Copyright © 2011 by the author(s). Published here under license by the Resilience Alliance.

Etienne, M., D. R. Du Toit, and S. Pollard. 2011. ARDI: a co-construction method for participatory
modeling in natural resources management. Ecology and Society 16(1): 44. [online] URL: http://www.
ecologyandsociety.org/vol16/iss1/art44/

Research, part of a Special Feature on Mental Models

ARDI: A Co-construction Method for Participatory Modeling in Natural
Resources Management

Michel Etienne 1, Derick R. Du Toit 2, and Sharon Pollard 2

ABSTRACT. The outcomes of a series of tests of the ARDI (Actors, Resources, Dynamics, and Interactions)
method in complex cases or conflict-ridden situations is presented. ARDI is part of a companion modeling
approach that makes it possible to engage a broad spectrum of stakeholders in the design and development
of land and water management plans. It is based on participatory workshops that set out to collaboratively
imagine a future open, dynamic management system, capable of adaptation and anticipation, by gathering
the various stakeholders in a partnership to examine conservation of the natural resources and promoting
a sustainable development. Its originality lies in the co-construction of a “conceptual model” of the
functioning of the territory, according to an overarching, negotiated development question. The approach
is based on the collective articulation of the key elements of a context or territory by stakeholders such as
managers, representatives, socio-professional technicians, nongovernmental organizations, experts, and
scientists, and local policy makers. This sharing of representations is done by means of a series of collective
workshops during which Actors, Resources, Dynamics, and Interactions constituting the profile of the
territory are identified and clarified. This work of co-construction is conducted within a precise
methodological framework that we present in a step-by-step format. The method is grounded in concrete
experience gleened from tests conducted by the authors over the past five years. Finally, the requirement
for specific skills as well as pitfalls to avoid when applying the method are discussed.

Key Words: co-construction; collective mental model; conceptual model; facilitation; natural resources
management; participatory modeling

INTRODUCTION Environmental issues are often technically complex

and deeply embedded in the minds, hearts, and
The application of simulation models in practices of society. As such, their resolution
collaborative decision making for the management requires the collaboration of public authorities,
of natural resources is a characteristic of adaptive private business, scientific experts, groups of users,
management (Holling 1978, Walters 1986). But the social interest groups, non-governmental organizations,
use of these models to stimulate the participation of and representatives of stakeholders in the particular
stakeholders in the development of management ecological domain (Bouwen and Taillieu 2004).
scenarios is much rarer (Costanza and Ruth 1998, Methods and approaches that facilitate the
Bousquet et al. 2002, Ozesmi and Ozesmi 2004, development of shared problem perspectives have
Bosch et al. 2007). The progressive shift from improved their effectiveness by melding different
management planning based on an authoritative or kinds of knowledge and different competencies.
rationalist approach towards the incorporation of They do this by involving multiple actors or
tools for mediation based on democratic approaches stakeholders in collectively constructing a social
(Van den Belt 2004) calls for the emergence of new learning process. These processes result in shared
tools that focus on co-construction of meaning and or collective mental models of the issues or problem.
the sharing of information and understanding Recent papers demonstrate that these participatory
regarding a particular context that is to be managed. processes produce better outcomes than classical
top-down processes (Lynam et al. 2007, Reed 2008,

1
INRA-Ecodevelopment Unit, 2AWARD
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Voinov and Brown 2008, Voinov and Bousquet which Actors, Resources, Dynamics, and
2010), particularly in terms of experiential learning Interactions (ARDI) are identified and clarified by
about the issue, about solutions to technical the participants themselves. The work of co-
problems, about other stakeholders, or about construction is conducted within a precise
communicational and organizational learning (Daré methodological framework which we present and
et al. 2010). The role of these participatory processes illustrate by means of concrete examples from the
in social change has barely been evaluated, but tests we conducted over the past five years.
recent work on companion modeling (Etienne 2010)
provides evidence of interesting changes in social
representation of environmental issues (Mathevet KEY QUESTION AND KEY PARTNERS
et al. 2011) or in social organization to face
environmental problems (Gurung et al. 2006). The success of participatory modeling depends on
three key choices that are made at the start of the
Following a series of tests of methods conducted in process. These choices have to be discussed among
multiple use and multi-stakeholder contexts (natural the mandatory partners and the facilitators of the
areas with multiple use, biosphere reserves, regional approach during one or more preparatory meetings.
or national parks) or in conflict situations (heritage The first choice involves clearly defining the
sites, urban-forest interfaces), a companion territory under question and identifying and
modeling approach was developed and applied, formulating the question to be addressed.
which made it possible to involve stakeholders in Environmental decision making may encompass
the design of land and water management plans conflicts over resource management or hidden
(Etienne 2006). This approach is based on sociological or ecological agendas and interests.
participatory workshops that were set up to envision Thus, formulating a precise and clear question that
an open, dynamic management approach, capable is easily understood by a wide set of stakeholders is
of adaptation and anticipation, by gathering the a crucial entry into the ARDI process. When the
various stakeholders together to promote demand for such an intervention arises directly from
sustainable development and conserve shared local stakeholders, the facilitator should engage
resources. Its originality lies in the co-construction with it, identify key stakeholders and participants,
of a shared “conceptual model” of the relationships and collaboratively define the central question or
and functioning of the different aspects of a issue to be addressed. If the demand comes from a
particular context or territory. This is usually scientific interrogation or research, the facilitator
conducted in relation to a specific and negotiated has to formulate the central question or objective in
development question. a way that will convince key stakeholders to
collaborate (Daré et al. 2009).
The approach focuses on encouraging participants
to describe, explain, and predict the purpose, form, The second key choice involves identifying one or
function, and state of a given system (Rouse and several facilitator(s). Their aptitude and legitimacy
Morris 1986), so as to elicit causal knowledge as to lead the process of design-validation-use of
defined by Jones et al. (2011). It is based on the ARDI tools has to be ascertained.
collective articulation of the key elements of a
territory and context by affected stakeholders such Thirdly, it is necessary to pay special attention to
as managers, representatives, socio-professional the composition of the working group: the choice
technicians, non-governmental organizations, experts of partners and meeting place, the periodicity of the
and scientists, and local policy makers. It frames the workshops, and the method of invitation are all
elicitation of individual learning that then leads to important contributors to the success of the method.
the emergence of collective learning (Röling 2002). This is mainly because the representativeness of the
This sharing of representations results in the participants and thus the richness and relevance of
progressive emergence of a collective “mental the conceptual model depend on the composition of
model” due to changes in the individual’s schemas the group.
that occur primarily through nonconscious
processes in response to experiential learning during The ARDI method has been tested under a variety
repeated exposure to novel ideas and relationships of conditions, key questions, and contexts. A
(Beratan 2007). This dialogical process is achieved diversity of territories were involved, including a
through a series of collective workshops during mountain pasture where contradictory resource
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

access rights led to a conflictive situation; a water Four types of situations were encountered during
catchment where compliance with the water law the testing process:
was questioned; a set of municipalities that were
concerned with the same forest fire risk; and a wide 1. Priority was given to a scientific
river delta where many human activities were representation of the system. The participants
overlapping. The conditions that led to the use of a were mainly researchers of topics related to
participatory modeling process included solving or the issue, and perhaps who had grounded
avoiding potential land use conflicts (Rouan et al. experience of the field or were from very
2010), imagining a more sustainable management similar situations.
scheme for a region (Simon and Etienne 2010,
Souchere et al. 2010)), and assessing the impact of 2. Priority was given to a global understanding
a regulation on water (Biggs et al. 2008) or of the system. The participants were chosen
biodiversity management (Anselme et al. 2010). from extension services whose local
experience legitimized their position to speak
The success of this approach to natural resources on behalf of the stakeholders that they
management lies in the relative independence of an frequently come in contact with. It is
external scientific agent who is familiar with and is important to ensure that all activities that
skilled in handling the methodological aspects. relate to the defined issue are covered but that
However, there is a distinct advantage to engaging over-representing any one activity is avoided
a researcher as facilitator who is skilled in both the (for example, inviting three foresters because
ecological sciences and social sciences and who has there are three forest companies working in
basic experience in facilitating debates between the territory).
researchers and managers. But a communication
expert can also easily play this role. That is the 3. Priority was given to the involvement of local
reason why the method was applied mainly by stakeholders who have an overall view of the
French researchers who were working in the field system. The participants were selected from
of companion modeling (Collectif Commod 2006), local stakeholder representatives who were
but several agents of regional natural reserves were chosen for their legitimacy (for example, they
trained to apply it in France, and mediators are were elected democratically or were a leader
currently being trained in western African biosphere of a professional organization) and for the
reserves. relevance of their activity to the initial
question.
Finally, several criteria should be considered when
choosing participants for the exercise. While the 4. Priority was given to the involvement of local
choice is flexible (it is possible to invite a new stakeholders while seeking to cover the
participant in the course of the exercise), the process diversity of the system. Emphasis was placed
gains from having access to an initial “core group” on selecting local stakeholders whose actual
that will be present throughout the process of co- practices were relevant to a particular
construction. Apart from the compulsory reflection question rather than selecting formal leaders
on the representativeness and legitimacy of the of groups of stakeholders.
participants, which is commonly discussed in any
participatory approach (Steyaert and Lisoir 2005), The position and status of researchers in the process
the composition of the core group can give priority was variable, and the significance of this is still
to different factors. Linkages between stakeholders being debated among the companion modeling
in resource management that occur at different community. The general rule is that researchers with
spatial and institutional levels may enhance some knowledge of the context and major processes
cross-scale interactions that can empower local (social, technological, economic, ecological, and
level user groups in creating social and political political) should be engaged. Some bring expertise
capital (Adger et al. 2005). Ensuring that less to the initial stage, while others will be engaged at
powerful stakeholders are involved as a means of a specific workshop (for example, the discussion on
promoting an equitable outcome of the participatory system dynamics or the design of the interactions
process may also be a determining criterion diagram) if the participants feel there is a need for
(Barnaud et al. 2008). expertise on a particular topic. This differentiation
is relatively easy in the field of ecological sciences.
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

It is problematic in the field of social sciences where concerned with the key question, their management
the researcher may play the role of the “expert” who entities, the resources used, and the main processes
has a global view of the social relationships or that are driving changes that affect these resources.
economic flows but who will frequently be The group that takes part in the co-construction of
confronted with diverging points of view raised by the model must clearly address a key question, the
the stakeholders who may have greater legitimacy formulation of which is adapted to the issue at stake.
to speak about social or economic issues (Charles In the South African case study, the issue that was
et al. 2008). tackled was formulated as follows: What do you
think about water resources use and management in
Additional factors such as the choice of the venue the Crocodile Catchment? What is driving change
and the duration and frequency of the meetings in the flow of the Crocodile River? For the ARDI
depend on many issues that are external to the exercise, the issue was first split into the following
exercise itself. But some principles should be three questions:
negotiated and respected if the method is to be
successfully applied. For example, it is desirable 1. What are the main stakeholders that interact
that the location of meetings is easily accessible to with the river and its flow?
participants, and is on neutral ground. If not, this
should be acknowledged lest it become an overt or 2. What are the main resources of the catchment
hidden issue within the proceedings. Each meeting in relation to water flow?
should be planned to last at least two hours, and the
participants must remain focused on the 3. What are the main processes that drive
collaborative exercise. In order to permit changes in the Crocodile Catchment that
participants to keep a precise record of the decisions affect the river flow?
made and to follow up easily the step-by-step
running of the ARDI method, the ideal is to conduct Depending on the extent and complexity of the
all the workshops over a period not exceeding one territory concerned and on the level of detail
month. The meetings may be held in one of the required, the collective response to each of these
following formats: (a) in a two-and-a-half-day three questions can take between 1 and 3 hours. It
workshop, (b) during one half-day per week, or (c) is important that the order of questions be respected
over three separate days. Ideally, the choice should and the facilitator take care that each participant has
be negotiated with the participants. the opportunity to voice an opinion. In the sessions
we facilitated, the following simple procedure was
adopted: (1) a drawing was made on an interactive
THE ARDI METHOD white board that was visible to and understandable
by all the participants, (2) for each element of ARDI,
ARDI is an acronym of the four French words, each participant had the opportunity to respond, and
"Acteurs", "Ressources", "Dynamiques", and (3) only one concept was be proposed at a time. This
"Interactions", that identify the four steps the formal procedure is used to systematically elicit a
method uses to elicit stakeholder mental models of representation of the system and to avoid premature
the system they are working or living in. Thus, it digression and discussion.
allows the progressive emergence of a shared
representation of the components and dynamics of To facilitate the sharing of mental models and
the system by describing stakeholders, the representations, the answers to the questions are
resources, the processes, and the interactions formulated as lists of words with a minimum of
between them. The steps involved in conducting an coding, which makes it possible to easily classify
ARDI process are described in detail below. the information and represent it on a diagram. The
workshop is generally led by two people: a
facilitator and an observer/recorder.
Co-constructing a common representation
A key role of the facilitator is to ensure clarity and
The first step of the companion modeling approach general agreement about the terms or concepts used,
following the ARDI method focuses on collectively such as when a response is either too generic (e.g.,
identifying the principal stakeholders who are manager), or uses a term that can lead to confusion
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

(e.g., ash can be a plant material, a woody plant, or Identifying key actors (“A”)
the residue from a fire). The role of the observer/
recorder is to keep track of the exchange between The first stage of the ARDI process culminates in
members of the group or between one participant the “actors” diagram, which is composed of the list
and the facilitator. Among the key interactions to of stakeholders and the corresponding management
monitor, three are particularly important: entities and the links between them (Fig. 2). The
exercise proceeds in three stages. Initially, the
1. By observing the attitudes of the participants participants simply list the stakeholders who they
to each other (who is supporting whose idea, consider to be associated with the question. As long
who is arguing against whose idea), some as new suggestions for stakeholders are proposed,
social links can be revealed. the facilitator goes on with the next participant or
begins a new round from the floor. Each actor
2. Writing down and analyzing the nature of proposed must be a direct stakeholder (people who
arguments developed to support or to use or whose practices have a direct impact on key
contradict a proposal permits the analyst to resources of the territory), or an indirect stakeholder
identify the domain of justification (people whose actions will encourage the direct
(Boltansky and Thévenot 1991) or the way stakeholders to change their practices). Generally,
participants frame the issues (Bouwen and the indirect actors refer to people who either bear a
Taillieu 2004) when they evaluate any public policy, enforce a regulation, or provide
submitted item. financial support. Even if they are likely to be
outside of the territory, it is the way to account for
3. Keeping track of the reasons advanced for decisions made at higher scales than the scales
changing a previously accepted concept or defined by the case study spatial limits. The
term is a good way to follow up the group facilitator adds each input to the interactive board
dynamics. by using a new label and colors to distinguish the
category to which it belongs (for example, black
The first aspect facilitates identification of social case for the direct ones, blue for the indirect ones).
networks, the second facilitates better understanding The facilitator may suggest subclasses to increase
of individual mental models, and the third facilitates the precision with which certain types of actors are
keeping track of the path followed to reach an defined (e.g., farmers may be subdivided into sugar
agreement. cane and citrus farmers) or challenge the assignment
to a category if there is not consensus in the room.
For instance, in the Crocodile River process, after A typical example of this type of intervention is the
each ARDI exercise, the facilitator and the observer status given to the entity "herd". Certain participants
discussed their observations and identified, for each will position it as a resource, others will regard it as
point that required discussion and argumentation, an actor. When the grazing impact on grassland
which justification domain (Boltansky and dynamics is a significant process, the facilitator may
Thévenot 1991) the participants appealed to, in ask whether participants think that the herd is
order to assert their choice. During the step used to autonomous (it decides where, when, and how much
identify the key processes that drive changes that it will graze), or if it depends mainly on the decisions
affect the river flow in the Crocodile Catchment, 12 of the shepherd/farmer. In the first case, the herd
processes were mentioned by the 10 interviewees. will be retained as a stakeholder; in the second case,
Among these, biodiversity laws, climate change, it will be listed as a resource managed by the
forestry development, nutrient leaching, stream shepherd.
flow, water rights reallocation, and chemical
modification were identified twice; drought, Next, the facilitator asks the participants to specify
sediment load, tourism development, and crop the links that exist between the identified
market were identified three times; and population stakeholders and to clarify this relationship in a
increase was identified seven times. The analysis of simple way. Progressively, arrows are added
the ways these assertions were justified identified according to suggestions made by the participants.
science, market/economics, and legal domains as The facilitator progressively shapes the diagram by
being the most frequent (Fig. 1). bringing closer the stakeholders who have many
relations and moving those away that do not have
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Fig. 1. Frequency of justification domains used by workshop participants to justify their selection of key
processes that drive changes in the Crocodile Catchment.

any. When the participants consider that the main Figure 2 shows that when “Water Abstraction” was
interactions between actors have been represented, mentioned during the Interaction step of the ARDI
the facilitator can identify or question incongruities process, “Rural Community” was acknowledged as
and gaps (e.g., no link between the municipalities not being an important stakeholder since the amount
and the developers) or point out stakeholders who abstracted was considered to be insignificant. The
have no relation with any others. In the latter case, rural community stakeholder concept was therefore
the facilitator launches a deliberation on the eliminated. The irrigation farmer class was split into
relevance of retaining this actor in the diagram, two categories in order to set apart commercial
while the observer keeps record of the decision farmers, who were identified as consuming much
made by the group and the justification for the less water than irrigation farmers. Foresters were
decision (the landowner is a typical example of a also split into two categories according to the level
stakeholder who does not have a link with anybody of compliance with the Water Act. However, this
but who is often retained in the diagram because he/ decision was reconsidered at the end of the ARDI
she can easily block the development of the process. National and provincial authorities were
activities of another stakeholder). aggregated because one was seen as being the arm
of the other. Finally, two new stakeholders appeared
Lastly, always adhering to the principle of the during debates on interactions around the action “to
negotiation, the participants must identify and pollute”: developers and urban residents. In such a
clarify the management entities used by each direct way, actors are added and removed from the
stakeholder. Those can be spatially explicit entities representation in a discursive manner, the main
(forest plot, water catchment, dam), or not (fish, point being that the category and its meaning are
cash). clarified as a social, dynamic process.

The diagram must clearly reflect the changes made

during the subsequent steps in the ARDI process. Identifying key resources (“R”)
Red letters are conventionally used for topics added;
italics are used for terms proposed by the facilitator. The second stage consists of listing the relevant
Words that are crossed out represent topics that were resources of the territory according to the key
eliminated when elaborating ongoing diagrams. stakeholders previously identified, the word
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Fig. 2. ARDI stage 1 at Crocodile River: “What are the main stakeholders that seem to be able to or
need to play a decisive role in managing the river flow?” Words or concepts written in red were added
during the later steps of the ARDI. Words or concepts written in italics were proposed by the facilitator
to get out of a deadlock. A crossed box means that the idea was finally rejected.

“resource” applying exclusively to goods or Several changes were made to the Resource diagram
products used by any of the stakeholders (Fig. 3). during the modeling process: when the process
During the collaborative construction of the list, the “Nutrient Leaching” was adopted at stage 3, the
principal types of resources are often grouped animals farmed near the river, such as trout,
within five main categories (infrastructure, water, crocodiles, or ostriches, were added. But the
minerals, plants, and animals). For each resource stakeholders had problems finding a generic word.
mentioned, the speaker is asked to justify his or her Finally, after a while, the facilitator’s suggestion of
choice and is encouraged to specify which indicator “Farmed Animals” was accepted.
seems to be the most relevant to making
management decisions regarding that resource. This
indicator can be quantitative or qualitative, and if Identifying key processes (“D for dynamics”)
there is debate or disagreement, several indicators
may be applied to a particular resource. Participants The third stage of the ARDI process consists of
are encouraged to explain which characteristics of listing the main processes that drive changes in the
the resource are being considered before making a territory in relation to the question (Fig. 4). These
decision about that resource. Because certain processes can deal with ecological dynamics (e.g.,
resources are temporary, one may have to specify vegetation transitions or water flow), economic
the period of existence (season, favorable year) and/ dynamics (e.g., market forces, amount of subsidies),
or length of existence (lifespan of a building, time or social dynamics (e.g., social cohesion,
for filling of a dam). The resources functioning as knowledge transfer). If the list is large, the facilitator
exogenous variables but whose characteristics are asks the participants to rank the 10 main processes
critical in operating the system can also be by assigning “10” to the most important one and “1”
mentioned (e.g., the rainfall in arid or dry zones). to the least. Then the facilitator sums up the scores
This set of indicators will be used afterwards, during given by each participant and selects the five
the model implementation and the development processes that get the highest score. For these
scenarios steps, to visualize and compare the processes, diagrams are drawn to reflect participant
stakeholders’ points of view (Etienne et al. 2003). perspectives on what forces are driving changes
with respect to which resources.
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Fig. 3. ARDI stage 2 at Crocodile River: “What are the main resources of the catchment and the key
information needed to support their management?” Words or concepts written in red were added during
the later steps of the ARDI. Numbers rank the items from the most (5) to the least important (1).

When dealing with ecological dynamics, “stream flow reduction activity”. When discussing
participants may agree on the successive states the importance of wetlands, two new processes
taken by the vegetation and specify the factors that arose and were added: “Flow Reduction” and
cause the transition from one state to another, “Water Purification”. The group could also not
including the time required to move from one state agree on a process that linked surface water to flora
to the next. The diagram can be designed either in and fauna. After much discussion, the facilitator’s
situ or as a response to a proposal designed by an suggestion of “Life Support“ was accepted.
expert. In the two options, the diagram must clearly
distinguish between the dynamics related to human
actions (effect of the techniques currently Eliciting interactions (“I”)
implemented) and natural dynamics (consequence
of some types of uses being abandoned). A similar The last stage of the ARDI method consists of
diagram can be applied to the dynamics of water. synthesizing answers to the three preceding
questions by stressing the interaction between users
At the end of this stage, it is advisable to review and and resources (Fig. 5). It is a pivotal process in the
revise the diagrams and to identify possible gaps. exercise since it leads to the conceptual model that
Three types of gaps may be identified: (1) an activity represents all interactions related to the key
or a resource was identified but no participant question. It is advisable to devote more time to this
possessed enough knowledge about it. The group phase since it generally takes one half-day for a
then agrees to call upon an expert and nominates a simple diagram (3–4 direct actors, 3–4 resources),
person to identify and mobilize the expert; (2) an and one day for a more complex diagram (5–8 direct
important actor was forgotten at the time of the actors, 5–10 resources). The group must then
preparatory phase, and the group was concerned by answer the following central question: How does
this absence. The group then agrees to invite the each stakeholder use the resources and modify the
person to the next phase; and (3) an actor, a resource, processes?
or a dynamic process were the subject of a total
disagreement between two or more participants. The facilitator will begin this stage by distributing
The group then agrees on the choice of an expert and summarizing the diagrams developed during
and the type of information required from that expert the ARDI process, making a particular effort of
in order to solve this deadlock. clarifying if new people were integrated into the
group. When the diagrams are relatively simple, the
Some comments were received and changes were facilitator directly invites the participants to
made to the Process diagram during the following collectively construct an interaction diagram. For
steps. When eliciting the impact of afforestation on that, the facilitator puts the main resource in the
the river flow, the acronym SFRA was chosen middle of the diagram and proposes to position the
because it corresponds to the terms of the law: direct stakeholders related to this resource. Each
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Fig. 4. ARDI stage 3 at Crocodile River: “What are the main processes that drive changes in the
Crocodile Catchment that affect the river flow?” Words or concepts written in red were added during the
later steps of the ARDI. Words written in italics were proposed by the facilitator to get out of a
deadlock.

participant, in turn, chooses to add an interaction In this case, the facilitator must take care that the
between a stakeholder and a resource or between a resources and the stakeholders mentioned by the
stakeholder and another stakeholder. The participants continue to relate well to the chosen
participant can either add a link on the collective stake, and in case of doubt, to clarify the considered
diagram or ask to add one of the stakeholders of the link. If key goals are not clearly identified, the
list, who has not yet been included on the collective facilitator proposes to group the resources into
diagram. Each new interaction suggested must categories, and then constitutes working groups on
include a verb that specifies the type of action that the three or four categories that appear to be most
generates the link. The proposer must justify their important to the participants. In this case, it is
choice and indicate, if they know, the type of necessary to add a phase of combining and
information used by the actors to make the comparing the three or four built diagrams.
corresponding decision (e.g., I reduce my use of
irrigation water because the water level in my dam The role of the facilitator during the “Interaction”
is less than 3 m; I will look for an agreement with step is particularly important and delicate since he/
the Water User Association because more than 30% she is responsible for constructing an easily
of the inhabitants complain about the lack of accessible and recognizable diagram at the same
drinking water). Finally, when all the arrows are time as facilitating interactions and inputs (taking
drawn, the participants locate on the diagram the care to avoid confusing representations and
key processes identified during the “D” stage of the “crossed arrows”, etc). The facilitator needs to pay
ARDI process by writing down their acronym attention to the clarity of inputs from participants
besides the arrow to represent an interaction that is (while avoiding polarization and putting them in
believed to strongly affect the resource or uncomfortable positions), and needs to regularly
stakeholder. revisit inputs that are not integrated into the diagram
(i.e., boxes without arrows) without forcing the
When the diagrams become too complex, it is participants too much. The facilitator simultaneously
preferable to proceed by dividing up the exercise seeks to achieve three objectives: (1) to gradually
into several manageable portions. Two options are prepare a common diagram that is comprehensible
possible. If several key goals were clearly identified to all, (2) to identify clear and indisputable
to be at stake during the co-construction process, interactions, and (3) to leave the possibility for
the facilitator proposes to create a diagram of addressing lapses of memory. Additionally, the
interactions for each of these stakes and leads the facilitator’s role is to ask each participant to
procedure described in the preceding paragraph as reformulate their input so as to avoid uninformative
many times as is necessary to complete the diagram. verbs (e.g., the forester manages the land, the farmer
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Fig. 5. ARDI stage 4 at Crocodile River: “How does each stakeholder use the resources and modify the
processes?“ White boxes indicate stakeholders, green boxes indicate resources, boldfaced letter codes
indicate processes, and blue text indicates actions. Words or concepts written in red were added during
the later steps of the ARDI.

farms the field, the manager manages the budget) process of constructing the four diagrams. There is
or to retain only the interactions that make sense specific value in knowing why and how a particular
according to the question (e.g., in an exercise on fire actor, resource, or interaction was mentioned,
prevention and urbanization, the interaction retained, eliminated, or transformed. It is possible
between the cereal farmer and his crop field was to use many means to reach this goal: audio
restricted to ploughing the stubble after harvest recording (very comprehensive but very time
because it was the only one that impacts land consuming to analyze), the use of a secretary who
sensitivity to fire). is dedicated to this task (very effective because they
can quickly give an account of the sequence
This interaction step is generally the richest and followed and how decisions were justified, but it
most interesting of the co-modeling process. To requires an additional person), or the use of an
capture this, it is essential to keep a record of the interactive table or a digital camera to take a series
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

of pictures of the diagrams as they are being In some aspects, the method has similarities with
constructed (very demonstrative but it requires problem structuring methods, such as the use of a
either particular equipment or a person partially model as a transitional object, the emphasis put on
dedicated to the exercise). the group process, and the importance of facilitation
skills (Eden and Ackermann 2006). As it is
concentrated on the preliminary issue conceptualization
Taking the perspectives further stage of modeling and on the visualization of a
shared mental model, it can also be compared with
The completion of these four stages leads to the or completed by other methods developed for
establishment of a conceptual model. This model is systems thinking and applied to natural resources
a critical output of the ARDI process because it is management, such as hexagon diagrams (Hodgson
a graphical representation of how the stakeholders 1992), group model building (Richardson and
perceive the system to function. This has Andersen 1995), influence diagrams (Cain et al.
fundamental implications for the next stages: 1999), causal loop modeling (Sterman 2001), or
designing and implementing a management plan for stock and flow diagrams (Takahashi 2008).
the territory based on the collaboratively established
understanding captured in the diagrams. Two The contribution of the ARDI method to
options arise for the working group that will take decentralized democratic approaches to natural
the ARDI steps to the next stage: (1) work out a resources management should be critically assessed
proposal for a management plan (e.g., concerted in the light of the work presented here (see D. Du
research plan, charter of sustainable development) Toit, H. Biggs, and S. Pollard unpublished
based on the conceptual diagram, or (2) develop a manuscript). Three sensitive aspects, identified as
computer simulation model that will assist in critical issues in recent reviews of methods to
decision making and dialogue. In the first case, the enhance stakeholders’ participation in environmental
thinking will be focused on the territory and its management (Lynam et al. 2007, Reed 2008,
priorities of development, education, and research. Voinov and Brown 2008), need to be highlighted.
In the second case, the thinking will focus on the First, highly skilled facilitation is required to ensure
implementation of a computer model or a role- mutual respect, conviviality, and psychological
playing game to help stakeholders transport safety in order to get to a common empowerment
themselves into the future and imagine and of the participants. Specific skills are required to
collectively envision adaptive co-management anticipate unexpected reactions due to the
scenarios. In both cases, the ARDI method is peculiarity of the process, to consider all types of
valuable and useful because it works with a knowledge as legitimate, and to be sensitive and
collectively established conceptualization of the responsive to the power relationships among the
territory and provides a concrete tool for applying people involved in the collaborative workshops.
the concepts of adaptive management.
Second, special attention has to be paid to the
The ARDI method can also be applied with the legitimacy of the process and representativeness of
specific research goal of comparing individual and the people involved. Thus, the method should be
collective mental models, or monitoring changes in tailored to a particular decision-making context, and
mental models during a collaborative process. In should take due consideration of the objectives,
this case, the approach can be split into two steps: nature, and status of participants and appropriate
first, a set of individual interviews where each level of engagement.
stakeholder’s mental model of the system’s form,
function, state, and dynamics is elicited; then a Third, the objectives of the participatory process
collective exercise of co-construction of what could must be clearly defined and agreed upon at the
be called a shared representation of the system beginning of the process, and regularly recorded and
(Mathevet et al. 2011). This particular approach revisited while proceeding. This goal is likely to be
requires additional logistical arrangements and frequently challenged by the complexity and
agreements with participants, and is useful as a tool uncertainty presented by the emerging mental
for collaborative reflection on the emergence of co- model.
management of a particular resource.
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Specifically, the application of the ARDI method Barnaud, C., G. Trébuil, P. Dumrongrojwatthana,
has demonstrated strengths in understanding and J. Marie. 2008. Area study prior to companion
stakeholders’ perspectives and values, and modelling to integrate multiple interests in upper
providing an effective way to get to a shared watershed management of northern Thailand.
representation of a complex system. It supports Southeast Asian Studies 45(4):559–585.
stakeholders’ participation in a process that
emphasizes empowerment, equity, trust, and Beratan, K. 2007. A cognition-based view of
learning (Reed 2008). It allows an integration of lay, decision processes in complex social–ecological
technical, and scientific knowledge by stressing the systems. Ecology and Society 12(1):27. [online]
choice of clear and simple terms, and providing URL: http://www.ecologyandsociety.org/vol12/iss1/
access to different ways of thinking about and art27/.
representing a situation. The method takes care that
practitioners not be concerned when results are Biggs, H., D. Du Toit, M. Etienne, N. Jones, A.
contradictory or unexpected, and helps initial Leitch, T. Lynam, S. Pollard, and S. Stone-Jovicich.
individual surprises to be transformed into 2008. A preliminary exploration of two approaches
unexpected collective discoveries (Lynam et al. for documenting mental models held by
2007). stakeholders in the Crocodile catchment, South
Africa. WRC Report KV 216/08, Water Resource
Commission, Gezina.
Responses to this article can be read online at:
http://www.ecologyandsociety.org/vol16/iss1/art44/ Boltansky, L., and L. Thévenot. 1991. De la
responses/
justification. Gallimard, Paris, France.

Bosch, O. J., C. A. King, J. L. Herbohn, I. W.

Acknowledgments: Russell, and C. S. Smith. 2007. Getting the big
picture in natural resource management: systems
The authors wish to thank the French Institute for thinking as ‘method’ for scientists, policy makers
Biodiversity and the UNESCO-MAB program for and other stakeholders. Systems Research and
the financial support provided to French and Behavioral Science 24(2):217–232. [online] URL:
Western African case studies, and the South African http://onlinelibrary.wiley.com/doi/10.1002/sres.818/
National Parks and Water Research Commission pdf.
for the South African case study. We also want to
acknowledge the participants in the many Bousquet, F., O. Barreteau, P. d’Aquino, M.
workshops that were organized in France and in Etienne, S. Boissau, S. Aubert, C. Le Page, D.
Western and South Africa to develop and validate Babin, and J. C. Castella. 2002. Multi-agent systems
this methodology. and role games: an approach for ecosystem co-
management. Pages 248–285 in M. Janssen, editor.
Complexity and ecosystem management: the theory
and practice of multi-agent approaches. Elgar
Publishers, Northampton.
LITERATURE CITED
Bouwen, R., and T. Taillieu. 2004. Multi-party
Adger, W. N., K. Brown, and E. L. Tompkins. 2005. collaboration as social learning for interdependence:
The political economy of cross-scale networks in developing relational knowing for sustainable
resource co-management. Ecology and Society 10 natural resource management. Journal of
(2):9. [online] URL: http://www.ecologyandsociety. Community and Applied Social Psychology 14:137–
org/vol10/iss2/art9/. 153.

Anselme, B., F. Bousquet, A. Lyet, M. Etienne, and Cain, J., C. Batchelor, and D. Waughray. 1999.
B. Fady. 2010. Modelling of spatial dynamics and Belief networks: a framework for the participatory
biodiversity conservation on Lure mountain development of natural resource management
(France). Environmental Modelling & Software 25 strategies. Environment, Development and Sustainability
(11):1385–1398. 1:123–133.
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Charles, M., F. Chlous-Ducharme, E. Faugère, and Lingmuteychu Watershed, Bhutan. Ecology and
M. Wintz. 2008. Science et démocratie dans la Society 11(2):36. [online] URL: http://www.ecolog
gestion de la nature : des ethno-sociologues pris yandsociety.org/vol11/iss2/art36/.
dans la modélisation d'accompagnement. VertigO 8
(2). [online] URL: http://vertigo.revues.org/4999. Hodgson, A. 1992. Hexagons for systems thinking.
European Journal of Operational Research 59
Collectif ComMod. 2006. Modélisation (1):220–230.
d’accompagnement. Pages 217–228 in F. Amblard
and D. Phan, editors. Modélisation et simulation Holling, C. 1978. Adaptive environmental
multi-agents : applications aux sciences de l’homme assessment and management. John Wiley, London,
et de la société. Hermès Sciences, Londres. UK.

Costanza, R., and M. Ruth. 1998. Using dynamic Jones, N. A., H. Ross, T. Lynam, P. Perez, and A.
modelling to scope environmental problems and Leitch. 2011. Mental models: an interdisciplinary
build consensus. Environmental Management synthesis of theory and methods. Ecology and
22:183–195. Society 16(1):46. [online] URL: http://www.ecolog
yandsociety.org/vol16/iss1/art46/.
Daré, W., R. Ducrot, A. Botta, and M. Etienne. 2009.
Repères méthodologiques pour la mise en oeuvre Lynam, T., W. de Jong, D. Sheil, T. Kusumanto,
d'une démarche de modélisation d'accompagnement. and K. Evans. 2007. A review of tools for
Cardère éditions, Avignon, France. incorporating community knowledge, preferences
and values into decision making in natural resources
Daré, W., A. Van Paassen, R. Ducrot, R. Mathevet, management. Ecology and Society 12(1):5. [online]
J. Queste, G. Trébuil, C. Barnaud, E. Lagabrielle. URL: http://www.ecologyandsociety.org/vol12/iss1/
2010. Apprentissage des interdépendances et des art5/.
dynamiques. Pages 223–250 in M. Etienne, editor.
La modélisation d'accompagnement: une démarche Mathevet, R., M. Etienne, T. Lynam, and C. Calvet.
participative en appui au développement durable. 2011. Water management in the Camargue
Quae éditions, Versailles, France. Biosphere Reserve: insights from comparative
mental models analysis. Ecology and Society 16
Eden, C., and F. Ackermann. 2006. Where next for (1):43. [online] URL: http://www.ecologyandsociety.
problem structuring methods. Journal of the org/vol16/iss1/art43/.
Operational Research Society 57(7):766–768.
Ozesmi, U., and S. L. Ozesmi. 2004. Ecological
Etienne, M. 2006. Companion modelling: a tool for models based on people's knowledge: a multi-step
dialogue and concertation in Biosphere Reserves. fuzzy cognition mapping approach. Ecological
UNESCO-MAB Technical Notes 1:44–52. Modeling 176:43–64.

Etienne, M., editor. 2010. La modélisation Reed, M. S. 2008. Stakeholder participation for
d’accompagnement : une démarche participative en environmental management: a literature review.
appui au développement durable. Quae éditions, Biological Conservation 141(10):2417–2431.
Versailles, France.
Richardson, G., and D. Andersen. 1995. Teamwork
Etienne, M., C. Le Page, and M. Cohen. 2003. A in group model building. Systems Dynamics Review
step-by-step approach to building land management 11(2):113–137.
scenarios based on multiple viewpoints on multi-
agent system simulations. Journal of Artificial Röling, N. 2002. Beyond the aggregation of
Societies & Social Simulations 6(2). [online] URL: individual preferences; moving from multiple to
http://jasss.soc.surrey.ac.uk/6/2/2.html. distributed cognition in resources dilemmas. Pages
25–66 in C. Leeuwis and R. Pyburn, editors.
Gurung, T. R., F. Bousquet, and G. Trébuil. 2006. Wheelbarrows full of frogs: social learning in rural
Companion modeling, conflict resolution, and resource management. Royal van Gorcum Assen,
institution building: sharing irrigation water in the The Netherlands.
 Ecology and Society 16(1): 44
http://www.ecologyandsociety.org/vol16/iss1/art44/

Rouan, M., C. Kerbiriou, H. Levrel, and M. Etienne.

2010. Sheep, turf and bikes: a co-modelling process
of social and natural dynamics on the isle of
Ouessant. Environmental Modelling & Software 25
(11):1399–1412.

Rouse, W. B., and N. M. Morris. 1986. On looking

into the black box: prospects and limits in the search
for mental models. Psychological Bulletin 100
(3):349–363.

Simon, C., and M. Etienne. 2010. A companion

modelling approach applied to forest management
planning. Environmental Modelling & Software 25
(11):1371–1384.

Souchère, V., L. Millair, J. Echeverria, F. Bousquet,

C. Le Page, and M. Etienne. 2010. Co-constructing
with stakeholders a role-playing game to initiate
collective management of erosive runoff risks at the
watershed scale. Environmental Modelling &
Software 25(11):1359–1370.

Sterman, J. 2001. Systems dynamics modelling:

tools for learning in a complex world. California
Management Review 43(1):8–25.

Steyaert, S., and H. Lisoir. 2005. Participatory

methods toolkit. A practitioner’s manual. King
Baudouin Foundation and the Flemish Institute for
Science and Technology Assessment, Brussels,
Belgium.

Takahashi, Y. 2008. Dynamic simulation modelling

using descriptive information in natural language.
International Journal of Simulation and Process
Modelling 4:215–222.

Van den Belt, M. 2004. Mediated modelling: a

system dynamics approach to environmental
consensus building. Island Press, Washington, D.C.

Voinov, A., and F. Bousquet. 2010. Modelling with

stakeholders. Environmental Modelling & Software
25(11):1268–1281.

Voinov, A., and E. J. Brown. 2008. Lessons for

successful participatory watershed modeling: a
perspective from modeling practitioners. Ecological
Modelling 216:197–207.

Walters, C. 1986. Adaptive management of

renewable resources. McGraw Hill, New York.