West Sussex. Webster, R., Oliver, M.A., 2001b. Geostatistics for environm ental scientists. John Wiley & Sons. Wen, C. H., Koppelman, F.S., 2001. The generalized nested logit model. Transp. Res. Part B Methodol. 35, 627 641. West, T.O., Marland, G., 2002. A synthesis of carbon sequestration, carbon emissions, and net carbon flux in agriculture: comparing tillage practices in the United States. Agric. Ecosyst. Environ. 91, 217 232. doi:10.1016/S0167 8809(01)00233 X West, T.O., Marland, G., 2003. Net carbon flux from agriculture: carbon emissions, carbon sequestration, crop yield, and land use change. Biogeochemistry 63, 73 83. White II, D.A., Welty Bernard, A., Rasmussen, C., Schwartz, E., 2009. Vegetation controls on soil organic carbon dynamics in an arid, hyperthermic ecosystem. Geoderma 150, 214 223. doi:10.1016/j.geoderma.2009 .02.011 White, W.A., 1970. The Geomorphology of the Florida Peninsula, Published for Bureau of Geology, Division of Interior Resources, Florida Department of Natural Resources. Designers Press, Orlando. Whitehead, J.C., Pattanayak, S.K., Houtven, G.L.V., G elso, B.R., 2005. Combining Revealed and Stated Preference Data to Estimate the Nonmarket Value of Ecological Services: An Assessment of the State of the Science (Working Paper No. 05 19). Department of Economics, Appalachian State University. Wilding, L.P ., Smeck, N.E., Hall, G.F., 1983. Pedogenesis and soil taxonomy. II, II,. Amsterdam; New York; New York. Williams, H.C.W.L., 1977. On the formation of travel demand models and econ omic evaluation measures of user benefit. Environ. Plan. A 9, 285 344. Wilson, M.A., Howarth, R.B., 2002. Discourse based valuation of ecosystem services: establishing fair outcomes through group deliberation. Ecol. Econ. 41, 431 443. doi:10.1016/S0921 800 9(02)00092 7
309 Wollenberg, E., Edmunds, D., Buck, L., 2000. Anticipating change: scenarios as a tool for adaptive forest management: a guide. CIFOR. Wooldridge, S., 2003. Bayesian Belief Networks. Presented at the Commonwealth Scientific and Industrial Resea rch Organisation (CSIRO), CSIRO Centre for Complex Systems Science. Wu, Q.B., Wang, X.K., Ouyang, Z.Y., 2009. Soil organic carbon and its fractions across vegetation yypes: Effects of soil mineral surface area and microaggregates. Pedosphere 19, 258 264. d oi:10.1016/S1002 0160(09)60116 8 Xiong, X., 2013. Geo spatial modeling of soil organic carbon and its uncertainty (Ph.D. Dissertation). University of Florida, Gainesville, FL. Xiong, X., Grunwald, S., Myers, D.B., Kim, J., Harris, W.G., Comerford, N.B., 20 12. Which soil, environmental and anthropogenic covariates for soil carbon models in Florida needed, in: B. Minasny, B.P. Malone, A. B. McBratney (Eds.), Digital Soil Assessments and beyond: Proceeding of the Fifth Global Workshop on Digital Soil Mapping 2 012, Sydney, Australia. CRC Press, London. Xiong, X., Grunwald, S., Myers, D.B., Kim, J., Harris, W.G., Comerford, N.B., 2014. Holistic environmental soil landscape modeling of soil organic carbon. Environ. Model. Softw. doi:10.1016/j.envsoft.2014.03.004 Y ang, Y., Chen, Y., Li, W., Chen, Y., 2010. Distribution of soil organic carbon under different vegetation zones in the Ili River Valley, Xinjiang. J. Geogr. Sci. 20, 729 740. doi:10.1007/s11442 010 0807 4 a? Rural urban differences in environmental attitudes. J. Environ. Psychol. 38, 39 48. doi:10.1016/j.jenvp.2013.12.009 Zelezny, L.C., Chua, P. P., Aldrich, C., 2000. New Ways of Thinking about Environmentalism: Elaborating on Gender Differences in Environm entalism. J. Soc. Issues 56, 443 457. doi:10.1111/0022 4537.00177 Zhao, L., Sun, Y., Zhang, X., Yang, X., Drury, C.F., 2006. Soil organic carbon in clay and silt sized particles in Chinese mollisols: Relationship to the predicted capacity. Geoderma 132, 31 5 323. doi:10.1016/j.geoderma.2005.04.026 Zomaya, A.Y., Kazman, R., 2010. Algorithms and Theory of Computation Handbook, in: Atallah, M.J., Blanton, M. (Eds.), Chapman & Hall/CRC, pp. 33 33. Zvomuya, F., Janzen, H.H., Larney, F.J., Olson, B.M., 2008. A lo ng term field bioassay of soil quality indicators in a semiarid environment. Soil Sci. Soc. Am. J. SSSAJ 72. doi:10.2136/sssaj2007.0180
310 BIOGRAPHICAL SKETCH Pasicha Chaikaew was born in Bangkok, Thailand in 1981 and moved to Chiangmai with her family i n 1983. She has three siblings, one of which is her twin sister. Pasicha attended elementary school at Sacred Heart College and high school at Montfort College in Chiangmai. She received her Bachelor of Business Administration with first honor degree at Ma ejo University, Chiangmai, Thailand. Her strong interest in nature and the environment led to her studying environmental science at Mahidol University, Bangkok, where she obtained her Master of Science in Natural Resource Management d egree. After completio n of her m environmental analyst for international firms. Nearly three years of work experience, Pasicha then received funding from the Royal Thai Government to pursue a higher education degree. She enrolled in the doctorate program at the University of Florida in 2009, where she earned her Ph.D. degree in Soil and Water Science in 2014. After graduation, Pasicha will be employed in the Environmental Science Department at Chulalongkorn University, Bangkok.