Landscape and Urban Planning 139 (2015) 115

Contents lists available at ScienceDirect

Landscape and Urban Planning

j our na l ho me pa g e: www.elsev ier .com/ locate / landurbplan

Research Paper

Identif es place-b

Greg Browna,, Kelly de Bieb,c, Delene Weberd

a School of Geography, Planning, and Environmental Management, University of Queensland, Brisbane, QLD 4072, Australiab School of BioSciences, The University of Melbourne, Parkville, VIC 3010, Australiac Parks Victoria, 535 Bourke St, Melbourne, VIC 3000, Australiad Barbara Hardy Institute, School of NBE, University of South Australia, Mawson Lakes, SA 5085, Australia

h i g h l

Identies a Measures s Implement Spatial res Need for m

a r t i c l

Article history:Received 5 JanReceived in reAccepted 5 Ma

Keywords:PPGISVGIParticipatory mPublic landsStakeholder an

CorresponE-mail add

(K. de Bie), del

http://dx.doi.o0169-2046/ i g h t s

nd classies stakeholder perspectives for public land management.patial stakeholder preferences for access, conservation, and development.s a model to identify level of spatial stakeholder agreement.ults were sensitive to method of aggregation and stakeholder weighting.ore research that integrates stakeholder analysis methods with spatial data.

e i n f o

uary 2015vised form 4 March 2015rch 2015

apping

alysis

a b s t r a c t

Public lands provide signicant environmental, economic, and social values to society across a range ofclassications and tenures. Stakeholders representing multiple interests are presumed to hold differentmanagement preferences for these lands. The purpose of this study was to demonstrate how stakeholderperspectives can inuence place-based management preferences for public lands. We developed a multi-dimensional public land preference scale and used cluster analysis of responses to classify individuals(n = 1507) into stakeholder groups using data collected from a large public participation GIS (PPGIS) sur-vey in Victoria, Australia. We analyzed the results of the two largest stakeholder groups (identied asPreservation and Recreation) to assess their spatial preferences for public land conservation, access,and development. We developed a method to assess the level of spatial stakeholder agreement, with theresults identifying geographic areas of both agreement and disagreement between stakeholder groups.To determine the effects of unequal stakeholder participation in mapping, we performed sensitivity anal-ysis by weighting the responses of the Recreation stakeholder group to approximate the mapping effortof the Preservation stakeholder group. The place-based management preferences changed signicantlyfor conservation/development and improving/limiting public land access, while preferences for increas-ing/limiting facility development were less sensitive to stakeholder weighting. The spatial mapping ofstakeholder preferences appears effective for identifying locations with high potential for conict as wellas areas of agreement, but would benet from further research in a range of land management applica-tions to provide further guidance on the analysis of stakeholder group responses that result from diversestakeholder group participation.

2015 Elsevier B.V. All rights reserved.

ding author. Tel.: +61 7 33656654.resses: greg.brown@uq.edu.au (G. Brown), khun@unimelb.edu.auene.weber@unisa.edu.au (D. Weber).

1. Introduction

Public lands provide a diversity of environmental, economic,and social values to society across a range of public land cate-gories (Brown, Weber, & de Bie, 2014a). The laws that govern publiclands often identify their purpose, but rarely provide specic guid-ance on how to balance the multiple and sometimes conicting

rg/10.1016/j.landurbplan.2015.03.0032015 Elsevier B.V. All rights reserved.ying public land stakeholder perspectivased land managementfor implementing

2 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

uses. Management of multiple-use public lands can generate con-troversy as they provide for recreation opportunities as well asresource usout the wochallenged ment (see eWestern &

Public lasive concepof a socialspecic legtutes the pmanagemetutes publicrelated to access, the opment of vdeveloped ament prefemanageme

Individupublic land stakeholdercorporate wor individupurpose (pmanagemeorganised oin a particution in socito the levelsize or aggrrelevant to egories of society (Greral stakehare affectedland stakehof stakeholpreferences

Reed etysis approaresearch. Thders, differinvestigatinfor identify(e.g., using we rst idemanagemeanalysis to gnative approrganized spublic landAustralia.

A commmaps stakpower/inuses do not aassess leveeffort. Rathlate into spland manation. We (PPGIS) melocation-spgroups.

1.1. Public participation GIS (PPGIS) and stakeholder analysis

lic participation GIS (PPGIS), participatory GIS (PGIS), andeered geographic information (VGI) refer to methods andses tal, aBrow). PPng agped cng meen sd song teifferes. Thcally

geogble, aividuappiatioe, mmenGIS/Pe usther

st PPd to oldersbergmay s, inteolderlly poups

usu Thusrveyng.

cha persted wviduudiesatialace-s (n =per af ec

stroor, Bwith PPGIatialorreltive uartic

usere hationpationt de(e.g. iffer

ecose and extraction. Even national park agencies through-rld, an important sector of public lands managers, areto nd a balance between conservation and develop-.g., Budowski, 1976; Newsome, Moore, & Dowling, 2012;Henry, 1979).nd management is inextricably linked with the elu-t of public interest that seeks to advance the welfare

collective over private interests. In the absence ofislative guidance, the determination of what consti-ublic interest falls on the agencies responsible for

nt of public lands. The resolution as to what consti- interest manifests in small and large-scale decisionsresource use versus protection, the level of publictypes of recreation opportunities provided, the devel-isitor facilities, and regulatory control. In this study, wen exploratory scale that measures public land manage-

rences across these multiple dimensions of public landnt.als and organizations that share common interests indecisions are commonly called stakeholders. The term

has numerous formal denitions. For example, in theorld, Freeman (1984) dened stakeholders as groups

als that can effect, or are affected by the organizational. 25). Another denition applied to natural resourcent considers stakeholders to be any group of people,r unorganised, who share a common interest or stakelar issue or system. . .who can be at any level or posi-ety, from global, national and regional concerns down

of household or intra-household, and be groups of anyegation (Grimble & Wellard, 1997, p. 176). Especiallypublic lands, stakeholders can include the nebulous cat-future generations, the national interest and widerimble & Wellard, 1997). A key distinction between gen-olders is those who affect decisions and those who

by decisions. In this study, we do not identify publicolder groups a priori but provide for the emergenceder groups through an inductive analysis of individual

for public land management. al. (2009) provide a typology of stakeholder anal-ches for participatory natural resource managemente typology consists of methods for identifying stakehol-entiating between and categorizing stakeholders, andg relationships between stakeholders. Our approaching stakeholders differs from the reviewed approachesfocus groups, interviews, or snowball sampling) in thatntify stakeholders based on preferences for public landnt collected through a survey, and then use clusterroup individuals based on these preferences. This alter-oach is pragmatic given there are many individuals andtakeholder groups whose interests span the diversity of

types found within our study area, the state of Victoria,

on stakeholder analysis technique identies andeholders in two-dimensional space consisting ofence by level of interest (Bryson, 2004). Our analy-ssess stakeholder power/inuence and only indirectlyl of interest as indicated by participatory mappinger, we focus on how stakeholder values trans-ecic place-based preferences for three key public

gement issuesaccess, development, and conserva-use data collected from public participation GISthods that identify spatial preferences to determineecic agreement or disagreement between stakeholder

Pubvoluntprocesregion2005; cationsplannidevelosamplically bto builmappihave dteristicspeciteeredassemby indtory mthe crepracticrecruitthe PPies havwhile o

Mobe saistakehSchlosPPGIS vidualstakehditionaand grdo not2009).than susampli

Theholderassociaof indieral stnon-spping plvisitorVan Ritions oversusbehaviences three non-spwere cextracwith prelated

Theconcepparticiagemegroup show dtion ofhat generate spatial information for a variety of urban,nd environmental planning applications (see Brown,n & Kytt, 2014; Sieber, 2006 for a review of appli-GIS has typically been implemented by governmentencies or academics to enhance public involvement inountries for urban and regional planning using randomethods and digital mapping technology. PGIS has typi-ponsored by NGOs in rural areas of developing countriescial capital using purposive sampling and non-digitalchnology (Brown & Kytt, 2014). These two methodsnt origins and applications, but many common charac-e key difference is around the denition of participation,

who participates and why. A related concept, volun-raphic information (VGI) refers to systems that create,nd disseminate geographic data provided voluntarilyals (Goodchild, 2007). The general term participa-ng describes any process where individuals share inn of a map and would include PPGIS, PGIS, or VGI. Inultiple sampling methods may be used for participantt. The language used to describe participants framesGIS/VGI process and explains why some PPGIS stud-ed the term stakeholder to characterize participants

studies have continued to use the term public.GIS/PGIS/VGI processes that inform planning mayinvolve stakeholders, given the broad denition of

that includes those affected by planning decisions. and Shuford (2005) describe how the term public ininclude decision makers, implementers, affected indi-rested observers, or the general publicin other words,s. However, stakeholder research and analysis, as tra-racticed, involves methods that identify key individuals

with interests within a specic policy domain andally include broad-based social surveys (Reed et al.,, stakeholder analysis is considered narrower in scope

research and involves purposive rather than scientic

llenge for analyzing PPGIS/PGIS/VGI data from a stake-pective is the ability to differentiate mapping behaviorith stakeholder group afliation, from the high degree

al variation found in general mapping behavior. Sev- have found that participants translate their personal,

attitudes and values into behavioral choices when map-specic attributes. For example, in a PPGIS study of park

323) to the Channel Islands National Park in the U.S.,nd Kyle (2014) found differences in the mapped loca-osystem values perceived by visitors holding neutralng environmental worldviews. In a study of mappingrown (2013) analyzed non-spatial values and prefer-place-based values and preferred resource uses acrossS studies of national forests and found that positive,

attitudes toward extractive uses of national forestsated with participant mapping of economic values andses while nonmaterial forest attitudes were correlatedipant mapping of amenity values and conservation-s.ve been several PPGIS studies using the more narrow

of stakeholder that targeted specic stakeholders forn in the mapping activity for natural resource man-cisions. When analysis disaggregates for stakeholderDarvill & Lindo, 2014; Garca-Nieto et al., 2014, resultsences in stakeholder perceptions of the spatial distribu-ystem services. These studies highlight the importance

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 3

of including different stakeholder groups to capture the diversityof knowledge sources, humanenvironment relations, and valuesystems (Garca-Nieto et al., 2014) and that the differences in stake-holder mapping are likely a complex combination of individualfactors suchurban livingstakeholderthematicallipatory mapstakeholderPark. They overlap in fagreement some agrgroups. Thisagreement

Other PPdisaggregatNdumbaro,Haines-You2011) or hwithout usilyzed mappstakeholderpublic land Alessa, andlic perceptiexpert mapand non-sciprot organof spatial cgeographic and Whitalhold samplin a nationagroup exprest use prefthe study coone group oanother forspatial inforshops (n = 7household s

1.2. The stu

In manyAustralia, 2Australia, 2eral and staarea (NRCMprise 15% ofof public lahistoric areulations, anoften includcountries sulegally embthus providbe expresse

The primof how stamation for purpose, wetied as res

(1) What are the core dimensions of public land managementand can these be operationalized and measured in a multi-dimensional preference scale?

(2) Can public land management preferences be used to identify clasat ty

stakese diw dovatioults en dkehokeho

seek a sta201

atorylysis

ng stemo

d selppinpingolder, accerminy wech thion 4or thng anpatorolder

thod

udy l

sett stat

poption

mos the s arend t laned na

lans fo

and ximaillion796,0and,

(DEP

ta co

reset-baed a Grag a as household income, individual needs, rural versus, and time spent living in the study region rather than

afliation per se (Darvill & Lindo, 2014). In a study mosty similar to ours, Eadens et al. (2009) conducted partic-ping workshops with 35 individuals representing six

groups for recreation planning in a Bahamian Nationalmodeled spatial agreement by examining the spatialuture activity zones mapped by the six groups. Strongwas dened as areas mapped by ve to six groups andeement was dened as areas mapped by three to four

method resulted in a park map showing areas of spatialfor protection, ecotourism, and hunting activities.GIS studies have used stakeholder terminology withouting results by stakeholder group (Fagerholm, Kyhk,

& Khamis, 2012; Palomo, Martn-Lpez, Potschin,ng, & Montes, 2013; Ruiz-Frau, Edwards-Jones, & Kaiser,ave examined the spatial results of different groupsng stakeholder terminology. The studies that have ana-ed data by groups highlight the potential inuence of

sampling and participation to spatial outcomes forplanning and management. For example, Brown, Smith,

Kliskey (2004) compared the results of general pub-ons of biological value (n = 542) with results from anping workshop consisting of individuals holding scienceence positions with government, educational, and non-izations (n = 31). The study found a moderate degreeoincidence between the two groups but also distinctareas of agreement and disagreement. Brown, Kelly,l (2014) compared responses from a random house-ing group (n = 144) with volunteer participants (n = 84)l forest planning application in the U.S. The volunteeressed stronger utilitarian values and consumptive for-erences, both spatially and non-spatially. Importantly,ncluded that if forest planning decisions were based onr the other, there would be very different outcomes. Inest planning application, Brown et al. (2014) found thatmation generated from community-based PPGIS work-1) differed from spatial data collected through randomampling (n = 244).

dy contextpublic land values and preferences

countries the public land estate is signicant. In3% of terrestrial land area is public land (Geoscience014) and in the United States the combined total of fed-te-owned public land encompasses 35% of total land, 2014). Parks and protected areas on public land com-

the worlds land area (WDPA, 2014). Typical categoriesnds include forests, national parks, wilderness areas,as, and nature reserves. The quantity, structure, reg-d management of public lands vary by country, ande a state or region-specic governance framework. Inch as Australia and the U.S., participatory processes areedded in the planning and management of public landsing a formal arena where values and preferences mayd and contested by stakeholders.ary purpose of this study was to advance understandingkeholders can inuence participatory mapping infor-public land planning and management. To achieve this

seek to answer the following research questions, iden-earch aims in Section 2:

and(3) Wh

dothe

(4) Hoserres

(5) Givstasta

Wepart ofet al., explorter anaresultisocio-dity, anthe madevelostakehvationto deteeffort bto matIn Secttions fplanniparticistakeh

2. Me

2.1. St

Thelargestmatedpopulasecond35% oftiguoustate, aCrownmanagwealthreserveParks (appro3.14 mcover ment lCrown

2.2. Da

Theinternetion uscould dsify multiple stakeholder groups?pe and quantity of public land values and preferencesholder groups identify in a PPGIS process and how arestributed by public land type and location?

different stakeholder preferences for public land con-n, access, and development inuence place-specicin a participatory mapping process?ifferences in public land mapped preferences amonglders, how does spatial aggregation and weighting oflder preferences inuence the results?

answers to these questions using data collected astewide PPGIS study of public lands in Victoria (Brown

4a). We apply a novel approach that combines an, multi-item public land preference scale with clus-

to classify participants into stakeholder groups. Theakeholder groups are proled and compared on theirgraphic characteristics, public land use and familiar-f-identied reasons for using public lands. We analyzeg behavior of the two largest stakeholder groups by

and applying a spatial model that assesses the level of agreement for place-specic preferences for conser-ss, and development. Sensitivity analysis is performede the potential effects of unequal stakeholder mappingighting the mapped responses of one stakeholder groupe mapping behavior of the other stakeholder group., we reect on the ndings that have strong implica-e use of participatory mapping methods in public landd management. We provide some guidance for futurey mapping processes that seek to integrate multiple

groups for decision support.

s

ocation and context

ing for this study was the state of Victoria, the sixthe in Australia with an area of 237,629 km2 and an esti-ulation of 5,768,600 (ABS, 2013). Most of the statesis located near the capital city of Melbourne, Australiast populous city. Public lands in Victoria comprise aboutterrestrial land area (DEPI, 2013) with the largest con-as located in the mountainous eastern third of thehe northwest sector. Public lands in Victoria, known asds, have a wide variety of classications including statetional parks, state forests, federally managed common-

ds, metropolitan and regional parks, and specializedr the protection of historic and cultural resources.conservation reserves make up 3.98 million hectarestely 50% of all Crown land), state forests comprise

hectares (approximately 40%), and other Crown lands00 hectares (10%) including Commonwealth Govern-metropolitan parks, and land held under lease from theI, 2013).

llection process

arch team designed, pre-tested, and implemented ansed PPGIS application for data collection. The applica-oogle maps interface or mashup where participantsnd drop digital markers onto a map of Victoria. The web

4 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

Table 1Landscape values and preferences used in this study with their operational denitions.

Operational denition

ValuesScenic/est attraRecreation ere I e

(e.g., cEconomic e natuLife sustai oduceLearning/e e place

Biological/ e a varHeritage/c nt na

wledgTherapeut me feSpiritual red, re

Intrinsic/e o mattWildernes d, uni

PreferencesIncrease c g., dueAdd recrea ails, pAdd touris signs,

Improve a to 4Wation

Improve bResource e or miResource u lectricDecrease o hiclesNo develo

interface is ated with tpanels contnine managcategories w& Reed, 20(Brown & FaThe typologics, historyecosystem ing/regulativalues), and(Brown & Ktively conta(Brown, Weitems weretected area for increaseincreased oand an opti

Before ththe postcodthe study. Tdrag smalland your puanywhere, were instrunecessary tFollowing cparticipantsnon-spatialdemographlands, frequ

From Dewere recrumethods asand 9 metrVictorias b

ed ato thpersationd to), a n

consagedebs

codeice oovidproa. A fhetic These areas are valuable because they contain These areas are valuable because they are wh

participating in outdoor recreation activities These areas are valuable because they providning These areas are valuable because they help prducation/research These areas are valuable because they provid

or study.conservation These areas are valuable because they providultural These areas are valuable because they represe

continue and pass down the wisdom and knoic/health These places are valuable because they make

These areas are valuable because they are sacrespect for nature here.

xistence These areas are valuable in their own right, ns/pristine These areas are valuable because they are wil

onservation/protection Increase conservation and protection here (e.tion facilities Add more recreation facilities (e.g., walking trm services/development Add new tourism services (e.g., guided tours,

visitor center) here (Please specify).ccess Improve vehicular access (i.e., from no access

increased walking trail access under the recreushre protection Improve bushre protection here.xtraction Engage in resource extraction such as loggingse Engage in resource use such as grazing, hyroer limit access Decrease or limit access here (e.g., close to vepment or change No development or change to land use here.

described in greater detail in a technical report associ-he project (see Weber & Brown, 2014). Two differentained markers representing 11 landscape values andement preferences (see denitions in Table 1). The valueere derived from a landscape values typology (Brown

00), also called social values for ecosystem servicesgerholm, 2014; Sherrouse, Clement, & Semmens, 2011).y includes cultural values such as recreation, esthet-

/culture, and spiritual values, values for provisioningservices (economic/subsistence value), and support-ng ecosystem services (biological and life sustaining

has been used in more than 15 published PPGIS studiesytt, 2014). The public land estate in Victoria collec-ins the full range of values found in the values typology

prepara link spokesCorportribute(VNPAnatureencourstudy waccessthe chowas prling apsampleber, & de Bie, 2014a, 2014b). The public land preference identied and selected in consultation with the pro-management agency, Parks Victoria, and included itemsd conservation/protection, resource use and extraction,r limited access, increased facilities, bushre protection,on for no development or change.e mapping activity, participants were asked to identifye of their home location and how they learned abouthe application instructions requested the participant to

icons onto a map of Victoria to identify places you valueblic land preferences. . .although you can place markersthe focus of this study is on public lands. Participantscted to place as few or as many markers as they deemedo express their values and management preferences.ompletion of the mapping activity (placing markers),

were directed to text-based survey questions to assess public land management preferences (16 items), socio-ic characteristics, self-identied familiarity with publicency of use, and motivation for visiting national parks.cember 2013 to February 15, 2014, study participantsited using both purposive and convenience sampling

follows: (1) visitors to 16 national parks, 5 state parksopolitan parks were contacted on site as part of Parksiennial visitor satisfaction survey; (2) Parks Victoria

of the samp

2.3. Public

We deveences for pdiversity ofto multipleWe hypothconsisting items), resoaccess/contment (2 itesemantic-deither end items. The i

Public laBelow arThinkingyou placanswer. have a ppositionctive scenery including sights, smells, and sounds.njoy spending my leisure timewith family, friends or by myself,amping, walking, or shing).ral resources or tourism opportunities., preserve, clean, and renew air, soil, and water.s where we can learn about the environment through observation

iety of plants, wildlife, and habitat.tural and human history or because they allow me or others toe, traditions, and way of life of ancestors.el better, physically and/or mentally.ligious, or spiritually special places or because I feel reverence and

er what I or others think about them.nhabited, or relatively untouched by European activity.

to encroaching development, feral animals/weeds, illegal use).laygrounds, picnic ground) here.

brochures, apps) or development (e.g., trail head, toilet block,

D access or from 4WD road to 2WD road). Note: please map facilities icon.

ning here. energy, or wind energy here.

or 4WD).

nd distributed a press release about the study, placede study URL on the agencys website, and an agencyon promoted the study in an Australian Broadcasting

radio interview; and (3) a recruitment letter was dis- members of the Victoria National Parks Associationon-governmental organization (NGO) that promoteservation in Victoria. Participants in the study were also

to refer friends, relatives, and acquaintances to theite. Any member of the public could request an instant

to participate in the study. To encourage participation,f a modest non-cash incentive equivalent to about $10ed the to the rst 1500 study participants. This samp-ch resulted in a geographically representative statewideull evaluation and discussion of the representativeness

le can be found in Brown et al. (2014a).

land management preference scale

loped an exploratory scale to measure general prefer-ublic land management that would accommodate the

public land types ranging from national parks/reserves-use lands such as state forests to local/regional parks.esized and pilot-tested ve preference dimensionsof multiple items related to recreation experience (7urce extraction (2 items), renewable resource use (3),rol of public lands (2 items), and facilities develop-ms). Each survey item was measured on a seven pointifferential scale with contrasting preferences located atof the scale. See Fig. 1 for the specic wording of thenstructions provided to participants were as follows:

nds (not just parks) can be managed many different ways.e some contrasting views about public land management.

about your own preferences for public lands, where woulde yourself on the scales below? There is no right or wrongA position in the middle (number = 4) indicates you dontreference leaning either direction on the scale, whereas thes 1 and 7 suggest you have a very strong preference.

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 5

Preferences for recr eatio n experi ences . I prefer to spe nd ti me in an environ ment that ha s.. .

Primarily na tura l e nvironment Primarily bu ilt environm ent

Little social int eractio n (solit ude ) Lot s of social int eracti on (be around othe rs)

ction..

e...

.

..

Fig. 1. Groups from scaled on a sem

The recrgesting thaby the phyrecreation oBrown, 197on historicaincluded pring, wind eaccess/contto public lalands. The services and

2.4. Analyse

2.4.1. PubliIn the a

preference v.22) on thinitial groutify potentextraction mof the obsemum varianby eigenvaing factors winterpretatirelated.

2.4.2. ClustCluster

on the simpotential grferences, wand implemThe prograber of clustwith differe

s arehich

ime. clus

st clu resp7 parpre) solndivethoh of ant

the factoHigh ris k/ad ventur e activit ies

Minimal moto rized ac cess (explora tion requires walking, canoe ing, etc. )

Type o f acti vity i s most i mpor tant

If staying overnigh t, prefer hotel/mote l

Minimal signs of manag eme nt (e.g ., sign s, inform ation, staff)

My pre feren ces for resour ce extr aMining allowed

Logging of nati ve fore sts al lowed

My preferen ces for resource usGrazing livestock allowed

Wind en ergy al lowed

Forestry plantation s allowed

My preferences for acces s.. Limit ed public access

Governm ent cont rol of land s

My preferences for facilities .Many tourism services (e.g., visitor centres,

edu cation prog rams, lar ge tou rs)

Larg e rang e of tourism accommo dation

means of four groups (Preservation, Recreation, Socialization, Utilization) derivedantic-differential scale.

eation experience items were based on research sug-t the quality of the recreation experience is inuencedsical, social, and managerial setting that comprise apportunity spectrum (Clark & Stankey, 1979; Driver &8). The resource extraction and use items were basedl and contemporary use of public lands in Victoria andeferences for logging of native forests, mining, graz-nergy development, and plantation forestry. The tworol survey items assessed preferences for ease of accessnds and a question about who should control publictwo facilities items assessed preferences for tourism

accommodation associated with public lands.

s

model(BIC) wsame tand thethe besis, then = 150

Integroupsof the itical mfor eacthe semtion ofof the c land management components (aim 1)bsence of theory about the structure of public landdimensions, we ran exploratory factor analysis (SPSSe 16 preference items (1) to determine whether ourping of the items held in the results, and (2) to iden-ial latent variables. We used principal componentsethod which forms uncorrelated linear combinations

rved variables where the rst component has maxi-ce. The number of factors extracted was determined

lues greater than one (Kaiser, 1960) and the result-ere rotated using direct oblimin (non-orthogonal) for

on under the assumption that the components could be

er analysis and interpretation (aim 2)analysis is a method for grouping individuals basedilarity in responses to survey questions. To identifyoups of study participants based on public land pre-e used a model-based clustering program developedented in R (Mclust v.4.4) by Fraley and Raftery (1998).

m does not assume prior knowledge about the num-ers and evaluates a number of different cluster modelsnt geometric properties and parameters. The different

(PCA) wereformed homthe demogrand compato survey qpublic landlyzed by stparticipantstakeholder

2.4.3. AnalyWe ana

placed by pof variancemarkers machi-squaredwhether thgory differestakeholder

To examand stakehlic lands (coforests, maSafe/secur e ac tiviti es

Extensive mot ori zed ac cess (e.g., 2WD an d 4WD vehicles)

Physical setti ng is most impor tant

If st aying overnight, pre fer cam ping

Obvious manageme nt presence (e.g. , signs, infor mation , and staff)

. No mining at all

No log ging of na tive for ests

No grazing all owed

No wind energ y allowed

No for estry plan tation s

High/ eas y public access

Private le ases of la nd, less government control

Few tourism se rvices (e .g., sma ll to ur grou ps, emphasis on self -suffi cient travel )

Minimal touri sm accommoda tion

cluster analysis plotted on 16 public land preference items that were

compared using the Bayesian Information Criterion allows comparison of more than two models at theThus, the problem of identifying the number of clusterstering method are solved simultaneously by identifyingster model according to the BIC. In our cluster analy-onses to the 16 public land preference questions fromrticipants were used as input to the cluster procedure.tation and validation of the four cluster (stakeholderution was determined by examining the characteristicsiduals classied into the groups using a number of statis-ds: (1) the means of the 16 public land preference itemsthe stakeholder groups were analyzed and plotted onic differential scale to visually show the relative posi-groups on each of the preference items, (2) the meansr scores derived from principal components analysis analyzed to determine whether the stakeholder groupsogenous subsets on the public land components, (3)

aphic characteristics of the four groups were analyzedred to available census data, (4) participant responsesuestions about their level of familiarity, frequency of

use, and motivation for using public lands were ana-akeholder group, and (5) the different sources of hows learned about the study were compared across the

groups.

sis of mapping behavior by stakeholder group (aim 3)lyzed the number of value and preference markersarticipants in each stakeholder group. We used analysis

(ANOVA) to determine whether the mean number ofpped differed by stakeholder group and we generated

statistics with standardized residuals to determinee proportion of markers by value or preference cate-d signicantly from what would be expected across the

groups.ine the potential association between public land typeolder group, we selected 11 major categories of pub-astal reserves, community/metro/regional parks, staterine parks and sanctuaries, national parks, natural

6 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

Fig. 2. Possibstakeholder grindices. P = preI = increase, anthe reader is r

features resstate parks,for over 90%counts witstandardizeber of markin each pub

2.4.4. ModeWe exam

for the Preswith sufciidenticatioof preferencces were exto public lause. In simplocation foroutcomes aexample. Taccess in thaccess, but

To illustdisagreemea GIS grid-bopment pre1 km cell siand facilitietion versusdensities wunder the resource us(concurrenctwo groupssimilarity invation groucell densiti(RI). Howevspatial agrebution betwincrease accwhich by dthe intensitthe ratio of tion. As the index tabul

ratio becomes small (i.e., only one group mapped preferences at thelocation), the agreement index tabulates a smaller proportion of thecombined density. Thus, the formula for agreement on increasingaccess would be as follows:

ent

PI = ; RI =.al spoxesundhis and dhe ndicencesg theWheetw

is spes ann pre

on ent peatnd cing cith mrs foth ms/de

naults patoinedvidu

mory m

vatio we ing ltipleticalar phe sa

we eprefele outcomes related to preferences for public land access for twooups, with the formula for calculating the agreement/disagreementservation group marker density, R = recreation group marker density,d L = limit. (For interpretation of the references to color in the text,eferred to the web version of the article.)

erves, nature conservation reserves, wilderness parks, and water production and reservoir areas) that account

of Victorian public lands, and cross tabulated markerh the stakeholder groups. Chi-squared statistics andd residuals were used to determine whether the num-ers differed signicantly from what would be expectedlic land category.

ling group preferences by location (aim 4)ined the spatial distribution of public land preferenceservation and Recreation groups, the only two groupsent data for this particular analysis. This allowed forn of similarities and differences in the spatial locationes, not just the number and type of markers. Preferen-

amined for three public land management issues: accessnds, facility development, and conservation/resourcele terms, two groups can agree or disagree on the spatial

the three preference issues, resulting in four possibles illustrated in Fig. 2 which uses public access as anhe two groups can both agree to increase or decreasee same location or the two groups can disagree on

for different reasons.rate preferences spatially, we generated agreement andnt indices for each of the four possible outcomes usingased modeling approach. For access and facilities devel-ferences, we generated point densities (kernel) using aze with a 3 km search radius. This assumes that accesss development are reasonably place specic. Conserva-

resource use preferences were modeled using kernelith 2 km grid cell with a 5 km search neighborhood

Agreem

whereaccessaccess

Totment bwas fored). Tment aarea. Tment ipreferetractinindex. ment bthere providment iplottedagreemwas rement aincreasible wMarkeble wiservice

Thethe resparticidetermof inditainedas manPresergroup,increasWe muhypothin simiUsing tabove,tional assumption that preferences for conservation versuse cover larger areas on public lands. Spatial agreemente) was determined by aggregating the densities of the

in each 1 km or 2 km cell. For example, to identify limiting public land access, the cell densities of Preser-p markers for increasing access (PI) were added to thees of Recreation group markers for increasing accesser, this aggregation accounts for the intensity or level ofement but does not account for the evenness of distri-een the two groups. It is possible that all the markers toess in a given location were mapped by just one group,enition, would not constitute agreement. Therefore,y of agreement (combined densities) was multiplied bymapping evenness between the two groups at the loca-ratio of evenness approaches one (completely even), theates the actual combined densities. When the evenness

3. Results

3.1. Particip

A total placed one or partiallyping activitcollection, in Victoria.markers wistudy, we otory mappimapping ac to increase access = MIN (PI, RI)MAX(PIRI)

(PI + RI)

density of Preservation group markers for increasing density of Recreation group markers for increasing

atial agreement was found by adding the two agree- in Fig. 2 (shown in green). Total spatial disagreement

by adding the two disagreement boxes (shown inggregation approach identies areas of spatial agree-isagreement for the three preferences across the studynal step was to combine the agreement and disagree-s to identify where spatial agreement or disagreement

dominate in a given location. This was done by sub- total disagreement index from the total agreementre the difference score is positive, there is spatial agree-een the two groups and where the score is negative,atial disagreement. The magnitude of the difference

indicator of the intensity of the agreement or disagree-ferences at the location. The preference outcomes werea map to show the locations where agreement and dis-occur for the three preference issues. This procedureed to generate preference maps for facility develop-onservation/resource use on public lands. Markers foronservation/protection were assumed to be incompat-arkers for resource extraction and resource use.

r no development/change were assumed incompati-arkers to add recreation facilities and add tourismvelopment.l analysis performed was to determine how sensitiveare to group size and/or mapping behavior (aim 5). Inry mapping, the quantity of preference markers will be

by the participant group size and the mapping intensityals comprising the group. The Recreation group con-e participants, but these participants mapped only halfarkers. Given the quantity of markers placed by then group were about twice that placed by the Recreationmodeled potential changes to location preferences bythe weight of markers placed by the Recreation group.ied the point densities of the Recreation group by two, al scenario that assumes the Recreation group engagedreference mapping behavior to the Preservation group.me agreement/disagreement spatial analysis describedxamined how this weighting of markers changed loca-rences for the three public land issues.

ation

of 1905 participants accessed the study website andor more markers. Of these participants, 1624 (85%) fully

completed the survey questions that followed the map-y. A total of 35,347 markers were mapped during datawith 30,194 (85%) of these attributable to public lands

Approximately 85% of the markers placed were valueth the remaining 15% being preference markers. For thisnly included respondents that engaged in participa-ng and answered the survey questions following thetivity (n = 1507).

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 7

Table 2Descriptive statistics and factor analysis results for public land preference survey items. The number of responses per item ranged from n = 1620 to n = 1636. Factors wereextracted using principle components and were rotated using direct oblimin (non-orthogonal). The highest loadings on each factor are shown (highlighted, underlined) withCronbachs alpha calculated for these items. Cronbachs alpha is also shown when asterisk * item is not included in factor scale.

ype s )

Natural vs. buLow social in(recreation) Adven ture/ris(recreation) Non-motorizeActivity vs. seLodging vs. cMinimal manman age ment Mining all owe Logging nativGrazing li vesWind ene rgy Forestry plan Limited publicGovernmen t control Many tourismExten sive t outou rism accoCronba chs A If item de

3.2. Public l

The explitems resulranging frocumulativesampling adple quality by Hutchesappear in Tfor resourceaccess and (2 items), aloadings melatent variamotorized vlation with moderate toall survey iranging fronumber of i

The loadated somewservices anexperience subscale. Gbut was mothan the re(activity verall variabilitseemingly uting when lodging wh

3.3. Cluster

The R clmodel (ellipreference models witmation Critwas determ

puan pland

rsts of nmen

solition forees anlic lablic y do

prefant fd on

seco Relic latheyr motodeItem Mea n Std. Dev.

Resou rce extraction preferences (=4.2, 26%)

Recrea tion tpreference(=3.2, 20%

ilt environment (recreation) 3.0 2.1 .280 .637teraction vs. high social interaction

3.7 1.9 .12 2 .75 2 k activities vs. safe/secure activities

4.3 1.7 .254 .74 6d use vs. motorized use (recreation) 3.6 2.0 .459* .427tting importance (recreation) 4.6 1.7 .114 .24 6 amping (recreation) 4.9 1.8 -.058 -.21 8 agement vs. obvious signs of (recreation ) 4.0 1.8 .197 .649d vs. no mining 5.4 1.8 .82 3 .06 4 e forests allowed vs. no log ging 5.2 1.8 .88 6 .09 7 tock all owed vs. no grazing 4.8 1.9 .71 0 .000 all owed vs. no wind energy 3.9 1.9 .01 3 .06 9 tation s allowed vs. no plantations 4.3 1.9 .266 .030 access vs. easy public access 4.3 1.7 .085 .014

control of lan ds vs. privatization/less 3.0 1.8 .25 3 .22 7

services vs. fewer tourism services 4.6 1.8 .188 .604*rism accommoda tion vs. minimal

mmodation 4.7 1.8 .27 1 .591 * lpha .52 .51 leted * .84 .78

and management components

oratory factor analysis of the 16 public land preferenceted in the extraction of ve factors with eigenvaluesm 0.98 (rounded to 1.0) to 4.2, explaining 71% of the

variance. The KaiserMeyerOlkin (KMO) measure ofequacy for factor analysis was 0.80 indicating the sam-falls in the meritorious range of values as describedon and Sofroniou (1999). The loadings for each factorable 2. Structurally, the factors measured preferences

extraction (4 items), recreation experience (6 items),regulatory control (2 items), renewable resource usend activity versus setting preferences (2 items). Factorasure the correlation between the survey item and theble represented by the factor. With the exception of theersus non-motorized item, which had moderate corre-multiple factors, loadings were 0.6 or greater indicating

high correlation with the factor. Cronbachs alpha fortems was 0.72 with individual factor reliability scoresm 0.52 to 0.72. Reliability scores are dependent on thetems and usually increase with more items.ing of the 16 items on the ve extracted factors devi-

16-itemthe mepublic

Theconsistenvirotial forextracttation facilitito pubThe puactuallwouldimportfocuse

Thelabeledon pubment, greateThey mhat from our initial grouping in the survey. The tourismd accommodation items loaded with the recreationfactor but decreased the reliability of the recreationrazing is generally regarded as a renewable resourcere highly correlated with the resource extraction factornewable resource use factor. The last factor extractedsus setting preference) was weakest in capturing over-y with the least intuitive interpretation as it loaded twonrelated items on the importance of activity versus set-using public lands, and preferences for camping overen visits involve overnight stays.

analysis results

uster analysis (Mclust v.4.4) identied a four clusterpsoidal, equal shape) using the 16-item public landscale. This model was selected over other potentialh fewer or more clusters based on the Bayesian Infor-erion. A general description of each stakeholder groupined by examining the distribution of responses to the

native foresshow slightgroups, theconcerned tion group.camping ovwith fewer

The thirization anenvironmenized accessPreservatioand use, blimiting or prefer morelands.

The fourlabeled Utation preferon the activRegu latory preferences (=1.7, 10%)

Rene wab le resou rce

preferences (=1.4, 9%)

Activity vs. sett ing

preferences (=1.0, 6%)

.138 .352 .163

.14 2 .18 4 .11 5

.083 .094 .047

.263 .120 .363 -.079 .087 .669-.043 .060 .854

.060 .234 .350

.07 6 .01 8 .14 3

.02 5 .01 0 .04 6 .05 5 .39 2 .06 8 .04 7 .82 9 .09 4 .083 .772 .066

.850 .228 .034

.639 .162 .196 .478 .239 .197

.40 2 -.19 5 -.25 3 .52 .72 .52

blic land scale and other survey variables. Fig. 1 showsosition of each of the four stakeholder groups on the

semantic-differential preference scale. group (n = 394, 26%) was labeled Preservation andindividuals that prefer public lands for their naturalt, non-motorized recreation opportunities, and poten-ude. They are strongly opposed to all forms of resourceand use (native forest logging, grazing, mining, plan-stry, wind energy) and they prefer minimal tourismd services. They prefer to see decreased or limited accessnds with government maintaining regulatory control.land setting is more important than the activities they

and if they are to stay on public lands overnight, theyer camping over lodging. In summary, public lands areor their natural values, with activity and use preferencesappreciating and preserving these natural values.nd and largest stakeholder group (n = 656, 44%) was

creation and are individuals that appear more focusednds for recreation. While preferring a natural environ-

want more opportunities for social interaction withorized access than the Preservation stakeholder group.rately oppose resource extraction (mining, logging of

ts), are ambivalent about grazing on public lands, and

support for wind energy and plantation forestry. Of ally prefer to see easier access to public lands but are lessabout government control compared to the Preserva-

Similar to the Preservation group, they slightly preferer lodging when visiting public lands and prefer placestourism services and facilities.d stakeholder group (n = 263, 17%) was labeled Social-d comprise individuals that prefer a primarily builtt, social interaction, safe activities, extensive motor-, and obvious management presence. Similar to then group, they oppose all forms of resource extractionut with less conviction. They are ambivalent aboutincreasing general access and government control and

tourism services and facilities associated with public

th and smallest stakeholder group (n = 194, 13%) wasilization and are individuals that show similar recre-ences to the Recreation group, but place more emphasisity rather than setting, and have a moderate preference

8 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

for lodging over camping. Similar to the socialization group,they prefer more tourism services and facilities. A distinguishingattribute of this stakeholder group is their preference toward allforms of resource extraction and uses of public lands.

3.4. Interpr

Factor scholder groupost hoc tesurvey itemsignicantlunique homexceptionfrom the Somight be ethe greatesand tourismtion groupswith the Utthe Recreatmost differexperiencesanalyses of ically consis

There wcompositioThe Preservhigh level ohold incomUtilization the Melboueducation ailies with clower levelcally male the other ggroup was trather than

The Prenational/stalic land use.parks but rpublic landfrequent usfrequency oidentied fathe least freiarity (95% =lowest leve

The survtify their pThe choicespark visitorZanon, Hallresponse foin a trail-baSocializatiohigh percenization as trelatively hrest/relax a

The naistics was astudy. The Parks Assoc

of information (25%) with another 14% identifying bushwalkingclubs. The Socialization group primarily learned of the study fromthe Parks Victoria website and the Utilization group learned of thestudy through friend referral (55%). The Recreation group had the

st div

akeh

map catestud

thandentnd S

prop mornomiries

(resipondtion al = 7

numhe nualsarkark

sing on pvatio

pre of

tion in msed ecrea

maping

es useser

natisanct/reg

oirs. reatistaters in The arke

factunity

atial

spats in Frs (nzatioiliza7 anr Me

and Rhout

distretation and validation of the stakeholder groups

ores for the ve public components for the four stake-ps were analyzed using ANOVA with Tukeys HSDsts to conrm the results for individual public lands. All mean factor scores for the four groups were

y different (ANOVA, Tukey HSD, p < 0.05) and formedogenous subsets for all public land factors with onethe Recreation group was not signicantly differentcialization group on the resource extraction factor. Asxpected, the Preservation and Utilization groups hadt difference in factor scores on the resource extraction

development factors. The Socialization and Utiliza- were most different on the renewable resource factorilization group more inclined toward resource use. Onion factor, the Preservation and Utilization groups wereent from the Socialization group, favoring recreation

in less managed natural environments. Overall, thefactor scores for the ve public land factors appear log-tent with results from the individual preference items.ere some signicant differences in the demographicn of the four groups. See descriptive statistics in Table 3.ation group was the oldest of the four groups with af formal education but with the least amount of house-e and the greatest diversity of family structure. Thegroup tended to be younger males with families fromrne area who had attained a high level of both formalnd income. The Socialization group tended to be fam-hildren (95%) in households with higher income ands of formal education. The Recreation group was typi-with income and education levels in the mid-range ofroups. A distinguishing characteristic of the Recreationhe high proportion of individuals from regional Victoria

the capital city of Melbourne.servation group was most active in the use ofte parks and reported a high frequency of general pub-

The Socialization group was least likely to visit nationaleported the highest frequency of use of other types ofs. The Recreation and Utilization groups were the leasters of public lands of the four groups. The reportedf public land use was not necessarily related to self-miliarity with the lands. The Utilization group reportedquent use, but the highest level of self-identied famil-

good or excellent). The Recreation group reported thels of public land familiarity.ey included a question that asked participants to iden-rimary reason for visiting national parks in Victoria.

were based on cluster analysis of extensive national research conducted by Parks Victoria (n = 11,700) (see, Lockstone-Binney, & Weber, 2014). The most frequentr the Preservation and Recreation groups was to engagesed activity while the most frequent response for then and Utilization groups was to sightsee. A relativelytage (31%) of the Socialization group identied social-heir primary reason for visiting national parks while aigh percentage of the Utilization group (23%) selecteds their primary reason.l survey items providing insight into group character-

question asking how participants learned about thePreservation group identied the Victorian Nationaliation (an organized stakeholder group) as their source

greateclubs.

3.5. St

Theing thein the valuesresponation ahigherone orof ecocategologicalby resRecrea(residu

Thethan tIndividence mthose mdecreament Preseraccessportionprotecences increaand d

TheexaminferencThe Prers inparks/munityreserv

Recparks, markeparks. few mby thecomm

3.6. Sp

TheappearmarkeSocialiThe Ut(R = 0.0greatevationthrougspatialersity of sources and notably included 4 wheel-drive

older group mapping behavior

ping behavior of the groups was assessed by analyz-gories of values and preferences mapped by each groupy. The Preservation group mapped signicantly more

the other groups, about 36 markers on average per compared to about 14 and 5 markers for the Recre-ocialization groups, respectively. See Table 4. Further, aortion of respondents in the Preservation group placede markers in all value categories with the exceptionc value. The largest proportional differences in valuewere associated with wilderness (residual = 16.7), bio-dual = 11.9), and intrinsic (residual = 9.9) values mappedents in the Preservation group. Respondents in thegroup mapped proportionately more economic values.2) than the other two groups.ber of mapped preferences was signicantly less

umber of values markers for all study participants. in the Preservation group also placed more prefer-ers, on average, than the other groups, in particularers associated with increasing conservation/protection,or limiting access, and preventing further develop-ublic lands. Proportionately fewer individuals in then group mapped resource extraction, use, and improvedferences. The Recreation group had a higher pro-individuals express preferences to increase bushreand to extract resources. The largest group differ-apping behavior were associated with preferences forconservation/protection, no development/change,sing/limited access.ping behavior of the groups was further analyzed bywhere the four groups mapped their values and pre-ing 11 major categories of public lands. See Table 5.vation group placed disproportionately more mark-onal parks, nature conservation areas, and marineuaries and proportionately fewer markers in com-ional parks, state forests, historic/cultural areas, andIn contrast, theon group placed more markers in community/regional

forests, reservoirs, and wilderness parks and fewernational parks, nature conservation areas, and marineSocialization and Utilization groups placed relativelyrs in all public land categories, and are distinguished

that they placed disproportionately more markers in/regional parks than any other public land category.

distribution of stakeholder preferences

ial distribution of all markers placed by the four groupsig. 3. The Preservation group placed signicantly more

= 16,251) than the Recreation group (n = 8562), then group (n = 608), and the Utilization group (n = 208).tion and Socialization markers were highly clusteredd R = 0.12) with the majority of markers mapped in thelbourne area in community/regional parks. The Preser-ecreation group markers were more spatially dispersed

Victoria (R = 0.21 and R = 0.23) reecting the generalibution of public lands in the state.

G.

Brown

et al.

/ Landscape

and U

rban Planning

139 (2015)

115

9

Table 3Study participant and cluster group demographics with comparison to Victoria census data (ABS, 2011). Participant familiarity, use, and motivations for using public lands.

Variable All StudyParticipants

ABS Census 2011 Cluster 1 Preservation(n = 394)

Cluster 2 Recreation(n = 656)

Cluster 3 Socialization(n = 263)

Cluster 4Utilization (n = 194)

Age (median)** 36 37 55 37 35 30Gender*

Male 57% 49% 50% 63% 43% 72%Female 43% 51% 50% 37% 57% 28%

Education (highest level completed)*

Bachelors degree 39% 16% 37% 34% 38% 75%Postgraduate education 16% 11% 38% 13% 0% 4%

Household income (annual)*

Median $110,000 $63,200 $70,000 $110,000 $130,000 $150,000Less than $20,000 2.6% 3.1% 6% 2% 0% 2%$140,000$160,000 15.1% 7.3% 6% 10% 18% 44%$160,000$180,000 9.4% 4.2% 3% 10% 13% 13%$180,000$200,000 5.0% 1.9% 2% 6% 7% 3%$200,000+ 2.9% 3.1% 4% 4% 0% 0%

Household structure*

Families with children 41% 46% 30% 58% 95% 77%Mature couple/no children at home 35% 17% 0% 18%Mature single 17% 7% 0% 0%Young couple/no children 11% 13% 3% 0%Young single 7% 6% 2% 4%

Live in Greater Melbourne area*** 46% 75% 53% 34% 51% 65%

Visits to national/state parks last 12months (mean/median)**

28/12 20/10 7/6 14/12

Self-rated knowledge of public lands*

Excellent 20% 22% 46% 25%Good 58% 46% 50% 70%Average 21% 29% 4% 5%Below average/poor 2% 3% 0% 1%

Frequency of public land use*

Daily 21% 9% 23% 3%At least once week 39% 26% 36% 17%At least once month 31% 47% 35% 71%Less than once month 9% 18% 6% 10%

Most common reason(s) for visitingpublic lands*

Trail activity (44%) Trail activity (33%) Sightsee (35%) Sightsee (47%)

Sightsee (32%) Sightsee (19%) Socialize (31%) Rest/relax (23%)Stay/camp (9%) Rest/relax (15%) Trail activity (20%) Trail activity (14%)Other activity (6%) Stay/camp (14%) Rest/relax (10%) Socialize (10%)

Most frequent recruitment sources VNPA (25%) Friend (27%) PV Website (71%) Friend (55%)Friend (23%) PV website (17%) VNPA (10%) PV website (30%)Bushwalking clubs(14%)

VNPA (15%) Park contact (6%) Park contact (8%)

4WD Club (8%)

Note: ABS income percentages are estimates to match survey income categories.* Proportional differences are statistically signicant (chi-square, p 0.05).

** Mean differences are statistically signicant (ANOVA, p 0.05).*** Percentages are estimates based on postcodes provided by study participants.

10 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

Table 4The total and average number of landscape value and management preference markers placed by four stakeholder groups and the percent of individuals within each clustergroup placing one or more markers for the given value or preference category. Chi-square analysis and residuals analysis was performed on the proportion of participantswith cell color indicating whether the value or preference category was signicantly over-represented (green) or under-represented (pink) based on standardized residualsbeing greater than +2.0 or less than 2.0.

Marke r ca tegory by Groupa Preservation (n=394 ) Rec rea tion (n=656 ) Sociali zation (n=263 ) Utili zation (n=194)Number (rank) Mea n Pct

Std. Res.a

Number (rank)

Mea n Pct Std. Res.Number (rank) Mea n Pct

Std. Res.

Number (rank)

Mea n Std. Pct Res.

Scen ic value 2488(2) 6.3 15.5% 2.2 1441(2) 4.1 17.1% 0.5 129(1) 2.8 21 .3% 56(1) 3.4 0.3 19.3% 1.4 Recrea tion value 2536(1) 6.4 15.8% 3.3 2194(1) 17.6 26.1% 0.3 81(2) 20.0 13.3% 32(2) 3.6 0.2 11.0% 3.5 Econ omic value 208(11 ) .5 1.3% 0.4 283(9) 13.7 3.4% 0.2 56(4) 8.2 9.2% 29(5) 11.6 0.1 10.0% 8.9 Life sustaining value 1207(5) 3.1 7.5% 0.8 532(5) 3.1 6.3% 31(3) .7 6.4% 0.1 39(3) 3.5 0.2 10 .7% 2.4 Learning value 845(8) .6 5.8% 0.1 35(5) 1.1 5.0% 0.6 424(7) .0 5.3% 2.1 30(4) 0.2 10.3% 3.9 Biological value 2345(3) 6.0 14.6% 0.8 555(3) 19.3 6.6% 17 .6 0.1 33(6) 5.4% 20(7) 4.8 0.1 6.9% 2.5 Heritag e value 863(7) .6 4.8% 0.1 29(7) .3 5.3% 0.7 442(6) .6 5.4% 2.2 13 (9) 0.1 .6 4.5% The rapeu tic value 615 (9) 2.3 4.1% 0.1 25(8) .0 3.9% 0.5 328(8) .7 3.8% 1.6 1(6) 0.1 7.2% 3.0 Spiritual value 485(10 ) 1.2 3.0% 0.3 171(11) 3.4 2.0% 0.1 25(8) 4.9 4.1% 17(8) 2.1 0.1

0.10.1

5.9% 3.3 Intrinsic value 964 (6) 2.4 6.0% 0.4 234(10) 11.2 2.8% 10 .8 12(10) 1.4 3.6% 0.1 22(10) .6 4.1% Wilde rness value 1211(4) 3.1 7.5% 0.8 537(4) 4.8 6.4% 0.0 5(11) 2.5 0.8% 11(11) 6.0 3.8% 2.1 Total values 1376 7 36.4 85 .7% 84.9%14.1 7141 93.8%

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 11

Fig. 3. Spatial distribution of mapped values and preferences by cluster group with nearest neighbor ratio for (a) preservation; (b) recreation; (c) socialization; and (d)utilization groups. Smaller R ratios indicate greater clustering within the study region while larger R ratios indicate greater dispersion.

disagreement, for example, in places like the Great Otway andAlpine National Parks. Thus, conservation and resource preferenceswere less inuenced by sampling and mapping behavior comparedto public land access preferences.

The nal model examined preferences for new facility develop-ment versus preferences for no development or change. Comparedto preferences for conservation, there were signicantly more loca-tions and greater variability in agreement (see Fig. 6c). One location

Fig. 4. Mappefor limiting ord locations of public land access preferences by group: (a) Preservation and (b) Recreation. improving public access while map (d) is a weighted model with Recreation group densi Map (c) shows the location and intensity of agreement/disagreementties doubled.

12 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

Fig. 5. Mappe ) Presof agreement/ eighte

with signiwhere bothdevelopmefurther devthe preferenthe two grdevelopmeis an iconicof Victoria ahighly photPort Campbnew facilitistretch betwhere someThe weighthad little otwo prefere

4. Discussi

The purmanagemetify and clastakeholdercollected inpublic landservation anidentify statant in claspreferenceswere equald locations of public land conservation versus resource use preferences by group: (adisagreement for increasing conservation versus resource use while map (d) is a wcant agreement was Wilsons Promontory National Park Preservation and Recreation groups favor no furthernt. In contrast, there was signicant disagreement aboutelopment at Mt. Buffalo National Park, but in this case,ces were divided as much within each group as between

oups. Another interesting result was preferences fornt of facilities along the Great Ocean Road. This road

scenic highway that follows the rugged southern coastround Cape Otway peninsula to the Twelve Apostles, aographed coastal feature of eroded limestone rocks atell. Overall, there is general agreement not to developes along the scenic route with the exception of theween the two communities of Lorne and Apollo Bay

individuals in the Recreation group favor new facilities.ing of Recreation preferences for facilities developmentbservable effect on the results compared to the othernce issues.

on

pose of this research was to identify key public landnt preference components, use this information to iden-ssify public land stakeholder groups, and examine how

groups could inuence spatially explicit information a PPGIS process. Given the historical conict over

management, we anticipated that preferences for con-d development of public lands would be important to

keholder perspectives. While these factors were impor-sifying stakeholder groups for public lands in Victoria,

for the type of recreation experience on public landsly or more important in identifying stakeholder groups.

We identipreferencesstructure, f

The Preces for mainnature-baseand prefereinant focusand reservemembers bgroup repreferences, uimportant uals that spVNPA and fforest landsUtilization communitystrongest plands, theseior. The Socprotection also did not

The pubation groupof map maof agreemeconservatioences existnot new, buseful for cervation and (b) Recreation. Map (c) shows the location and intensityd model with Recreation group densities doubled.ed four stakeholder groups using cluster analysis of and found differences in their socio-demographicamiliarity, and use of public lands.servation stakeholder group showed strong preferen-taining natural values on public lands and engaging ind recreation activities, mapping twice as many valuence locations as the largest group, Recreation. The dom-

of the Preservation group is protecting national parkss with many Victoria National Park Association (VNPA)eing part of this group. The Recreation stakeholdersents a broad spectrum of public land values and pre-nied mostly by the belief that public lands providerecreation opportunities. This group contains individ-an multiple organized stakeholder groups such as theour wheel-drive clubs, with primary interests in state

and community/regional parks. The Socialization andgroups were smaller groups with dominant interests in/regional parks. Although the Utilization group had thereferences for resource use and development of public

preferences did not manifest in actual mapping behav-ialization group held strong preferences for public landsimilar to the Preservation group, but these preferences

manifest in actual mapping behavior.lic land preferences of the Preservation and Recre-s were reected in the quantity, type, and location

rkers placed, resulting in identiable geographic areasnt and disagreement in preferences for public landn, access, and development. That differences in prefer-

for how public lands should be managed is certainlyut rendering these differences spatially is potentiallyonsidering future management options. The capacity

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 13

Fig. 6. Mappe by grand intensity o odel w

for participence maps in public lagate stakehreect the ing one stalocation-spically, prefesample sizeior, or a comissue of aggstakeholder

Public equilibriumnormal statare largely icy subsystinterested such as catresources copportunitytions (Sabin public lais charactepolitical agdominated participatioshowed PPdisrupt thealternative into the pld locations of public land facility development versus no development preferencesf agreement/disagreement for facility development while map (d) is a weighted matory mapping to generate spatially explicit prefer-raises an important question for its prospective usend planning and management. How should one aggre-older values and preferences to identify outcomes thatpublic interest? In this study, we showed that weight-keholder group over another changes some of theecic outcomes regarding level of agreement. Specif-rences for public land access appear sensitive to the

of the stakeholder group, the group mapping behav-bination of the two factors. To address the importantregation, it is necessary to briey discuss the role ofs and politics in the domain of public land management.land management generally ts a punctuated

policy model (Baumgartner & Jones, 1993), where in ae of equilibrium, public land decisions (many of whichadministrative or technical in nature) are made in pol-ems dominated by public land management agencies,parties, and stakeholder groups. Rare, major eventsastrophic wildre or the discovery of major energyan sometimes provide the punctuation, window of (Kingdon, 1995), or the formation of advocacy coali-

atier & Jenkins-Smith, 1993) needed for major shiftsnd management. Generally, public land managementrized by relatively low public salience on the widerenda, resulting in relatively stable policy subsystemsby a limited number of stakeholder groups. The highn rate and diversity among respondents of this studyGIS processes are capable of having the potential to

subsystem by systematically seeking and includingvoices in the form of unorganized stakeholder interestsanning and decision process. But elsewhere, Brown

(2012) hasuse PPGIS mlack of inceexperienceknowledgedestabilizat

If publiclarger-scaledamental pspatial resusion to increwould be cRecreation ditional stastakeholderinterests. Sple, be assiBut stakehocated in theis a reasonato participathreats to pnecessary, outcomes. Wthetic towathe mappinPreservatioagement plmany publities are notbecause somoup: (a) Preservation and (b) Recreation. Map (c) shows the locationith Recreation group densities doubled. identied the reluctance of public land agencies toethods for a range of institutional reasons including

ntives to engage non-traditional stakeholders, lack of in participatory process, and distrust of non-expert. To this list, one could also add the potential forion of the public land policy subsystem.

land management decisions are expanded to include public participation as described in this study, the fun-roblem of aggregation and potential weighting of thelts must be addressed. For example, an agency deci-ase public access to Wilsons Promontory National Parkontroversial as evidenced in our spatial model whengroup preferences were given greater weight. In tra-keholder analysis, the relative inuence or power of

groups suggests a strategy for managing stakeholderpatial weighting of mapped results could, in princi-gned based on an assessment of stakeholder power.lder power is a subjective judgment and is not indi-

participatory mapping results per se. Mapping effortble proxy for the salience of public land managementnts, but salience is not equal to inuence. Perceivedublic lands require mobilization where salience is a

but insufcient condition for inuencing public lande posit that the Socialization group which is sympa-

rd public land protection, though not highly engaged ing component of the study, could be mobilized by then stakeholder group for a signicant public land man-an or decision. The two groups would be natural allies onc land development issues. The same preference afni-

present between the Recreation and Utilization groupse types of resource use would be considered a threat to

14 G. Brown et al. / Landscape and Urban Planning 139 (2015) 115

public land recreation. Without weighting, the preferences of thePreservation group dominate the aggregated mapping results. Thismay, in fact, represent the actual Victorian political landscapewhere public land preservation interests secured expansion ofnational pamultiple-usIn Barmah National Pa

4.1. Study l

The expmeasure puclassicatioreasonably scale wouldThe scale wvation/devemore nuancwere simpl

The quaPPGIS procecovered a lWhen specdata becomdata adequasmaller scadata showepublic land2014b). Forunits, suchspatial data

The spament is buspatial stakstakeholderresearch in agement (sviewed as alternativelBrown and ferences to has been ap(Brown et a

Developrationale foimportant fCarlo can bebased on inical insightperceptionsThe defaultipants are their mappThis may soland shouldracy. Howevproblems antory GIS proare not reprBased on ousuch samplrun, withouwe suggestorganized srience with

develop defensible spatial aggregation rules and supporting ratio-nale for the weighting of stakeholder responses.

nces

an Buusdatendortner, ago: UG. (2urce M), 12G. (20ning: .

G. (20e-baseG., &

servi016/j.G., & Kon GIS221., & R

g part., & R

st planG., Smiologicraphy., Kelarticiems foageme., We

lic par789G., Weuationning. J. M. (2

and aki, G. e, or s

N., & Slanninrt (PN., & Li

eholdes. Ecoepartm

area le/00ssed 2. L., &

eation resourgriculion (GL. M., eholdnd Nalm, N. knowices. E., & Rers v

., & ChVictor, R. E. ieto,

Marrole oser.20ce Auationld, M.ourna, R., &agemculturaon, G.icatiorks in the Otways region at the expense of former,e state forest lands (OREN, 2014; Parks Victoria, 2009).too, there is a movement to create the Great Forestrk in the central highlands of Victoria (Rees, 2014).

imitations and future research

loratory public land preference scale was designed toblic land preferences across a full range of public landns and tenures. The semantic-differential scale waseffective in segmenting stakeholder groups, but the

benet from further development and renement.as most effective in segmenting recreation and conser-lopment preferences and least effective in identifyinged preferences about public land tenure and access thatistically presented in the scale.ntity of spatial data collected was the largest of anyss to date (over 35,000 mapped locations) but this studyarge area with a high diversity of public land types.ic public land units are examined in detail, the spatiales quite sparse, along with condence that the spatialtely captures stakeholder values and preferences at thele. Assessment of the spatial data quality of the studyd that the statewide mapping process missed smaller

units and less densely populated areas (Brown et al., several of the larger and more popular national park

as Alpine and Wilsons Promontory, the quantity of appears adequate to model stakeholder preferences.tial model presented for assessing stakeholder agree-t one of many alternative approaches. Our focus oneholder agreement/disagreement was selected because

agreement has been the subject of stakeholderother policy domains including park and resource man-ee Eadens et al., 2009). Stakeholder agreement may bean indicator of conict potential and we could havey modeled spatial conict using methods described byRaymond (2014) which combine both values and pre-generate a conict index. This conict mapping methodplied to public land units, but not stakeholder groupsl., 2014a).ing defensible guidelines for spatial aggregation and ther potentially weighting stakeholder responses are anuture research need. Simulation methods such as Monte

used to identify the range of possible spatial outcomesput probabilities, but the more pressing need is crit-

into stakeholder dynamics, power relations, and the of stakeholders held by public land decision makers.

position for spatial analysis is that stakeholder partic-similar in importance and inuence, and accordingly,ed values and preferences should be treated equally.und appealing to those holding the position that public

be managed by the principle of participatory democ-er, such a position would require that existing samplingd non-response bias be overcome given that participa-cesses result in low response rates and participants thatesentative of larger populations (Brown & Kytt, 2014).r experience in multiple PPGIS studies, it is unlikely thating deciencies can be overcome, at least in the shortt a signicant investment in sampling effort. Therefore,

more research with public land managers, as well astakeholder groups, to glean their knowledge and expe-

stakeholder subsystems as a potential path forward to

Refere

Australicens2?op

BaumgaChic

Brown, Reso18(1

Brown, plan516

Brown, plac

Brown, tem10.1

Brown, pati46, 1

Brown, Gusin

Brown, Gfore

Brown, of bGeog

Brown, Glic psystMan

Brown, Gpub53, 7

Brown, evalplan

Bryson, tion

Budowstenc

Clark, R.for pRepo

Darvill, Rstakvalu

DEPIDlandpdf Acce

Driver, Brecrableof AStat

Eadens, Stakety a

Fagerhodersserv

Fraley, CAnsw

Fraser, Ithe

FreemanGarca-N

C., &The j.eco

Geoscieneduc

GoodchiGeoJ

GrimblemanAgri

HutchesPublreau of Statistics. (2013). QuickStats. Available from http://www.a.abs.gov.au/census services/getproduct/census/2011/quickstat/cument&navpos=220 Accessed 27.04.14F., & Jones, B. (1993). Agendas and instability in American politics.niversity of Chicago Press.

005). Mapping Spatial Attributes in Survey Research for Naturalanagement: Methods and Applications. Society & Natural Resources,

3.12). Public Participation GIS (PPGIS) for regional and environmentalReections on a decade of empirical research. URISA Journal, 25(2),

13). Relationships between spatial and non-spatial preferences andd values in national forests. Applied Geography, 44, 111.Fagerholm, N. (2014). Empirical PPGIS/PGIS mapping of ecosys-

ces: A review and evaluation. Ecosystem Services, http://dx.doi.org/ecoser.2014.10.007ytt, M. (2014). Key issues and research priorities for public partici-

(PPGIS): A synthesis based on empirical research. Applied Geography,36.aymond, C. (2014). Methods for identifying land use conict potentialicipatory mapping. Landscape and Urban Planning, 122, 196208.eed, P. (2000). Validation of a forest values typology for use in nationalning. Forest Science, 46(2), 240247.ith, C., Alessa, L., & Kliskey, A. (2004). A comparison of perceptionsal value with scientic assessment of biological importance. Applied, 24(2), 161180.ly, M., & Whitall, D. (2014). Which public? Sampling effects in pub-pation GIS (PPGIS) and Volunteered Geographic Information (VGI)r public lands management. Journal of Environmental Planning andnt, 57(2), 190214.ber, D., & de Bie, K. (2014a). Assessing the value of public lands usingticipation GIS (PPGIS) and social landscape metrics. Applied Geography,.ber, D., & de Bie, K. (2014b). Is PPGIS good enough? An empirical

of the quality of PPGIScrowd-sourced spatial data for conservationLand Use Policy, 43, 228238.004). What to do when stakeholders matter: Stakeholder identica-

nalysis techniques. Public Management Review, 6(1), 2153.(1976). Tourism and environmental conservation: Conict, coexis-ymbiosis? Environmental Conservation, 3(1), 2731.tankey, G. H. (1979). The recreation opportunity spectrum: A frameworkg, management, and research. USDA Forest Service, General TechnicalW-98).ndo, Z. (2014). Quantifying and mapping ecosystem service use acrosser groups: Implications for conservation with priorities for culturalsystem Services, http://dx.doi.org/10.1016/j.ecoser.2014.10.007ent of Environmental and Primary Industries. (2013). Victorian crown

statement. Available from http://www.depi.vic.gov.au/ data/assets/06/199068/FactSheet CrownLandVictoria 20130821FINAL.pdf2.04.14

Brown, P. J. (1978). The opportunity spectrum concept in outdoor supply inventories: A rational. In Proceedings of the integrated renew-ces inventories workshop (pp. 2431). Ft. Collins, CO: U.S. Departmentture, Forest Service, Rocky Mountain Forest and Range Experimenten. Tech. Rep. RM-GTR-55).Jacobson, S. K., Stein, T. V., Confer, J. J., Gape, L., & Sweeting, M. (2009).er mapping for recreation planning of a Bahamian National Park. Soci-tural Resources, 22(2), 111127., Kyhk, N., Ndumbaro, F., & Khamis, M. (2012). Community stakehol-ledge in landscape assessmentsMapping indicators for landscapecological Indicators, 18, 421433.aftery, A. E. (1998). How many clusters? Which clustering method?ia model-based cluster analysis. The Computer Journal, 41(8), 578588.isholm, T. (2000). Conservation or cultural heritage? Cattle grazing inia Alpine National Park. Ecological Economics, 33(1), 6375.(1984). Strategic management: A stakeholder approach. Boston: Pitman.A. P., Quintas-Soriano, C., Garca-Llorente, M., Palomo, I., Montes,tn-Lpez, B. (2014). Collaborative mapping of ecosystem services:f stakeholders proles. Ecosystem Services, http://dx.doi.org/10.1016/14.11.006stralia. (2014). Land tenure. Available from http://www.ga.gov.au/

/geoscience-basics/land-tenure.html Accessed 28.04.14 F. (2007). Citizens as sensors: The world of volunteered geography.l, 69(4), 211221.

Wellard, K. (1997). Stakeholder methodologies in natural resourceent: A review of principles, contexts, experiences and opportunities.l Systems, 55(2), 173193.

, & Sofroniou, N. (1999). The multivariate social scientist. London: Sagens Ltd.

G. Brown et al. / Landscape and Urban Planning 139 (2015) 115 15

Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Edu-cational and Psychological Measurement, 20(1), 141151.

Kingdon, J. (1995). Agendas, alternatives, and public policies. Boston: Little, Brown.Newsome, D., Moore, S. A., & Dowling, R. K. (2012). Natural area tourism: Ecology,

impacts and management (2nd ed.). Clevedon, UK: Channel View Publications.NRCMNatural Resource Council of Maine. (2014). Public land ownership by state.

Available from http://www.nrcm.org/documents/publiclandownership.pdfAccessed 28.04.14

OREN. (2014). Overview OREN/Otway forest campaign (19952008). Available fromhttp://www.oren.org.au/campaign/intro.htm Accessed 22.12.14

Palomo, I., Martn-Lpez, B., Potschin, M., Haines-Young, R., & Montes, C. (2013).National parks, buffer zones and surrounding lands: Mapping ecosystem serviceows. Ecosystem Services, 4, 104116.

Parks Victoria. (2009, December). Great Otway National Park Management Plan. Gov-ernment of Victoria., ISBN 9-7807-3118-384-5. Available from http://parkweb.vic.gov.au/ data/assets/pdf le/0019/313282/great-otway-np-mp.pdfAccessed 22.12.14

Reed, M. S., Graves, A., Dandy, N., Posthumus, H., Hubacek, K., Morris, J., et al. (2009).Whos in and why? A typology of stakeholder analysis methods for naturalresource management. Journal of Environmental Management, 90(5), 19331949.

Rees, S. (2014). A new national park. Interaction, 42(4), 812. Available fromhttp://search.informit.com.au/documentSummary;dn=776222132814300;res=IELHSS Accessed 22.12.14

Ruiz-Frau, A., Edwards-Jones, G., & Kaiser, M. J. (2011). Mapping stakeholder valuesfor coastal zone management. Marine Ecology Progress Series, 434, 239249.

Sabatier, P., & Jenkins-Smith, H. (1993). Policy change and learning: An advocacycoalition approach. Boulder, CO: Westview.

Schlossberg, M., & Shuford, E. (2005). Delineating public and participation in PPGIS.URISA Journal, 16(1), 1526.

Sherrouse, B. C., Clement, J. M., & Semmens, D. J. (2011). A GIS application forassessing, mapping, and quantifying the social values of ecosystem services.Applied Geography, 31(2), 748760.

Sieber, R. (2006). Public participation geographic information systems: A literaturereview and framework. Annals of the Association of American Geographers, 96,491507.

Van Riper, C. J., & Kyle, G. T. (2014). Capturing multiple values of ecosystem servicesshaped by environmental worldviews: A spatial analysis. Journal of Environmen-tal Management, 145, 374384.

Weber, D., & Brown, G. (2014). Identifying & mapping the values of Victorian publiclands. University of South Australia and University of Queensland. Available fromhttp://www.landscapemap2.org/publications/StatewidePVreport2014.pdfAccessed 03.03.15

Western, D., & Henry, W. (1979). Economics and conservation in third world nationalparks. BioScience, 29(7), 414418.

WDPA. (2014). World database on protected areas. Available from http://www.wdpa.org/resources/statistics/2013 MDG Regional and global stats Indicator7 6.xlsx Accessed 19.04.14

Zanon, D., Hall, J., Lockstone-Binney, L., & Weber, D. (2014). Development of a wholeagency approach to market segmentation in parks. Journal of Leisure Research,46(5), 563592.

Identifying public land stakeholder perspectives for implementing place-based land management1 Introduction1.1 Public participation GIS (PPGIS) and stakeholder analysis1.2 The study contextpublic land values and preferences

2 Methods2.1 Study location and context2.2 Data collection process2.3 Public land management preference scale2.4 Analyses2.4.1 Public land management components (aim 1)2.4.2 Cluster analysis and interpretation (aim 2)2.4.3 Analysis of mapping behavior by stakeholder group (aim 3)2.4.4 Modeling group preferences by location (aim 4)

3 Results3.1 Participation3.2 Public land management components3.3 Cluster analysis results3.4 Interpretation and validation of the stakeholder groups3.5 Stakeholder group mapping behavior3.6 Spatial distribution of stakeholder preferences

4 Discussion4.1 Study limitations and future research

References