R

Tm

PBa

b

c

h

•••••

a

ARRA

KSPRDA

1

aoP

als

0h

Landscape and Urban Planning 119 (2013) 147– 157

Contents lists available at ScienceDirect

Landscape and Urban Planning

jou rn al hom ep age: www.elsev ier .com/ locate / landurbplan

esearch paper

aking into account farmers’ decision making to map fine-scale landanagement adaptation to climate and socio-economic scenarios

énélope Lamarquea,∗, Aloïs Artauxa, Cécile Barnaudb, Laurent Dobremezc,aptiste Nettierc, Sandra Lavorela

Laboratoire d’Ecologie Alpine, LECA, UMR 5553 CNRS Université Joseph Fourier, BP 53, 2233 Rue de la Piscine, 38041 Grenoble cedex 9, FranceINRA-SAD, UMR Dynafor, 24 Chemin de Borde Rouge, Auzeville, CS 52627 31326 Castanet Tolosan cedex, FranceIrstea Grenoble, Domaine universitaire, BP 76, 38402 Saint-Martin d’Hères cedex, France

i g h l i g h t s

Participatory regional scenarios combining climate and socio-economic change.Role-playing game to map quantitatively small-scale changes in land management.Drought frequency interacts with socio-economic context to determine farm adaptation.High mountain conditions strongly constrain available options for farm adaptation.Recurring drought threatens mountain agriculture and farmers livelihoods.

r t i c l e i n f o

rticle history:eceived 28 September 2012eceived in revised form 23 July 2013ccepted 24 July 2013

eywords:cenariosarticipatory approachole-playing gamerought

a b s t r a c t

Mountain grassland ecosystems are particularly vulnerable to direct climate impacts and to indirect cli-mate change impacts through farmers’ management adaptation. We modelled expected spatio-temporaltrajectories of land management of a mountain grassland landscape in the French Alps under a range ofshort-term climate and socio-economic scenarios which were constructed using an advanced partic-ipatory approach with a variety of stakeholders. First, regional experts from nature conservation andagricultural extension were involved in the co-development of detailed qualitative climate and socio-economic scenarios, expressed as coherent storylines. Second, to map land management adaptation tothese storylines, we used a role playing game whereby farmers were put in an imaginary future situa-tion and asked to make decisions under scenario constraints. For each scenario, game outcomes were

daptive grassland management used to map future land management at parcels to landscape scales. Main adaptations were conversionfrom mowing to grazing and increasing manured area, with varying proportions and locations for thesetwo types of changes differing across scenarios, though overall small. These results highlight the limitedadaptability of current farmers given a strongly constraining natural and social context. Beyond researchoutputs, this framework generated interesting outcomes for stakeholders and raised their awareness

l syst

about the socio-ecologica

. Introduction

Mountain ecosystems are highly vulnerable to climate changend extreme events (Engler et al., 2011), such as the increasedccurrence of drought observed over the last decades (Lemaire &flimlin, 2007) and predicted in the future (IPCC, 2012). Because

∗ Corresponding author. Tel.: +33 476635661/+33 479350685.E-mail addresses: penelope.lamarque@gmail.com (P. Lamarque),

.artaux@gmail.com (A. Artaux), cecile.barnaud@toulouse.inra.fr (C. Barnaud),aurent.dobremez@irstea.fr (L. Dobremez), baptiste.nettier@irstea.fr (B. Nettier),andra.lavorel@ujf-grenoble.fr (S. Lavorel).

169-2046/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.landurbplan.2013.07.012

em’s vulnerability to future changes.© 2013 Elsevier B.V. All rights reserved.

mountain livestock farming relies mostly on local grasslandproduction (Lemaire and Pflimlin, 2007), the vulnerability of man-aged grasslands to climate change results from a combination ofdirect responses of vegetation, and of indirect effects through theimpacts on vegetation of management adaptations (e.g., grazing ormowing) (Vittoz, Randin, Dutoit, Bonnet, & Hegg, 2009). Therefore,and because agricultural decision-making is influenced by a varietyof drivers (e.g., policies, societal values) (Nettier, Dobremez, Coussy,& Romagny, 2010), the socio-economic and ecological contexts inwhich climate problems occur are likely to be as important as the

climate shock itself (Fraser et al., 2011).To enhance the understand-ing of the complex interactions and the dynamics of change of allthese parameters, environmental and land-use scenarios have beendeveloped and used.

1 Urba

tt(ohDAsnAam2t2

mdasaPitt2stJshs

lsm(ebamimmpc&sdctsa2rsh&rot&fmV

production of calves and heifers (three farms, mean = 67 LU, 55 ha),(3) mixed sheep and cattle production (two farms, mean = 54 LU,48 ha). During summer, the alpine meadows are managed by ashepherd who looks after the local sheep herds along with his own

Table 1Current characteristics of grasslands types of Villar d’Arène (minimum–maximum).

Grassland type Altituderange (m)

Slope(degree)

Forageproduction(T/ha)

Terraces mown and manured 1584–1944 0.16–37 3.07–4.63Terraces mown 1554–1938 1.12–61.3 3.00–4.87

48 P. Lamarque et al. / Landscape and

Scenarios are “plausible and often simplified descriptions of howhe future may develop based on a coherent and internally consis-ent set of assumptions about key driving forces and relationships”Millennium Ecosystem Assessment, 2005). “Predictive” scenariosf the type “What-if” address the specific question: “What willappen, on the conditions of specific events?” (Borjeson, Hojer,reborg, Ekvall, & Finnveden, 2006), as we did in this study.nswers to this question can be attained by quantitative (e.g.,imulation modelling; Schroter et al., 2005) or qualitative sce-arios (e.g., SRES storylines) or both (e.g., Millennium Ecosystemssessment, 2005; Walz et al., 2007). Each method has advantagesnd drawbacks but qualitative scenarios allow to take into accountore variability and uncertainty than quantitative models (Coreau,

009) and are more understandable for communicating informa-ion about the future than numeric tables or graphs (Alcamo et al.,006; van Vliet, Kok, & Veldkamp, 2010).

Usually, scenario building follows a framework composed of fiveain steps: (1) defining the focal question, (2) identifying the key

rivers, (3) determining the scenario logics, (4) describing scenariossumptions using qualitative scenario storylines, and (5) assessingcenario outcomes or developing quantitative scenarios based on

numerical model (Metzger, Rounsevell, Van den Heiligenberg,erez-Soba, & Hardiman, 2010). Qualitative and/or regional scenar-os tend to involve stakeholders and experts in a set of procedureshrough which they work together to develop the storylines (steps 1o 4) (Alcamo et al., 2006; Volkery, Ribeiro, Henrichs, & Hoogeveen,008). Their involvement is particularly valuable due to their exten-ive local knowledge (Swetnam et al., 2011; Walz et al., 2007) andhe fact they are very likely to be actors themselves (Bohunovsky,ager, & Omann, 2010). Several methods are used in participatorycenario development such as interviews or focus groups, stake-olders panel workshops, gaming workshops, policy exercises, ortory and simulation approaches (Alcamo et al., 2006).

Quantitative land-use change models used to integrate story-ines to visualize alternative land-use configurations are typicallytatistically-calibrated models (Verburg et al., 2002), economicodels (Janssen & van Ittersum, 2007), or agent-based models

Valbuena, Verburg, Veldkamp, Bregt, & Ligtenberg, 2010). Whileconomic models assume perfect rationality of stakeholders, agent-ased modelling (ABM) has the advantage of allowing spatialnalyses of the interactions between agents and their environ-ent (Matthews, Gilbert, Roach, Polhill, & Gotts, 2007), by taking

nto account explicitly a greater complexity of agent-decisionsaking processes (Valbuena et al., 2010). In ABMs, the decision-aking process of agents is generally based on rules defined and

arametrized with either artificial or empirical data. As it requiresomplex and intensive data gathering (Smajgl, Brown, Valbuena,

Huigen, 2011; Valbuena, Verburg, & Bregt, 2008), only a limitedet of parameters is usually taken into account. Moreover, in ABMs,ecision-making is often the only cognitive component explicitlyonsidered to explain how agents change their land-use prac-ices under changing conditions. The assumption of a simple andtable set of preferences and decisions that underlies modellingpproach is insufficient in case of non-linear change (Meyfroidt,012). Similarly to ABMs, role-playing games (RPG) focuses on theole of agents of land-use changes (as opposed to drivers as intatistically-calibrated models), but RPGs, by putting real stake-olders in a close to real situation during the game (Castella, Trung,

Boissau, 2005), allow measuring and analysing impacts of theeal complexity of human cognition and social interactions. More-ver, stakeholders can debate whether the model represents howhey played the RPG, and how the RPG differs from reality (Janssen

Ostrom, 2006). Because of the additional knowledge gainedrom RPG compared to surveys, RPG can also be used to develop

ore realistic ABM (Janssen & Ostrom, 2006; Smajgl et al., 2011).alidation of modelling results dealing with the complexity and

n Planning 119 (2013) 147– 157

uncertainty of human decision-making requires further interviewswith stakeholders involved or experts to verify the plausibility ofthe simulated results (Valbuena et al., 2010).

Considering the advantages and shortcomings of various exist-ing approaches, this study aimed at developing and testing aframework combining a role playing game with ancillary techni-cal information and spatial statistics on current land-use to modelspatially-explicit, fine-scale land-use adaptation to climate andsocio-economic change, taking into account real farmers’ decisionmaking, beliefs and values. For that purpose, we used a three-stepparticipatory scenario approach combining: (i) the development ofregional expert-based qualitative storylines at regional scale, (ii)identification and quantification of local land-use adaptation usinga RPG with farmers and (iii) a systematic mapping of these land-useadaptations on the real landscape at field scale. The integration ofthese three steps, combining qualitative and quantitative methodsfrom different disciplines, allows to integrate and to translate qual-itative driving forces and farmers land-use decisions into spatialmap of land-use change at small-scale.

This framework was tested on a municipality hosting a long termsocio-ecological research platform (LTSER) where detailed data onvegetation, climate and farming systems have been collected for adecade. These highly detailed data and local knowledge allowed usto easily test and put into practice the framework.

2. Study area

The study site (45◦03′N, 6◦24′E) is located within the munici-pality of Villar d’Arène, located in the Ecrins National Park (CentralFrench Alps). The total area of the south-facing slopes used for live-stock production is 13 km2 and the elevation ranges from 1552 to2500 m asl. Climate is subalpine with a strong continental influ-ence due to a rain shadow with respect to dominant westerlywinds. Mean annual rainfall is 956 mm and mean monthly tem-peratures of −4.6 ◦C in January to 11 ◦C in July (at 2050 m above sealevel). Only, 40% of annual rainfall occurs during the growing sea-son (April–September). The alpine meadows above 2200 m havebeen grazed extensively in summer for centuries, but the formerarable land on terraced slopes (1650–2000 m) was converted atthe beginning of the 20th century into grasslands that are nowgrazed or mown where they are accessible to machinery (Girel,Quétier, Bignon, & Aubert, 2010). All grassland types (Table 1) havebeen managed at low intensity with low stocking rates, very lowmanure inputs (every two or three years) and a single annual haycut. Farmers try to be fodder self-sufficient for the long winter sea-son (6 to 7 months) given the cost of fodder. Eight farmers remaintoday, of which five farm full-time, one farms part-time and two areretired but continue to farm. The eight farms can be classified intothree categories according to their production systems: (1) lambproduction (three farms, mean = 21 livestock units (LU), 19 ha); (2)

Terraces grazed 1539–1794 0.39–59 2.98–4.86Unterraced grasslands mown 1854–2013 1.65–34.2 3.57–4.64Unterraced grasslands grazed 1702–2024 1.28–43.7 3.12–4.90Alpine meadows 2228–2710 0.33–63.5 3.16–4.18

Urba

spsfttaiTn

heff1mlufmmparfitf

3

oefc

Fm

P. Lamarque et al. / Landscape and

heep (around 1500 sheep). Only two farmers sell a part of theirroduction by direct sales. In the region, European subsidies con-titute a significant share of farm income as a means to make upor low productivity. In this municipality, agricultural parcels ofhe multiple landowners are pooled into a communal organiza-ion (Association Foncière Pastorale (AFP)) which allocates parcelsmong farmers in order to distribute land more equitably accord-ng to technical constraints associated with the mountain terrain.ourism is another dominant economic activity in the region recog-ized for its aesthetic, cultural, conservation and recreation values.

The main dataset used in this study is a land management (usedereafter instead of the commonly used term land-use, due to thexclusive use of grassland for livestock production) map derivedrom different surveys conducted between 2003 and 2009 witharmers and analysis of aerial photographs and cadastral maps since810 (Girel et al., 2010; Quétier, Thebault, & Lavorel, 2007). Thisap contains six land management classes (Table 1) at a reso-

ution of 20 m and has been given a nominal date of 2009. Wesed additional spatial information: elevation and slope obtainedrom a 10 m Digital Elevation Model under ArcGiS 9.2, ESRI; settle-

ents, farms and roads digitized from the 1:25000 topographicap (IGN); fodder production modelled according to altitude, soil

arameters and plant functional traits (Lavorel et al., 2011). Inddition, we used results from a survey of farmers’ responses toecent droughts (Nettier et al., 2010). Finally we used long-termeld observation and on-site experimental data (Benot et al., 2013)o quantify expected drought effects on vegetation composition andodder production.

. Research framework

The step-wise framework presented hereafter aims at devel-

ping, with a variety of stakeholders, local climate and socio-conomic scenarios to understand adaptive management ofarmers to changing contexts and to map land managementhanges at parcels, farms and landscape scales (Fig. 1).

ig. 1. Framework of climate and socio-economic scenarios (light grey) and land manageent (Metzger et al., 2010).

n Planning 119 (2013) 147– 157 149

3.1. Co-construction of regional storylines with experts

The first four steps of scenario development (Fig. 1) aimed atbuilding detailed and relevant short-term (around 2020) qualita-tive scenarios expressed as coherent storylines at regional scale.Participants and scientists agreed to construct four scenariosfollowing a prospective approach (de Jouvenel, 2002) couplingtwo climate alternatives with two socio-economic alternatives.Socio-economic alternatives were downscaled from national andinternational scenarios (Agrimonde®, 2009; Millenium EcosystemAssessment (MEA) Carpenter, Pingali, Bennett, & Zurek, 2005; LesNouvelles Ruralités (NR) Mora, 2008): One reflected continuedglobalization (see ‘Global orchestration’ in MEA, G0 in Agrimonde®and Scenario 1 in NR) and the other one explored an increasein regionalization of policy and economics with flexible gover-nance (see ‘Adapting mosaïc’ in MEA and scenario 4 in NR). Climatealternatives were formulated on the basis of downscaled projec-tions of climate change models for 2050 (Boe, Terray, Habets, &Martin, 2006) and represented as a five year window.

Participants, chosen for their local expertise (regional experts),were involved in order to draw contrasted and plausiblealternatives, reflecting local features and being relevant to localstakeholders (Alcamo et al., 2006). The selection of experts (10experts from land management and nature conservation bodies and9 researchers in agronomy, forest science and ecology) ensured adiversity of knowledge which is important to avoid a single visionof the future (Volkery et al., 2008). Co-construction with expertswas conducted during two workshops (1/2 day + 1 day) structuredfollowing the usual steps of scenario building (step 1.1 to 1.4 inFig. 1) with additional individual interviews to refine storylinesafter the second workshop. A particular attention was devoted togiving, along with general trends (e.g., % precipitation decrease) as

in usual storylines, concrete scenario elements for presentation tofarmers such as expected effects on vegetation composition, fod-der production and water availability across the five scenario years(step 1.4 in Fig. 1).

ment scenarios (dark grey) building according to usual phases of scenario develop-

1 Urba

aeos

3m

nb2haimfamst(t&p2t2ctte

hdoMmdRaiuiwvbbbr

wtttsw2moutwt

50 P. Lamarque et al. / Landscape and

Discussions of the second workshop also focussed on possiblenticipation of adaptations based on recent responses to climate orconomic events (Nettier et al., 2010) so as to ensure acceptabilityf scenarios to farmers and to avoid in so far as possible eitheretback or win–win scenarios.

.2. Development of the role playing game to map local landanagement adaptation

To help farmers propose coherent adaptations, making the sce-arios complexity quickly understandable is a challenge that cane addressed by role-playing games (RPG) (Martin, Felten, & Duru,011). RPG is a model putting players in a given situation with theelp of different supports (Castella et al., 2005). Our choice to use

RPG rather than more traditional methods such as semi-directiventerviews or a focus group was motivated by six objectives: pri-

arily (1) to make scenario content easily understandable byarmers (Martin et al., 2011), (2) to develop a spatial support as

means to downscale quantitatively and spatially land manage-ent change (Washington-Ottombre et al., 2010) according to the

cenarios, (3) to take into account and identify decisions and fac-ors that drive land management change (Pak & Brieva, 2010), and4) to stimulate collective strategies and to understand interac-ions among farmers (Barnaud, Van Paassen, Trébuil, Promburom,

Bousquet, 2010) which are likely to be important in this munici-ality given the presence of the AFP collective mechanism (Section). In addition to these expected outcomes, RPG also allows (5)o elicit stakeholders’ representation of the system (Castella et al.,005) as players bring along their own habits and strategies to makehoices within contexts given by the game (Naivinit, 2009), and (6)o share knowledge between scientists (drought effects on vegeta-ion) and farmers (effects on farm and land management) (Martint al., 2011).

A major stake for livestock farmers is to manage to feed theerd during the vegetation season (May to November) and to pro-uce enough stock of forage for the entire winter. This central issuef forage sufficiency was the basis for the game design. Similar toartin et al. (2011), the aim for players was thus to successfullyeet this challenge on an annual time step given grassland pro-

uction under climate constraints. The value and originality of thePG in this research lies in its pivotal role to map land managementdaptations in the context of scenarios (step 2 in Fig. 1), by takingnto account decision-making processes of real farmers. For this, wesed a two-dimensional board game composed of cells represent-

ng a fictitious landscape where farmers, playing their own role,ere asked to put pieces representing their herd and fodder har-

est according to rules translating the effect of climate scenarios oniomass production for each type of grasslands (Table 1, Fig. 2). Theoard retained the actual proportions of grassland types managedy each farmer, which was essential to allow us to translate RPGesults quantitatively to the landscape map.

The game was designed by the scientists and adapted to fitith the Villar d’Arène socio-ecological system composed of the six

ypes of grasslands (Table 1) managed by eight farmers. However,he details of the game could be redesigned according to specifici-ies of other sites (Section 5.2). Knowledge and data on multipleocial or technical parameters obtained during surveys conductedith farmers in 2009 were used to build the game (Lamarque,

012). The game in itself was a translation of the storylines of cli-ate alternatives which described the impacts of drought events

n grassland production, while socio-economic conditions were

nchanged from the current context. The game was organized inwo sessions corresponding each to a climate alternative. A sessionas divided into five rounds representing a year each, therefore the

ime step was one year and the time horizon was five years later.

n Planning 119 (2013) 147– 157

Rules on forage production determined for each round, accord-ing to climate conditions, how many pieces of each type (cows,sheep, fodder and/or manure) could be placed on the different cellsof the board representing the land management types (Fig. 2). Giventhe annual time step, and that most fields can only be used once ayear due to the absence of regrowth under the constraining climateof the site, two different shapes for pieces were used to distin-guish spring from summer grazing. During the game farmers wereallowed to request more pieces or pieces representing any othertype of animals or forage production (e.g., cereals) to support theiradaptation to climate change. If farmers could not place the totalityof their livestock pieces on the board they had to either purchasethe complement of fodder pieces or sell the corresponding numberof livestock pieces.

The game started with the board game filled with piecesallocated according to each farmer’s current farm characteristics(corresponding amount and type of land and herd). Each round fol-lowed five steps: (1) reading of storylines by the game coordinator,(2) distribution of cards giving practical information on constraintsto grassland production, expressed as number of pieces allowed bycell type, (3) during the action itself players may move, remove oradd pieces from the board and talk together, (4) summary of fod-der and herd outcomes for each farmer into a purpose-designedtable, (5) group debriefing about main actions. The first round wasgiven a favourable climate year enabling farmers to understand thegame and to move pieces around in case they disagreed with theinitial state as compared with their own vision of their farm sys-tem. Climate of the subsequent rounds followed the sequences ofthe corresponding storyline.

Each game session ended with a debriefing session whichallowed players to debate about the relationships between theactions taken during the game and the real world (Barreteau,Le Page, & Perez, 2007). The entire game ended with a gen-eral discussion of how socio-economic alternatives may favour orconstrain adaptation of each farmer in response to each of the cli-mate alternatives. We presented the parameters of socio-economicalternatives to farmers based on graphs and landscape drawings,and their effects on adaptations to each climate alternative werediscussed and summarized on a paper board. This final step allowedus to tease out climate from socio-economic effects in our analy-sis of adaptation pathways. The full game, including debriefings,lasted 5 h (2 h the morning and 3 h the afternoon), and was videorecorded.

3.3. Validation of game adaptation and land managementchanges

Due to the complexity and uncertainty of human decision-making, plausibility of the game results for farmers adaptationswas assessed by cross-checking them with information from mul-tiple sources (step 3 in Fig. 1). First, on the day following the gamethe relevance of game actions was discussed individually with thefarmers to place them out of the idealized space offered by thegame. These interviews also allowed us to understand individualrationales underlying the actions during the RPG and to discuss thequestion of the difference between the RPG and reality (Castellaet al., 2005). Available narrative extracts from the RPG video wereused to identify spatially-explicit biophysical or socio-economicdrivers of land-use decisions (Washington-Ottombre et al., 2010).Second, feasibility of proposed adaptations was assessed with tech-nicians from local agriculture and environment extension services.Third, statistical analyses were carried out to validate adaptations

quantitatively and spatially. The possibility of mechanization ofparcels according to slope was checked for mowing and fertilizationby statistical analysis of the spatial distribution of current practises.The currently mechanized parcels (manured and/or mown) have a

P. Lamarque et al. / Landscape and Urban Planning 119 (2013) 147– 157 151

F ds typa

31Twseodstpbet

3

ttt(aagtn

cmppam

ig. 2. RPG components. The coloured cells of the board game represent grasslanccording to rules of climate effects on forage production for each grassland types.

rd quartile slope of 18◦ and an average of 15◦, hence a slope below8◦ was considered mechanizable and easily mechanizable at 15◦.he maximum potential of fertilized surfaces (organic manure onlyere used) were calculated for each farm based on their herd

ize in each scenario with the help of technician from agriculturalxtension services (the reference for calculation was an averagef 20 T/ha of manure spread every three years and an annual pro-uction of manure of 4.5 T/livestock unit). Other decision criteriauch as distance to the farm, time of travel or difficulties to reachhe parcels were not taken into account quantitatively in the sim-le model presented here due to the amount of data required andecause they were considered less restrictive than slope by farm-rs. In addition, any mention to a specific parcel by a farmer wasaken into account.

.4. Generation of local scale land management scenario maps

Final state (round 5) of the game board for each session wasranscribed into a GIS (ArcGIS 10, ESRI) (step 4 in Fig. 1). First, prac-ices applied in each cell were translated into land managementypes (Table 1) based on a land-use state and transition modelQuétier et al., 2007). Modifications or conversions of land man-gement types were translated as a transition from one type tonother (e.g., conversion from a mown unterraced grasslands to arazed unterraced grasslands), except for the fertilization on grazederraces and unterraced grasslands outside terraces which lead toew type of land management (Figs. 3 and 4).

Second, we rescaled the board to the actual landscape by cal-ulating the proportion between the number of cells of each landanagement type by each farmer on the board and real land

arcels. Third, we applied rules based on the statistical validationrocesses (Section 3.3) to adjust the manured area with the avail-ble manure quantity, and the manured and mown area with theaximum slope allowing mechanization of parcels.

es on which farmers put pieces that represent their herd and the hay harvested

4. Results

4.1. Climate and socio-economic scenarios

The two workshops with experts lead to four scenarios combin-ing two socio-economic alternatives and two climate alternativeswith relevant details for the study area (Table 2 and Appendix A)in terms of general drivers and also on expected consequenceson grasslands. The “intermittent” climate context expressed anincrease in the frequency of spring or summer drought periodsalternating with wetter years, consistent with the recent situationin some regions of the Alps. The “drastic” climate context includedfour consecutive years of springtime drought. Effects of drought onvegetation were considered very weak in the “intermittent” contextconsistent with the high resilience of alpine vegetation to droughtsand experimental drought simulations.

In the “international” socio-economic context, stakeholdersfaced continuing globalization and urban concentration, with agri-culture being supported only for its role as a producer of globalenvironmental services (e.g., carbon storage, open landscapes). Inthe “local” context, as citizens showed a growing interest in theirgeographic area and its activities, and subsidies to mountain farm-ing were maintained allowing it to remain a producer of qualityfood in line with strict environmental requirements.

4.2. Projections of land management change

The two RPG sessions lead to contrasted landscape config-urations in terms of land management (Fig. 3). Two core landmanagement changes were proposed by farmers, shift from mow-

ing to grazing and increase in manuring (Fig. 4).

In the “drastic” climate alternative almost all farmers ceasedmowing due to the strong decrease in grassland production andto the priority given to sustaining grazing. Remaining mown

152 P. Lamarque et al. / Landscape and Urban Planning 119 (2013) 147– 157

Fig. 3. (a) Current land management on the Villar d’Arène landscape; (b) Scenarios maps obtained from the analyses of the board games and rule-based decisions on slopeconstraints to mechanization. Pie charts give the percentage of each land management type in the landscape.

Fig. 4. Land management transitions under the four scenarios. Climate alternatives are described by the upper and lower rows of boxes and numbers give the percentage(a) and (b) of land in the landscape respectively for: (a) the “local” alternative and (b) the “international” alternative (differences in the total are due to rounding). Arrowsshow transition from one land management type to another.

P. Lamarque et al. / Landscape and Urban Planning 119 (2013) 147– 157 153

Table 2Scenarios drivers and related assumptions. Example of the corresponding storylines for the “drastic and local” scenario is presented as supplementary material (AppendixA).

Drivers Climate alternatives

Drastic IntermittentTemperature rise No NoSeason of drought and occurrence Spring drought during four consecutive years Spring or summer drought every two yearsEffects on vegetation Change in species composition. Development of

species adapted to drought (eg., Festuca paniculata,Carex sempervirens)

No change

Effects on biomass production Decrease by more than 50% Decrease by 15% during drought yearsEffects on water quantity (springs) Decreased flow of all springs, even quenching of the

less productive onesDecrease flow of the weaker springs

Socio-economic alternativesLocal International

Consumption demand Local and high quality products Cheapest pricesAim of agricultural subsidies To maintain both an agriculture with quality

production and a high level of ecosystem services andbiodiversity conservation. High subsidies but morerestrictive in term of expected outcomes than in the“International” alternative.

To maintain open landscapes and production ofenvironmental services such as carbon sequestration.Lower subsidies than on the local alternative, but lessrestrictive. A minimal income is guaranteed to farmers

Agricultural input prices (fodder, straw) Variations due to climate Variations due to climateAgricultural product prices 15% decrease only for conventional products 15% decrease even for quality labelled product and

−10% for organic productsagri-to

siswd“etomdt1s

issstt1(

tmtmeetdtf(nanh

mb

Part-time job opportunities On-farm pluri-activity linked to

urfaces made up 8% of the landscape in the “drastic and local” (onlyn unterraced grasslands) and 12% in the “drastic and international”cenario (9% in terraces, 3% in unterraced grasslands), comparedith the current 28% (Fig. 4). Farmers had to compensate for fod-er supply by purchases, regardless of economic difficulties. Theintermittent” alternative was seen as business-as-usual by farm-rs (Fig. 4: lower part) given frequently recurring droughts overhe last decades. The annual time step of the game allowed us tobserve that the extent of mowing was reduced in response to sum-er rather than spring droughts. Farmers replenished fodder stocks

uring good years to buffer supply in drought years. Overall, underhe “intermittent” alternative mowing was reduced to 27% and8%, respectively for the landscape in the local and internationalocio-economic alternatives, respectively.

Increased manuring was seen by farmers as a means ofncreasing grassland production and thereby fodder yields andtocking rates in both scenarios. The proportion of manuredurfaces increased from 7% (current) to 16% under the “local”ocio-economic alternative whether on grazed surfaces underhe “drastic” alternative or mown surfaces under the “intermit-ent” alternative. Similarly, manured surfaces reached 15% and3% in the “intermittent-international” and “drastic-international”almost exclusively mown surfaces) scenarios, respectively.

Socio-economic alternatives appeared to affect adaptation par-icularly regarding the decision to maintain mowing in less

echanizable parcels, and herd size. In the “international” alterna-ive, less restrictive Common Agricultural Policy subsidies on the

inimum stoking rate and on required practices allowed farm-rs to decrease their herd and to restrict mowing to the moreasily mechanizable parcels. In contrast, in the “local” alterna-ive, herd size could be decreased thanks to the value added byirect sales. Moreover, the context of the “international” alterna-ive was not favourable to new farmers taking over land of retiredarmers, hence parcels were redistributed among current farmersredistribution of 15% area in the “drastic and international” sce-ario; 5% in the “intermittent and international” or the “drasticnd local” scenario; and 1% in the “intermittent and local” sce-ario), allowing them to maintain or only slightly decrease their

erd.

Beyond effects on the farm functioning in itself, these landanagement changes translated to strong landscape effects

y modifying the proportion and location of different land

urism Off-farm pluri-activity job opportunities

management types. Changes occured mainly in the lower part(terraces), and in the international context also in mown unterracedgrasslands (Fig. 3). The “drastic and local” scenario incurred themost dramatic change with all the terraces managed only by graz-ing (about a doubling, with grazing on terraces increasing from 12%to 27% of the total area, of which 14% fertilized).

5. Discussion

5.1. Methodological relevance and opportunities

The main objective of this study is to test a framework for gen-erating spatially-explicit projections of land management undercombined climate and socio-economic scenarios. The originalityof the framework lies in the consideration of three features offarmers’ decision-making and adaptations that differ from exist-ing simulation-based approaches to land-use change modelling.First, quantitative land-use models usually assume that eachagent uses only one decision-making system because multipledecision-making schemes are combined by applying a typologyof decisions-makers (here farmers) (Smajgl et al., 2011; Valbuenaet al., 2008). Second, behaviours and attitudes of farmers are con-sidered constant during the duration of scenarios and their possiblechange is not anticipated, except in models including several stepsof feedback loops (Le, Seidl, & Scholz, 2012). Third, boundaries ofpossible adaptations set by the model design are based on current orpast land management obtained through census data or interviewsand specified a priori by researchers. The strength of our frameworkto overcome these limits rests mainly on the pivotal role playedby the RPG and the attainment of its six objectives (described inSection 3.2 and mentioned below).

In our framework, flexibility of behaviours, decision-makingand adaptations were intrinsically considered by the directinvolvement of farmers during the game. Indeed, in comparison toother participatory tools such as interviews based on a hand-drawnmap the RPG had the advantage of putting people in the futurecontexts and at the same time testing adaptation in a “concrete”

situation, instead of only imagining adaptations out of context (RPGobjective 1).The farmers’ experiences and knowledge were usedto focus on conditions that are important for them rather thanthose assumed by the researcher for which data are readily

1 Urba

artuArttoaaatgidtonei

tetoa

tvpmfwps

bepiociinMat6tPsTcwopwdsis6

t

54 P. Lamarque et al. / Landscape and

vailable. No assumption or categorization were made byesearchers because the pivotal role of RPG and the directranscription of its results on landscape maps do not require sim-lation modelling of farmers’ decisions (RPG objective 2). Like inBMs, the game was designed for specific adaptations expected byegional experts. But the game included numerous possible adap-ations for grassland management (e.g., modification of livestockype and/or size, stocking rate, fertilization rate, location and areaf management type; transition from one management type tonother; stopping management or leaving the farm, buying fodder)s well as any other type of adaptation that farmers would proposend would be discussed during the game (e.g., off-farm activities,ransition to arable crops). Moreover, discussions during theame, collective debriefings and individual post-game interviewsncreased understanding of the rationale behind game actions andeterminants of adaptation options (RPG objective 3). Althoughhe game takes into account group decisions, none of themccurred despite discussions among farmers about the opportu-ity of such collective decisions. But, expected neighbourhoodffects occurred such as discussion between players, advices, andmitation between farmers, much like in real life (RPG objective 4).

As the RPG is not coupled with an ABM, technical skills and staffo run the model and to interpret outputs are not required (Alcamot al., 2006). But, data and resources as well as stakeholders neededo implement the framework are important, but no more than anyther participatory tools used to implement land-use model suchs ABM.

Therefore, in the same way as previous studies about participa-ive approaches involving several stakeholders (Alcamo et al., 2006;an Vliet et al., 2010), it is important to assess to which extent theroposed framework meets the four criteria of scenarios assess-ent (Alcamo et al., 2006): relevance (how relevant are scenario

or end users?), credibility (is the scenario plausible and consistentith existing information?), legitimacy (does the scenario reflectoints of view that are perceived to be fair?) and creativity (do thecenarios challenge the current view of the future?).

Relevance is an important issue in participatory researchecause it increases stakeholders’ willingness to participate (Walzt al., 2007) and reduces the risk to disappoint them during therocess (Barreteau et al., 2007). Our iterative framework (Fig. 1)

s appropriate for this purpose because it helps the different typesf potential end-users to get relevant scenarios according to theironcerns (Alcamo et al., 2006). Regional experts were interestedn the consequences of land management changes; our results fornstance stimulated their discussions about tools of local gover-ance (e.g., financial incentives, animation to increase awareness).ore than half of the farmers considered that the RPG made them

ware of the potential vulnerability of their farm system and ofhe effects of adaptations for the whole landscape (RPG objective). A farmer said afterwards “when I came back home, I thoughto all these things we discussed during the RPG. It was formative!”.olicy-makers could be interested by looking at the influence of theocio-economic alternatives on farmers’ adaptations to droughts.he understanding of whether or not farmers’ adaptations to thelimate context are influenced by policies or consumers’ demandas enhanced by focussing the game initially on climate change

nly and discussing the effects of socio-economic alternatives onlanned adaptations in a second step. Finally, this original frame-ork can be useful for researchers in land-use science (see aboveiscussion about strengths and weaknesses). Moreover, as in othertudies (Martin et al., 2011; Volkery et al., 2008), the RPG helpsncrease our (researchers) understanding of the system which sub-

tantially improves our model for future uses (RPG objectives 5 and).

Credibility was strengthened by the original method construc-ing land-use scenarios in two steps, each involving stakeholders

n Planning 119 (2013) 147– 157

who know best what is appropriate, feasible and acceptable intheir respective contexts: (1) climate and socio-economic scenar-ios development with experts and (2) effects of scenarios on landmanagement with farmers by RPG. Highly detailed storylines andtranslation into the RPG increased credibility of climate and socio-economic scenarios because simulation of future contexts makesstakeholders understand easily which and how changes occur.Indeed, farmers entered quickly into the RPG and took care to repro-duce realistic management despite the complexity of the multiplecolours and pieces on the fictitious board game considered rep-resentative of their farm systems. Presenting scenarios to farmerswith a yearly time step increased their understanding (RPG objec-tive 1) (Martin et al., 2011) and scenarios credibility. For example,while the “intermittent” alternative was quickly considered as busi-ness as usual by farmers, the “drastic” alternative, first consideredmore unlikely due to difficulties in imagining adaptations, wasprogressively seen as more plausible because some changes oradaptations proposed during the game reminded them of real pastsituations. Numerical and spatial information generated by the RPGenabled checking the consistency of storylines. Land managementadaptation maps were indeed found credible by experts duringthe validation step (Section 3.3). Experts only asked us to verifythe plausibility of increasing the manured area in accordance withlivestock size (Sections 3.3 and 3.4).

Scenarios creativity was efficiently triggered by involving adiversity of regional experts (as in van Vliet et al., 2010; Volkeryet al., 2008) with a diversity of technical and empirical knowledgeand covering a variety of disciplines and sources. The longer onelooks into the future, the larger the distance with experience ofthe surrounding environment and the uncertainty of assumptionsabout the future. Such grey areas can be reduced by combining adiversity of backgrounds (Volkery et al., 2008) or by shortening thetime horizon as we did, although a shorter time horizon can alsoreduce creativity. Stronger transitions would have probably beenobserved during the RPG with a time horizon longer than five years.

Legitimacy was strengthened by the iterative process involv-ing a diversity of stakeholders. Such a process avoids a one-sideassessment and fosters communication among stakeholders abouttheir different values and beliefs. Both farmers and regional expertsfound results consistent with their vision of the socio-ecologicalsystem. Moreover, this kind of participatory framework buildstrust between participants and researchers (Castella et al., 2005).Farmers said at the end of the RPG “this game is better than justpresentation and discussion, we felt like actors of the study” (RPGobjective 6).

The addition of a rigorous protocol (Etienne, 2010) of assess-ment by stakeholders of these four criteria and of outcomes of theprocess could benefit to our framework.

5.2. Perspectives

The small size of the study site allowed to test the frameworkwith all the farmers. Nevertheless, application to other or largerareas with non-exhaustive participation of farmers like in otherRPG (Barnaud, Trebuil, Dumrongrojwatthana, & Marie, 2008) isworth investigating. This could be tested either by repeating thegame with several groups of farmers or by sampling farmers. Thesechanges should lead to two additional steps in the framework: (1)developing a sampling strategy of farmers and (2) up-scaling thesampled data to land management adaptation representative ofthe whole population (Smajgl et al., 2011) by developing an agent-

based typology of farmers’ behaviours (Valbuena et al., 2008). Seefor example, the typology described by Gibon, Sheeren, Monteil,Ladet, & Balent (2010) for the Pyrenees which is consistent to alarge extent with farmers’ behaviours in our study area.

Urba

dgitwFbr

5V

ahcsdsct(Th2nwfabt

cobBcaeelaammeitosC

saosaCif(slf

P. Lamarque et al. / Landscape and

We considered a yearly average rate of change of forage pro-uction. For livestock system with year-round management ofrasslands (e.g., grassland mown three times a year), it could benteresting to add information on the effects of climate change onhe seasonal patterns of biomass production in order to observeithin-year land management adaptation (Martin et al., 2011).

inally, although this RPG focused on drought impacts, it could alsoe adapted for any other change affecting forage resources such asecent vole outbreaks.

.3. Adaptation strategies and land management changes inillar d’Arène

Since the beginning of the 20th century, a total conversion ofrable crops (two thirds of the area around 1810) to grasslandsas taken place at the site (Section 2). In spite of these dramatichanges, during scenarios farmers felt uneasy about going beyondimple land management modifications. Adaptations carried-outuring the RPG were mostly tactical and reflect their experience inimilar recent drought events, e.g., with the purchase of fodder toomplement stocks (Nettier et al., 2010). Some other strategic adap-ations were actually underway in the current economic contexte.g., decrease of livestock size by added value from direct sales).he presence of a large area of lightly stocked alpine meadowselped buffer inter-annual climate variations (Lemaire & Pflimlin,007) in the “intermittent” alternative, but not in the “drastic” alter-ative where even their production decreased. The adaptationsere also influenced by socio-economic alternatives through dif-

erence in types of subsidies. For example in the “local” alternativegri-environmental measures provided high support to maintainiodiversity in unterraced grasslands, therefore farmers consideredhe opportunity cost of converting them to grazing.

Overall, farmers appeared quite conservative and limitedhange as much as possible. While in some areas the possibilities offf-farm activities or non-agricultural activities on farms enabledy tourism result in a decrease in herd size (Garcia-Martinez,ernues, & Olaizola, 2011), pluri-activity was in our case more aonsequence of income loss following forced decrease in herd sizes a result of lack of fodder, than a motivation in itself. But non-conomic motives are also important in land-use change (Frasert al., 2011). Therefore, adaptations lead mainly to modifications ofand management and only to one type of conversion of land man-gement (but not land-cover), from mowing to grazing, again mores a result of natural environmental constraints (topography, cli-ate, short growing season, low productivity) induced by the highountain context (Mottet, Ladet, Coque, & Gibon, 2006) than by

ndogenous motivation. While some studies have shown that graz-ng represents an intermediate option between hay making andotal abandonment (Mottet et al., 2006), abandonment was neverbserved during the game because farmers consider themselves astewards of the meadows’ good state (Quétier, Rivoal, Marty, Dehazal, & Lavorel, 2010).

Our analysis showed that drought occurrence affects farmingtrategies and land management. However, consistent changescross scenarios regarding the specific location of changes pointedut the most vulnerable areas in the current landscape and farmingystem. For example, areas where mowing is ceased consistentlycross scenarios are more likely to be abandoned in the future.onversion from mowing to grazing, or increased manuring can

n the mid- to long-term affect grassland floristic composition and

unctioning (Quétier et al., 2007) and thereby ecosystem servicesLavorel et al., 2011; Quétier et al., 2010). Environmental policyhould thus consider such vulnerabilities, as the cultural and regu-ating services provided by these grasslands underpin current andoreseeable support to mountain regions.

n Planning 119 (2013) 147– 157 155

6. Conclusion

This paper describes and tests a framework to map land man-agement change according to various climate and socio-economicscenarios taking into account the diversity and complexity offarmers’ decisions making. The qualitative storylines provide anunderstandable way for communicating the complex content ofthe scenarios and the RPG converts both qualitative knowledge ofstorylines and the complexity of farmers’ decisions making intoquantitative maps of land-use change. Mapping land managementadaptation allows scientists and experts to check the consistencyof the different assumptions of the storylines. This framework sup-plies the numerical and spatial data often needed in land-use andenvironmental studies. Beyond useful outcomes for stakeholders,our framework also offers opportunities to develop participatorymethodologies to conceive local land-use adaptations to globalchange.

Acknowledgement

We thank all participants of workshops and the role playinggame for their contributions and the time they devoted to our study.We thank Claire Eveilleau for her help and work on storyline writ-ing. We also thank Fabien Quétier for his useful advices and helpduring the RPG session, Patrick Meyfroidt for his advices, and theJoseph Fourier Alpine Station for logistical support. This researchwas conducted on the long term research site “Zone Atelier Alpes”, amember of the ILTER-Europe network (ZAA publication no. 28). Thisresearch was funded by the ERA-Net BiodivERsA, with the nationalfunders ANR, part of the 2008 BiodivERsA call for research propo-sals and by the French Environment Ministry (MEEDDM) throughthe GICC-2 SECALP project.

Appendix A. Storylines of each climatic andsocio-economic alternative

A.1. “Intermittent” alternative

Both dry years and intermittent wet (normal) years areobserved. The first year, the weather is normal with wet winterand spring. But the second year spring is very dry (like in 2004 inFrance). This leads to a 25% (even 50% locally) decrease of yields inmown grasslands and a decrease of the same magnitude of grassquantity in alpine meadows alongside with an earlier phenology(flowering stage started when grazed). Yet weak water springs rundry quickly and the others dwindle during the summer season.Late summer rains (from mid-July) spur vegetation regrowth inalpine meadows and a good aftermath in mown grasslands. Duringthe third year no drought occurs and forage is good in quality andquantity. But the fourth year drought starts during summer aftera wet spring (like in 2009 in France). Mown grasslands are welldeveloped but grass dries quickly in early summer and low after-maths prevent from mowing a second time. In alpine meadows,agronomic situation start to be difficult from mid-August. Usually,woody grasslands show better conditions of grass production thanthe other during drought. Climate of the following years goes onwith the same alternation.

A.2. “Drastic” alternative

The next four years are marked by a series of droughts during

the spring season which affect the different types of grasslands.Over the years, grass grows less and less in alpine meadows andautumn regrowth does not occur anymore. After two or three years,from mid-July, alpine meadows resources are not sufficient for the

1 Urba

aoagesafbsf

A

eadpaPcdo–tddtomam

A

afiipataerneswme

R

A

A

B

B

56 P. Lamarque et al. / Landscape and

ctual stoking rate. Fertilized mown grasslands show a loss of 40%f forage production and an increase of dandelion (Taraxacum)bundance as well as other basal-rosette plants. Unfertilized mownrasslands are being progressively dominated by rosette- and drynvironment, their fodder production is too low to be mown. Inter-eason pastures also show a lower quality due to an early floweringnd a shift in species abundances like a decrease of leguminousodder plants and flower species with large leaves (decrease of Ono-rychis sppl., Lathyrus sppl. vs. increase of Plantago major, Hieraciumppl.). The lack of fodder in several French regions leads to very highodder purchase prices.

.3. “International” alternative

Simultaneously with this occurrence of drought, a socio-conomic shift appears. Urbanization growth continues. Mountaingriculture is more and more expected to meet urban society’semand for environmental and landscape outcomes rather than itsroductive role. City-dwellers seek for open areas to practice leisurectivities, but without concern for biodiversity or local activities.opulation prefers to eat at cheaper prices, therefore global marketompetition between products is promoted and leads to a globalecrease of local products prices already falling due to withdrawalf milk quotas [milk price: −20% conventional, −20% labelled, and10% for organic/meat: −15% ovine or sheep and bovine]. In ordero support agricultural activities, subsidies decoupled from pro-uction continue to be paid to farmers even if their amount isecreasing [−20%]. Nevertheless, a minimum income is guaran-eed. Moreover, it is possible for farmers to be paid if they adoptr maintain activities enhancing the provision of global environ-ental services such as carbon storage, water quality, biodiversity

nd so on. Pluriactivity in the form of off-farm jobs is possible asountain rural areas are very touristic.

.4. “Local” alternative

Along with this occurrence of drought, a socio-economic shiftppears. Urbanization growth continues but rural areas benefitrom an increasing demand on local products which reflect anmage of quality and territorial anchorage. Prices of direct sell-ng products and organic or labeled (PDO or PGI European label)roducts are kept constant while prices of conventional productsre falling. Rural areas are also appreciated for their leisure oppor-unities relying on natural and cultural heritage, which promotesgritourism. Following the new CAP reform, farmers no longer ben-fit from first pillar subsidies. Instead, in a context of increasingequirements towards environmental impact of agriculture, anyew aid is now coupled with environmental requirements (Agri-nvironmental measures) and provides payments to farmers whoubscribe, on a voluntary basis, to environmental commitmentsith results obligation. Local administrations and associations pro-ote rural development and help fund new infrastructures or

quipment necessary for agricultural and pastoral activities.

eferences

grimonde® (2009). Agricultures et alimentations du monde en 2050: scénarios etdéfis pour un développement durable. (Agriculture and food in 2050: scenarios andchallenge towards a sustainable development) Note de synthèse. INRA et Cirad(ed.), Paris.

lcamo, J., Kok, K., Busch, G., Priess, J. A., Eickhout, B., Rounsevell, M., Rothman, D.S., Heistermann, M., Lambin, E. F., & Geist, H. (2006). Searching for the future ofland: scenarios from the local to global scale land-use and land-cover change. Berlin,Heidelberg: Springer.

arnaud, C., Trebuil, G., Dumrongrojwatthana, P., & Marie, J. (2008). Area study priorto companion modelling to integrate multiple interests in upper watershedmanagement of Northern Thailand. Southeast Asian Studies, 45, 559–585.

arnaud, C., Van Paassen, A. M., Trébuil, G., Promburom, T., & Bousquet, F. (2010).Dealing with power games in a companion modelling process: lessons from

n Planning 119 (2013) 147– 157

community water management in Thailand highlands. Journal of InternationalAgricultural and Extension Education, 16(1).

Barreteau, O., Le Page, C., & Perez, P. (2007). Contribution of simulation and gamingto natural resource management issues: An introduction. Simulation & Gaming,38(2), 185–194.

Benot, M.-L., Saccone, P., Vicente, R., Pautrat, E., Morvan-Bertrand, A., Decau, M.-L., Grigulis, K., Prud’homme, M.-P., & Lavorel, S. (2013). How extreme summerweather may limit control of Festuca paniculata by mowing in subalpine grass-lands. Plant Ecology & Diversity, 1–12.

Boe, J., Terray, L., Habets, F., & Martin, E. (2006). A simple statistical–dynamical down-scaling scheme based on weather types and conditional resampling. Journal ofGeophysical Research—Atmospheres, 111(D23).

Bohunovsky, L., Jager, J., & Omann, I. (2010). Participatory scenario developmentfor integrated sustainability assessment. Regional Environmental Change, 11(2),271–284.

Borjeson, L., Hojer, M., Dreborg, K. H., Ekvall, T., & Finnveden, G. (2006). Scenariotypes and techniques: Towards a user’s guide. Futures, 38(7), 723–739.

Carpenter, S. R., Pingali, P. L., Bennett, E. M., & Zurek, M. B. (2005). . Ecosystemsand human well-being: scenarios: the millenium ecosystem assessment (vol. 2)Washington, DC: Inland Press.

Castella, J. C., Trung, T. N., & Boissau, S. (2005). Participatory simulation of land-usechanges in the northern mountains of Vietnam: The combined use of an agent-based model, a role-playing game, and a geographic information system. Ecologyand Society, 10(1).

Coreau, A. (2009). Dialogue entre des chiffres et des lettres. Imaginer et construire desfuturs possibles en écologie (Words and numbers: imagining and building possiblefuturs in ecology). Ph.D. Thesis. Université Montpellier II, p. 524.

de Jouvenel, H. (2002). La démarche prospective. Un bref guide méthodologique[Strategic planning: A short methodological guide]. Futuribles, 247, 1–24.

Engler, R., Randin, C. F., Thuiller, W., Dullinger, S., Zimmermann, N. E., AraÚJo, M. B.,Pearman, P. B., Le Lay, G., Piedallu, C., Albert, C. H., Choler, P., Coldea, G., De Lamo,X., DirnbÖCk, T., GÉGout, J.-C., GÓMez-GarcÍA, D., Grytnes, J.-A., Heegaard, E.,HØIstad, F., NoguÉS-Bravo, D., Normand, S., PuS CaS , M., SebastiÀ, M.-T., Stanisci,A., Theurillat, J.-P., Trivedi, M. R., Vittoz, P., & Guisan, A. (2011). 21st centuryclimate change threatens mountain flora unequally across Europe. Global ChangeBiology, 17, 2330–2341.

Etienne, M. C. (2010). (Companion modelling: A participatory approach to supportsustainable development) La modélisation d’accompagnement: Une démarche par-ticipative en appui au développement durable (Quae Editions, pp. 368). France:Versailles.

Fraser, E. D. G., Dougill, A. J., Hubacek, K., Quinn, C. H., Sendzimir, J., & Termansen, M.(2011). Assessing vulnerability to climate change in dryland livelihood systems:Conceptual challenges and interdisciplinary solutions. Ecology and Society, 16(3),01.

Garcia-Martinez, A., Bernues, A., & Olaizola, A. M. (2011). Simulation of mountaincattle farming system changes under diverse agricultural policies and off-farmlabour scenarios. Livestock Science, 137(1–3), 73–86.

Gibon, A., Sheeren, D., Monteil, C., Ladet, S., & Balent, G. (2010). Modelling and sim-ulating change in reforesting mountain landscapes using a social–ecologicalframework. Landscape Ecology, 25(2), 267–285.

Girel, J., Quétier, F., Bignon, A., & Aubert, S. (2010). Histoire de l’agriculture en Oisans.Hautes Romanche et pays faranchin. Villar d’Arène, Hautes-Alpes (History of agri-culture in Oisans. Villar d’Arène, Hautes-Alpes). In: Cahiers Illustré du LautaretStation Alpine Joseph Fourier Grenoble, France, p. 79.

IPCC. (2012). Managing the risks of extreme events and disasters to advance cli-mate change adaptation. In: Field, C., Barros, V., Stocker, T., Dahe, Q., Dokken, D.,Ebi, K., Mastrandrea, M., Mach, K., Plattner, G., & Allen, G. (Eds.), Special reportof the intergovernmental panel on climate change, Cambridge University Press,Cambridge, UK and New York, NY, USA, p. 582.

Janssen, M. A., & Ostrom, E. (2006). Empirically based, agent-based models. Ecologyand Society, 11(2), 37.

Janssen, S., & van Ittersum, M. K. (2007). Assessing farm innovations and responsesto policies: A review of bio-economic farm models. Agricultural Systems, 94(3),622–636.

Lamarque, P. (2012). Ecosystem services in a mountain grassland socio-ecological sys-tem. Ph.D. Thesis. Grenoble: Laboratoire d’Ecologie Alpine, Université JosephFourier. p. 256.

Lavorel, S., Grigulis, K., Lamarque, P., Colace, M. P., Garden, D., Girel, J., Pellet, G.,& Douzet, R. (2011). Using plant functional traits to understand the landscapedistribution of multiple ecosystem services. Journal of Ecology, 99(1), 135–147.

Le, Q. B., Seidl, R., & Scholz, R. W. (2012). Feedback loops and types of adaptation inthe modelling of land-use decisions in an agent-based simulation. EnvironmentalModelling & Software, 27–28(0), 83–96.

Lemaire, G., & Pflimlin, A. (2007). Les sécheresses passées et à venir: quels impactset quelles adaptations pour les systèmes fourragers [Past and future periods ofdrought: What are their effects on the forage systems and the possible adapta-tions?]. Fourrages, 191, 163–180.

Martin, G., Felten, B., & Duru, M. (2011). Forage rummy: A game to support theparticipatory design of adapted livestock systems. Environmental Modelling &Software, 26(12), 1442–1453.

Matthews, R. B., Gilbert, N. G., Roach, A., Polhill, J. G., & Gotts, N. M. (2007). Agent-

based land-use models: A review of applications. Landscape Ecology, 22(10),1447–1459.

Metzger, M. J., Rounsevell, M. D. A., Van den Heiligenberg, H., Perez-Soba, M., & Hardi-man, P. S. (2010). How personal judgment influences scenario development: Anexample for future rural development in Europe. Ecology and Society, 15(2.).

Urba

M

M

M

M

N

N

P

Q

Q

S

P. Lamarque et al. / Landscape and

eyfroidt, P. (2012). Environmental cognitions, land change, and social–ecologicalfeedbacks: An overview. Journal of Land Use Science.

illennium Ecosystem Assessment. (2005). Ecosystems and human well-being: sce-narios. Washington, DC: Island Press., 515.

ora, O. C. (2008). Les nouvelles ruralités à l’horizon 2030. Des relations villes cam-pagnes en émergence? (New patterns of urban and rural development in 2030.Emergence of urban–rural relationship?), Quae ed., Paris.

ottet, A., Ladet, S., Coque, N., & Gibon, A. (2006). Agricultural land-use change andits drivers in mountain landscapes: A case study in the Pyrenees. AgricultureEcosystems & Environment, 114(2–4), 296–310.

aivinit, W. (2009). Modélisation d’accompagnement pour l’analyse des interactionsentre usages des terres et de l’eau et migrations dans le bassin versant de laLam Dome Yai au Nord Est de la Thaïlande (Companion modelling to analysethe land/water use and labour migration interactions in Lam Dome Yai watershed,Lower Northeast Thailand). Ph.D. Thesis. In: Géographie humaine, économiqueet régionale, Université Paris Ouest Nanterre-La Défense. p. 344.

ettier, B., Dobremez, L., Coussy, J. L., & Romagny, T. (2010). Attitudes of livestockfarmers and sensitivity of livestock farming systems to drought conditions inthe French Alps. Revue De Geographie Alpine, 98(1), 383–400.

ak, M. V., & Brieva, D. C. (2010). Designing and implementing a role-playing game:A tool to explain factors, decision making and landscape transformation. Envi-ronmental Modelling & Software, 25(11), 1322–1333.

uétier, F., Rivoal, F., Marty, P., De Chazal, J., & Lavorel, S. (2010). Social represen-tations of an alpine grassland landscape and socio-political discourses on ruraldevelopment. Regional Environmental Change, 10, 119–130.

uétier, F., Thebault, A., & Lavorel, S. (2007). Plant traits in a state and transitionframework as markers of ecosystem response to land-use change. EcologicalMonographs, 77(1), 33–52.

chroter, D., Cramer, W., Leemans, R., Prentice, I. C., Araujo, M. B., Arnell, N. W., Bon-deau, A., Bugmann, H., Carter, T. R., Gracia, C. A., de la Vega-Leinert, A. C., Erhard,

M., Ewert, F., Glendining, M., House, J. I., Kankaanpaa, S., Klein, R. J. T., Lavorel,S., Lindner, M., Metzger, M. J., Meyer, J., Mitchell, T. D., Reginster, I., Rounsevell,M., Sabate, S., Sitch, S., Smith, B., Smith, J., Smith, P., Sykes, M. T., Thonicke, K.,Thuiller, W., Tuck, G., Zaehle, S., & Zierl, B. (2005). Ecosystem service supply andvulnerability to global change in Europe. Science, 310(5752), 1333–1337.

n Planning 119 (2013) 147– 157 157

Smajgl, A., Brown, D. G., Valbuena, D., & Huigen, M. G. A. (2011). Empirical characteri-sation of agent behaviours in socio-ecological systems. Environmental Modellingand Software, 26(7), 837–844.

Swetnam, R. D., Fisher, B., Mbilinyi, B. P., Munishi, P. K. T., Willcock, S., Ricketts,T., Mwakalila, S., Balmford, A., Burgess, N. D., Marshall, A. R., & Lewis, S. L.(2011). Mapping socio-economic scenarios of land cover change: A GIS methodto enable ecosystem service modelling. Journal of Environmental Management,92(3), 563–574.

Valbuena, D., Verburg, P. H., & Bregt, A. K. (2008). A method to define a typology foragent-based analysis in regional land-use research. Agriculture, Ecosystems andEnvironment, 128(1–2), 27–36.

Valbuena, D., Verburg, P. H., Veldkamp, A., Bregt, A. K., & Ligtenberg, A. (2010).Effects of farmers’ decisions on the landscape structure of a Dutch ruralregion: An agent-based approach. Landscape and Urban Planning, 97(2),98–110.

van Vliet, M., Kok, K., & Veldkamp, T. (2010). Linking stakeholders and modellersin scenario studies: The use of fuzzy cognitive maps as a communication andlearning tool. Futures, 42(1), 1–14.

Verburg, P. H., Soepboer, W., Veldkamp, A., Limpiada, R., Espaldon, V., & Mastura,S. S. A. (2002). Modeling the spatial dynamics of regional land use: The CLUE-Smodel. Environmental Management, 30(3), 391–405.

Vittoz, P., Randin, C., Dutoit, A., Bonnet, F., & Hegg, O. (2009). Low impact of climatechange on subalpine grasslands in the Swiss Nothern Alps. Global Change Biology,15.(209-220).

Volkery, A., Ribeiro, T., Henrichs, T., & Hoogeveen, Y. (2008). Your vision or mymodel? Lessons from participatory land use scenario development on a Euro-pean scale. Systemic Practice and Action Research, 21(6), 459–477.

Walz, A., Lardelli, C., Behrendt, H., Grêt-Regamey, A., Lundström, C., Kytzia, S., &Bebi, P. (2007). Participatory scenario analysis for integrated regional modelling.Landscape and Urban Planning, 81(1–2), 114–131.

Washington-Ottombre, C., Pijanowski, B., Campbell, D., Olson, J., Maitima, J., Musili,A., Kibaki, T., Kaburu, H., Hayombe, P., Owango, E., Irigia, B., Gichere, S., & Mwangi,A. (2010). Using a role-playing game to inform the development of land-usemodels for the study of a complex socio-ecological system. Agricultural Systems,103(3), 117–126.