Research

INTRODUCTION

Pastoralism—as transhumant (seasonal movements across rainfall or elevation gradients) or nomadic (opportunistic movements following pasture resources; Degen 2011)—is practiced by many peoples across West, Central, East, and Southern Africa. It is regarded as critical to African food system resilience, connecting biomass for livestock forage and manure for organic fertilizer across space and time (Ayantunde et al. 2014), as well as to food security, with pastoralism providing 60% and 90% of animal protein in West and East Africa, respectively (Adzande 2019). Pastoralism is a system forged in change and variation, adapted over long periods by those experiencing among the most variable climates on Earth (Shettima and Tar 2008). Mobile pastoralists support herds of livestock on scarce, scattered forage between rains by carefully navigating the dry seasons, moving to agricultural lands post-harvest (where agricultural residues support herds in exchange for critical fertilizer from manure), or to lower elevations where water accumulation enables dry season grass (Degen 2011, Tilahun 2020).

Critically, pastoral systems are built on highly adaptive, flexible rules. The Borana pastoral systems of Southern Ethiopia and Northern Kenya organize rangelands across multiple scales in households, settlements, encampments, and grazing associations, each with specific rules of access (Abebe 2016, Belayneh 2016, Dabasso et al. 2012, Debela et al. 2019). These polycentric, flexible, and adaptive rules (Ostrom 1990, 2010) allow groups of pastoralists with varying degrees of kinship and connection to share scarce and only partially predictable forage resources over wide areas.

However, a range of stressors, drought, population growth, and government intervention, have compounded in recent decades to shift how well pastoral systems across Africa can adapt and function (Wario et al. 2016). The encouragement of agricultural expansion (Ayantunde et al. 2014, Zampaligré et al. 2019), where people enjoyed better market access and suffered less from drought and disease stressors (Burian et al. 2019), has also encouraged land enclosure and appropriation, disrupting norms for movement and grazing of livestock (Benti 2018, Mgalula et al. 2023). Disrupted access to mobility and grazing land shifts the importance of kallo forage reserves in Borana communities (Tache and Oba 2010), which in turn favor the resilience of households at the expense of their communities (Abebe 2016, Wario et al. 2016). Much land that historically had been under transhumant, wet- or dry-season use is now used perennially, in part because of population expansion (Wario et al. 2016) and government intervention. For example, the construction of ponds in Borana dry-season grazing areas led to year-round use and expanded unregulated settlement (Degen 2011). Livestock density, reduced access to grazing lands, and bans on bushfires compound to lead to bush encroachment (Belayneh 2016, Yassin 2019). In sum, these compound stressors encourage increased, perennial pressure on reduced grazing resources; in cases where herders adapt to these stressors by increasing pressure on remaining resources (by increasing livestock density, or reducing forage access through enclosures or by cropping land), they close a set of reinforcing feedbacks that threaten the sustainability of pastoral systems.

However, not all adaptations by herders are like this. Where enclosures or local rules allow for wet season forage to be preserved (possibly as hay) for the dry season, for instance, they substitute for lost grazing opportunities away from the community (Birhanu et al. 2017, Iticha and Husen 2019, Tilahun 2020). Where Borana communities shift away from their preferred cattle to camels or goats, browsers, rather than grazers, they shift both their reliance on grasses as well as grazing pressure (Wario et al. 2016, Birhanu et al. 2017, Ouédraogo et al. 2021). Considered all together, this landscape of adaptations forms a set of interconnected processes around forage access (Fig. 1). Adaptations that reduce pressure on forage form balancing feedbacks that help to stabilize forage access within pastoral communities. However, in any real pastoral system, it is challenging to know how these various feedback processes are interacting, in large part because the measurement of people and animals in these remote contexts is prohibitive (Liao et al. 2017, 2018).

As one part of identifying how these different adaptations (herd size and species, shared grazing areas, reserve areas, and forage production) relate to another, we developed a multi-part intervention for measuring and observing preferences in pastoral systems and seeing how these preferences might respond to experiential learning from a game. Understanding pastoralists’ preferences can help identify entry points for targeted interventions to support pastoral communities in climate change adaptation. Although we intend for our mixed-method intervention to function as a common “toolkit” for assessing adaptation decision making across pastoral systems, our activity reported in this paper focused on Borana communities in Moyale County, in northern Kenya.

Games for research and learning in human-environment problem solving

Wherever a decision may hang strategically on what other people might do, or how an uncertain or ambiguous process might play out, games are useful both as tools for observation (Cárdenas and Ostrom 2004) and for learning (Janssen et al. 2023). Growing out from the field of game theory in the mid-20th century, experimental games (as “framed field experiments” or “lab in field experiments”) have been a mainstay in experimental economics for decades (Levitt and List 2007, Camerer 2015). Prominent among these tools are coordination games with defined payoffs (e.g., the prisoners’ dilemma; see Bruns and Kimmich 2021 for the landscape of 2x2 games) or voluntary contribution games (such as public goods games, where players decide how much of a resource endowment to commit to a shared benefit; e.g., Pfaff et al. 2019); these games commonly have well-known equilibria or benchmarks, so that researchers can measure how different groups of people deviate from these expected behaviors as a useful signal. For example, Hoel, Hidrobo, and others compare spouses’ willingness to contribute tokens to a cup where their value would be doubled but shared equally with their partners (as opposed to retained privately by the player) as a measure of intra-household bargaining power (Hoel 2015, Hidrobo et al. 2021, Hoel et al. 2021). Researchers have applied variations and elaborations of these archetypal games across a range of disciplines and (importantly for this readership) human-environment problems, with framings around specific resource dilemmas (e.g., Sargent et al. 2022) and the incorporation of real resource stewardship challenges including real-time competition (Knapp and Murphy 2010), uncertainty (Alpizar et al. 2011), or time constraints (Brozyna et al. 2018). With these additions come greater realism for participants and a richness of behavioral responses to examine and learn from, but at the expense of the equilibria and benchmarks that ground research in the earlier, archetypal games. For instance, while equilibrium behavior in a simple public goods game is easily determined, it is much more difficult to estimate when this game is repeated, when the benefits of the public good are distributed via a common pool resources game (where players take as much as they choose, subject to possible asymmetric access), and where actions are subject to time constraint, as is the case in the irrigation game developed by Janssen and Anderies (e.g., Janssen et al. 2011). Such tailored, multidimensional games raise a question of “what are we measuring” in game play, formalized in the concept of construct validity (Esterling et al. 2023), or a demonstration that the variables measured in the instrument (game decisions) have a meaningful bearing on the true variables of interest (resource decisions, such as crop choice or irrigation behavior). To a degree, this question is best informed by building out evidence, exploring the conditions and group characteristics that shape game play across different contexts as Janssen et al. have done (spanning many contributions including, Janssen et al. 2011, Janssen et al. 2012, Anderies et al. 2013, Janssen and Anderies 2013, Baggio et al. 2015, Janssen et al. 2015).

Games have found purpose in other disciplines beyond observation and research, with recent literature highlighting the potential for games to trigger learning in resource contexts, with examinations of the longer-term effects of games interventions on resource and community decision making (Meinzen-Dick et al. 2014, 2018). As Janssen et al. (2023) highlight, game interventions for research and inference often differ strongly from those for learning; Falk et al. (2023) articulate a conceptual framework for how the dimensions of game intervention structure (e.g., narrative, rules, visual aids, facilitation) shape potential for experiential learning. In particular, they and others have highlighted community debrief and discussion as a critical component of an intervention in triggering a shift in perceptions and mental models (Crookall 2023, Falk et al. 2023).

Our own engagement with games (both in this study and in prior work) lies largely in the space of tailored, multi-dimensional human-environment games. Filling a gap in existing computer gaming platforms (which typically required technical literacy and a lab setup; Janssen et al. 2014), we developed a games framework using the NetLogo modeling platform that would easily allow players to join highly visual, flexible multiplayer games without relying on lab setups or network access, playable at a table or in a field or under a tree (Bell et al. 2023). Beginning with NonCropShare, a coordination game examining willingness to engage ecosystem services as an alternative to pesticides (Bell et al. 2016, Bell and Zhang 2016), we have continued to develop games tailored (or tailorable) to a range of problem domains, with published examples including human-wildlife conflict and non-lethal control (GooseBump; Rakotonarivo et al. 2021a, Sargent et al. 2022) and rural property rights (SharedSpace; Rakotonarivo et al. 2021b). In this current manuscript we present a novel game and intervention framework that bridges the research and learning goals that Janssen et al. (2023) show as lying somewhat in opposition. In addition to the well-documented critical mechanisms for debrief (Crookall 2023), we add opportunities for participants to shape the structure of the game by specifying how the game board should be used; and add assessment of shifts in attention or perception via discrete choice experiments (DCE), as early elements of learning, which to our knowledge have not been used for this purpose. More recent work, such as ElDidi et al. (2024), highlights the need for deeper follow-up engagement with the communities where the games were played to reinforce (and even continue) learning and support the capacity strengthening of communities to develop their own solutions.

Study objectives

Awareness and capacity building are among the gaps identified for the livestock sector in the Kenya National Adaptation Plan 2015–2030 (GoK 2016). However, learning about climate responses requiring coordination and collective action, as in the case of pastoral adaptation, is more effective when a group of actors learn together about the need and ways to achieve collective outcomes (Falk et al. 2023). There is growing attention to the potential of collective action games to facilitate sustainable management of common pool resources (such as water or rangelands; Cárdenas and Carpenter 2008, Meyer et al. 2021, Falk et al. 2023). By simulating the effects of multiple rounds of decision making and biophysical changes, and delineating with participants the ways in which aspects of the game align or differ with their realities, games create a relatively low-risk forum for experiencing and discussing complex social-ecological systems (Speelman et al. 2017, Muhamad and Kim 2020). In particular, games and debriefings that link experiences in the game to their real life situation can affect mental models and norms about behavior with regard to a resource (Meinzen-Dick et al. 2018, Falk et al. 2023). A secondary objective of this study is to explore whether the game has triggered learning (change in perceptions) at individual or community level, and which experiences or insights from the game people resonate with.

Our focal area of study was shaped by partnerships and opportunity, including initial focus group discussions and planning sessions in Southern Ethiopia, and a field exercise following in Moyale, Northern Kenya. Unfortunately, delays related to COVID-19 meant we were only able to complete a small field study and not a full intervention within the scope of our funded work. The field work we report on in this paper took place in October and November 2021, which typically covers the shorter of two rainy seasons for the region. Unexpectedly, the 2021 short rainy season was particularly weak, and we looked for impacts this may have had on our participants during the course of this small field study (Fig. 2). The guiding questions for this study were as follows:

1. How do pastoralists consider herd size, losses, and the use of their time?
2. Does an intervention that focuses on coordinating pasture and forage reserves shift this?
3. Do people relate to the game?
4. Do we see changes over time along the field study?

In the following sections we introduce the multi-part intervention we developed to inform these questions, and discuss possible implications of our findings from this small fieldwork activity.

METHODS

Preliminary work and game development

We began developing plans for this game intervention in 2019 as a component of a CGIAR Research Program on Policies, Institutions, and Markets (PIM), and conducted several focus group sessions with Borana communities near Yabello town in Southern Ethiopia. The COVID-19 pandemic disrupted all components of this work, though we continued development of our game through online meetings over 2020–2021. With the PIM program coming to a close in late 2021, and with our team still unable to undertake fieldwork in Ethiopia, we adjusted our plans to tailor our instruments to conditions among Borana herders in Moyale County, northern Kenya, and to undertake a first field test before the close of PIM. Though (like many other projects planned over the same time period) our goals of conducting a larger-scale intervention went unrealized, we were fortunate to have had the opportunity to conduct a meaningful small field exercise. In advance of identifying a follow-up funding opportunity for this work, we present key findings from this first field work in the current manuscript.

Theory of change

Falk et al. (2023) build a theory of change for experiential learning through games and community debrief interventions that proposes an “experiential learning action situation” (of rules, narrative, and player characteristics) in the intervention, within which participants make and observe decision making that may update their mental models for problem domains they experience in “real life action situations” (of rules, environments, and characteristics of community). Importantly in their theory of change, the interventions allow for development in mental models without the potential for risk or damage from a real life action situation, nor the real world consequences of decisions. We propose a focus on a part of this theory of change, with all other assumptions of Falk et al. (2023) holding, but where we emphasize the role of the intervention in focusing participant attention on specific issues or hazards that may arise in a real world context, driving possible action situations:

This adaptation of Falk et al.’s theory of change posits that exposure to some issue or hazard, which may lead to positive or negative outcomes, draws our attention to that issue as an initial step in issue-specific learning that may in turn build our capacity to respond to such issues in future and (plausibly) increasing the likelihood of experiencing positive outcomes. We suggest that a game intervention built with a focus on the specific issue can enhance attention to the issue in the absence of such real world exposure and possible negative consequences. In line with Crookall (2023), who points to games as a source of experience, and debriefing sessions as opportunities to learn from that experience, our intervention includes a game exercise, run in parallel with a focus group discussion exercise, with both activities feeding in to a community debriefing. Insights about the community and local context from the focus groups, linked to experiences in the game by the debriefing participants, provide opportunity for issue-specific learning. Additionally, where learning itself (the development of mental models in Falk et al.’s work) makes the issue more salient to the participant, game interventions may catalyze learning feedbacks over longer periods of time, with or without the real life experience of the issue.

Critical for this change to occur is that the game intervention must meaningfully engage the kinds of decisions participants make in the face of the real world issue, that is, what is varied or measured in the game must be a valid construct (Esterling et al. 2023). In this first examination of the group game GreenReserve intervention, we look for signals of GreenReserve’s construct validity across the different components of the intervention (circle 1 in Fig. 3). We also compare the results of the two DCE experiments to look for signals that the game has shifted attention to the issue (circle 2 in Fig. 3), and examine results from our debriefing to identify any signals of learning or learning potential (circle 3 in Fig. 3).

Study area and sample

We conducted a test of a two-day intervention to learn about preferences for time use and herd size across 13 villages located in Borana pastoral areas near Moyale in northern Kenya (Fig. 4). We sampled villages from an overall sampling frame of 48 villages near Moyale with guidance from a representative of the National Drought Monitoring Authority (NDMA), prioritizing in this order: (i) safety and security to visit; (ii) inclusion of representative livestock types for the area of camels, cattle, and small ruminants (sheep and goats, commonly referenced jointly in Kenya as “shoats”); and (iii) including locations that spanned the 2–3h travel distance away from Moyale city that was safe to visit. In each sampled village, we asked village leadership to invite participants for our focus group and games activities (18 participants in total); we were not permitted to sample participants randomly. Instead, we agreed with village leadership on the inclusion of household heads with a balance across those in leadership positions in the village and those not, women and men, and a balance across herding activities of different animals (Table 1).

We adopted a mixed methods approach using both quantitative and qualitative data collected from the multi-part intervention. For the qualitative analysis, we conducted thematic coding using “Taguette,” an open-source tool to analyze the translated open-ended responses from focus group discussions and transcribed player interactions (verbatim) made during game sessions and community debriefing. Coding was conducted iteratively to generate codes and themes that summarized and captured the essence of the responses and remarks. The mixed methods approach allows us to quantitatively assess the relationships between farmer and group characteristics and key outcome variables to answer the research questions, while insights from the qualitative analysis help deepen or further the understanding.

Intervention overview

Our intervention began on the first of two days with an initial focus group discussion with 10 members of the village to learn about pastoral activities in the village, perceived changes over time, as well as characteristics of the village such as local norms and governance (Fig. 5). At the same time as the focus group, a separate sample of eight village members participated in a short individual survey, including a discrete choice experiment framed around time use and herd characteristics, a short module of questions about beliefs regarding their community, and then a training protocol (including practice rounds and a comprehension check) of the group game GreenReserve. These activities typically took until late morning to complete, at which point the team adjourned for the day, reconvening with the game participants the following morning. On the second day of the intervention, we met again with the players of the game to play a single game treatment, consisting of two “years” of pastoral activity, with each year including eight rounds of play that included a longer and shorter wet period, with dry weather in between each. After completing this game, players repeated the discrete choice experiment and beliefs module undertaken on the previous day. At the end of the experiment sessions, all players, focus group participants, and any other interested village members were invited to participate in a community debrief about the two-day intervention. Focus group participants were compensated with 325 Kenyan Shillings (Ks) for their time. Game participants earned a base payment of 700 Ks for their two day contribution, plus a performance payment of 40 Ks for each of their animals that was surviving at the end of their game play (for a total payment of up to 1100 Ks). Game participants were told that if they for any reason were unable to complete both days, they would receive a payment of 300 Ks, though all participants returned for all activities.

Focus group protocol

We conducted one focus group discussion (FGD) of max 10 participants each at each study site/village on the first of the two days. The FGD aimed to understand the historical experience of pastoral communities with collective action and how these communities have over time responded to environmental and socioeconomic changes (such as drought and conflict) through collective action. As we aimed to gain balanced perspectives, each FGD included at least one elder, one female, one youth, one person in the leadership committee, and a few active herders. Each FGD session lasted 40 minutes on average and were run concurrently with the activities for game participants. The FGDs were conducted in Kiborana and Swahili, and audio-recorded with participants’ consent. Excerpts of transcripts and selected quotes were translated into English.

Survey protocol

We conducted a short survey, developed in Android Open Data Kit (ODK) and implemented via Kobo-toolbox, undertaken in two parts (“DCE and Survey Materials” folder in the GreenReserve V1.12 repository: see link in Data Availability section). The first part of the survey included short modules on household, herd size, losses, time use, and beliefs, and was followed immediately by the pre-game discrete choice experiment (DCE). The results of this survey define the “status quo” option (the reality that participants currently live in) against which the hypothetical options presented in the DCE below are compared. The second part of the survey occurred after the game was played, and included an evaluation of the strength of relationship each player had with other players in their game, as well as a repeat of the beliefs module, before moving to the post-game DCE.

DCE protocol

We designed a DCE with five attributes: ( i) fraction of time invested in pastoral activities for own benefit, (ii) fraction of time invested in pastoral activities for community benefit, (iii) herd size, (iv) herd losses in a “good” year, and (v) herd losses in a “bad” year. We developed a d-efficient design of 20 choice sets, with 4 choice sets being selected randomly for each participant (Fig. 6). The experiment was always framed around the main animal in the respondent’s herd (camel, cattle, or shoat) and the herd size and loss attributes were framed relative to the respondent’s actual herd size. To accomplish this, we generated variants of the 20 choice sets for each of the three animal types and for each of a range of herd sizes (from 1, 2, 5, 10 animals through to several hundred animals), with survey items automatically selecting the appropriate choice set (matching animal and herd size) to be shown to the participant. To avoid the error of enumerators displaying the incorrect choice set, each choice card included a randomized 3-letter code that had to be verified by our survey software before enumerators could enter responses for the choice set. We undertook a pre-test before data collection with 25 respondents in villages not included in the main sample (each responding to 4 choice sets), which we used to estimate a main-effects only utility model (of the 5 attributes listed above) and perform a Bayesian updating of our overall choice set design.

GreenReserve protocol

Using a games platform based in NetLogo that we have used in other problem domains (Bell et al. 2023), we developed GreenReserve: a multiplayer game of herd and grass management that begins with several set-up phases in which players can tailor elements of the game landscape to match their own pastoral system, followed by round-wise play in which players make decisions about where to graze their cattle, and whether to invest effort in harvesting grass for later use (Fig. 7). Players begin the game with an endowment of 10 animals that may survive along the 16 rounds of the game by having sufficient grass forage, becoming more likely to die with forage insufficiency. The focus on living animals as the sole metric avoids any possible challenges with numeracy, as well as the need to identify rates of substitution that may be poorly defined or variable (such as relative costs of animals, inputs, insurance, crops, etc.). Land that is designated as private, or as shared reserve, requires permission of other players before herd animals may use it for grazing or consumption of harvested grass, so that a central dilemma for players is to coordinate on designation and shared use of reserve lands in order to maximize resource availability for their herd animals. As applied in this study, GreenReserve speaks most closely to the connections among forage access, enclosures, and forest harvest in Figure 1 (feedback loops in lower left of the diagram). Each player uses a tablet to record their decisions on the game board, with decisions of each player visible to other players. Communication among players is free and unrestricted through all game phases described in this activity.

A detailed description of all GreenReserve game phases, submodels, and parameter settings for this application, as well as a detailed training protocol, is provided in the “Protocols” folder in the GreenReserve V1.12 repository (see link in Data Availability section). Other games in this family and associated publications are available for viewing and download from https://www.elm-lab.org/.

Community debrief

To explore how participants’ experiences from playing the games relate to their real-life experiences of pasturing practices and understand motivations for broad decision strategies in the game, we further conducted group debriefing meetings with the game participants and other community members upon completion of the game sessions and post DCE survey. Debriefing sessions can serve as fora for community members to discuss the challenges pertinent to the game topic and actions needed, as well as (where there is experiential learning from playing the game) potential “spillover” of learning effects (from players to non-players; Falk et al. 2023).

The community debriefing data were also audio-recorded with participants’ consent. Excerpts of transcripts and selected quotes were translated into English.

Analytical approach for the quantitative analysis

We estimated sample-level utility coefficients for the DCE results using two different specifications of a mixed logit main-effects only model of utility using Stata: (1) one in which we separated out time use for own herd labor and one for shared community labor, and (2) one in which we pooled together both time use variables as overall labor time:

(1)

where u_ijt is the utility outcome for individual i, on choice alternative j within choice set t; x_ijt is the vector of choice attributes observed by individual i on choice alternative j within choice set t; β_i is a vector of utility coefficients on the observed attributes specific to individual i; and ε_ijt is an error term. We did not find coefficients on either time use attribute in the first specification to be significantly different from zero, suggesting respondents were not attentive to this detailed breakdown of their time (compare “dce_regressions_self_comm” and “dce_regressions_pooled_time” folders in GreenReserve V1.12 repository: see link in Data Availability section). Therefore, in our analysis we work with the second specification pooling these as overall work time. We generate participant-level estimates of utility coefficients, and use labor time as a cost measure to allow us to estimate willingness to pay / willingness to accept (WTP/WTA) measures as the marginal rate of substitution of labor time for the other attributes of herd size and good and bad year losses; that is, the ratio of the utility coefficient on herd size or losses (in animals^-1) over the utility coefficient on labor time (in time^-1) yielding a measure of time accepted/given per additional animal.

We analyzed game results at the level of the game round, including data from rounds from 3 to the final 16th round in each of our completed games as outcome variables (the preceding two rounds of rainfall data were included as covariates in the analysis, precluding the analysis of rounds 1 and 2 as outcomes). Outcomes expressed (amounts of grass, number of squares used, number of animals, etc.) are all summed or averaged across the landscape. All analyses explaining outcomes (utility coefficients or game parameters) are robust ordinary least squares regression with standard errors clustered at the village level (for DCE results) or at the game session level (for game results). All regression outcomes are presented with model coefficients (significance marked at 1%, 5%, and 10%) and model fit statistics as appropriate; standard errors are not presented in manuscript tables but are included in complete tables included in the “Data tables with SE” folder in the GreenReserve V1.12 repository (see link in Data Availability section).

RESULTS

Findings from the DCE: pre-game, post-game, and comparison

Measuring preferences on herds, losses, and time use in the pre-game DCE

Overall, we observed pastoralists to derive utility (prefer) from big herds, and to derive disutility (not prefer) from losses of cattle and the use of their time for work (Table 2), providing face validity for participants’ understanding of attributes in the choice experiment. Across the field study, we saw those whose herds were mainly cattle or camels (against a baseline of shoats) showing stronger positive preference for large herds, those from households with more working-age males showing less disutility on good-year losses, and female respondents showing less disutility on labor time (Table 3).

How willing are pastoralists to invest time to reduce losses in the pre-game DCE?

We did not observe very much explanatory power over WTP and WTA, with the note that older players appeared more willing to invest time to have larger herds and reduced good year losses in the pre-game DCE, while players from households with more working-age females were less willing to spend time to reduce losses in bad years (Table 4).

Comparing stated preferences between pre-game and post-game DCE

Overall estimates in the DCE after our game experiment showed decreased utility on having large herds, and greater disutility on losses during good seasons and the overall use of time for labor (Table 2). Changes in the ratio of utility coefficients before and after the intervention is instructive; for example, comparing the ratio of preferences for good-year losses over those for herd size suggests that along the intervention, participants move (on average) from being willing to give up 1.76 animals in herd size to avoid losing 1 animal, to being willing to give up 2.36 animals to avoid losing 1 animal after the intervention. This is broadly consistent with experimental expectations around losses (people dislike losses about twice as much as they like gains; Kahneman 2003), and suggests participants are more acutely concerned with the loss of animals following the intervention. Higher preferences for large herds in the post-game DCE were associated with female players and older players, whereas lower preferences for large herds were associated with those with bigger herds to begin with (Table 3). Older players and female players were also associated with greater disutility on spending time for labor in the post-game DCE (Table 3). We did not observe many of our player explanatory variables significantly explaining variation in WTP or WTA of labor time for animals in the post-game DCE, with one exception: camel herders were more willing to invest time for larger herds (Table 4). We also observed significantly less agreement with the statement “People should be able to raise however many animals they want, without any rules restricting how the animals are grazed” after players played the game (Table 5).

Examining shifts in pre-game and post-game DCE responses along the study period

We observed the disutility of/from losses during bad seasons to increase (i.e., become increasingly negative) along the period of our study, for both pre- and post-game DCEs (Table 3). We also observed that the difference in willingness to invest time for large herds and reduce losses in good years got smaller over the period of the field work (Table 4); that is to say, the change in DCE estimates from pre-game to post-game was smaller, suggesting a smaller influence from the intervention. Though we cannot rule out the possibility that our enumerator team shifted systematically in their framing of the DCE during that period, or that we visited villages in sequence of systematically worse experiences with losses or stronger beliefs about losses, we have no indications or causes for concern that these occurred; rather, the characteristic of our field work most directly associated with time during October and November 2021 in Moyale was the increasing stress on herders and animals alike of the failed short rainy season (Fig. 1).

Findings from the game: pre-game, post-game, and comparison

How did people play the game?

We played 22 game sessions with identical parameters, including the identical patterns of rainfall (Fig. 8a), each consisting of 16 pasture turns. Each player began the game with 10 animals, and under the game parameterization of our study, typically finished the game with around four surviving animals. Grass capacity was calibrated to allow survival of all animals in theory, though survival was probabilistic from round to round, and players may not have coordinated perfectly. The particular parameterization we chose for this study was a “best guess” at reasonable conditions and we make no claim that this specific set of outcomes (averaging more than 50% losses over two years) had spatial or temporal validity to the study site; it was the set of parameter values we chose, and kept constant across game sessions to maximize statistical power.

What shaped the designation of private and reserve areas?

We observed no variation in the amount of private land specified; all players in all games took the one square allowed, but we did see some variation in shared reserve areas, with most games choosing 4 or 6 squares, but some games choosing as many as 12 squares (there was no fixed maximum of squares that could be designated as shared reserve). We are limited in our capacity to explain this by a small number of cases, but do examine the role of some basic player characteristics (age, sex, household and herd structure). We observed in our best simple model (Table 6, Model 2, minimizing AIC and BIC) that (controlling for the number of living animals in the game) the younger the participants and the more female participants, the more that reserve areas were made use of in the game.

How did different people play differently?

Overall, much of the variation in key game outcomes (grass use, standing biomass, and surviving animals) is explained by game conditions; current turn and lagged rainfall, and progress through the rounds (Table 7, “Game Conditions Only” models). Notably here, because rainy seasons are short, only 1–2 turns, current and lagged rainfall can be very different, with differing roles in shaping in-game behavior. Beyond these basic in-game explanatory factors, characteristics of the groups also explain significant variation in game outcomes (Table 7, “Game Conditions + Group Characteristics” models).

Greater use of open grass (as opposed to grass in shared reserve squares) is associated with older, more male groups, representing households with more working-age males, who show less disutility on bad season losses, and more disutility on the use of their time for labor. By contrast, greater use of reserve grass is associated with younger groups, with more crop growers and with less disutility on good year losses. Grass cutting is associated with younger, more male groups, representing households with more working age males. Further, it is associated with groups with greater disutility on good and bad-season losses, and with groups with less disutility on spending time to cut more grass. Lastly, groups with more women had fewer “conflicts,” cases in which players canceled (did not permit) the efforts of others to use shared reserve squares.

Examining shifts in game play along the study period

We observed that game groups used less open grass, and made greater use of shared reserves, as the month of our field work went on, leading to more animals surviving at the end of game sessions. As discussed in the case of the DCE, though we cannot rule out shifts in enumerator trainings or systematic differences in the villages as sequenced, increased salience of the failing rainy season along the month is the pattern most plausibly represented by the date variable.

Participant interactions during GreenReserve sessions

Communications during the gameplay also revealed important dynamics of pastureland management. For example, just as in the game settings, participants commonly agreed on the use of shared reserve areas in times of crisis, such as the late arrival of rains or when facing a lack of preserved or live grass, thereby preserving other rangeland areas from overgrazing. These were illustrated by the following quotes:

I am worried a lot with how the environment looks. There is yellow color all over ... Let us move into the community reserve because it is the only place that still has grass.

It is raining now. No one shall be allowed to enter the shared land.

No long-distance herding now, it is the dry season.

While making decisions in the games, participants also regularly inquired about the number of animals each one has left. The disutility on losses during bad seasons was most apparent as participants progressed with the pasturing turns in the game, and they have exhausted the grassland carrying capacity. For example:

My animals have reduced. ... There is no grass. ... There is no grass. ... Oh no, my animals have died.

Participants also found the dynamic features of the games to be very engaging and enjoyable. The games provided a relaxed and safe atmosphere for expressing preferences that can be sensitive or emotionally charged:

What I enjoyed most about the games is how we were herding. Sometimes, when you lose animal, you just laugh about it.

We were happy because it [the game] relates to our lives.

Participant A to participant B: Why would you want to enter my private space? Get out! [Both laugh]

That was interesting, thank you for having us.

Although many aspects of reality translate into gameplay, some participants could not fully associate their real-life experiences with the game environment. In particular, these participants noted the inability of the games to capture the influence of migrants or “outsiders” and population growth, perceived as the greatest threats to the sustainability of rangeland management, to be frustrating, as the following sentences demonstrate:

The game was a bit hard because in our traditions, we do not graze through a machine but in the fields ... And in the fields, it’s mostly the outsiders that degrade and destroy our community reserve land, making it no longer viable.

These days, the population is bigger and pastoralists from everywhere like coming here and when it’s raining, everyone comes. These people are now so many and have overpowered us and even after chasing them away today, they come the following day without permission.

One participant also perceived tree species invasion as one of the key issues in rangeland management, and suggested that it should be included in future versions of the game:

Also, the invasive species trees are destroying the environment, they suppress the grass growth. The conditions will deteriorate, and the future will be hard.

Drought perceptions in focus group and debrief data

Outside of our experimental instruments, comments from participants in our focus group discussions and debrief support the notion that the current drought was salient in participants’ minds, although we did not observe changes in how those concerns were expressed along the field study.

We foresee a lot of difficulties soon. For example, today is 19th October and we still haven’t gotten any rain. A few years back, at such a date, this entire area would be green. If the trend continues, scarce rain will lead to animal loss and things will be very difficult.

Presently, it should have rained in October, and we still haven’t experienced any rain. So everyone is now worried about where they will graze and move their animals.

Most participants perceived drought and its increasing severity to be a primary cause of the decline in household livestock assets, and as a rising threat to food security.

In the past, there were places we would move to during drought. But this drought is worse and all places are affected in the same way.

Animals have been experiencing hunger for around one and a half years. The last rainy season was short and the one before was short and this one as well. So it seems animals are just dying.

We are also experiencing hunger and lack of water. Without water, you cannot cook even the little food available.

In most places, droughts have also contributed to the erosion of traditional practices such as the establishment of community reserve lands.

In the past this is exactly how life used to be but now because of drought and many problems within the community, some things such as setting aside community land are difficult for us to do. So now we don’t have community land.

Some participants felt helpless as droughts have become more severe and their capacity to cope with the social-ecological changes is weakening. Insurance schemes aimed at compensating for any livestock losses caused by extreme weather, and thereby increasing communities’ resilience, have been introduced in the study area, but most participants experienced disappointment on their effectiveness.

DISCUSSION

Did GreenReserve capture meaningful dimensions of pastoralism?

A key question for a pilot activity such as GreenReserve is whether the instrument captures aspects of participants’ livelihood that they find meaningful. Our focus group discussions and debriefings revealed that the key dilemmas presented in GreenReserve have analogs in our participants’ livelihoods. Participants traditionally establish community reserve lands and rules regulating its access and use, especially during episodes of heightened droughts. Some of these rules are still enforced in the present. This is illustrated by the following quotes from FG participants.

Whenever it rains, we hold a meeting and decide that livestock should not move to this specific place, we call it Mazingira. Livestock is not allowed in this place. We use [it] during the dry seasons when pasture runs out in other areas. The livestock is captured and the trespasser is fined Ksh. 5000.

If you enter the shared community reserve without permission, we fine.

We are one community and when it rains, we decide together which place we should move to. ... Also during drought, we decide together on how we should herd and where we should settle to be able to access water.

Traditional ecological knowledge enables participants to survive and adapt in their unpredictable environments, thereby increasing the sustainability of their practices. Pastureland management includes migration to neighboring grazing areas during the rainy season to allow grassland regeneration.

Animals depend on the environment. And when it rains, we move the animals away, even to our Ethiopian neighbors; the grass regenerates and water is reserved.

Although the game settings were necessarily simplified, they were perceived by most participants as a faithful representation of their realities.

This game represents our daily lives. We have a community reserve and during the rainy season no one is allowed to herd near it. We also have helpers just like in the game. When it rained in a certain area, we discussed among ourselves and decided how we would move. When there was no rain, we returned to our own private reserves or near a water source or the community land to graze.

The incorporation of the temporal and spatial dimension helped capture the dynamic features of pasturing practices in the study areas. For instance, by moving livestock around, pastoralism diversifies strategies for adapting to a changing climate.

The ways it reflects our current lives is that during dry season and when there is no grass, you send someone (a scout) to go look for areas that have grass, and then move to areas with grass and water.

The community debriefing further indicated that the game can engage players to reflect upon challenges related to climate adaptation decision making and rangeland management strategies, and, hence, facilitate learning.

We have learned that during dry seasons like now, one way of keeping your animals from dying is using the community reserve. When it rains, we advise moving the animals. Also, it’s important to know the size of your herd. If you have many animals, they die easily during the dry periods, and with few animals, they are easy to manage. We have learned a lot.

We saw the animals survived more during the rainy season and when the animals were not moved far from home. And moving into areas that are greener. Also making sure not to share space with another herd.

Did GreenReserve highlight salient dimensions of current environmental concern and change?

Individual enclosure of pastureland and the preservation of grass are new pasturing practices that have emerged as drought coping strategies in the past few years.

Fencing of individual family pastureland is also new because before, there used to be enough pasture for everybody.

Problems caused by change in seasonal patterns have led us to preserve grass.

When the pasture is plenty, we are allowed to go and harvest and store the grass from the community reserve, to prevent exhaustion by animals from outside our community.

Some participants even foresee an important weakening of the role of pastoralism in supporting local livelihoods in the near future as droughts become more intense and the number of animals decreases. Only the wealthiest members of the pastoral communities are expected to survive in these contexts.

Individuals who are able, their animals are still healthy. Since they have cars, they transport water to the animals and spray them. Those are the people who will still have animals five years from now.

Although cooperative relationships used to exist between seasonally migrant populations that herd livestock, drought has disrupted these arrangements, sometimes causing conflicts with the residents, and eroding communal management.

There’s something that we used to do a long time ago with our grandfathers but now we’re unable. We used to have grazing land for different communities and when the Ethiopians came, they used to follow certain rules. When we go to Ethiopia and Isiolo, we also have to ask for permission. Nowadays people just come without asking. People will fight and there will be conflict.

They are usually armed and getting into confrontation may not end well.

The most important thing to do right now is to set aside land for community land ... and everyone outside this location wants to come and graze here. This is what prevents us from setting aside community land for use later.

Across these observations we see the local, within-community coordination challenges made easier by adaptations such as grass harvesting (mechanisms observable in our piloted GreenReserve game), as well as the regional, between-community coordination challenges that are growing as pastoralists from different groups increasingly overlap as they struggle to meet their herds’ needs. Our game as implemented in the current study does not tackle this challenge, but could do so with only small changes in parameters: a larger green space, and variation in the rules by which players could coordinate within and across groups. Such a structure would allow study of the within-group and between-group interaction: e.g., how does improved grass harvesting coordination relax between-group conflict?

What explained game play?

We also see some evidence of reality shaping responses in the experiment, of real concerns around drought and implications for herd animals shaping preferences in the DCE and decision making in GreenReserve. People bring many layers of behavior to the game, their response to the rules of the game, to the dynamic of the group, to the novelty of a tablet; the other priorities they have for their day, beyond their beliefs and preferences related to the resource domain (Cárdenas and Ostrom 2004). The observed apparent influence of current drought conditions on perceptions and behavior around herding support the idea that (among other layers of behavior) in a well-designed game, the linkage to real-life decision making is there to be found.

Gender balance shapes many dimensions of herding thinking: female herders value large herds less overall, but were more likely to have their perception of large herds improved through the intervention. In the games, groups with more female herders were less likely to use open grass as well as less likely to cut grass for harvest; further, they were less likely to have conflicts over the use of the reserves. Age shapes game play as well. In fact, younger players were more likely to make use of shared reserve squares, as well as explore the practice of grass harvesting, whose real-world analog is a relatively recent practice. Although some of this difference is likely a signal of younger game players adapting to the rules of a game more quickly, the joint results of age and gender in game play suggest important research questions to explore in future work beyond the pilot scale: will youth and improved space for women in decision making help (and be necessary for) pastoral systems to integrate the innovations (such as fencing and grass harvesting) that will keep them resilient to climate and resource pressure? Despite their importance to the long-term sustenance of pastoralism, the voice of pastoral youth remains unheard (Maru 2017). Our results demonstrate the importance of understanding pastoral youth’s preferences and behavior and engaging them in adaptation decision making. Our results about age, gender, and education also call for further research on possible socio-cultural shifts as a result of drought-induced threats and what such shifts mean for the pastoral system. Marty et al. (2023), for example, documented an important socio-cultural shift among transitioning Maasai pastoralists in southern Kenya during the diversification processes away from customary governance practices, which favored elder men, to one in which formal education and knowledge take precedence.

Lastly, we see shifts in preferences and priorities following game play (via the DCE and beliefs questions) that suggest the games intervention shifted participants’ thinking or attention. The game environment pushes players to think about their spatial and temporal coordination in a way that may relate to (and have spillover to) their own herding interactions. Follow-up exercises at later times (such as reported by Meinzen-Dick et al. 2018), coupled to control exercises or new villages added to the intervention, could identify if any longer-term legacies of this exercise remain. Results at this first fieldwork stage suggest that, in addition to being a useful research tool, the intervention (including games and community discussion) may also have value as a component of herding extension learning.

Construct validity, learning, and next steps

Synthesizing across the three threads above, we found evidence that the problem domain (i.e., the experiential learning action situation) we developed in GreenReserve as well as the specific decisions (what land to allocate as shared reserve, how to allocate scarce time to cattle grazing vs. dry forage harvesting) resonated with participants as capturing key elements of the real life action situation. We also found evidence that the way participants made decisions in GreenReserve had meaningful linkages to perceptions and decisions they made in real life, perhaps most acutely in the way rising anxiety over drought losses in the real world transferred into better pasture resource use and animal survival in GreenReserve. These signals are early support, hopefully to be built upon, that GreenReserve provides a valid construct for local pastoral decision making. As the strength of this signal grows, we will be better able to put GreenReserve findings, such as the gender- and youth-differentiated behavior, in appropriate context.

Should we have the opportunity to follow with GreenReserve what Janssen et al.’s many publications have accomplished with their irrigation game and explore a range of contexts and institutions, there are several directions the work could be expanded. First, we chose one very specific game parameterization within our small pilot in the interest of maximizing statistical power. However, within the existing GreenReserve structure, we have much to learn by varying rainfall conditions (scarcity, year-to-year variation, etc.), game length, number of players (varying the coordination challenge), pasture quality and heterogeneity, as well as household and herd structure and heterogeneity (and through these capturing other processes in the conceptual model presented in Fig. 1). Second, our evaluation of attention and perceptions via the debrief and repeated DCE would be enhanced by longer-term observation and return visits (precluded in the present case by COVID-19 delays and the end of the project), as well as the randomization of pastoralist participants into “treatment” groups receiving our games and debrief intervention, and “control” groups engaging only in the DCE, as a better means of demonstrating internal validity, that observed shifts in attention and perception are attributable to our intervention. We would perhaps also benefit from conducting the DCE with those pastoralists who participate only in the debrief, as a means of disentangling the roles of game and debrief in reshaping mental models outlined in Falk et al. (2023). Third, the scope of GreenReserve could be extended beyond the within-community coordination challenge to address the increasingly important between-community coordination challenge described by participants, capturing the increasing degree to which declining pasture quality and availability draw pastoralists further into open lands and coordination challenges with people they do not know. In undertaking any of these extensions, we would hope to be adding to prior examples like Janssen et al.’s irrigation game of the process of building out novel, validated instruments for understanding and learning from human-environment problems.

CONCLUSION

We developed a mixed methods intervention in pastoral communities in Northern Kenya that included focus group discussion, a discrete choice experiment, a novel experimental game called GreenReserve, and a community debrief. Together, these methods enable a hybrid inductive (generated in qualitative data), deductive (tested in quantitative data) approach to thematic analysis.

We found GreenReserve to capture dimensions of human-environment dilemmas faced by pastoralists, and for playing GreenReserve to shift (in the period immediately following game play) preferences around herd size, losses, and the use of time, as measured through a repeated discrete choice experiment. We found game groups with younger players and with more female players to include more exploration of pastoral adaptations such as the use of grass reserves and the harvesting of grass, as well as (for groups with more female players) to have less conflict within the game. We also observed both preferences (in the DCE) as well as game strategy to shift across different game groups along the length of the study; as the region moved further into a drought and failed short rainy season, players were more conscious of large herds, bad rainfall-year losses, and made better use of reserves in game play. These findings point to two key dimensions for expanded work: (i) expansion of the intervention to a larger sample, facilitating analysis of a greater breadth of explanatory factors and inclusion of additional treatments such as variation in climate, pasture, or household and community conditions, in particular, future research, should shed light on how national policies and climate adaptation programs (such as the County Climate Change Fund) can support the adaptation responses and resilience building of pastoral communities based on their preferences and needs, and (ii) follow-up visits with participants at later points in time, alongside new participants, to identify whether mixed methods approaches such as this helpfully function as extension learning. In particular, group games such as GreenReserve are promising experiential learning tools to simulate key elements of reality and group dynamics, allowing participants to “experiment” formulating rules and “experience” the outcome of rules and cooperation (or the lack of it) in a safe environment. Such learning can complement other adaptation capacity building efforts to support the resilience building of vulnerable populations in a climate crisis.

Importantly, we see the potential for GreenReserve to capture variation across players in the importance of different adaptive strategies. This first version included the capacity to coordinate across space, to differentiate rules for different spaces, and to preserve forage as hay for dry periods. Future iterations could also incorporate livestock diversification or cropping, engagement over larger spaces with separate communities, and vary the number and structure of households and herds. In this way, differentiated applications of GreenReserve could identify contrasts and common threads in decision making across the differing nomadic and transhumant pastoral systems that face different stressors across African spaces. Tools such as GreenReserve offer opportunities for observation, for exploratory policy investigation, and as inputs to policy simulations, for experiential learning among resource users as well as for role play and learning among decision makers and policy designers. A longer-term goal is that through tools like GreenReserve, we can identify what interventions best shift thinking and behaviors toward adaptations that strengthen balancing feedbacks and the stability of pastoralism as a critical food systems component across African landscapes.

LITERATURE CITED

Abebe, D. 2016. Resilience and risk in Borana pastoral areas of Southern Ethiopia: recent trends in diversified and alternative livelihoods. Page in P. Little, editor. Resilience and risk in pastoralist areas: recent trends in diversified and alternative livelihoods. USAID, Washington, D.C., USA.

Adzande, P. 2019. Migration of pastoralists in Africa: reflections on practical and policy Implications. African Human Mobility Review 5(2). https://doi.org/10.14426/ahmr.v5i2.885

Alpizar, F., F. Carlsson, and M. Naranjo. 2011. The effect of ambiguous risk, and coordination on farmers' adaptation to climate change — a framed field experiment. Ecological Economics 70(12):2317-2326. https://doi.org/10.1016/j.ecolecon.2011.07.004

Anderies, J. M., M. A. Janssen, A. Lee, and H. Wasserman. 2013. Environmental variability and collective action: experimental insights from an irrigation game. Ecological Economics 93:166-176. https://doi.org/10.1016/j.ecolecon.2013.04.010

Ayantunde, A. A., R. Asse, M. Y. Said, and A. Fall. 2014. Transhumant pastoralism, sustainable management of natural resources and endemic ruminant livestock in the sub-humid zone of West Africa. Environment, Development and Sustainability 16(5):1097-1117. https://doi.org/10.1007/s10668-014-9515-z

Baggio, J. A., N. D. Rollins, I. Pérez, I., and M. A. Janssen. 2015. Irrigation experiments in the lab: trust, environmental variability, and collective action. Ecology and Society 20(4):12. https://doi.org/10.5751/ES-07772-200412

Belayneh, D. 2016. State development interventions versus indigenous resource management institution: Whose reality count? Evidence from Borana pastoral system of Southern Ethiopia. Munich Personal RePEc Archive 70780.

Bell, A. R., O. S. Rakotonarivo, A. Bhargava, A. B. Duthie, W. Zhang, R. Sargent, A. R. Lewis, and A. Kipchumba. 2023. Financial incentives often fail to reconcile agricultural productivity and pro-conservation behavior. Communications Earth and Environment 4:27. https://doi.org/10.1038/s43247-023-00689-6

Bell, A., and W. Zhang. 2016. Payments discourage coordination in ecosystem services provision: evidence from behavioral experiments in Southeast Asia. Environmental Research Letters 11:114024. https://doi.org/10.1088/1748-9326/11/11/114024

Bell, A., W. Zhang, and K. Nou. 2016. Pesticide use and cooperative management of natural enemy habitat in a framed field experiment. Agricultural Systems 143:1-13. https://doi.org/10.1016/j.agsy.2015.11.012

Benti, T. 2018. Factors influencing indigenous knowledge of rangeland management practices in Borana pastoral areas, Southern Ethiopia. Journal of Science and Sustainable Development 6(2):62-78.

Birhanu, Z., A. Ambelu, N. Berhanu, A. Tesfaye, and K. Woldemichael. 2017. Understanding resilience dimensions and adaptive strategies to the impact of recurrent droughts in Borana Zone, Oromia Region, Ethiopia: a grounded theory approach. International Journal of Environmental Research and Public Health 14(2):118. https://doi.org/10.3390/ijerph14020118

Brozyna, C., T. Guilfoos, and S. Atlas. 2018. Slow and deliberate cooperation in the commons. Nature Sustainability 1(4):184-189. https://doi.org/10.1038/s41893-018-0050-z

Bruns, B., and C. Kimmich. 2021. Archetypal games generate diverse models of power, conflict, and cooperation. Ecology and Society 26(4):2. https://doi.org/10.5751/ES-12668-260402

Burian, A., R. Karaya, J. E. V. Wernersson, M. Egberth, B. Lokorwa, and G. Nyberg. 2019. A community-based evaluation of population growth and agro-pastoralist resilience in Sub-Saharan drylands. Environmental Science and Policy 92:323-330. https://doi.org/10.1016/j.envsci.2018.10.021

Camerer, C. F. 2015. The promise and success of lab-field generalizability in experimental economics: a critical reply to Levitt and List. Pages 249-295 in G. Frechette and A. Schotter, editors. Handbook of experimental economic methodology. First edition. Oxford University Press, Oxford, UK. https://doi.org/10.1093/acprof:oso/9780195328325.003.0016

Cárdenas, J. C., and J. Carpenter. 2008. Behavioural development economics: lessons from field labs in the developing world. Journal of Development Studies 44(3):311-338. https://doi.org/10.1080/00220380701848327

Cárdenas, J. C., and E. Ostrom. 2004. What do people bring into the game: experiments in the field about cooperation in the commons. Agricultural Systems 82(3):307-326. https://doi.org/10.1016/j.agsy.2004.07.008

Crookall, D. 2023. Debriefing: a practical guide. Pages 115-124 in L. Angelini and R. Muñiz, editors. Simulation for participatory education: virtual exchange and worldwide collaboration. Springer, Cham, Switzerland. https://doi.org/10.1007/978-3-031-21011-2_6

Dabasso, B. H., G. Oba, and H. G. Roba. 2012. Livestock-based knowledge of rangeland quality assessment and monitoring at landscape level among Borana herders of northern Kenya. Pastoralism 2:2. https://doi.org/10.1186/2041-7136-2-2

Debela, N., D. McNeil, K. Bridle, and C. Mohammed. 2019. Adaptation to climate change in the pastoral and agropastoral systems of Borana, South Ethiopia: options and barriers. American Journal of Climate Change 8(1):40-60. https://doi.org/10.4236/ajcc.2019.81003

Degen, A. A. 2011. Transformation of Borana from nomadic pastoralists to agropastoralists and shift of livestock from cattle to include more goats, camels and sheep in Southern Ethiopia. International Journal of Business and Globalisation 6(3-4):292-312. https://doi.org/10.1504/IJBG.2011.039389

ElDidi, H., W. Zhang, I. Blackmore, F. Gelaw, C. de Petris, N. Teka, S. Yimam, D. Mekonnen, C. Ringler, and R. Meinzen-Dick. 2024. Getting ahead of the game: experiential learning for groundwater governance in Ethiopia. International Journal of the Commons 18(1):66-81. https://doi.org/10.2499/p15738coll2.136723

Esterling, K. M., D. Brady, and E. Schwitzgebel. 2023. The necessity of construct and external validity for generalized causal claims: a critical review of the literature on quantitative causal inference. I4R Discussion Paper Series 18, The Institute for Replication (I4R).

Falk, T., W. Zhang, R. Meinzen-Dick, L. Bartels, R. Sanil, P Priyadarshini, and I. Soliev. 2023. Games for experiential learning: triggering collective changes in commons management. Ecology and Society 28(1):30. https://doi.org/10.5751/ES-13862-280130

Government of Kenya (GoK). 2016. Kenya National Adaptation Plan 2015-2030. Government of Kenya, Nairobi, Kenya.

Hidrobo, M., J. B. Hoel, and K. Wilson. 2021. Efficiency and status in polygynous pastoralist households. Journal of Development Studies 57(2):326-342. https://doi.org/10.1080/00220388.2020.1762863

Hoel, J. B. 2015. Heterogeneous households: a within-subject test of asymmetric information between spouses in Kenya. Journal of Economic Behavior and Organization 118:123-135. https://doi.org/10.1016/j.jebo.2015.02.016

Hoel, J. B., M. Hidrobo, T. Bernard, and M. Ashour. 2021. What do intra-household experiments measure? Evidence from the lab and field. Journal of Economic Behavior and Organization 188:337-350. https://doi.org/10.1016/j.jebo.2021.05.019

Iticha, B., and A. Husen. 2019. Adaptation to climate change using indigenous weather forecasting systems in Borana pastoralists of Southern Ethiopia. Climate and Development 11(7):564-573. https://doi.org/10.1080/17565529.2018.1507896

Janssen, M. A., and J. M. Anderies. 2013. A multi-method approach to study robustness of social-ecological systems: the case of small-scale irrigation systems. Journal of Institutional Economics 9(4):427-447. https://doi.org/10.1017/S1744137413000180

Janssen, M. A., J. M. Anderies, and J. C. Cardenas. 2011. Head-enders as stationary bandits in asymmetric commons: comparing irrigation experiments in the laboratory and the field. Ecological Economics 70(9):1590-1598. https://doi.org/10.1016/j.ecolecon.2011.01.006

Janssen, M. A., J. M. Anderies, I. Pérez, and D. J. Yu. 2015. The effect of information in a behavioral irrigation experiment. Water Resources and Economics 12:14-26. https://doi.org/10.1016/j.wre.2015.09.001

Janssen, M. A., F. Bousquet, J.-C. Cardenas, D. Castillo, and K. Worrapimphong. 2012. Field experiments on irrigation dilemmas. Agricultural Systems 109:65-75. https://doi.org/10.1016/j.agsy.2012.03.004

Janssen, M. A., T. Falk, R. Meinzen-Dick, and B. Vollan. 2023. Using games for social learning to promote self-governance. Current Opinion in Environmental Sustainability 62:101289. https://doi.org/10.1016/j.cosust.2023.101289

Janssen, M. A., A. Lee, and T. M. Waring. 2014. Experimental platforms for behavioral experiments on social-ecological systems. Ecology and Society 19(4):20. https://doi.org/10.5751/ES-06895-190420

Kahneman, D. 2003. Maps of bounded rationality: psychology for behavioral economics. American Economic Review 93(5):1449-1475. https://doi.org/10.1257/000282803322655392

Knapp, G., and J. J. Murphy. 2010. Voluntary approaches to transitioning from competitive fisheries to rights-based management: bringing the field into the lab. Agricultural and Resource Economics Review 39(2):245-261. https://doi.org/10.1017/S1068280500007279

Levitt, S. D., and J. A. List. 2007. What do laboratory experiments measuring social preferences reveal about the real world? Journal of Economic Perspectives 21(2):153-174. https://doi.org/10.1257/jep.21.2.153

Liao, C., P. E. Clark, S. D. DeGloria, and C. B. Barrett. 2017. Complexity in the spatial utilization of rangelands: pastoral mobility in the Horn of Africa. Applied Geography 86:208-219. https://doi.org/10.1016/j.apgeog.2017.07.003

Liao, C., P. E. Clark, M. Shibia, and S. D. DeGloria. 2018. Spatiotemporal dynamics of cattle behavior and resource selection patterns on East African rangelands: evidence from GPS-tracking. International Journal of Geographical Information Science 32(7):1523-1540. https://doi.org/10.1080/13658816.2018.1424856

Marty, E., R. Bullock, M. Cashmore, T. Crane, and S. Eriksen. 2023. Adapting to climate change among transitioning Maasai pastoralists in southern Kenya: an intersectional analysis of differentiated abilities to benefit from diversification processes. Journal of Peasant Studies 50(1):136-161. https://doi.org/10.1080/03066150.2022.2121918

Maru, N. 2017. Youth perspectives on pastoralism: opportunities and threats faced by young pastoralists. Nature & Faune 31(2).

Meinzen-Dick, R., R. Chaturvedi, L. Domènech, R. Ghate, M. A. Janssen, N. D. Rollins, and K. Sandeep. 2016. Games for groundwater governance: field experiments in Andhra Pradesh, India. Ecology and Society 21(3):38. https://doi.org/10.5751/ES-08416-210338

Meinzen-Dick, R., M. A. Janssen, S. Kandikuppa, R. Chaturvedi, K. Rao, and S. Theis. 2018. Playing games to save water: collective action games for groundwater management in Andhra Pradesh, India. World Development 107:40-53. https://doi.org/10.1016/j.worlddev.2018.02.006

Meyer, S., P. Santos, and F. Yang. 2021. Economic games can be used to promote cooperation in the field. Proceedings of the National Academy of Sciences of the United States of America 118(47):e2026046118. https://doi.org/10.1073/pnas.2026046118

Mgalula, M. E., U. Richter, O. Hensel, O. V. Wasonga, and C. Hülsebusch. 2023. Mapping cropland evolution based on biophysical data and cropping history in the Borana rangelands, Southern Ethiopia. Regional Environmental Change 23:80. https://doi.org/10.1007/s10113-023-02075-6

Muhamad, J. W., and S. Kim. 2020. Serious games as communicative tools for attitudinal and behavioral change. Pages 141-162 in H. D. O'Hair and M. J. O'Hair, editors. The handbook of applied communication research: Volume 1. Wiley, Hoboken, New Jersey, USA. https://doi.org/10.1002/9781119399926.ch9

Ostrom, E. 1990. Governing the commons: the evolution of institutions for collective action. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9780511807763

Ostrom, E. 2010. Beyond markets and states: polycentric governance of complex economic systems. American Economic Review 100(3):641-672. https://doi.org/10.1257/aer.100.3.641

Ouédraogo, K., A. Zaré, G. Korbéogo, O. Ouédraogo, and A. Linstädter. 2021. Resilience strategies of West African pastoralists in response to scarce forage resources. Pastoralism 11:16. https://doi.org/10.1186/s13570-021-00210-8

Pfaff, A., L. A. Rodriguez, and E. Shapiro-Garza. 2019. Collective local payments for ecosystem services: new local PES between groups, sanctions, and prior watershed trust in Mexico. Water Resources and Economics 28:100136. https://doi.org/10.1016/j.wre.2019.01.002

Rakotonarivo, O. S., A. Bell, B. Dillon, A. B. Duthie, A. Kipchumba, R. A. Rasolofoson, J. Razafimanahaka, and N. Bunnefeld. 2021b. Experimental evidence on the impact of payments and property rights on forest user decisions. Frontiers in Conservation Science 2:661987. https://doi.org/10.3389/fcosc.2021.661987

Rakotonarivo, O. S., I. L. Jones, A. Bell, A. B. Duthie, J. Cusack, J. Minderman, J. Hogan, I. Hodgson, and N. Bunnefeld. 2021a. Experimental evidence for conservation conflict interventions: the importance of financial payments, community trust and equity attitudes. People and Nature 3(1):162-175. https://doi.org/10.1002/pan3.10155

Sargent, R., O. S. Rakotonarivo, S. P. Rushton, B. J. Cascio, A. Grau, A. R. Bell, N. Bunnefeld, A. Dickman, and M. Pfeifer. 2022. An experimental game to examine pastoralists’ preferences for human-lion coexistence strategies. People and Nature 4(5):1263-1278. https://doi.org/10.1002/pan3.10393

Shettima, A., and U. Tar. 2008. Farmer-pastoralist conflict in West Africa: exploring the causes and consequences. Information, Society and Justice 1.2:163-184.

Speelman, E., M. van Noordwijk, and C. Garcia. 2017. Gaming to better manage complex natural resource landscapes. Chapter 33 in S. Namirembe, B. Leimona, M. van Noordwijk, and P. A. Minang, editors. Co-investment in ecosystem services: global lessons from payment and incentive schemes. World Agroforestry Centre (ICRAF), Nairobi, Kenya.

Tache, B., and G. Oba. 2010. Is poverty driving Borana herders in Southern Ethiopia to crop cultivation? Human Ecology 38(5):639-649. https://doi.org/10.1007/s10745-010-9349-8

Tilahun, F. 2020. Major climate change adaptation and coping strategies in Borana, Southern Ethiopia: a review. Journal of Resources Development and Management 69:1-8.

Wario, H. T., H. G. Roba, and B. Kaufmann. 2016. Responding to mobility constraints: recent shifts in resource use practices and herding strategies in the Borana pastoral system, Southern Ethiopia. Journal of Arid Environments 127:222-234. https://doi.org/10.1016/j.jaridenv.2015.12.005

Yassin, I. M. 2019. Bush encroachment in Borana rangeland in the case of Southern Ethiopia: causes, impacts and management implications. International Journal of Agriculture Innovations and Research 7(4):2319-1473.

Zampaligré, N., D. Ouedraogo, C. Chikozho, L. Sawadogo, and E. Schlecht. 2019. Changes in livelihood strategies and animal husbandry practices of pastoralists in the sub-humid zone of West Africa. African Journal of Agricultural Research 14(30):1311-1325. https://doi.org/10.5897/AJAR2019.13878

Zhang, W., J. Gowdy, A. M. Bassi, M. Santamaria, F. DeClerck, A. Adegboyega, G. K. S. Andersson, A. M. Augustyn, R. Bawden, A. Bell, I. Darknhofer, J. Dearing, J. Dyke, P. Failler, L. Galetto, C. C. Hernández, P. Johnson, S. K. Jones, G. Kleppel, A. M. Komarek, A. Latawiec, R. Mateus, A. McVittie, E. Ortega, D. Phelps, C. Ringler, K. K. Sangha, M. Schaafsma, S. Scherr, M. S. Hossain, J. P. R. Thorn, N. Tyack, T. Vaessen, E. Viglizzo, D. Walker, L. Willemen, and S. L. R. Wood. 2018. Systems thinking: an approach for understanding ‘eco-agri-food systems.’ Pages 17-55 in TEEB for Agriculture & Food: Scientific and Economic Foundations, Geneva, Switzerland.