Modelling Tourism in the Galapagos Islandsjasss.soc.surrey.ac.uk/17/1/14/14.pdf · Monitoring...
15
©Copyright JASSS Francesco Pizzitutti, Carlos F. Mena and Stephen J. Walsh (2014) Modelling Tourism in the Galapagos Islands: An Agent-Based Model Approach Journal of Artificial Societies and Social Simulation 17 (1) 14 <http://jasss.soc.surrey.ac.uk/17/1/14.html> Received: 29-Jan-2013 Accepted: 08-Sep-2013 Published: 31-Jan-2014 Abstract Currently tourism is the main driver of change in the Galapagos Islands, affecting the social, terrestrial, and marine sub-systems. Tourism also has direct and indirect consequences for the unique archipelago's natural habitats and for the human well-being. Describing the mechanisms that drive and affect most the tourism development in Galapagos is a preliminary condition to developing a better understanding of the interaction structure of factors that shape the Galapagos archipelago as a social-ecological complex system. In this paper, we present a first attempt to represent the touristic market in Galapagos trough an Agent Based Model (ABM) of touristic activity, focusing on touristic offers, reservations, and touristic activities. The model is based on an individual-based representation of tourists' consumption preferences and touristic accommodation offers in the Galapagos Islands. Tourist agents are created to mimic the real world by assigning average characteristics of individuals who visit the Galapagos Archipelago of Ecuador. The accommodation offers (i.e., hotels and cruises) are generated in accordance with actual conditions derived from data collected through field surveys. The model includes a market agent that can change the prices, create and delete accommodation offers following an evolutionary algorithm. We carried out preliminary simulations that show a close agreement between real world data and model outputs. Furthermore we used the model to generate three "what if" scenarios in order to study how emergent patterns in the touristic market in Galapagos are affected by changes in the archipelago environment. In this way we illustrate how the model can be used as a useful tool to help public policy makers to explore the consequences of their decisions. Keywords: Spatial Agent Based Model and Simulation, Galapagos Islands, Tourist Destination Dynamics Introduction 1.1 Over the last few decades, tourism in the Galapagos Islands, like in most intact and pristine biodiversity sites around the world, has experienced an explosive growth in the number of international visitors and in the development of infrastructure to support the consumptive needs of tourists (Epler 2007). This remarkable development is inducing changes in nearly every aspect of life in the Galapagos Islands (Watkins & Cruz 2007). For instance, local communities in the Galapagos have experienced a rapid increase in the residential population caused by the migration of Ecuadorians from the mainland in search of jobs in the lucrative tourism industry, with negative implications for the unique and fragile natural environment of this World Heritage Site (Bensted-Smith 1999; Trueman et al. 2010). Migrants are being "pushed" from the Ecuadorian mainland by a generally poor economy and relatively low wages, and "pulled" by the higher paying jobs in tourism that are mandated by Ecuadorian law and accentuated by the robust growth in the island's tourism industry. The development of the touristic market, in spite of the efforts carried out by public and private conservation institutions (Parque Nacional Galápagos 2007), seems to be occurring in a relatively uncontrolled and conflictive manner. The perception is that all the attempts to regulate the rapid tourism expansion have been conducted in an intermittent and unconnected way, proposing solutions to acute social and ecological crises without a long-term vision or strategic plan (González, Montes, Rodríguez, & Tapia 2008). Attempts to regulate tourism and to create a sustainable island economy that explicitly links to nature have been fleeting and episodic. Figure 1. The study area 1.2 The Galapagos archipelago (see Figure 1 for a map of the archipelago), viewed as an ecotourism destination, is an emblematic case of the challenges of a coupled human-natural system (CHANS) (Liu et al. 2007) in which a natural resource (the Galapagos island ecosystem) is exploited by humans (tourists) producing feedback mechanisms that influence social-ecological interactions, dynamics, and trajectories. Endogenous factors and exogenous forces interact in complex ways to shape and reshape human-environment interactions. In coupled human-natural systems, at the beginning of the linking process, the http://jasss.soc.surrey.ac.uk/17/1/14.html 1 15/10/2015
Transcript of Modelling Tourism in the Galapagos Islandsjasss.soc.surrey.ac.uk/17/1/14/14.pdf · Monitoring...
©CopyrightJASSS
FrancescoPizzitutti,CarlosF.MenaandStephenJ.Walsh(2014)
ModellingTourismintheGalapagosIslands:AnAgent-BasedModelApproach
JournalofArtificialSocietiesandSocialSimulation 17(1)14<http://jasss.soc.surrey.ac.uk/17/1/14.html>
Received:29-Jan-2013Accepted:08-Sep-2013Published:31-Jan-2014
Abstract
CurrentlytourismisthemaindriverofchangeintheGalapagosIslands,affectingthesocial,terrestrial,andmarinesub-systems.Tourismalsohasdirectandindirectconsequencesfortheuniquearchipelago'snaturalhabitatsandforthehumanwell-being.DescribingthemechanismsthatdriveandaffectmostthetourismdevelopmentinGalapagosisapreliminaryconditiontodevelopingabetterunderstandingoftheinteractionstructureoffactorsthatshapetheGalapagosarchipelagoasasocial-ecologicalcomplexsystem.Inthispaper,wepresentafirstattempttorepresentthetouristicmarketinGalapagostroughanAgentBasedModel(ABM)oftouristicactivity,focusingontouristicoffers,reservations,andtouristicactivities.Themodelisbasedonanindividual-basedrepresentationoftourists'consumptionpreferencesandtouristicaccommodationoffersintheGalapagosIslands.TouristagentsarecreatedtomimictherealworldbyassigningaveragecharacteristicsofindividualswhovisittheGalapagosArchipelagoofEcuador.Theaccommodationoffers(i.e.,hotelsandcruises)aregeneratedinaccordancewithactualconditionsderivedfromdatacollectedthroughfieldsurveys.Themodelincludesamarketagentthatcanchangetheprices,createanddeleteaccommodationoffersfollowinganevolutionaryalgorithm.Wecarriedoutpreliminarysimulationsthatshowacloseagreementbetweenrealworlddataandmodeloutputs.Furthermoreweusedthemodeltogeneratethree"whatif"scenariosinordertostudyhowemergentpatternsinthetouristicmarketinGalapagosareaffectedbychangesinthearchipelagoenvironment.Inthiswayweillustratehowthemodelcanbeusedasausefultooltohelppublicpolicymakerstoexploretheconsequencesoftheirdecisions.
Keywords:SpatialAgentBasedModelandSimulation,GalapagosIslands,TouristDestinationDynamics
Introduction
1.1 Overthelastfewdecades,tourismintheGalapagosIslands,likeinmostintactandpristinebiodiversitysitesaroundtheworld,hasexperiencedanexplosivegrowthinthenumberofinternationalvisitorsandinthedevelopmentofinfrastructuretosupporttheconsumptiveneedsoftourists(Epler2007).ThisremarkabledevelopmentisinducingchangesinnearlyeveryaspectoflifeintheGalapagosIslands(Watkins&Cruz2007).Forinstance,localcommunitiesintheGalapagoshaveexperiencedarapidincreaseintheresidentialpopulationcausedbythemigrationofEcuadoriansfromthemainlandinsearchofjobsinthelucrativetourismindustry,withnegativeimplicationsfortheuniqueandfragilenaturalenvironmentofthisWorldHeritageSite(Bensted-Smith1999;Truemanetal.2010).Migrantsarebeing"pushed"fromtheEcuadorianmainlandbyagenerallypooreconomyandrelativelylowwages,and"pulled"bythehigherpayingjobsintourismthataremandatedbyEcuadorianlawandaccentuatedbytherobustgrowthintheisland'stourismindustry.Thedevelopmentofthetouristicmarket,inspiteoftheeffortscarriedoutbypublicandprivateconservationinstitutions(ParqueNacionalGalápagos2007),seemstobeoccurringinarelativelyuncontrolledandconflictivemanner.Theperceptionisthatalltheattemptstoregulatetherapidtourismexpansionhavebeenconductedinanintermittentandunconnectedway,proposingsolutionstoacutesocialandecologicalcriseswithoutalong-termvisionorstrategicplan(González,Montes,Rodríguez,&Tapia2008).Attemptstoregulatetourismandtocreateasustainableislandeconomythatexplicitlylinkstonaturehavebeenfleetingandepisodic.
Figure1.Thestudyarea
1.2 TheGalapagosarchipelago(seeFigure1foramapofthearchipelago),viewedasanecotourismdestination,isanemblematiccaseofthechallengesofacoupledhuman-naturalsystem(CHANS)(Liuetal.2007)inwhichanaturalresource(theGalapagosislandecosystem)isexploitedbyhumans(tourists)producingfeedbackmechanismsthatinfluencesocial-ecologicalinteractions,dynamics,andtrajectories.Endogenousfactorsandexogenousforcesinteractincomplexwaystoshapeandreshapehuman-environmentinteractions.Incoupledhuman-naturalsystems,atthebeginningofthelinkingprocess,the
http://jasss.soc.surrey.ac.uk/17/1/14.html 1 15/10/2015
exploitationincreasesexponentially.Thenaperiodofdegradationofthenaturalresourcequalityandavailabilityisobserved,withasubsequentdeclineinecosystemgoodsandservices,potentiallyleadingtoradicalchangesoracollapseoftheexploitiveactivities.ThissituationcanleadtoanevolutionoftouristdestinationsthathasbeenextensivelydescribedbyPlog(1974)andButler(1980).Typically,CHANSaredifficulttostudybecauseoftheirnon-linearitycharacteristics,multiplespace-timescales,complexfeedbackmechanisms,anduniquelocalcontextsthatarelinkedtoforcesofglobalization.Thisfrequentlymakesitdifficulttoexportprocessunderstandingfromoneplace-basedstudytoanother(Ostrom2007).Obviously,fromamanagementanddecision-makingpointofview,themainchallengeofCHANSissustainability,thatis,theattainmentoftheequilibriumbetweenprolongedandcontinuoushumanactivitiesandthewell-beingofthenaturalenvironment.UnderstandingwhysomesystemsaresustainableandotherscollapseisthemainpurposeoftheCHANSperspective(Ostrom2009).
1.3 Withouttourism,theGalapagoswouldremainarelativelypristine,unique,andbiodiversity-richislandarchipelagoinwhichagricultureinthehighlandsandfishingalongthecoastwouldcontinuetobetheprimaryhouseholdlivelihoods,nowsubsumedbytourism.Inhabitedbyfarfewerpeople,thehumandimensionwouldlikelyberelativelywellintegratedintotheoceanicenvironmentsettingoftheGalapagos,firstencounteredbytouristsarrivingintheGalapagosinthe1950s(Epler2007).Butevenbeforethearrivalofinternationaltourists,theGalapagoswasexploitedforitsnaturalresourcesofbothterrestrialandmarinespecies.Determiningthehistoricallegacyofexploitation,thespace-timelagsofimportantendogenousandexogenousfactors,andtheadaptivecapacityofthetourismindustry,aswellaslocalorinternationalactors,whocollectivelyshapesocial-ecologicalinteractions(Baggio2008)isakeystepinunderstandingthedynamicsandtrajectoriesofthearchipelagoasitreactsandadjuststotheconsumptivedemandsofaburgeoningtourismindustry.
1.4 Inthelastdecades,AgentBasedModels(ABMs)wereincreasinglyusedinawiderangeofstudies(Grimmetal.2005).InABMs,agentsareintegratedinacomputer-basedmodel.Everyagentcanrepresentindividualssuchastouristsorgroupsofindividualssuchasinstitutions.Toeverysingleagent,asetofbehavioralrulesareassociatedthatpermitagent-to-agentandagent-to-environmentinteractionstobesimulated.IntheABMsetting,itispossibletostrategicallyandcontrollablychangetheagents'featurestobuildasetof"whatif"scenarios,studytheresultingchangesinthesimulationsoutputandexaminesystembehaviorandagentlearning.ThisfinalaspectofABMscanbeusedtoanalyzeandmanagetouristicdestinations,butrarelyhaveABMsbeenusedwithsuchanobjectiveinmind.Indeed,despitefewpreliminaryworks(JohnsonandSieber2011;BalbiandGiupponi2010;Yin2007;CecchiniandTrunfio2007),noattempthasbeenmadetobuildacomputationalmodelofatouristicdestinationthatintegratestheeconomic,social,environmentalandpsychologicalfeaturesofalltheactorsinvolved.
1.5 ThispaperrepresentsafirststeptowardsthebuildingofanABMofthetourismmarketintheGalapagosIslandsandinwidertermstowardsthebuildingofanABMofatouristdestination.Theaimofthisstudyistoidentify,amongthevastnumberofmicroscopic,individualbasedelementthatcreate,throughacomplexnetworkofinteractionsthetourismmarketsysteminGalapagos,alimitedsetoffeaturesthatcanbeusedtoeffectivelybuildamodelabletoreproducesomeoftheemergentpatternsoftherealworldsystem.Theresultingmodel,aswedescribeinthelastsectionofthispaper,couldbeusedtobuilddifferentkindof"whatif"scenariostotesttheimpactofpossibleinterventionsandpolicies.FromthispointofviewthemodelthatwepresentcanbeusedbypublicpolicymakersliketheGalapagosNationalParkadministratorsasatooltoevaluatetheconsequencesoftheirdecision.
1.6 Thepaperisdividedasfollows:inthefirstpart,aspectsofthetourismmarketintheGalapagosaredescribedthatarerelevanttothederivedmodel,inthesecondpart,theABMisdescribed,andinthethirdpart,resultsfromsimulationarepresentedanddiscussed.
TheGalapagosTourismMarket:Uncontrolled,ExplosiveExpansion
2.1 Asmentionedabove,tourisminGalapagoshasexperiencedanexplosivegrowthinthelastfewdecades.AreviewoftheavailabledatafromtheGalapagosNationalPark(ParqueNacionalGalápagos2012)showsthattouristarrivalsinGalapagosgrewexponentiallyfrom1995–2011,withanaverageannualgrowthrateofapproximately9%(seeFigure2).
Figure2.NumberoftouristsvisitingGalapagosfrom1975to2011(Source:GalapagosNationalPark)
2.2 FollowingButler(1980),therepresentationofthelife-cycleofatouristicdestinationcanbebasedonaSshapedcurvedescribingsixstagesthatthedestinationwouldgothroughastourismdevelops:exploration,involvement,developmentshowinganexponentialgrowth,consolidation,stagnationandtheendingwitheitherrejuvenationordecline.AsshowedinFigure2,theexpansionoftourisminGalapagosisstillinthe"development"phase,notyetapproachingthe"stagnation"phase.TheGalapagostouristicmarketevolutionprocessinthelastfewyearsfollowedthesamegeneralpatternofgrowthseenontheEcuadorianmainlandandintheneighboringcountriesofPeruandBolivia(WorldTourismOrganization2006).AnothercharacteristicoftheGalapagosIslandsisthatthearchipelagoisrelativelyremoteandisolated,factorgenerallyresponsibleforitshighnaturalendemismthatinturnisthesourceoftheattractionofinternationaltourism.Also,thearchipelagogeographyremotenessresultsinageneraldisconnectionfromtheusualtouristicroutes,creatingtheneedforahightransportationinvestmentintheinfrastructureintheislandsbylocalandregionalgovernmentsandhighercostsofvisitingtouristsincomparisontoothercompetingtouristicdestinations.Consequently,thenormaltouristrarelyvisitsGalapagosmorethanonce,oftenseeingthearchipelagoasa"bucketlist"destination.ThesurveyconductedbytheGalapagosTourism
http://jasss.soc.surrey.ac.uk/17/1/14.html 2 15/10/2015
MonitoringCenteroftheEcuadorianMinistryofTourismin2011(ObservatoriodeTurismoGalapagos2011)pointsoutthat92.66%oftheinterviewedtouristvisitedGalapagosforthefirsttime.ThisimpliesthatalmosteverytouristicexperienceinGalapagosisafirsttimeexperience,probablynevertoberepeated.AnotherimportantaspectoftheGalapagostouristicmarketisthatvisitorstypicallybookanaccommodationofferbasedonthechoiceofsecond-handsourcessuchasinternationaltravelagenciesand/orcompaniesthatadvertiseontheinternet.TheseorganizationspresenttotheinternationaltourismmarkettheGalapagosasapristine,natural,anduniquearchipelago.Thetouristacceptsthisrepresentationandthecorrespondingofferswithoutadirectknowledgeofwhathe/sheisactuallybuying.Weinfusetheseconditionsinourmodelbyrelyinguponparametersthatcharacterizetouristicoffers,suchas,whetherthehotelisnearorfarfromabeachorthecategoryofacruiseboat.
2.3 ThetourismexperiencesinGalapagoscanbedividedintotwomainsegments:theland-basedtourismandtheboat-basedtourism.TouristsreservingacruisetoexploretheGalapagosIslandsaredefinedas"boat-based,"because,ingeneral,theyspendmostofthetimeoftheirtouristicexperienceonboard–sleeping,dinning,anddisembarkingforshortislandactivities.Thistypeoftouristsisperceivedbyresidentsasnotcontributingtothelocaleconomy(Epler2007),asfundstendtosupportnationalandinternationalboatcompaniesandlargetouringorganizations.Land-basedtouristsaretouriststhatspendallormostoftheirtouristicexperienceinlocalcommunities,sleepinginhotels,arrangingtouristactivitieswithlocalvendors,andeatingattheirhotelsorinlocalrestaurantsandtheyareperceivedbylocalpeopleasmorereliantonlocalproducts,workers,andservices.Thedifferencebetweenlandandboat-basedtouristsishighlyrelevantfromthepointofviewoftheimpactonthesocial,terrestrial,andmarinesub-systems,asland-basedtourismimpliesgreaterurbandevelopmentandsupportinginfrastructure,useofpublictransportationandprivatevehicles,increasedrateofexoticspeciesintroductionasaconsequenceoftheimportationofmostfoodandproductsfromthemainland,depletionofdrinkingwaterreserves,consumptionofenergy,generationofwaste,andtheneedsofanenhancedwaterandsanitationsystemsthroughoutthearchipelago.Overthelast20–30years,boat-basedtourismdominated,buttoday,thenumbersofland-andboat-basedtouristsarenearlyequal.
2.4 ToabetterunderstandingoftheGalapagostouristsystemenvironment,itshouldbenotedthatintheGalapagos,onlyfourislandsarepopulatedandthusequippedwithtouristiclandinfrastructures:SanCristobal,SantaCruz,Isabela,andFloreana(seeFigure1).SanCristobalishometotheprovincialgovernment,SantaCruzisthehuboftourismwhileIsabelaistheepicenteroftheconflictsbetweenresourceconservationandeconomicdevelopment.Floreanahasthefewestresidentsandtourists,constrainedbythelackofavailabilityfreshwaterandthelimitedsupportinfrastructurefortourism.OnlySantaCruzandSanCristobalhostairportsthatdirectlyconnecttheseislandswiththeEcuadorianmainland.
2.5 ThemodelwepresentinthispaperreproducestheseGalapagostouristmarketfeaturestorepresentthereservationprocessaswewillshowinthenextsections.
TheAgentBasedModel
3.1 ThisABMisdescribedfollowingthe"'Overview–Designconcepts-Details"(ODD)protocolproposedbyGrimmetal.( 2006;2010).
3.2 Shortly,themodelisdesignedasfollows:whenatouristagentiscreated,theagentestablishesasortedlistofalltheavailableaccommodationoffersbasedupontheagentindividualcharacteristicsandthefeaturesoftheallavailableaccommodationoffers.Scrollingthroughthesortedlistofaccommodationoffers,thetouristagentbooksthefirstofferthatisaccessiblebypricerangeandbedsavailability.Accordingtothetouristagent'sdecisions,theaccommodationofferscanaccumulateaneconomicprofitforthetouristicprovideragent.Theoffersthataccumulateaprofitfasterareduplicatedwiththesubsequentcreationofanewindependentaccommodationoffer.Theduplicationprocessisimplementedincludingsmallchanges(mutations)inthepackagingoftheparentduplicatedaccommodationoffer,therebycreatinganevolutionaryselectionprocesssimilartotheevolutionaryalgorithmsusedinmetaheuristicoptimizationproblems(Gen2010).
3.3 Furthermore,apartialrepresentationoftheGalapagosgeographicalspaceisincludedinthemodeltorepresentthecostsintimeandmoneythatarerequiredforatouristagenttomovefromoneislandtoanother.Inthisway,itispossibletoincludethemovementsoftouristsbetweenislandstotakeincountthefactthateverytouristwantstomaximizethenumberoftouristicattractionsvisited,travelingfromislandtoislandminimizingthetransportationcostintimeandmoney.
3.4 Inwhatfollows,wedefineanaccommodationofferasatouristicofferthatincludesaccommodationamongtheirservices.Wedefineahotelasanykindofland-basedaccommodation(e.g.,hotels,hostals,resorts),andcruisesasanykindofboat-basedaccommodation(cruiseboats,yachtsandothers).Moreover,weusetheexpression"touristicattraction"toindicateallthepossibletouristicactivitiestowhichthetouristcouldaccessduringhis/herstayintheGalapagosIslands.Amongtouristicattractions,weincludedaytours,visitstointerestingplaces,watersports,trekkingtrails,snorkelingsites,useofbeaches,andothers.
Overview
Purpose
3.5 Thepurposeofthemodelistoexaminethetourists'decision-makingprocessesthatrelyuponthecharacteristicsoftheaccommodationoffersinfrastructure,uponthetourist'sindividualpreferences,aswellasuponthespatialconfigurationcharacteristicsoftheGalapagosIslands.ThereforeoneofthegoalsofthemodelistoidentifyalltheindividualbasedcharacteristicsoftheprocessthatarerelevantindeterminingtheoverallemergentdynamicsofatouristicdestinationliketheGalapagosIslands.Inthisperspective,thecomparisonbetweenthemodeloutputandtherealworldempiricaldatawouldpermitustoevaluateiftheindividualbasedelementsusedtobuildthemodelaresignificantornot.AnotherobjectiveofatouristdestinationABMliketheonepresentedinthispaperistoprovidetopublicpolicymakersatooltotesttheconsequencesoftheirmanagementdecisions.
Entitiesstatevariablesandscales
3.6 Themodeliscomposedoffivemaintypesofentities:touristagents,accommodationofferagents,themarketoperatoragent,thetouristagentgenerator,andthegeographicalenvironmentcircumstancesthatrepresenttheGalapagosIslands.
http://jasss.soc.surrey.ac.uk/17/1/14.html 3 15/10/2015
Figure3.Themixed,spatialexplicitandgraph-likerepresentationofthegeographicalenvironmentoftheGalapagosarchipelago.EveryIslandisthevertexofatransportationgraph.Onlytheareainsideeveryislandistreatedexplicitly
3.7 Figure3showsthegeographicalenvironmentrepresentation,characterizedbymixinganexplicitspatialrepresentationwithatopological,graph-basedrepresentation.ThefourinhabitedislandsoftheGalapagosArchipelago,i.e.,SanCristobal,SantaCruz,Floreana,andIsabela,arerepresentedasverticesonagraph.Theedgesareusedtorepresenttransportationconnectionsbetweenislands.Twoislandsareconnectedbyanedgeifanavailabletransportationoperationexistsbetweenthem.Weightsaregiventoedgestospecifythecostandtheamountoftimeoftransportation.Someoftheislandsaredefinedasentrancehubs,becausetheyhostanairportfromwheretouristsenterthearchipelagofromtheEcuadorianmainland(i.e.,SanCristobalandSantaCruzthatareaccessedthroughthenearbyBaltraIslandairport).Notethataninter-islandairlineserviceconnectsSanCristobal,SantaCruz,andIsabelaislandsusinglowcapacity,smallaircraftbutlocalcommercialboats(i.e.,"fibras")arethemostcommonwayofmovingbetweenislands.Thereforeinthemodelonlytransportationsbyboatwillbeconsidered.Thespatiallyexplicitrepresentationoftheinnerterritoryofeveryislandisdenoted,includingthenetworkofroadsandthetouristicoffers,representedaspointslocatedalongtheroadnetwork.Moreover,weassociateacertainnumberoftouristicattractionstoeachoneofthepopulatedislands.
3.8 Thetouristagentsaredescribedbythefollowingstatevariables:age,budget,stay,reservationflexibility,reservationanticipation,group,andcharacter.Thevariable"budget"istheamountofmoneythatthetouristagentspendsduringhistouristicexperience.Thevariable"stay"isthelengthofthetouristicexperience,whereas"reservationanticipation"isthetimeintervalbetweenthedateofbookingandthedatethetouristagentbeginsthetrip.The"reservationflexibility"representstheextenttowhichthetouristagentcanchangetheanticipation,andthevariable"character"isaderivedvariablethatreferstothetouristagentpreferencefordifferentaspectsofatouristicexperienceintheGalapagosIslands.Thevariable"group"indicateshowmanypeopletraveltogetherwiththetouristagent–whenthevariablevalueisgreaterthan1thetouristisconsideredtobeagroupofagentmakingthesamedecisions.The"character"variablehasfivedimensions:worldheritage,activities–sports,sun–beach,all-inclusive,andexclusivity.Eachdimensionhasvaluesbetween[0,2]–0forlowinterest,2forhighinterest.The"worldheritage"dimensionindicateshowmuchthetouristagentisinterestedinseeingGalapagoswildlife."Activities–sport"isthetouristwillingnesstoaccesslandandwatersports(e.g.,horsebackridingandkayaking)andothertouristicattractionswiththeexclusionofwatchingwildlife,while"sun–beach"isthewillingnesstoexperiencebeachlife(e.g.,swimmingandsunbathing)."All-inclusive"indicateswhetheratouristagentisinterestedinbuyinganall-inclusivetouristicexperience(i.e.,generallyatourledexperiencethatincludesalllandandwateractivities,arrangedmealsandlodging,andtransit),andlastly"exclusivity"indicatestheextenttowhichthetouristagentisinterestedinstayinginanexclusivetouristicfacility(i.e.,afullservice,resort-typeaccommodation).
3.9 Theaccommodationoffersarecharacterizedby:type,time,capacity,category,price,location,andfeatures."Type"couldtake"hotel"or"cruise"values."Time"istheminimumamountoftimeduringwhichtheofferappliesforatouristagent.The"Capacity"isdefinedashowmanypersonscanbeaccommodatedinanofferatthesametime,whilethe"Category"istherankassociatedtotheoffersaccordingtotheiroverallquality."Price"isthecostofahotelnightoraboatcruisethroughtheislands(refertotheTouristreservationsubmodeltoseehowcruisesandhotelspricesarecompared)."Location"isspecifiedonlyforland-basedaccommodationtypesandindicatesthegeographicpositionofhotelsonthecorrespondingisland."Feature"isanofferstatevariablethatcorrespondstothe"character"touristagentstatevariable,owningthesamefivedimensionsandvalueranges.Theoffer"features"indicatestheextenttowhichtheoffersatisfiestheexpectationofatouristagentinthefivedifferentdimensionsofthevariable.Theactualnumberoftheaccommodationoffersinthesimulationandthevaluesoftheirstatevariablesareassignedtothetouristicoffers,inthemodelinitialization,toreproducetherealworldstateintheGalapagosArchipelagoofEcuador.
3.10 Themarketoperatoragentisdefinedbythefollowingstatevariables:periodicity,priceincreasethreshold,priceincreaserate,pricedecreasethreshold,pricedecreaserate,numberofrandomoffercreations,operatingexpensesparameters,offercostparameters,offerdepreciationrate,numberofyearsofthenetpresentvalue,yearlydiscountrate,reinvestmentrate,andbankruptcythreshold.ThemarketoperatorstatevariablesaredescribedinTables6and7inAppendixA.
3.11 Thetouristagentgeneratorisnotstrictlyanagent,asitcouldbeconsideredamoduleinthemodelwherealltheparametersneededtogeneratetouristagentsarestored.Itisdefinedbythefollowingstatevariables:periodicity,numberoftouristagentsperyear,numberofageclasses,andseasonaltuning.ThetouristagentgeneratorstatevariablesaredescribedinTable12inAppendixA.
Processoverviewandscheduling
3.12 Thetimestepofthemodelisoneday.Monthsaredefinedascomposedby30days,andyearsaredefinedascomposedby12months.Hotelsandcruisesoperatewithadailytimestep,registeringthenumberoftouristagentstheyhost.Themarketoperatoragentfollowsatimestepdefinedbyitsperiodvariablethat,inthisstudy,isalwayssetatamonth.
3.13 Whentouristagentsaregeneratedbythetouristagentgenerator,theyimmediatelyattempttoreserveanaccommodationoffer.Iftheydonotsucceedinfindinganaccommodation,theyexitfromthesimulation.Ifthetouristagentsucceeds,theagentwaitsuntilthedatescheduledbythereservation,thenswitchestothe"trip"stateandtheagentisthenregisteredintheofferthatwasbooked.Inthecasewherethetouristagentbookedahotel,eachsimulationdaytheagentsearchesforanewaccommodationforthenextday.Wenotethat,inthisway,thetouristagentscannotreserveahotelformorethanoneday.Ofcourse,thisisnotwhatactuallyhappensintherealworldasincertaincasesatouristwantstobookinthesameplaceformorethanoneday.However,thereservationprocesswasdesignedinthiswaybecausewehavenoinformationaboutthepercentageoftouristsbookinghotelsinadvanceinGalapagosformorethanoneday.Alsoallowingtouristagentstobookformorethanonedayinadvancewouldimplythatthetouristagentcouldcancelthereservationincasehewantstochangeplacebeforeofthereservationends.Inthiscircumstancethetouristagentisforcedtopayafinebutthenagainwehavenodataaboutcancellationpoliciesappliedinthehotels.
3.14 Everysimulationday,theland-basedtouristagentssubtract3touristicattractionsfromthelistoftheunvisitedtouristicattractionoftheislandwheretheagentsareactuallylocated.Thissimulatethefactthattheland-basedagentgothrough,asdailyaverage,3activities,forexamplevisitinganaturalenvironmentsite,takeawalkalongahighlandtrack,andtakeasunbathonalocalbeach.Inthisway,whilethetourist
http://jasss.soc.surrey.ac.uk/17/1/14.html 4 15/10/2015
agentremainsinthesameislandIA,thenumberofunvisitedtouristicattractionassociatedtoIAwilldecreaseeverydaybythreeunits.Afteracertainnumberofdays,forthetouristagent,inanotherislandIBofthearchipelago,thenumberofunvisitedtouristicattractionswillbehigherthanintheislandIA.Aswewillseeinthe"Touristreservationsubmodel"section,thetouristagentstendtoreserveaccommodationofferslocatedintheislandwiththehighernumberofunvisitedtouristicattractions.ThereforewhenthenumberofunvisitedtouristicattractionsintheislandIAwillbelessthanthenumberofunvisitedtouristicattractionintheislandIB,theagentwillbepushedtobearthecosts–intimeandmoney–ofmovingfromanislandtoanotherisland.
3.15 Ifthetouristagentbooksacruise,theagentfollowsthecruisejourneythatincludesimplicitly3dailyactivities.Attheendofthecruisejourney,andifthelengthofhis/herscheduledstayisgreaterthanthelengthofthecruisejourney,theagentattemptstobookahotel.Notethatthestopoverdestinations,thenumberofstops,andthenumberofdaysofeverycruisejourneyarestrictlyregulatedbytheGalapagosNationalParkandcannotbechangedbytheboats'owners.Moreover,thecruisesstopoversneverincludenightsoffboardsotheboat-basedtouristagents,whiletheyareincruiseneveruselandaccommodationoffers.
3.16 Themarketoperatorcalculatestheprofitorlossandtheaveragenumberofhostedtouristagentsforeveryaccommodationofferinthemodel,anddecideswhethertocreatenewoffers,removeoldoffers,orincrease/decreaseprices.
http://jasss.soc.surrey.ac.uk/17/1/14.html 5 15/10/2015
Figure4.Flowdiagramoftheprocessessequenceofagentsinthemodelintheirlifecycle.
3.17 AflowdiagramofthelifeprocessofeveryagentinthemodelisshowninFigure4.NotethatthemaximumnumberofcruiseboatswithregularoperatingpermissionisstrictlyregulatedbytheGalapagosNationalPark.Forthisreason,weimposedthatthetotalnumberofcruisesduringthesimulationisnotgreaterthantheinitialnumberofcruisesintheinputdata.ThetotalnumberofhotelsisnotsostrictlyregulatedbytheGalapagosNationalParksothenumberofhotelsinthesimulationcanchange.
Softwareandhardwareenvironment
ThemodelhavebeenimplementedintheMASON(Lukeetal.2005)environment.MASONisafree,Java-based,discrete-event,multi-agentsimulationlibrarycoreusedtoreducetherepetitivecodewritingeffortnecessarytodevelopanABM.Everyrunof15yearswith200,000generatedtouristsperyearrequiresapproximately4CPUhoursonanIntelPentiumDual2.40GHz.
DesignConcepts
Emergence
3.18 Manymacroscopicpatternsemergefromthetourist,offer,andmarketoperatoragents'individualbehaviors–themodelreproducetherelativeproportionofland-andboat-basedtouristsandthespatialdistributionofaccommodationoffersandtouristagentsacrosstheGalapagosArchipelago.
Adaptation
3.19 Themarketoperatoradaptstheaccommodationofferstotheavailableenvironmentconstitutedbytouristagents.Theprocessispermittedbythevariationofoffersprices,replicationofmostbookedoffers,andtheeliminationofun-bookedoffers.Onlytheoffersthatsatisfymostthetouristagentpreferenceswillsurviveandreplicate.
Objectives
3.20 Thetouristagents'aimistofindthemostsatisfyingaccommodationofferandmaximizethenumberofpossibletouristicattractionsenjoyedduringthetrip.Thepurposeofthemarketoperatoristoproduceanoffers'populationthatsatisfytouristagentsmost.
Knowledge
3.21 Touristagentsandmarketoperatorhaveanunrestrictedaccesstoinformation.Thismeansthatineverymomenttheyknownexactlythecharacteristicsofeveryaccommodationoffer.Evenifitisknownthattheaccesstoinformationintherealworldtourismmarketcouldbeasymmetricandnotcompletewemakethisassumptionforthesakeofmodel'ssimplicity.
Interaction
3.22 Touristagentsinteractwithofferstolearnaboutthehotel'sbedsavailabilityaswellasboatberths,andtogetinformationabouttheoffers'features.Marketoperatorsinteractwithofferstodeterminewhetheritisnecessarytointroducechangesintheoffers'environment.
Stochasticity
3.23 Alltheparametersdefiningthetouristagentcharacteristicareconstructedasstochasticvariables,astheyareextractedrandomlyfromempiricalprobabilitydistributions.TheseempiricaldistributionsaretheresultoffieldsurveysinwhichasampleoftouristsvisitingGalapagosIslandswasinterviewedthroughapre-testedsurveyinstrument.Moreover,astochasticfactorisextractedtogettheextenttowhichtouristicoffersaredifferentinrespecttotheirduplicates.
Observation
3.24 Duringsimulation,wemonitorthenumberofland-andboat-basedtouristagentsvisitingtheGalapagosIslandsandthenumberoftouristagentsineachofthepopulatedislands.
Details
3.25 Inthissection,wedescribetheinitializationprocessandtheinputdatausedtoinitializethestatevariables.Furthermore,thesubmodelsthatcontaintheequationsandrulesofthemodelarepresented.
Initialization
3.26 Atthebeginningofeachsimulationruntheislandparametersarereadfromtheconfigurationfiles.First,thearchipelagographischarged.InTable2inAppendixA,theparametersdefiningthearchipelagotransportationgrapharereported.Next,theshapefilecontainingthecoastlineofeveryislandisread.ThecoastlinesdatawereobtainedfromtheINGALA,NationalInstituteofGalapagos,nowrenamedtheGovernmentCouncilofGalapagos.Then,thegeographicalcoverageoftheroadsnetworkisread(dataobtainedfromtheEcuadorianInstituteofStatisticsandCensus,INEC),followedbytheshapefilecontainingthehoteldata.TheinformationabouthotelsinGalapagosisobtainedfromtheGalapagosTourismMonitoringCenter.Cruisesarealsoinitializedinthemodel,informationobtainedfromtheGalapagosNationalParkandtheGalapagoscruisespromotionorganization(www.crucerogalapagos.com).Pricesofhotelswereobtainedthroughatelephonesurvey.Afterthegeoreferencedinformationisread,thesimulationbegins.ThenumberoftouristicattractionsofeveryislandinthearchipelagoisreportedinTable3inAppendixA.NotethattouristicactivitiesarestrictlydelimitedbytheGalapagosNationalParksoweknowexactlyhowmanyactivitiesareofferedtotouristsineveryisland.Theactualvalueofeverystatevariableofeveryindividualtouristagentisextractedfromempiricaldistributionsfromthe2011fieldsurveyoftheGalapagosTourismMonitoringCenter(ObservatoriodeTurismoGalapagos2011).Notethattheempiricaldatawehaveusedindicateonlythetotaltouristagent'sexpenditurefortheGalapagostrip,includingtheairticketfromtheEcuadorianmainlandandtheparkentrancefee.BecausetheairticketpricesandtheGalapagosNationalParkentrancefeesaredifferentfornationalandforeigntourists,wecalculatedtheactualbudgetavailablefortourismintheGalapagostoeverytouristagentbysubtracting$450.00(USD)fortheairticketand$100.00(USD)fortheparkentrancefeeappliedtothe78.16%ofthegeneratedtouristagentthatareconsideredasforeigntouristagents,and$350.00(USD)fortheairticketand$16.00(USD)fortheparkentrancefeeappliedtothe21.84%ofthegeneratedtouristagentsconsideredasnationaltouristagents.Thevaluesofthetouristagent'sstatevariablearereportedinTables3and4inAppendixA.Theoffers'variablefeatureisdeterminedbythecharacteristicsoftheoffer.Forhotels,WorldHeritagevaluesaresetto0,1or2dependingonthedistancefromthecentroidoftheislandroadsnetwork–2fordistanceslargerthan10,000m,1fordistancesbetween10,000mand8,000m,and0forhotelswithadistancebelow8,000m.ThesameWorldHeritagevalueissetto2forcruises.Theactivities-sportvalueisassignedtohotelsandcruisestobeproportionaltotheCategory.Sun-beach,forhotels,aresetto1or2dependingonthedistancefromtheislandcoastline–2fordistancessmallerthan100m,and1fordistancesgreaterthan100m.Sunandbeachissetto0forcruise.Wenotethat,inthisway,theislandsthatshowroadswithmoreextensionalongtheseafronttendtobemoreattractiveforsunandbeachtouristagents,whileislandsthatshowroadsthatbringtouristagentsfarfromtheurbancenterstendtobemoreattractiveforworldheritagetypetouristagents.AlltheoffersfeaturevariableassignmentsarereportedinTable6inAppendixA.
3.27 Thetime-seriesofthetouristagentsgeneratedforonesimulationyearisproportionaltothehistoricalseriesofvisitingtouristsinGalapagos,extractedfromtheGalapagosNationalParkstatistics(ParqueNacionalGalápagos2012).
Submodels
Touristreservationsubmodel
3.28 Thissubmodeldefinesthewaytouristagentschoosetheaccommodationthattheybook.Whenatouristagentiscreatedinthesimulation,thefirstactiontheagentperformsistotrytobookanaccommodationoffer.Iftheagentbooksahotel,thetouristagentisdefinedasaland-basedtouristagent.Thissubmodelisdescribedasfollows:
1. Thetouristagentusestheoffercomparatormoduletobuildasortedlistofalltheavailableoffers.Ifthisistheinitialapplicationofthecomparatormodule,allpossibleland-andboat-basedoffersareincluded.Iftheagentwishestorenewaland-basedreservation,theagentincludesonlyhoteltypeoffers.
2. Startingfromthetopofthesortedlist,thetouristagentexaminestheaccommodationoffersandbooksthefirstofferwithapricewithintheagent'sbudgetandwithavailableroomsduringthedatesofthedesiredtrip.Iftheendofthelistisreached,theagentexitsfromthesimulation.
ThetwostepsprocesscorrespondstothediagrampresentedinFigure5.
http://jasss.soc.surrey.ac.uk/17/1/14.html 6 15/10/2015
Figure5.Flowdiagramofthereservationsubmodel
3.29 TheOfferComparatormoduleworksinthefollowingway:thetouristagentgeneratordividestheintervalfromtheminimumtothemaximumtouristagents'budgetintonsubintervals.Wedefineas"lowbudget"and"highbudget"thegeneratedtouristagentswithabudgetbelongingtothelowerandhigherlevelsoftheconsideredbudgetsubintervals,respectively.Theremainingtouristagentswillbedefinedas"mediumbudget"touristagents.Dependingonthebudgetsegmentthetouristagentbelongs,theagentwillapplyadifferentstrategytosorttheaccommodationoffers:lowbudgettouristagentswillpreferlowestpriceoffers,thehighbudgettouristagentwillpreferthehighestqualityoffersandthemediumbudgettouristagentswillfollowabestqualityforminimumpricestrategy(Piana2004).Indetail,thelowbudgettouristagentswillsortoffersaccordingtothefollowingcondition:theofferO1precedestheofferO2ifP1<P2wherePiisgivenby:
Pi=pbi+Tci
HerepbiisthepriceforaroompernightfortheofferOiandTciisthetransportationcosttoreachtheislandwheretheofferOiislocatedfromtheclosestentranceisland(iftheofferOiislocatedinanentrancehubthenTci=0).CruisesalwaysembarktouristagentfromSantaCruzorSanCristobalsoTci=0forcruises.InthecasewheretwoormoreofferspresentthesameP,theOfferComparatorModulewillorderthematrandom.Toallowcomparisonsbetweentheaccommodationpriceofanall-inclusivetouristicoffer,likeacruiseandanoffersuchasahotelforwhichtheaccommodationpriceonlyincludestheexpensesforthenightstay,itisnecessarytoconsiderthepriceofthehotelasifthehotelwhereanall-inclusivetouristicoffer.Toaccommodatethatinthemodel,aspecificparameterisdefined–thedailyexpenserate,aratiobetweenthetotalaveragedailyexpensesofaland-basedtouristagentandtheaveragepriceforonenightinahotel.Bymultiplyingthepriceofahotelbythedailyexpenserate,itispossibletoobtainacorrectedhotelpricethatiscomparabletocruiseprices.Below,allthehotelspricesareintendedascorrectedpricesandthedailyexpensesrateissetto3(Epler2007).
3.30 Thehighbudgettouristagentwillconsideroffersthroughthefollowingcondition:theofferO1precedestheofferO2iftheoverallqualityQ1isgreaterthanQ2.Theoverallqualityisdefinedas:
Herefijisthej-thelementofthefeaturevariableoftheofferOi,whilecjisthej-thelementofthetouristagent'scharactervariable.Moreover,NiisthenumberoftouristicattractionsassociatedtotheislandwheretheofferOiislocated.TtiisthetransportationtimetotraveltotheislandwheretheofferOiislocatedfromanentranceisland.Lastly,Aistheagefactorassociatedtothetouristagent–asinthecaseoftouristagents'budget,thetouristagentgeneratorseparatesthegeneratedtouristagentsbyageintoacertainnumberofageintervals,definingas"youngaged"and"oldaged"theagentsthatcorrespondtothelowestageandhigherageintervals,respectively.Theremainingtouristagentswillbedefinedas"middleaged"touristagents.TheagefactorAisdefinedtobeequalto1,2or3fortouristagentsbelongingtotheyoung,middleandoldagedclasses,respectively.ThethreefactorsCf,CN,andCAareparametersthatwillusedtocalibratethemodel.
3.31 Forthemediumbudgettouristagent,theofferO1precedestheofferO2ifR1>R2.Riisthequalitypriceratioanditisdefinedas:
Ri=Qi/Pi.
Whenthetouristagentisalreadyaccommodatedinahotelandtheagentislookingforanewaccommodationforthefollowingday,theexpressionusedtoindicatetheoverallqualityofthetravelexperienceisslightlydifferent:
HereNεiisequaltothenumberofthetouristicattractionsnotvisitedbythetouristagentontheislandwheretheofferOiislocated.Theislandswithmorenotvisitedtouristicattractionwillbemoreattractiveforthetouristagent.Theparameterεitisalwaysequalto0andisequalto1onlyiftheofferOiisnottheofferwherethetouristagentisactuallyaccommodated.
Marketoperatorsubmodel
3.32 Thissubmodeldefinesthemarketoperatordecision-makingrules.Themarketoperator(MO)usestheprofitgeneratedbyanoffertocreatenewofferssimilartoasuccessfuloffer.ThestepsfollowedbytheMOtodecideifanoffershouldbe"replicated"areasfollows:
1. Calculatethetotalprofitaccumulatedbytheofferanddetermineifitishighenoughtocovertheinvestmentneededtobuildacopyofthestartingoffer.2. CalculatetheNetPresentValue(NPV)ofthenewoffer.AnewofferwillbecreatedonlywhenNPVispositive.
3.33 ThetotalprofitPoftheoffersisdefinedasthesumofallthedailyprofitspiovertheperiodfromthelastreplicationorfromthecreationoftheoffertothepresenttimeminustheaccumulatedrenovationcosts:
HereNdaysisthenumberofdaysbetweenthecreationorthelastreplicationoftheofferandthepresenttime,whileDcisthecostduetotherenovationaccumulatedbytheofferinthesameperiod.Inaddition,oiandpbiaretheoccupancyandpriceperbedorberthoftheoffercorrespondingtothedayi.ThevariablesCcapandboerepresentthecapacityandtheoperatingexpensesper"bed"respectively.Indetail,theoccupancyisthenumberofoccupiedbedsandtheoperatingexpensesarethecostperday,perbedtomaintaintheoffer.boe,isafunctionofthecharacteristicsoftheofferasfollows:
whereCatcouldtakethevalues0,1,2and3,fiisthei-thelementoftheofferfeaturestatevariableandαe,βeandγearetheparameterspresentedinTable10forHotelsandTable11forCruises.
3.34 Notethatboeautomaticallydefinesminimumrentablepricefortheoffer:
prent=boeprentinturndefinestheminimumacceptablepricepmin:pmin=prent·fmp
http://jasss.soc.surrey.ac.uk/17/1/14.html 7 15/10/2015
Wherefmpisafactorthatissettobe1.2.
3.35 Theinvestmentneededtoduplicatetheoffer,inthecaseofhotels,isproportionaltothefeaturesandcharacteristicsoftheoffer:
whereABistheaverageareaperbedasfunctionofthecategoryCat,Pm2isthebuildingpricepersquaremeterasfunctionofthecategory.ThevaluesoftheparameterspresentedinthislastequationhavebeendeterminedthroughafieldsurveyofdepartingtouristsonSanCristobalandSantaCruzIslandsandtheresultingvaluesarepresentedinTable8inAppendixA.ForCruisestheinvestmentexpressionisasfollows:
Ic=Cap·αi(Cat)
whereαi(Cat)isaparameterdependentontheCruisecategory.ThevaluesforαIarepresentedinTable9inAppendixA.
3.36 WenotethattherenovationcostDcintheaboveexpressionoftheprofit,isequalsto:
Dc=a·dr·IH.C
Herea,istheageoftheofferinyearsanddristhedepreciationannualrate.Asresultofourfieldsurvey,wesetdr2.5%forhotelsand3.5%forcruises.
3.37 Oncetheconditionatpoint1issatisfied,theMOevaluatesconditionnumber2,calculatingtheNetPresentValue(NPV)ofthenewofferthatrepresentsthecost-effectivenessoftheinvestmentforanewoffer.TheNPVisexpressedas:
whereNisthenumberofyearsinwhichtheinvestmentmustberecovered,<O>isthedailyaverageoccupancyoftheofferthatisduplicated,calculatedinthelastyearofactivity,pbisthepriceperbedpernightoftheduplicatedofferatthetimeofduplication.FurthermoreDRTisthediscountratefortheyeartandIistheinvestmenttocreatethenewofferpresentedabove.Thediscountrateistherateofreturnthatcouldbeearnedonaninvestmentinthefinancialmarketswithsimilarrisk.AlltheoffersthatfulfilltheconditionsP·Rr>IandNPV>0arereplicated.Whenthereplicationisdone,thecumulatedprofitofthereplicatedofferissettozeroandthesameofferwillnotbereplicateduntiltheconditionsatpoints1and2aresatisfiedagain.Thenewofferwillbenotanexactcopyofthereplicatedone–allthefeaturesandcharacteristicsofthereplicatedofferwillbechangedbyasmallstochasticfactorSc=10%togiverisetoaslightlydifferentnewoffer.Inthatway,wecreateanofferecosystemwherenotonlythefittestoffersarereplicatedi.e.,theoffersthatcanaccumulateprofitfastest,butwherewealsointroduceerrorsintheprocessofreplicationtopermittotheoffers'ecosystemtoadapttothetouristagentfeaturesenvironmentfasterandmoreefficiently.Whenahotelisduplicated,anewhoteliscreated.Thisnewhotelisplacedinarandompositioninsidetheroadsnetworkofthesameislandoftheparenthotel,withadistancefromtheparenthotelbetween100and1600m.Alsonewhotelshavenottobecloserthan20mtootherestablishedhotels.Ifanaccommodationoffershowsapassiveprofit(P<0),andthepassiveabsoluteisequaltothebankruptcythresholdmultipliedforthecost(I)oftheoffer,themarketoperatorwilldestroytheoffer.
3.38 Theseconddecisionmakingprocessofthemarketoperatoristheofferpricevariation.Iftheoffer'saverageoccupationisabovethepriceincreasethresholdpresentedinTable7inAppendixA,theMOwillincreasethepriceoftheofferbyafactordefinedbythepriceincreaserate.Ontheotherhand,iftheaverageoccupationisbelowthepricedecreasethreshold,theMOwilldecreasethepriceoftheofferbyafactordefinedbythepricedecreaserate.Thepriceofanofferwillbeneversetbelowtheminimumacceptablepricepmindefinedabove.
Modelverification,CalibrationandValidation
Verification
4.1 Theverificationprocessconsistsofobtainingaconfirmationthattheimplementedprogramcodebehavesasplannedinthedesignphase.Toverifyourmodel,weappliedthefollowingstrategy:first,weexaminedeverycodelinetoconfirmitsfunctionandcorrectness;second,weperformedanextremevaluetestforeveryparameterinthemodel,includingempiricaldata,provingthatthesystemrespondsintheexpectedway,andthird,webuiltanoversimplifiedtestmodelwithonlyafewtouristsandaccommodationoffersagentsandwecloselyexaminetheresults,checkingthateveryagentbehavedintheappropriatemanner.
CalibrationandValidation
4.2 Thecalibrationistheprocessofadjustingthevalueofsomeoftheindependentparameterstochangetheresponseofamodelinsearchofanempiricaladequacy.Thevalidation(afterCrooksetal.2008;Ligtenbergetal.2010)consistsofcontrollingthattheoutputfromamodelproducesresultsthataresufficientlyclosetothedataofthecorrespondingrealworldsystem,withinthelimitsandtheobjectivesofthemodeling.Calibrationandvalidationarequiteinterconnectedprocesses,becauseduringthecalibrationprocess,theparameterspaceisrestrictedtoobtainamodelthatcanbeconsideredasvalidated.
4.3 OurmodelwasdesignedtoreproducemacroscopicfeaturesoftourismintheGalapagosIslands,suchastherelativeproportionbetweenland-andboat-basedtourismandthespatialdistributionoftouriststhroughoutthearchipelago.Therefore,thevalidationprocessconsistsinverifyingthattheland-andboat-baseddistributionsandthespatialdistributionoftouristagents,predictedbythemodel,isinagreementwithdatasetsfromtheGalapagosNationalParkandtheGalapagosTourismMonitoringCenter.
4.4 Weperformedthecalibrationprocessbyvaryingthevalueofthreeparameters:Cf,CN,andCA.Thesethreecalibrationparametersarepresentedinsection"Submodels"intheexpressionoftheaccommodationoffersqualityQandareusedtobalancethecontributiontotheofferquality,originatingfromtheofferfeaturesterm,theislandstouristicattractionsterm,andtheageterm.Beinginvolvedintherepresentationofthetourist'sdecision-makingprocessand,therefore,intherepresentationofmentalprocesses,thesethreefactorsareusedascalibrationparameters,becauseitwouldbeverydifficulttoobtainameasureofthem.Weperformedasystematicsearchinthisthree-dimensionalparameterspaceandweselectedthevectorvaluethatgivesusthebestapproximationoftheobservedproportionbetweenland-andboat-basedtouristandofthedistributionoftouristsinthefourinhabitedislandsoftheGalapagosArchipelago.Asthemodelincludesdifferentstochasticfactors,distinctrunsproducegenerallydistinctresultsintermsofthesimulatedsystemhistory.Forthisreason,weaveragedtheoutputsfrom10independentrunsof15year.Thenwecomparedtherealworlddatawiththeaveragevaluesoverthelast7simulationyearsoftheresultingaveragedsimulatedhistory.Inthisregard,weconsideredthefirst8simulationyearsasatimeintervalduringwhichthesystemcanreachalocalequilibrium,allowingtheaccommodationofferstoadjusttheirpricesandnumberstothecharacteristicsofthetouristagents.Thecalibrationisonlyaqualitativeprocessbecauseweknowempiricaldataonlyapproximatelyandwedonothaveaccesstoempiricaltime-seriesdata.
4.5 Fromthe2011GalapagosNationalParkstatistics,weknowthatapproximately43%oftouriststouredthearchipelagoaspartofacruiseandthat23%ofpeopleenteredthearchipelagothroughSanCristobalIsland.Moreover,fromtheGalapagosTourismMonitoringCenterreport,weknowthatapproximately20%oftouristsreservedacruiseandthenarrangedforahotelattheendofthecruise.AsurveyfromtheTourismAssociationofIsabela(ArucariaXXI2007)showsthatin2007nearly12,000touristsvisitedIsabelaIsland,whichaccountsfor8%ofthetotalnumberoftouristsvisitingGalapagosin2007.Usingtheseabovepercentagereportsofuseandgeneralaccess,wecalibratedthemodel.
Calibrationresults
4.6 InFigure6,weshowthefinalcalibrationoutputscorrespondingtothevalueofthecalibrationvector(Cf,CN,CA)=(1,2,0.2).Afteraninitialequilibrationperiod,thecurvesrepresentingthemonthlynumberoftouristagentsreservingcruisetripsorvisitingthefourinhabitedislandsreachaplateau.TheplotsinFigure6alsoshowatypicalseasonalbehaviorbetweenthemonthsofSeptemberandOctober,theperiodwithlowesttouristicpresence.InFigure7weshowsomesnapshotsfromthecalibrationsrun.InthisFigureisrepresentedasmallportionofthetotalsimulationspace:theareaofPuertoAyorathat,asshowedinFigure1,isthemaininhabitedcenteroftheSantaCruzIsland.Aswecanseeinthetimeseriesofthethreereportedsnapshot,duringthesimulationtime,somehotelsareremovedfromthesimulation,whileotheracreated.
http://jasss.soc.surrey.ac.uk/17/1/14.html 8 15/10/2015
Figure6.Calibrationmodeloutput.Intheuppergraphareshowedthemonthlytotalnumberoftouristagentreservingacruise(blueline)orahotel(redline)asfirstoption.Thegreenlineinthesamegraphisthemonthlynumberoftouristagentreservingahotelafterreservingacruiseasfirstoption.ThebottomgraphshowsthemonthlytotaltouristicpresenceinthefourinhabitedislandsoftheGalapagosarchipelago.
http://jasss.soc.surrey.ac.uk/17/1/14.html 9 15/10/2015
Figure7.Snapshotpicturestakenfromthecalibrationrun.InthethreesnapshotsisrepresentedthecentralpartofPuertoAyorathatisthemainurbanareaoftheSantaCruzIslandwherethemajorityoftheisland'shotelsarelocated.Roadsarerepresentedtroughblacklines.Greensquaresarehotel.Bluesquaresarenewhotelcreatedduringthesimulationtimebetweenthetimeofthecurrentsnapshotandthetimeoftheprecedingsnapshot.
Table1:OutputsfromtheABMsimulations–fractionsofthetotalnumberoftouristagentgeneratedinthelast10-yearofthesimulationruns.
Boat-based Land-based Landafterboat SanCristobal IsabelaEmpiricaldata 0.42 0.58 0.20 0.23< 0.08
ReferenceRun 0.44 0.56 0.26 0.16 0.08Whatifscenarios
AirportinIsabela 0.43 0.57 0.25 0.17 0.08Tourismcollapse 0.44 0.58 0.26 0.16 0.06Parkticketincrease 0.76 0.24 0.59 0.08 0.07
4.7 InTable1,weshowtheresultsofthelast7-yearrunaverages.Inthistable,wereportthefractionofthetotalnumberoftouristagentsenteringthesimulationandreservingacruiseastheirfirstchoice(i.e.,boat-basedtouristagent),thefractionoftouristagentsreservingahotelastheirfirstchoice(land-basedtouristagent),thefractionoftouristagentsthatreservedacruiseandthenreservedahotelroomimmediatelyafterthecruise(land-andboat-based),thefractionoftouristagentsvisitingSanCristobalIsland,andthefractionoftouristagentsvisitingtheIslandofIsabela.Wenotethatafterthefirstequilibrationperiod,thenumberandthespatialdistributionoftheaccommodationoffersisdifferentwithregardstotheinitialconfiguration.Indetail,thenumberofcruiseboatschangedfromaninitialvalueof66vesselstoanaveragevalueof47.2inthelast7-yearsofthesimulationruns.Furthermore,thenumberofhotelsinSantaCruz,SanCristobal,Isabela,andFloreanachangedfrom43,24,35and2to34.5,7.3,4.3and0respectivelyforthelast7-yearsofthesimulation.Inthecontextofourcalibrationprocess,thesechangesarenotsignificantbecauseweareinterestedinfindingastateofourmodelthatcanreproducethedistributionoftouristsamongland-andboat-basedoffersandamongtheislandsofthearchipelago,andnotthenumberanddistributionoftouristicoffers.
ABMexperiments
5.1 Themodelresultingfromthecalibrationprocesswasusedtobuildthree"whatif"scenariostostudytheresponseofthemodeltostatevariablechanges.Thethreescenarioswetestedwereasfollows:
a. OpeningofanewairportonIsabelaIslandwithdirectflightsto/fromtheEcuadorianmainlandb. Tourismcollapse–areductionofvisitingtouristsfrom180,000peryear(asrecordedin2011)to100,000peryearc. Increaseintheparkfeesfrom$6.00(USD)and$100.00(USD)fornationalandforeigntourists,respectively,to$400.00(USD)forallthevisitingtourists
Forallthethreetestedscenarios,weperformed1015-yearruns,permittingthesystemtoequilibrateforthefirst7-years,afterwhichwechangedthestatevariablesaccordinglytothe"whatif"experiment.
5.2 Indetail,theset-upofthe3"whatif"scenarioswere–forscenario(a)attheendofthe7thsimulationyear,wepermittedtouristagentstoenterthesimulationdirectlyalsofromIsabela.Togeneratescenario(b)attheendofthe6-hsimulationyear,wedecreasedthenumberofgeneratedtouristagentsby0.75peryeartogetthetotalnumberofgeneratedtouristagentsto112,500after2-yearsandthenwemaintainedas
stablethenumberofgeneratedtouristagents.Tocreatethescenario(c)attheendofthe7thsimulationyear,wesharplyreducedthebudgetavailabletoeverytouristagentby$400.00(USD).
ABMExperimentsResults
5.3 Theresultsofthe"whatif"scenariosshouldbeinterpretedasaninformationabouthowourmodelreactstocertainchangesinthestatevariables.Newempiricaldataareneededandadditionalvalidationdataarerequiredtodeterminewhetherthemodelissufficientlyrobusttosuitablyrepresentthedecision-makingofagentandsomeofthestochasticprocessesassociatedwithexogenousforcesandendogenousfactorsthatcombineincomplexwaystoaffecttouristicbehaviorsinresponsetomarketconditionsandadynamicenvironment.
5.4 InTable1,theaveragefractionsoftouristagentsobtainedinthecalibrationprocessarereported.Wenotethatforscenarios(a)and(b),themodeldoesnotshowanysignificantchangeregardingthereferencecalibrationrun.Scenario(c)showsthatchangesinthemodelstatevariablesgenerateastrongerinfluenceonthesimulationoutputs.
5.5 Wecanassumefromtheseresultsthatifwedonotchangetheindividualbasedcharacteristicsofthevisitingtouristagents(suchasthetouristagents'budget),onlychangesomeaspectsofthetouristicinfrastructure(e.g.,anewairport),andchangethetotalnumberofarrivingtouristagents,thiswillnotaffecttheobservedemergentpatterns.WithanewairportinIsabelaandwithoutanupgradedtouristicinfrastructureonthesameislandthatisabletoattractandhostanincreasednumberoftouristagents,intheshortterm,nochangeinthetouristagentdistributionwillresultamongtheislandsofthearchipelago.ThismayoccurbecauseSantaCruzIsland,duetohisstrategicandcentralgeographicalpositioninthearchipelagoandasthehubofthetourismindustry,willremainthefocalpointoftourismintheGalapagosIslands.Assuch,SantaCruzIslandsubstantiallyinfluencestourismmobilityaroundthearchipelagoand,therefore,willstillattractthelargestportionofthevisitingtourists.Wecanexpectasimilarpatternforasuddendecreaseintourists'arrivals–withoutanyotherchange,thevisitingtouristswillbedistributebetweenland-andboat-basedtoursandamongthearchipelagoislandsinanunalteredmanner.
5.6 Thecircumstanceschangecompletelywhenwealtersomeofthecharacteristicsofindividualtouristagents.Inscenario(c),theincreaseintheentrancefeestotheGalapagosNationalParkcompletelyaltersthetouristagent'sabilitytogainaccesstothedifferentsegmentsoftheGalapagostouristicmarket.Inthislastcase,theproportionofland-basedtouristagentscollapsesto24%ofthearrivingtouristagents,whiletheboat-basedtouristagentsassume76%ofthevisitingtouristagents.Wenotethatinthisscenario,theaveragenumberoftouristagentssucceedinginreservinganaccommodationofferoneverysimulationmonthfallsfrom14,880ofthecalibrationreferencemodel,to8,322registeringadecreaseofabout56%.Clearly,thetouristagentsthatareremovedbythereservationprocessarethelowbudgettouristagents.
Conclusion
6.1 ThedynamicsandchangesofatouristicdestinationlikeGalapagosIslandsdependonahighnumberofcombinedfactorsthat,inmostcases,actalsoattheindividualscaleoftouristsandtouristicoffers.Inthispaper,wehaveshownanapproachtodevelopanAgentBasedModeloftouristicoffersandreservationintheGalapagosIslands.Theresultingmodelis,asusualinanABM,amultidisciplinaryassembly,whereelementsfromtourismmarket,economy,psychology,andgeographyarecombinedtocreateasystemthatrepresentssomeofthecharacteristicsofthetourismmarketoftheGalapagosIslands.ThemodelincludesapsychometricrepresentationoftouristsvisitingtheGalapagos,togetherwithadetailedrepresentationoftouristicaccommodationoffersandtouristicattractionsintheGalapagos.Weparameterizedthemodel,exploitingtothemaximumextent,theavailableempiricaldatasets.Someofthemodel'sparameterswhereusedtocalibrateitandtoreproducecloselyemergentpatternsoftourisminGalapagos.Themodeloutputsarequiteconsistentwiththeavailableempiricaldatasetsshowingthattheindividual-basedmechanismandprocessesdefinedtobuildthemodelcancloselyreproducesomeofthekeyaspectsoftheGalapagostourismmarket.Asindicatedabove,thismodelrepresentsafirststepinbuildinganABMthatwillincludeothersocialandenvironmentalaspectsofGalapagosIslandssuchasfisheries,agriculture,landcoverchange,andmore.Moreover,additionalcollectionandderivationofempiricaldataarenecessarytostrengthenthevalidationofthemodelandtoreproducemoreaspectsofthetouristicmarketintheGalapagosIslands.
Acknowledgements
TheresearchconductedinthegenerationofthismodelwasmadepossiblethroughagrantfromtheMcDonnellFoundationtotheUniversityofNorthCarolinaatChapelHill,USAandasub-contracttotheUniversidadSanFranciscodeQuito,EC.Inaddition,theauthorswishtothanktheGalapagosScienceCenter,apartnershipbetweentheUniversidadSanFranciscodeQuito,ECandtheUniversityofNorthCarolinaatChapelHill,USAforuseoffacilitiesandassociateddatasets.WeparticularlywishtothanktheGalapagosNationalParkfortheirassistanceandcooperation.
References
ARUCARIAXXI.(2007).SATISFACCIÓNDETURISTASQUEVISITANPUERTOVILLAMIL,ISABELA.ISLAISABELA–GALÁPAGOSJULIO2007(pp.1–25).
BAGGIO,R.(2008).SymptomsofComplexityinaTourismSystem.TourismAmalisys,13(1),1–20.[doi:10.3727/108354208784548797]
BALBI,S.,&Giupponi,C.(2010).Ca'FoscariUniversityof,(05).
BENSTED-SMITH,R.(Ed.).(1999).ABIODIVERSITYVISIONFORTHEGALAPAGOSISLANDS.CharlesDarwinFoundationandWorldWildlifeFund,PuertoAyora,Ecuador.
BUTLER,R.W.(1980).theConceptofaTouristAreaCycleofEvolution:ImplicationsforManagementofResources.TheCanadianGeographer/LeGéographecanadien,24(1),5–12.[doi:10.1111/j.1541-
http://jasss.soc.surrey.ac.uk/17/1/14.html 10 15/10/2015
0064.1980.tb00970.x]
CECCHINI,A.,&Trunfio,G.A.(2007).AMultiagentModelforSupportingTourismPolicy-MakingbyMarketSimulations.Computationalscience:7thInternationalconference,ICCS2007:proceedings,partI(pp.567–574).NewYork:Springer.[doi:10.1007/978-3-540-72584-8_75]
CROOKS,A.,Castle,C.,&Batty,M.(2008).Keychallengesinagent-basedmodellingforgeo-spatialsimulation.Computers,EnvironmentandUrbanSystems,32(6),417–430.[doi://dx.doi.org/10.1016/j.compenvurbsys.2008.09.004]
EPLER,B.(2007).Tourism,theEconomy,PopulationGrowth,andConservationinGalapagos.CharlesDarwinFoundation,PuertoAyora,ProvinceofGalapagos,Ecuador,(September).
GEN,M.andYu,X.(2010).IntroductiontoEvolutionaryAlgorithms.(R.Roy,Ed.)(second.).NewYork:Springer.
GONZÁLEZ,J.A.,Montes,C.,Rodríguez,J.and,&Tapia,W.(2008).RethinkingtheGalapagosIslandsasacomplexsocial-ecologicalsystem:implicationsforconservationandmanagement.EcologyandSociety,13(2).Retrievedfromhttp://dlc.dlib.indiana.edu/dlc/handle/10535/2456
GRIMM,V.,Berger,U.,Bastiansen,F.,Eliassen,S.,Ginot,V.,Giske,J.,Goss-Custard,J.,etal.(2006).Astandardprotocolfordescribingindividual-basedandagent-basedmodels.EcologicalModelling,198(1–2),115–126.[doi:10.1016/j.ecolmodel.2006.04.023]
GRIMM,V.,Berger,U.,DeAngelis,D.L.,Polhill,J.G.,Giske,J.,&Railsback,S.F.(2010).TheODDprotocol:Areviewandfirstupdate.EcologicalModelling,221(23),2760–2768.[doi:10.1016/j.ecolmodel.2010.08.019]
GRIMM,V.,Revilla,E.,Berger,U.,Jeltsch,F.,Mooij,W.M.,Railsback,S.F.,Thulke,H.-H.,etal.(2005).Pattern-orientedmodelingofagent-basedcomplexsystems:lessonsfromecology.Science(NewYork,N.Y.),310(5750),987–91.[doi:10.1126/science.1116681]
JOHNSON,P.A.,&Sieber,R.E.(2011).Anagent-basedapproachtoprovidingtourismplanningsupport.EnvironmentandPlanningB:PlanningandDesign,38(3),486–504.Retrievedfromhttp://www.envplan.com/abstract.cgi?id=b35148[doi:10.1068/b35148]
LIGTENBERG,A.,VanLammeren,R.J.a.,Bregt,A.K.,&Beulens,A.J.M.(2010).Validationofanagent-basedmodelforspatialplanning:Arole-playingapproach.Computers,EnvironmentandUrbanSystems,34(5),424–434.[doi:10.1016/j.compenvurbsys.2010.04.005]
LIU,J.,Dietz,T.,Carpenter,S.R.,Alberti,M.,Folke,C.,Moran,E.,Pell,A.N.,etal.(2007).Complexityofcoupledhumanandnaturalsystems.Science(NewYork,N.Y.),317(5844),1513–6.[doi:10.1126/science.1144004]
LUKE,S.,Cioffi-Revilla,C.,Panait,L.,Sullivan,K.,&Balan,G.(2005).MASON:Amultiagentsimulationenvironment.Simulation,1–18.Retrievedfromhttp://sim.sagepub.com/content/81/7/517.short[doi:10.1177/0037549705058073]
OBSERVATORIODETURISMOGALAPAGOS.(2011).Boletinn1sept2011.
OSTROM,E.(2007).Adiagnosticapproachforgoingbeyondpanaceas.ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,104(39),15181–7.[doi:10.1073/pnas.0702288104]
OSTROM,E.(2009).Ageneralframeworkforanalyzingsustainabilityofsocial-ecologicalsystems.Science(NewYork,N.Y.),325(5939),419–22.[doi:10.1126/science.1172133]
PARQUENACIONALGALÁPAGOS(2007).PlanManejoParqueNacionalGalapagos.
PARQUENACIONALGALÁPAGOS.(2012).PNGtouristsstatistics.Retrievedfromhttp://www.galapagospark.org/onecol.php?page=turismo_estadisticas
PIANA,V.(2004).CONSUMERDECISIONRULESFORAGENT-BASEDMODELS.EconomicsWebInstitute.Retrievedfromhttp://www.economicswebinstitute.org/essays/consumers.pdf
PLOG,S.(1974).Whydestinationareasriseandfallinpopularity.CornellHotelandRestaurantAdministrationQuarterly,14(4),55–58.Retrievedfromhttp://www.mendeley.com/research/destination-areas-rise-fall-popularity/[doi:10.1177/001088047401400409]
TRUEMAN,M.,Atkinson,R.,Guézou,A.,&Wurm,P.(2010).Residencetimeandhuman-mediatedpropagulepressureatworkinthealienfloraofGalapagos.BiologicalInvasions,12(12),3949–3960.[doi:10.1007/s10530-010-9822-8]
WATKINS,G.,&Cruz,F.(2007).GALAPAGOSATRISKASocioeconomicAnalysis.CharlesDarwinFoundation,PuertoAyora,ProvinceofGalapagos,Ecuador,(May).
WORLDTOURISMORGANIZATION.(2006).TourismMarketTrendsWorldOverview&TourismTopics.
YIN,L.(2007).AssessingIndirectSpatialEffectsofMountainTourismDe-velopment:anApplicationofAgent-basedSpatialModeling.TourismandRegionalScience,37(3),257–265.Retrievedfromhttp://jrap-journal.org/pastvolumes/2000/v37/F37-3-8.pdf
AppendixA:DataTables
Table2:Parametersdefiningthearchipelagograph.Islandsmarkedwithan"h"areentrancehubstothearchipelago.*Datafromfieldsurveys
Connection Price(USD)* Time(hours)*SantaCruzh-SanCristobalh 30 2.5
SantaCruzh-Isabela 30 2.5
SantaCruzh-Floreana 30 2.5
Table3:NumberoftouristicattractionsassociatedwiththeArchipelago'sIslandsandtheircruises.*DatafromGalapagosNationalPark
Island Numberoftouristicattractions*SantaCruz 17SanCristobal 12Isabela 16Floreana 7Cruises 18
Table4:Firstpartofthetouristagent'sstatevariables1thevaluesreportedinthistablemustbedecreasedby$560.00(USD)forforeigntouristandby$366.00(USD)forEcuadoriantourists
http://jasss.soc.surrey.ac.uk/17/1/14.html 11 15/10/2015
Table5:Valuesofthefivedimensionsofthetouristagent'scharacterstatevariable
Table6:Thefivedimensionsoftheaccommodationoffersfeaturestatevariable
http://jasss.soc.surrey.ac.uk/17/1/14.html 12 15/10/2015
Table7:Statevariablesofthemarketoperatoragent
http://jasss.soc.surrey.ac.uk/17/1/14.html 13 15/10/2015
Table8:InvestmentcostsparametersforHotels.Datafromfieldsurvey
Category AB(m
2) P
m
2(USD) α
First 30 800 100Second 23 700 50Third 15 600 10Fourth 8 400 0
Table9:InvestmentcostsparametersforCruises.Datafromfieldsurvey
Category αI(USD)
Luxury 156250First 143750Medium 131250Economy 118750
Table10:OperationexpensesparametersforHotels.Datafromfieldsurvey
Table11:OperationexpensesparametersforCruises.Datafromfieldsurvey.
>
Table12:Statevariablesofthetouristagentgenerator
Variable Description ValuesPeriodicity DefinethetimeschedulingoftheAgent dailyNumbertouristagentsperyear
Definethenumberoftouristagentgeneratedeverysimulationyear 200000(tourist)
Numberofbudgetclass Definethenumberclassesinwhichthegeneratedtouristagentswillbedividedusingtouristagentbudgetascriterion
7
Numberofageclasses Definethenumberclassesinwhichthegeneratedtouristagentswillbedividedusingtouristagentage 3
http://jasss.soc.surrey.ac.uk/17/1/14.html 14 15/10/2015
ascriterionSeasonaltuning Changethenumberofgeneratedtouristagentdependingonthemonthoftheyear FromGalapagosNationalPark
Statistics
http://jasss.soc.surrey.ac.uk/17/1/14.html 15 15/10/2015
