FLORES: Helping people to realize sustainable futures

Sub-Ple"aryPapen and Abstracs FLORES: Ilelping Peopleto RealizeSustainableFufures... . b y Jeny ' K Yarcla/, RobedMu€eelfeldf Mandy tt ggi6l "na r,ut*oi" Bo,r.qoa' 'Soulern CrossUniveoity, PO Box 157, LismorENSW 2480,Austtrlir E-mril: JvmclNy@u.e&.ru 'UniveisityofFdinbug\ ldayficld Rm4 Elinbursh EI|9 3JU,UK E-nuit R.M*lfeldt@edac.uk this energing netwoft dlmwh worksbopsand lechnical suFort will cnl|ance FLORES by offering a befi€r udeBtanding of the concepq arld by allowing no(e peoplq espociallythosc from &velopirg courthieq to influcnce the &vel@[Ent of FITRES and the issuesthat canbc cxpltrcd wi$in it Kcl,rordr: Docisior slppon syBErL Adaptive modeuing t and usedtemativcsrPolicy alr j/srs Introduction 'waul.cecs, I Ind|rr'orc, s[uy, B..uly, Invcncss M 7lL (,|K E-mril Hag@woddfocsrs.org tRAD, BP 5035,34032Mq@llier CedeL Frrncc E-{!!ril: Bousquet@ind.ft Abstrect Peeopfeusually blol,t how thq *ant their sitmtiotr ,t chatge ta st a hetter futue but dEy dD nof.^ly.ats l<nouthttw to dtarge thcn sihmtion. lnitierivcs inEndedto sec|ttEa better firture do not alwqa wo& as interdo4 and rnay have unintcded si& eFects. Computerrbdek cen help advocatesexlrlote consequenoes of Foposed inidstive& so they catr makc infonn€d s€lectims of altemativeq s€cure in the koowledge draf conseqHces have bccn thorougbty hvestigaEd By eocouragirlt people to cxplore sceMaios, modclsetrpolvet peopleto be mor€ inDor,afive and tcss dqrqldent on technmats, New softwarc solvcs tcclttrical limitadoN, but lhe rEal issuc is ftrt softwarE, hx rder tle pmvision of a supportivc fi"mcwort within wbich peoplccatr cxprEssad €xpadmentwilh idcas. FLoRES. the Forrst t atrd Oricdted Rcsouce Envisiqdng Systen, providcszucha framcwor* to stimulate htcdisciplirwy collaboratiotr bctwe€ resedchus. prEctitimers and clients. A rccerf Fotoq,pe detr|onsfated the feasibility of FI,oRIS. However, FIORES is not aboutsoftw.re; it is about providi[g the mearls to cxplorc the cons€quencs of ahemative scenarios. uhinstely, FLORES is not a physical packagg but a user group &rd orc hteracto||s they have amongsrthernselvB, and with th€ people bvolved in policy-matitrg. Fo6t ring Policiessnd inccrrivcs to Ffinore sustairable fo{Btsy ed bctlcf, laDd wc do not ilwayB achicvc thc dcsired cfrccf |@ rarcly fo@sccdl drc cq$€quffi, ad d|at rho$€ bcst ablc to offar aftamarive views Day bc uDableto contibrrc to ti. decision{rkiDg proccss. This lcads to incffici€n! and somctitrlcscourt..-cfrcclivc' initialivcs. How can wc bctter oqub potcy mater$ and drcil advisors to envisage ftlly tlrc effic{cy arld corsqtrtEces of initiatives? Orc way is to providc a simuLtor that helps people to viBualize poosiblc outcome3 of Foposed initiatives-FL,ORESis rn attcnpt to build such a simularor. Woft oo FI-rORES|is di[ itr progEss, so e lrrditional nefiodsresutts fomnt is irrpfopriaic, etrd we offer e Dar.tive highlighting trials, tibulations and itrsightsgaincd- Whrt is I'IORES, rtrd where did the idea come from? FI-ORES,thc Focst fal|d Oric €d Rlsotlrce EtrvisioNrilg Systafl airDs to ilnplove otf m&rsbtrdhg of lmd ulc pafiettrsin time ard space,espcciallyfu fqEs€d ledscryes, ad ro frcilime rigcons analpcs of policy optioos intendcdto manipuLtc6e3e pafiems. Thc idea for FLORES aros€ ftom several infuiarives,amo|rg which were the desircs to cx€atca platforE lbat would allow r€scrrchers to htega$e lhcir r6€arch, to mEkeit possible for then to u,o* togdher to rcved lhe bigg6 pictuG, and to Fovilc lhe ability to test rigomusly within a realisric fr"mewsk. Theseremrilr imponant influ€trc€s in the dcvelopmentof FI-rORES.Accordingly, FITORESis spstially .xplici! and operat6 at SuLPlqrary Papen and Abs.raca thc lr.&c{pe scalc, spanningboth forcsr and sgdcultnrxl lends. AgdcUlitrel lsds aDd villa8cs forrr a criticel tornpot€ot of 0E lurdscqe aad nust bc modelled to firly urderstadl thc procels6 at vqt h and trear ihe forcs, relyitrg o{r rcar-view mirrors for infqlutior. Up.lo-darc infonation (cf. loo&ingout of {rc sid€ mirrds !o $ee|[e $ide ofthc rood) hc!6, but ue cen ol y &ivc sofcly Bllcn we catr lrc f6*ards (cf. pr€dictingfir$rc otrtco@es). We cdr also draw a uscfitl cotrtrssrwfth air lavel, whqr nak6 iir rrdlsFlt so saft and pilot e'tor so red cood design, carcfirl phoniDg, diligc maint odlcc ud cqryelent $p€rvision .I! frctoE tut pilot F.inirg is cocisl. BGforEcrcw mcorbcrsra&cthe c@ttols of . cofirrcrcid.i ircr, dry *ill have mdid tbe dEory of fiitt( Frimd h light airEran, spctrt hous iD a fligbt simulatm, .nd flowr with morc cxFiclccd coleagucB-ltpy loos bow to rcad ltc hdicafors, what cv6y butl@ aDdcvcry lcvcr doar, od u,tan dd how ftcsc conrols should bc usqf ThEy how instincrively how to rcspoodwhen sonEthing goes wrq& d nLr b do if thc ptaoe dcvirtcc frm its pLm€d oorlsc. Atrd lt y rrr€ly nccd to {se thair rrainiDg b€c.rsc our loon'ldgc of fligh h.s h.en sy hdi*cd irlo el au@ilot lh.t L&esc.t€ of md sihutions- The basic coicc?ts of this wort rtc trol n€w; what is ocw is tic wf,y conceptsdE integrared ald 4plied- FLORES seerB mo6t cloecly rcld.d lo sqt by Bousqr*i .t ol. (1993, 1994), wlro corsrru.ted a multi-ngent simulrtioo (MAS) model of rn inlrod fslEy in dre ccotal NigEr Dcltr as r basfu for fa1|sing dtucl|ssioa €v.h|.tirg oDtio6 .rd Thdc is en fomulatiry rccm.nd.tios" st bctwcen FITRES iDlcrl*ing cotr ad MAS: both erc cmceoed wilh 48€ra that crn nodi& .rd r€spod !o rhcir avirwnen! ht lhc €qh|sir difftrs. G€oe!'aly,MAS dcdFb !o fird dE siryk*t sct of rulcs tiat can rqrod-c e Fnic| rr per fifln a d.fipd scqErio. ln css€ncc,tie u$d Sucgion f@ MAS is: whd rrr thc nrlcs thrl mi8bt cr(pLir thtu Frle d|rr vr havc ob6.rv€d? FLORES; ffii&ls lbc convcrrc: giycd slBt sc trorv aboorhroar bchavio||r, cn we Fcdict fimrc orrcmcs for r rugc of sceoaiod?Crcrelly wa do rot l@9 whaf frfrtr€ oufcoic dor d hok lit , .f,c.pt itr a fcw sp.cific c.6.s ttat ru)' bc uled b rd thc trrodel.FLOR.ES.&o rl otliccs that pooplc nay heve cqplq rtrsm fc dEit bchaviqt, rod stErpts !o tlpr€s.'rt o|r prts€nt urrlcrdatdir8 of dBc rEaroc, r$ar lba lc+irg tbr sinplerr rulc$ fbt rny rr7ro6rc a giva paern Now cdEa$ this uith or|I nmagcrncnt of foll tr: . Do c/€ knorv rfrat lo do ybEo Oitrgs go wrqg? . C.tr Be Etl rtco thirS6 rrc begimilg to go vrqE? Do vc l.'N vhich cdols usc !o chag! 6iDgs? we c.t Do P€ hos, Pt{ lbc codtols arc, etcr€ to trrd lh.rL erd how to .clirhb 6dn? Cen nE rEcogniscaDd interp(lt frc indicalors? xlb' d@'t re h.vc an 'ruopilofl to give rdvicc? Why do ve needFLORESI? Wr oficr sdE .n fogi6 to illus.fic why FIORES is.iAdhlt Arr'ole who ber plrycd Firh Bflts'. lbc Bc.r Gm.f o . 3fudlr meag(r[enf gaE slnrtd epprecie&rtc rcd fq W-to-datc itrfofm.tid" With Fish Ba s, a t.rE rbout or|.ifuble resource utilization, play€B oftE'l &stry a fi*ery bccaurc tlrca rcly m infamerirl fiom e prcviori genre cycle- t is otrb' ei€n ph!,tts ktrtr !o Fldict qlllqrf ard fr!ture filh stocts d|,] dEf crn .chiele a sur|ailEblc ourcone. Usirg old fufdrmfi''t for rtsout,cc rnat|agarnantis fte &ivitg a car wiftout forward visioo, aod Why fu it that so rnany .moDgd thooe who rnakc i|Dlrcht decisioos abour the wqtd's forEsls haw newr raised . IEe. tended a grlde4 galhemdfood from lhc fdffq or uted a sirnulilG to erplorc thc irplicalions of ar impendingdeci6in? wo ld a forEstlandscepc simulitor makea difierelG? I hF/www.uoh.erfrr/iFsrt t FishBaifl|ld ' lt$J Gdning.lrit..dltpn^ool/tc.r-htnl 124 Sub-PleurryPape^ and Abstracts The computer game SimCity' provides an inl€restinganalogyfor a user inlerfacethat we would like to developfor FLORES.The Max$ Corporationprovidcs a simulator itr the lolm of a gamc. The game ofrers the player an "aerial vief'of a city, a menu ofpolicies and incentives (c-9., expendituie on educarron, tt'.nspon, sanitarion, etc.), and indicatoG of pcrfomance (e.g., unemployrnent, GNp, Fllution, etc.). Scenariosare available fteelv on rhe Inlernet', and range liom rcal cities to fantasies- was prcduced (Muetrelfeldt a al. 1997). Significant progress' was nade during 1999, and at the time of writing, the FLORIS team arc about to embark on atrotherworkshop to test lie applicability ad ability to adapt the Rantau Pandar veNion to a Zimbabwe situation. Pmgressto date is lbe outcomc of collaborationbetwe€nthe Insrituteof Ecology and Rerouce Mamgementar the UniveEity of Edinburgh and dre C€ntlr for Inlernational Forcstry Resc{rch (CIFOR), and a group of fifty people'wbo clotributed to a model designworkshopheld itr Surruta in t999 with suppon fiorn lhe UK Depaitne for IntemationalDcveloprnent. I: ii: i,:, | '.. In FI-ORIS, we replace the ciryscjlp with a landscapeof forest and non-forcst land. Its menuincludesa rangeofoptiotrs to manipularc the forest and land usc pafterns and indicators cluld include Frfonnance biodiversity and nEal povcrty. It must l|ave a stong frctual b6sis, and must be able to bc custornisedto suit different si$rations.It will: . . . synthcsite €xisting knowledgeand identiry gaps and other deficiencies; expr€ss pres€nt knowledge conosely, complet€ly, explicitly and unarnbiguouslyas a model; cRate a fimework to Dmmote collaborative hterdisciplinary resemch: provide a basisfor shong empirical tcsts of hypothes€srclating to land usepolicy; o create a platrning tool to allow plamers and policy makers to cxplor€filnre sccmrios;and pmvide an educational garne to rmprove gercral knos,ledge of tropical forcsr envircnnents. Wh{t has b€en Achieved? The prEsenlversiotrof FTORES is still rather simplistic, but providesthe basisfor oo-going work over a loog pedod Tbis platfomr is noq and must not be, a 'black box", opaque to participa s. It is not enoughthat it should be tronspare ; it should be er ightenh& and should enEnwer paniciFtrts to make bett€r arElysesanddtaw nror€rev€alinginsightsttBn O|€ycould woddng in isolation-We havctried to provide rhis, atd hoperhatit will be usedas a basisfor testinga wid€ rangeofpropositiotrs, ad will be modified as neccssaryto make these tcsts and incorponre findings itrto the modcl- We must begin with sin{rle models, and should progressivelyenrich these as we rcfute inappropdatc simplifications- Models exc€l at cxpoGing counter-intuitive corB€quercesof simplc assumptions_ Even if idtial gototypes of the model are of little pr-acticalrclevance,lhey may offer vaturblc insights, and their main pupose may be to focusquestionsrathcr thanto provide atrswers. The challengeis to consrnrcta framewort that is bload enough to accommodatea wide variety of prcpositions, and sufficiently accessiblethat resetrche$ from a rangc of disciplines rre stimulated to collaborateard test thci. pmpositionsin this integratedway- The FITORESconoeptwas dcvelopeddurmg 1995(Vanclay 1995),but it x,as not udil 199? that work began in eamest and a pmtotyper ' S€ehtts:/lwww,maxis.com http://www-simcity,convexchange/erchang€-hrnl andhfiD://w.sd0m-conr/cities hnp://*ryw.ed.ac-ula.-€bft281/fl orcvvetsionl/indEx. htm htlp://heli06.tto.ed.ac.uk/ienr/fl or€s/RadauPandd, /index.hunl ' s hrF://wTw-cgiar.orglcifd/esearcvfl fticipants.htnl o orcEfl oreslx SuEPlenary Papers and Ahstracts Fortunately, many such models alrcady exrst (e.9., Vanclay 1994, Anon 1997), and somc arc amenable to calibmtion and iniegration wilhin ihe FLORES framework, How does itwwork and will it give the Right Answers? FLORES rclie.s on four basic assumptions, namely O|at: We have implemented FI-,ORES in AME to minimize lhe amolmt of computer code' in tlle hope thar we can ..9"g" potential participants who are not conversant with computer languagcs. AME!, the Agroforcsry Modelling Environment (Muetzelfeldt ad Taylor 1997), has a graphical interface that nakes the model acccssible lo res€archerswho are not fluenl i|r computer prograrnmin& while allowing access to the underlying code. Thus it offers a powertul and flexible platform that does not exclude less compuler-lilerac participarts itr lhe projecl. Therc are olher sdvantagcs itr using AME, some of which include the ability to: l- Land use pattems arc created by dc,ors, hdividrals or goups of individuals who collaboratc as families. clars" associatiors and corpomtions. 2. Thes€ acto6 make rationat decisioDs based on available informatior\ obligarions and €xlrectations, social as well as economic. Note that an act<'tr'spercepttutn is wh^r influenc€s decision-mrking. 3. When choo6ing an activity, actors erplore all options available to ther4 wilhin the constraints imposed by rcsources 0and, time, capilal, etc.). larcwledge. ard theil comfort zone (cultural attacbnerts, willingness to attempt novel activities, etc.). representrclationshipsas simple sketches, malbernaticalequations,or assetsof rules; substitutealtemativemodelseasily via rhe Windowscli&-and+ag facility; and 4- Actors tend to underbke activities that maximise exFcted ben€fits or miniDis€ anticipated dsts to thems€lves and their beneficiaries (frmilics, clans, sharcholdeG, etc.). It may be possible to model bolh benefir-sceking and risk-avoidhg behaviour by considering fisk-adjusted benefits. . create customised uscr interfaces f,,ith soffw t'helpers" fiar can be develop€d independendy ard 'Uuggcd in" later- So what doesit mern for Resource Mrnagers and Planners? Too many models languish, utder-utilise4 because drcy do not satisry the needs of potential uscn atd becaus€systemdevelopers did not explicitly conractclientJ,ascertainthelr needs. and stimulatc their i erEst To encourage Wtake, potential users must be involved in the deyelop[r€trt of the rnodel. Obviously, us€rstnay not bc irterEstedin ell aspectsof model designatrd constnrctiorLbut thcy shouldhavethe opporumityto participste in s?ecification and design of ihe us€r interface. Ir is not enough to ask lhem what they want and how ttey want it, TeaEl members have to etrgenderenthusiasmand involvem€nt lhrough munral understanding al|d mllabomtion. This m€ansdlat lhe model has ro be explainediD an accessibleway, and that simple prototjDes and mock-upsneed to be built so that ideas can tre derDnstiate4 testedand modified. The constraints implied by an actor's comfort mne and previous experiarrce Inean tlrat many aclo6 consider a ralher small number of activities, ofted only tho6e done in $e past, plus a few nex' aclivities puGued profitably by neighbours. However, there ar€ usually a few imovators who considei an extended list of activities and rnay attenrpt a div€rse mnge of enterprises. Typic|lly, innovators are morc willing to attempt isky enkrprises lhen therr morc consqvative fellows. Disposidon fu only one determinant of'willingness ro accept risk, and age, assets and income also fcature prominently in many explanatioDs. Decision-rnaking by acto6 is just one component of FLORES, and seveml other $rbmodels arc needed to predict the growlh of trees and crops, changes in the soil and water balanc€, interactions between key plant and animal s?€cie\ and olher ecosysk{n processes. ' S€e htlp://w.simile.co.uk 726 Sub-PlenaryPapersand Abstracts FLOR-ESwill provide a rarge of ourpursto slrit dilferent user requircments.One output will be the forEstedlandscapeofa SimForcsr implementalion.One gre{ contribution thal infomlation scierrce could make for conservationafld wise use of forcstswould be lo provide a virtual r€ality interfacefor forest managemmt plaming (Vatrclay 1993). This could allow a ministcr and his advisorsto put oo a virtual rcality heads€laod take a "magic calpet'' ride over a forcst estate-They could observethe spatial pafiem of th€ir forest and watch how it changesover time, and lmder dilferent sc€ ario6. They could 'toom in' to cxamine particular issues,and stard back to get a'l overall perspoctive.Thc technologyto d,othis exists, alrd it is pos.sibleto link for€st inventory systems,growth modch geognphic infonnation rystemsand vinual r€ality systcns in lhis way- However, it has oot beendon€ at this time, and awaits fi|rther software ard hardwarc develoFnent to nrake it mor€ allordable. In developing FLOR.ES,wc havc been mindlid that the cve$tual user inerface may well be a virtual rcality syst€n|'and we shoulddctbeiately designan openand flcxible sysremthat doesnot fq€close this po$ibilrty. However, thc Simcity-style inMace is adequatefor many alDlications,and would be particulady us€ftI for cducationalapplicatiotrs andgeneralinfomrationdisseminatiorl What's next? When will it be Useful to Others? Thqe are severalspecific probl€msthal need to be addrEssedbcforc this model can be realised as anything more than a sirnple prototype. Ma[y of these challengescan be addrcssedas s€paratetssks, and ar€ amenablc to rcsearchby otheG,including shrdents.Sorne ofthe more obviousissuesale listedbclow. In ti€ proposed model fodnulation, the underlying firnctional rclationships rnay be relatively simplc, hrt the daia rcquirem€ntsarc mther denranding- Most utility tunctions appearinnocentenough,but they rcquire a lot of data: anticiparcd yields and prices of all possible cmps under a rangc of situations, detailed tenure and demographicdatz, atrd a good understading of the socio-economc culturc of the community. This is a rnajor unde(aking, and may be onc limitation of the model-we envisagethat initial prototypeswill be resElcted to a limited gcographic area, allowing a completec€nsusof all inhabira s for thorcugh model testrng- However, subsequentopcrational implementationsmay samplc only s€lected actors to reduce the burdcnofdata acquisition-Crcp yields may be infen€d from models, but prices alld €lasticitiesmust be gleanedfrom ficld survey wort. This hsk may be poniculady on€mus for non-timber forcst products such as medicinalplatrtsSuperficially,the model app€arstraclable,bul it involves rltany chrllenges. ls it r€ally possible to qu&ti& the social profile of all actors in a community in srfficient detail to providc rn€aningli prcdicrionsftoln a simple utility function?There is no clear answer,and only an cmpirical test can elucidate if nunerical approximatiotrsof conplex social stuctures provide atr adequate basis for plaming. Two finther issues for meliodological r€searcharc evide$ at lhis stagc: whcthcr to model individual actors or classesof acloi$ and how to quadiry risk ard willingness of rctors to acc€ptrisk. Both are central to the FLORES approach,and in both caseE the issri€ is whcthe. lhe preliminary approach is a necessary and suffcienl repres€ntationof rcality. Therc arc some advantagcsin modelling irdividual actors:it is conceptuallyelegant .nd facilitates empncal testing, but it inpoc€s a substantial compuradonalload. Simulationbasedon a few classesof actors (e.g-, classifiedby age ard gender)x,ould sp€€dW simulations,aod rnay but it is oot clear easedatainplt rEquirEnrenfs, if this would lead to the same result as individual-basedmodelling.The issuernay bc b€st resolv€d through empiricat nials aad s€nsitivitytesrs. It is pres€fldy assumed that a[ actor's willingnessto acceptrist canbc quanlifie4 in pan ftmugh the histonc mriation in bercfits accruing from e particular activity, and from thc actor's age, tangible assetsard incomc. However, lhis ?F,sumptionwarrants closer scrutiny since attitud€sto risk lrave a major influenceon land usedecisions.Our ability to quafiiry risks and attiodes to risk will have a major influmce on the accumcyof FLOR-ES Dredictions. SuLPlenaryPapersand Abstrocts every parameter,atrd all assumptiotrsto !e how much influ€nce rhey hayc on pledllod ourputs-This is useful informationltat catrhc usedto direct funh€r developmentof a hod4 with a lower Fiority assignedto paratDaiqr and assumptionsrhat have lisle inlluencc o predictedoutpub. Satisfactory walas to value the intangibles involved with land usedecisionspocea rnajor challenge.One pErticularaspectthat n€€d$to be addr€ss€dis how lo valu€prEstigc-Pr€slige rnay take many forrns, and ltray explain land purchases at pric€s irconsistent with poduction (c.g-, prestigeof owning a bigger €state),herd sizes(e-g-,prcsige oflarge flocks l€3ds to overstocting, even though $naller flocks rnay ofrer equivalentrenrmsand lower risks), and po$s€ssionor productioBof cefiain elns. Thoroughbenchmarktcsting is anotherbig job that requires planDing and pepration It requirescoryrehensive data about a seriesof sites for at l€asttwo points in tine, pEferrbly over a realnnable infervaI. I&ally, the sibation at sorn€sites shouldremainmofE-{rless urchange4 wfiile subEtantialcharyca should be evidcnt at other sites. TlreE arE always diffcxlt issucsto be addr€ssedifthcsc sites involvc oly passive nonitorin& dd empiricaltestsarc stEngth€nedif exp€rimeofil data are available,In agriculturrl situatiuls; it is clrstonrary to use pair€d and rcplicated crpcrimcnB to codlp.Ie trEallr|enfs agrid control plots, Such data d€ Ino'r€dfficult !o obtain at the landscapescde atrdwh€n peopb are involvc4 so g€ater ingenuity is requirEd Suvey datapo€esl'ecialgoblems, sirce rlley factorsmay vary atrdit canbe dimclrlt b Datc reliable inferences-Ifl thetry, it is possiblclo conductexperimentsto galh€rrigomus dataio test FLORES, h therc ar€ €thicsl qu.sid6 that would needto be coNid€redcarcfitlly. Fc exarnplg ir is feasiblego ro r vilagc ard buy locally producedgmds at Fices hthcr the the prevailingme*et refe, and watch horr.d|e . Fottunot€lY, liis exp€rirnentis not nec€ssaD/, becausein ttrdy developing countrieg governrnenb cond0ct such'exFirnents" aI tbc tirne. For itrshcc, rcw hidges atrd roads can mrkedly changc transport costs. Thus dre dafa rcquircd fotr nodel testingmay be obreinedhy stategica[y choosingandmonitoringselectedcomnruniries over an extend€dperiod. A fifihct challengefor later versionswill be to nrodel selected species interachons in both plaDr and animrl species, especialy for apparenflypivotal or keystonespecies.It is not sufficie ro mod€l rh€ food web- becaus€ energyflows arc onb/ one of the aspocts.It is alsoimpdrant to crnsiderr€latior$hipssuchas myconhizal and other slnbiotic relationships, prollination snd fnnspon of sceds, microclimate end other rnodificalions of the envirmment lbat may facilitate the esrablislunentofplant and adrnal spccies-It is probably irqrossible ro rmdel all of these relrtionships ;n a tropical folrst, but it is importa ro recognis€and inclu& tbe pivotal rclalionshiF in our model. A FITORES-typcrno&l is eaErro conccivefor a srnall village, wherc we can simulateev€rJt individual actq. Howevcr, wb€n lve scile up oul efforts to Inodel laBer landscapes,it trlay becomc irrpractical !o examine decisionrnakint by all actols, and it Inay be necessary to extrapolatefrorn a sample of ec{ols- Thc choice of sa$ple may be critical ro th€ outcome,atrdsuitablesamplingstrdtegie,s must be investiqrcd before the appmach can be scaled-uplo the Fovirrcial d MtioDal level. A crucial pan of this invea{igationu.ill be to idcntiry the minirnum ess€ ial ser of Fime determinants. Perhapsth€ best t€st of a nodel is how rY€ll can the modsller aDswerdre questions'What do you knov now that you did not how beforc?' and 'How can you ftxd out if it is true?'. FLORES has nrany limitations, but il providesa fertile test-bedfor ideas"and ofiets arple scoF for firthering our krowledgc of poticies, inc€ntivesand land use pdtems in forcstedlatdscapes.We needlhe product tnd we needthe process.We need!o bdng toge6€r scientists fiom divers€ disciDlines to wd* FLORES seeks lo Drovidc a framewsk for testingand rcfining ideas-This m€als that tb€ basicftaD€work ofFITOR.ESmustbe carefuly tested. and lhat baseline data should be acquir€d for detailed empirical testing. Two components of tlres€ tests x,arnnt sp€cial attentionand prEparation:scnsitivity tests and benchmark Esrs (Vanclay and Skovsgaard 199n. ld@lly, a thorowh Fogram of sensitivity t€sting should examinceach input, 128 :* :*: Sub-Plewry PawB a d Abstracts towards a coll|mon goal- We also needto add morc rigour to forcst policy rcs€arch.FITRES cenhelp rEaliseit. Where can I get it and how crn I us€ it? FI-ORES, its docurne ation" and the AME softwarc are availablc freely via dE intemet fiqtr http/lElios-bto-edac-uvien/florex/You will Deed e rc tun ng Linux or Windows 95, 98 or NT. How crtr I help to mrke it bctter? FLORES is a continuing res€archprojec! thc produ.t of clos€ collabmation by nrany individualq and wc invite othersto participare. For mdr infomEtion oo the cunEnt statr|sof FITRES or ol how to beconre involvedcontactoneofthe authorE- Acknowledgements FLrOR-ES is drc rEsrlt of collaborationbetwm (5Ot) individuals .nd oany s€verdl institttions Spocial dEnks are due to Joh[ Palmer and lie uK Dcpartrnent for Intemalional Dcvclopment for their n{Don. This pnblicatiotr is an oulput ftom a rese{Ich pmject findcd by thc Depadrne for IntenralioMl Devclopment of d|e United Kingdon Howcver, the Dcpadrnent for Int€flational Developr€nt can accept no r€sponsibilityfor atry hform.tiotr Fovided or vicws cxpressed. R7315 FITRES Model DesignWork5hop- References Anonynrous, 1997. Crcp Yicld For€casting Medods- Procccdings of 0rc s€rrliffr, Vileft'dnche-$r-Mcr- 2+27 Ocr. 1994. Eurosht JRC-ISPRA.DC Vl and FAOECSC-ECEAEC,Luxanbolrg. Bousquct, F., Cambier, C., Mullon, C., Moril[4 P., QuetrsiereJ. and Pavc, 4., 1993-Simulatingth€ tnteractionBetweena Jounal Societyad r RenewableResn,UtrceofBiological Syste s,vol. l,W. 199-214. Bousqueg F., C$bier, C- and Moi'in4 P-, 1994.DistributedAnificial Intelligenceand Objcc kicntcd Mode[ing of a FisheryMathematical atl Conpuer Mode ing, vol. 20, pp. 97-107. Muetzefefdl R.I. and Tayld, t., 1991. Thr Suitability of AME for Agrofordtry Mcd.litng. Agrolqeslry Forum, vol. I, ro2,W.7-9. MueEelfeldt R-I., Taylor, J, and Haggith,M., 1997- DcveloprEnt of pFLORES, a Pmtotypc FLOR-ES Modcl Pmgress Rcpdt, Dec€rnber 1997. ItrstituE of Ecalogy and Rciouce MaMgemen! Edinburgh. Udve$ity of http://www.cd.ac.uvtbft28/fl ores/rcp9?12 ^Ep97l2.htttr Yield Varclay, J.L, 1993. Irleotoryd Fdcst Matrag€menl Pt€dictior for Natunl In: Thwaiteq LN- rnd Schaumber&B.J. (e6.), AusualasiatrFdestry and the Globol Environrn€nt. Institut! of Forcsters of Aurtrali4 Brisbane,pp- 163169Varclay, J.K", 1994-Mo&lling Fs€st Gowdr ad YieLrl Applicarionsto Mixed Tropical Forcsts. CAB Internatiorl walingfon4 UIL xvii+312p. Vanclay, J.K., 1995. Modeuing to ExploE I{xl Use Pattens at thc Fq€st Edge: Objectivesand Model Design.In: Biming, P., Bridgmarl H. ard Williams" B. (ede-), Proceedings, Intenational C-oqress on Modelling and Simulation (MODSIM95), 27-30November,Unive6ity of Newcastle, NSW,pp. l: ll3-116. Vanclay, J.L and Skovsgaard,J-P., 1997. E\raluathg For€s Growth Mod€ls. EcologicalModeni'g,eol- 98, pp, r-12-