‘We Are Not Bad People’- Bricolage and the Rise of Community Forest Institutions in Burkina Faso

From a critical institutionalism and institutional bricolage perspective, this article analyses what drives institutional change in the commons and the outcomes for forest and people. It builds on the comparison of three neighbouring villages in Burkina Faso that in 1989, expecting higher returns, agreed to release their common lands for the creation of a community forest called Chantier d’Aménagement Forestier (CAF) within an international forestry project. The project created new bureaucratic institutions to replace the pre-existing customary and socially embedded system. Decades later, the three villages display different institutional change pathways and outcomes: one village abandoned the CAF, converted, and sold its forest and land; another maintained the CAF; and a third operates in-between. Using qualitative research methods, we ask why and how these different change trajectories and outcomes occurred among villages of identical cultural and sociopolitical background. The results show that poor design and implementation of the new bureaucratic institutions, as well as their disrespect of customary and socially embedded rules, led to forestland disputes between the villages. The bureaucratic institutions failed to solve those disputes, effectively manage the forest, and share the benefits equitably. This caused local people’s discontent and prompted actions for change. Actors in diverse ways made use of their social networks, agency, and power relations within and between the villages to either reshape, re-interpret or reject the new forest institutions. These processes of institutional bricolage led to highly diverse trajectories of change. The findings demonstrate the crucial role of locals as agents of change from below and question universal claims in institutional theory on how institutions induce rule-guided behaviour and create path dependencies.


Introduction
In the management of common property's scholarship, the drivers, patterns and outcomes of institutional change remain controversial and one of the least understood areas. What drives change, how institutions evolve over time, the role of micro-politics, local subjectivities and the resulting institutional pathways are areas that still need more empirical investigation (De Moor et al. 2016;Agrawal 2003) and therefore constitute the focus of this article. So far, scholars in mainstream economics attempted to explain institutional change through three main approaches summarized by Coccia (2018): First, the theories on institutional design attribute institutional change to a coordinated and deliberate change in formal rules (e.g., Ostrom 2005). Second, new institutional economics ascribes institutional change to the aggregation of individual choices and the way that informal institutions are selected and evolve (e.g., North 1993). Third, the equilibrium approaches shift the drivers of change to the interactions between institutions and actors and the organisations making use of those institutions. If the first trend of thought neglects the role of informal institutions, the second in contrast underestimates the role of formal institutions, politics, and collective action. However, they all carry some notion of institutional path-dependency where the actors' perceptions and expectations over time converge toward unified goals and produce a relatively predictable change pathway (e.g., North 1993;Heinmiller 2009). There is more variety in new institutionalist studies stemming from Political Science (including on governing commons) (see e.g., Peters 1999;Hassenforder and Barone 2019), with Mahoney and Thelen (2010) providing a theory of gradual institutional change.
In our case, three neighbouring villages, called here as village V, village C and village L experienced similar conditions and forest management programs, yet present three distinctive change trajectories; thus, a more nuanced analysis of the patterns of institutional change is required. Indeed, in 1989, residents of V, C and L villages in southern Burkina Faso were asked by a United Nations Development Program (UNDP) funded project, led by the Food and Agriculture Organization of the United Nations (FAO) and the government, to release lands for the establishment of a Chantier d 'Aménagement Forestier (CAF), a community forest. The CAF refers to a social and ecological structure based on participatory and sustainable forest management and a fuelwood-based income for the local people. The project created new bureaucratic institutions to manage the forest and replace the existing customary and socially embedded institutions. However, as of 2017, the time of the empirical research for this paper, the neighbouring villages displayed highly distinct trajectories of change with village V complying, C challenging and L rejecting the initial new institutional arrangements to which they had subscribed. Therefore, we ask what drove these institutional changes and why did these neighbouring villages with seemingly similar structures and circumstances display these different change trajectories.
To investigate this question, we use a critical institutionalism and institutional bricolage perspective (Cleaver 2012), conscious that understanding changes in the socio-ecological system requires investigating the social, political and cultural spheres that govern those resources (Agrawal and Chhatre 2006;Leach, Mearns, and Scoones 1999). Critical institutionalism highlights institutions as mediating the relationships between the individual and natural resources and society. For Cleaver and De Koning (2015), it also recognizes the complex and uncertain character of institutions and the influence of agency and social networks to shape the actors' actions and outcomes. The concept of institutional bricolage for Cleaver (2002), describes the processes of the political agents, the bricoleurs who consciously or subconsciously make use of various tools at their disposal to realize a project. This implies the reaffirmation, recombination and/or rejection of the pre-existing and newly introduced institutions. Following others such as Friman (2020), Verzijl and Dominguez (2015), de Koning (2014), Sehring (2009), Funder andMarani (2015), our case sheds more light on the bricolage processes taking place between the socially embedded and formalized institutions for governing the commons. We first investigate what drives change in the rules-in-use. Second, we analyse how the three communities used those drivers differently to construct their arguments for or against the existing forest institutions and induce change. Third, we analyse what explains the resulting differences of change trajectories.
We argue that ineffective forest management by the bureaucratic institutions and their disrespect of customary and socially embedded rules were referred to in all three communities as what led the local people to initiate processes of change. We argue that the workings of networks, agency and power mediated those processes and determined the distinctive institutional change trajectories. The collected data illustrate how the interviewees identified seven categories of drivers of change that are identical or recognized by the other villages. However, the villagers re-defined, re-interpreted and re-used differently those common drivers to construct their dominant narratives and justify their actions. The data further corroborate how in each village the networks, agency and power operated to determine the change outcomes. Thus, we respond to the call of Cleaver and De Koning (2015) among others to empirically investigate how specific actors act, shape, sustain or undermine commons' institutions and the extent to which the institutions that are subject to bricolage can be transformative. Examining the systematic reasons why the actors undertake change helps us contribute to the current discussion on how critical institutionalism and bricolage perspectives can contribute to the governability of the commons and inform policy.
The next section introduces the theoretical framework and methods followed by the results section on the assemblage of the forest management institutions, the drivers of change and the way these played out in the three villages. A discussion on the mechanisms by which the networks, agency and power influenced the different trajectories of change ensues, followed by the conclusion.

Theoretical framework
Institutions are conceived as standardized systems of (in)formal rules and procedures regulating social relations between individuals and groups (North 1991). They are dynamic, fluid, constantly contested, and transformed according to people's practices, contexts, time, and circumstances. Critical institutionalism (Cleaver 2012) recognizes these characteristics of institutions but rejects their excessive ability to dictate and predestine peoples' actions. However, critical intuitionalism endorses a logic of practice (Bourdieu 1990); thus, it acknowledges the complexity, multipurpose and embeddedness of institutions in everyday life and the unpredictability of institutional change pathways (Haller 2010). In this sense, the drivers of behaviour and institutional change are embedded in the actors' daily experiences taking root in specific historical and social structures and contexts (Arts et al. 2012;Sandström, Ekman, and Lindholm 2017). The actors, the bricoleurs, are vital to this process. Consciously and sub-consciously, they intelligently mobilize, creatively choose and patch together principles, norms and resources to make and sustain their claims although they are not in full control of these means of production (Johnson 2012).
Scholars point to certain factors that are key to understanding any institutional change process, namely agency, power and networks. For Battilana (2006) and Arts et al. (2012), the role and situation of individual agency within the social structure/networks (as privileged or marginalized) is critical because it determines the degree that institutional opportunities and constraints are available to the agent of change. For Agrawal (2003), a closer attention to micro-politics, especially the ways in which local actors act and attempt to shape projects' interventions is essential for unpacking the issues of agency. Likewise, power relations are inherent to the process as they help to shape the dominant institutional logics while determining the institutional change trajectory (Cleaver and De Koning 2015). Therefore, the critical institutionalist's task is to investigate how the actors in their dealing with daily challenges reinvent the introduced forest institutions and induce changes. Building on these premises, we analyse the observed institutional dynamics as a bricolage process taking place between actors, organizations and institutions. Focusing on these main explanatory factors-agency, networks and power-we examine the strategic actions and tactics used by the actors in forest management that (deliberately or unconsciously) change the forest institutions and the rationales driving these.
The first factor is agency. It describes the social agent's ability to be political and take purposeful actions to realise various goals. Agency in institutional bricolage refers to the actors, the bricoleurs' continuous efforts of using elements at hand to construct narratives and give meaning to their reactions to daily challenges. The exercise of agency depends on the actors' opportunities and resources and the conventional codes of being and doing (Cleaver 2007;Funder and Marani 2015). Discursive and practical consciousness respectively based on the actors' reasoning and everyday life experiences guide the exercise of agency (Giddens 1984). Thus, agency is at the interface of the actors' situation and their assumed and creative practices (Arts et al. 2012).
Agency is not neutral but political, strategic and possibly emotional (Long 2001). Agency is also relational, and the actors' plural social identities, circumstances and leadership can influence the impact of their agency and thereby the institutional outcomes (Balooni et al. 2010;Cleaver 2007). Furthermore, because of the actors' improvisation, the institutional outcomes are unpredictable and may have unexpected or unintended consequences (Balkin 1994). In our case, we examine the opportunities and constraints that the status of residence and ethnicity present for the exercise of agency.
The second factor, the networks, pertains to the actors' representations of their interactions with their society, particularly their social and ecological systems, according to Bodin and Prell (2011) and Scott and Carrington (2011). In our understanding, networks extend to the social and political structures that define the rules and the resources (objects and symbols) enabling the actors to construct politically relevant and socially acceptable repertoires of bricolage (Giddens 1984). Far from being homogenous, networks are made up of various and complex spheres of governance where actors meet and interact. The bricoleurs' actions are not limited within a given sphere; rather, they gather new and old elements from within and outside the existing arenas to realize their goal (Carstensen 2011). Thus, the bricoleurs from one social network may contemplate strategies from a neighbouring network or outside of the local sphere, blend them with their own norms and feel pressured to undertake similar changes. Consequently, Campbell (2005) argues that the newer elements are borrowed and incorporated into the bricolage process, the more it produces revolutionary change yielding institutions that can be different from the original. We examine how the actors made use of the networks to convey their ideas within and across villages regarding the forest management.
Power, the third factor, is inherent to the operation of agency and networks. We conceive power as the actors' conscious use of social structures and resources for diverse purposes including enforcing an agenda to reduce other parties' authority. Here, power is conceptualized as the product of actors' control over strategic material and symbolic resources. Knowledge and discourses shape the actors' worldview and are media for exercising power (Arts et al. 2012). From a critical perspective, uncovering how power is exercised beyond the formal institutional setting (such as the CAF bureaucracy), and how the actors experience it, is important to understand the power relations and their effects on institutions (Cleaver and De Koning 2015). We examine how actors with different access and control over material and symbolic resources exercise power to enforce their ideas and curve the change trajectory.

Data collection and analysis
As a qualitative study, we opted for a nonprobability and purposive sampling approach that recommends the selection of interviewees based on defined criteria relevant to the topic and a flexible number of interviewees (a minimum of 16) to reach data saturation of the most common themes (Guest, Bunce, and Johnson 2006). However, for multisite and comparative research, roughly 20 to 40 interviews were necessary to reach data saturation for all themes according to Hagaman and Wutich (2017). To ensure a maximum variation sampling, we recruited our informants according to their knowledge of the CAF, their participation in the program, their residence status (migrants/autochthons) and their gender. Thus, we interviewed 116 participants, of which 24 were in village V, 54 in C and 33 in L; five national level actors were also interviewed. In total, there were more autochthon respondents (59%) than migrants (41%) and more men (70%) than women (30%). Beyond the informants' selection criteria mentioned above, these differences in number are explained by the various social groups' overall awareness of the CAF system. Through semi-structured and indepth interviews conducted from June-September 2017, we discussed the institutional arrangements at the establishment of the CAF; the observable changes over time and the reasons that prompted those changes. We also carried out 15 Focus Group Discussions (FGD) with separate male, female and youth groups in our initial villages and in a fourth village, D, because it had been repeatedly mentioned by the informants as being first to start the forest conversion in defiance to the CAF system. Using a timeline exercise, we explored the key triggers of change that occurred, how the actors interpreted them and their implications for the forest management.
We transcribed the interviews and anonymized the respondents' and villages' names. To analyse the data we adopted a grounded theory perspective that enables the discovery of theoretical patterns from systematically collected empirical data (Glaser and Strauss 2017). Concretely, in Atlas.ti 8 software, we used content-based coding and created quotations known as segments of text highlighted for their relevance to the topic. From the quotations, we constructed codes; each code represented a driver of institutional change mentioned in the interview transcripts. Thus, we obtained 24 codes or drivers of change. We clustered these codes by meaning and obtained seven sub-categories and two overall categories (see section 4.2). The most common codes or drivers of change were generated and finally the discourses and narratives around these were analysed.

Study site and context: setting up the forest institutional arrangements
To address the limitations of centralised forest management, deforestation and fuelwood shortages in the cities, the government of Burkina Faso in collaboration with the FAO and the UNDP initiated the community forest management project in the 1980s. The villages V, C and L, our study sites, were among the pioneering sites of the project. They share the same socio-political and cultural background and thereby offer a unique opportunity to assess the patterns of institutional change.
Aligned on a dirt road, the three villages are 14 to 23 km away from each other. Regularly, the area receives migrants searching for agricultural land. Three major ethnic groups with hierarchical relations to natural resources are found: the Nuni constitute the autochthons. Guardians of the traditions, they have exclusive customary authority and control over forests and lands, including land distribution among migrants (the Moose and Fulani ethinc groups). Consequently, the Nuni rose as an indisputable financial and socio-political authority. However, the establishment of the CAF disrupted this authority by introduicing a new and formal benefit sharing system that moved the control over forest decisions from the autochthons headed by a customary chief to the project's newly created bureaucratic structures.
Meanwhile, farmers still cultivated the forested lands chosen by the project for the establishment of the community forest CAF. Promising prosperity to the whole village, the project was able to get from the autochthonous land rights owners, the targeted land of 29,515 ha belonging to 25 villages. Drought and hunger dominated the local context. Therefore, the concerned farmers received new farmlands and food donations. At that time, the farmers did not offer any resistance because agricultural land was perceived as abundant and they were optimistic about the project for cash income. The new forest area was portioned into 15 sections for a total of 15 years' operation, i.e., each section corresponded to one year of fuelwood collection. In this process, the project ignored customary territorial limits by merging different villages' lands into one forest section for fuelwood harvest. This created land and territorial disputes between the villages over time. These elements are key as the actors used them later to reinterpret their commitments to the CAF.
The CAF's bureaucracy has two components: first, the local people's representatives composed of the village forest users' groups called Groupement de Gestion Forestière (GGF). They conduct forest management activities, and fuelwood logging and sale. The GGFs form a Union called Union des Groupements de Gestion Forestière (UGGF). At the end of the project in 1994, the exclusive authority over the CAF was transferred to the state, and afterward to the UGGF as the local people's representative. Second, the technical team is hired by the UGGF to oversee the implementation of the forest management plan. Other subnational and national state actors monitor the effectiveness of the forest management instruments.
The financial plan of the CAF is based exclusively on fuelwood sales. The national price of 4 USD/m 3 of fuelwood in 1997 was adopted in the CAF and shared as follows: the logger earns 50%, the forest management fund receives 27%, the state treasury is allotted 14% and 9% goes to the village development fund. This pricing has remained unchanged and attempts by the forest managers to change it-the last of which was in 2015-failed due to opposition from influential private fuelwood merchants and the Ministry of Environment. In the local perception, this benefit scheme is seen as unfair. The legally and customarily entitled local leaders (i.e., the local government, traditional chief, local forest officer) are excluded from the benefit sharing and the formal forest management organs. This is critical because the excluded actors used it to gradually revoke their support to the CAF while mobilizing grassroot revolts against the current establishment.

Reconfirming the CAF institutions in village V
"Had we copied those who convert and sell the forest, we would have been the same. We must not follow others and destroy our own home". A development councillor in village V (D70:51, 14-07-2017) In village V, certain discourses and narratives about forest management have become dominant. These focused on land tenure issues at the creation of the CAF, weak forest management and the emergence of new and attractive land sale practices in the area. It all began when the project merged V village's lands with its' neighbours of villages L and D into one forest unit for fuelwood collection without considering their respective customary territorial boundaries. At that time, villagers did not object because they knew their respective territory and had convergent interests towards supporting the CAF. The land merging became a contentious issue only when L and D villages decided to unilaterally withdraw their lands from the CAF and turn it into agricultural lands or sell it to private agribusiness. In the process, they exceeded their original customary forest territory limits and grabbed lands belonging to V. The people of L justified their trespassing by a customary ownership that they claim over V village's lands. However, the oldest man in L refuted this argument and confirmed the current customary limits of V as those recognised since the colonial period.
To solve these land and territorial disputes, the leaders in V village sought the mediation of the CAF bureaucracy but they were unsuccessful as the CAF managers said that they could not interfere in solving customary land disputes. They also confessed that they brought the police, who arrested and fined the transgressors, nonetheless failed to stop the forest conversion. The farmers would simply pay the fines and return to cultivate inside the forest. The forest managers could not take any further actions, explaining that they only have management rights over the forest and do not hold a formal document to claim property rights. Although the forest is under the management of the UGGF, the land remains community-owned and legally under customary tenure regimes and practices. Therefore, from the forest managers' perspective, only the central government has enough authority to persuade the local people to uphold the CAF's regulations and stop the forest conversion and sale.
Nonetheless, V villagers felt the encroachment of their land as an injustice that has occurred because of the ineffective enforcement of the forest management rules. They argued that if the forest managers had done their job properly, none of the forest conversion, leading to land conflicts, would have happened. Furthermore, the inability of the CAF managers to solve the disputes reinforced this argument and was perceived as a defeat and a proof that the CAF's implementation has ended. This was reinforced by the argument of 15 years deadline that the village leaders had agreed with the project for the implementation of the CAF. This narrative became popular when villagers officially heard that the forest management plan had expired. They interpreted this information as the end of the CAF and a signal that they could also remove their forestlands from the CAF and dispose of it. Meanwhile, the fuelwood stock declined in the forest, hence, decreasing the village development funds and peoples' incentives to maintain the forest. Two opposing camps arose within the village urging the customary chief to take back control over the forestland to either continue or end the CAF.
The first group, led by a municipal councillor who was also an autochthon and supported by the Moose migrants, advocated for the end of the CAF. Appealing to people's emotions, they invited the villagers to retaliate and join the forest conversion. Concretely, they allowed migrants to cultivate inside the forest to block the advancement or invasion of the neighbouring villages. In doing so, this group wanted to observe caution, test the waters as to which extent their forest conversion strategy would hold weight in the public eye. In their logic, if those migrants were not expelled from the forest, it meant that the autochthons' allies could also create new farms there with the intention of normalizing the forest conversion and eventually, impose it as the stance of the whole village. Otherwise, the group would abstain for the time being and readjust their tactics. Their strategy did not stick as the forest officers expelled the migrants from the forest and arrested and fined them on the recommendation of the second group opposing the forest conversion. This defeat discouraged others from following this approach.
The second group opposing the forest conversion was also led by an autochthon, member of the village development council and supported by the forest officers. This group appealed to peoples' common sense and the economic hardship associated with an unplanned forest conversion. They acknowledged the shortcomings of the CAF's bureaucracy and the resulting frustrations. However, they reminded the people of the numerous benefits (ecological, social and financial) that had come to them thanks to the CAF. They promised that with peoples' trust and support more could be achieved and that they would urge the CAF managers to make important reforms. They also explained that the people should think of alternative ways to express their discontent than to unwisely convert the forest. They even suggested that the village could still remove its forestland from the formal CAF and keep it as forêt villageoise without the CAF label instead of converting it to other land uses.
The two opposing parties, thus exercised their agency and campaigned to ally the customary chief and the public opinion to their position. Through their social networks in and outside of the village, they exchanged lessons, pressure and smear campaigns for or against the forest conversion. For example, the village V neighbours told them that they would die poor if they do not also convert or sell their forest. The V villagers responded by quoting the good and unfortunate experiences of neighbouring villages already engaged in the practice. Likewise, private agribusiness seeking to buy land promised higher prices to convince the village leaders to remove their lands from the CAF and sell it to them. Such active use of social networks circulated specific opinions and aimed to secure the interests of those who would benefit from the forestland use change.
It also aimed at building local coalitions to make a position dominant while impeding alternative opinions.
Finally, the forest conversion did not stick in V village. The customary chief's and public opinion shifted towards keeping the forest standing. Therefore, everyone wanted to be on the winning side, look good publicly and not miss out on future benefits. Moreover, when asked, the male autochthons explained: "We rejected agribusiness offers because of our loyalty to the CAF. We are also afraid of breaking the law […] here in (V village), we love the forest. (FGD, 73:4, 13-07-2017)". The autochthonous women also elaborated that they chose to maintain the forest for its contributions to their livelihoods. The herders highlighted the value of the forest in providing grazing for their livestock. However, all the actors unanimously denounced the progressive occupation and conversion of the forest by their neighbours as these actions provoked shortages of forest products and led to conflict between farmers and herders. Therefore, the customary chief formally asked the CAF managers to return part of their forestland for farming. Beforehand, to prevent imminent bloodshed. the customary chief of V village invited higher customary authorities, the Mayor and other leaders, to settle their territory dispute. These local leaders confirmed and marked the current borders between V and L village. They also commanded the illegal settlers of V forest to leave. On site, the people of L also asked the litigators to investigate their contested borders with their own neighbours of D and other villages.

Village C: Challenging and bending the CAF institutions
"The people convert or sell the forestland out of frustration because one cannot allow the forest managers to reap the benefits for themselves exclusively. The peasant is not stupid; he observes, he sees, he knows". A forest department director (D67:11, 12-08-2017) In C village, unfair benefit sharing and accountability issues were central to the discourses and strategies developed by the actors to bend the forest institutions and enact change. Three key leaders' actions influenced a coordinated grassroot movement against the CAF: the customary chief, the Mayor and the local foresters. The following introduces each of these actors' arguments. First, the customary chief of C village denounced his loss of authority in the CAF's decision making. For example, he complained that he used to decide on the usages of the village development funds since the beginning of the CAF. His relatives were also board members in the CAF. However, in the 2000s, a governmental decree declared the forest managers sovereign in managing the forest funds. Henceforth, the chief no longer received information and reports regarding the CAF although as he argued, he was the one who gave lands to create the CAF. This marked the beginning of grassroot revolts against the CAF; hence, showing the prominence of rural micro-politics and the clash between the bureaucracy created to manage the forest and the pre-existing customary logics.
Second, the Mayor's claim for accountability and financial benefit is based on the decentralisation law of 2004 that recognised the rights of democratically elected local governments to manage their natural resources including the CAF. However, the CAF's current benefit sharing scheme as detailed in section 3.2 above, does not include the municipality. Therefore, the local government is said to have incited and encouraged revolts against the current CAF structure.
Third, the local forest officers evoked the lack of transparency and accountability in the forest management to distance themselves from the CAF. This is critical because these officers as state representatives hold significant authority in people's eyes. Any association with them could have strengthened the CAF managers' authority in the villages. Consequently, the local people headed by the loggers subscribed to the emerging social movement, arguing that they do not perceive the benefits from the CAF and that the CAF managers are mismanaging and capturing the forest funds at the expense of the population. As acts of defiance, the loggers engaged in overexploitation, illegal logging and "black market" trade. In response, the forest managers inflicted various penalties that added to the growing protests.
When asked about the various complaints, the CAF's director responded that the forest management contract that they have signed with the state does not require them to be accountable to local actors. Nonetheless, to appease the tensions, the forest managers initiated a new tax revenue for the municipality, an annual grant for the customary chief, and a radio broadcast of a 10-years forest management report. However, the local government found its tax revenues small. The customary chief reported that he has received the grant only once and the population does not trust the reports heard on the radio. Therefore, the autochthons from different neighbourhoods mobilized diverse strategies and tactics to put more pressure on the forest managers. On the one hand, people from a neighbourhood in the village attempted to overthrow the CAF's board by creating a new forest management board. However, they failed to obtain an official recognition. Therefore, they converted their original complaints over forest management and benefits sharing into land claims; thus, illustrating the ways in which the actors' strategies and goals can change rapidly depending on the context, lived experiences and the opportunities at hand.
Learning from these shortcomings, actors from another neighbourhood pieced together rather disparate elements including (i) land tenure issues at the creation of the CAF, (ii) the country's popular uprising context and (iii) population growth to articulate and sustain their claims. First, they (i) argued that in 1989 the project incorporated their lands into the CAF area without the customary chief's consent. However, as the youth explained, "The elders of that neighbourhood noticed that their lands were originally spared. Without informing the customary chief, they directly asked the CAF project's leaders to also include some of their lands within the CAF because they expected to get rich through the CAF" (FGD, 5:25, 12-07-2017). Decades later, they were disappointed because they had not joined the logger's groups nor did they receive the anticipated benefits. That is when the (ii) country's popular uprising context played out. When local governments were discarded in 2015 in Burkina Faso, the prefect appointed in the village belonged to the neighbourhood of those re-claiming their forestlands. This leader is reported to have actively supported the revolt arguing that plus rien ne sera comme avant [nothing will be like before] and that people should intensify the contestations and get results before the normalization of the political context. These show how actors can actively use their social networks in and outside the village to learn from others' failure, and take advantage of national political context.
Lastly, to be more effective in their land claims, the actors combined the (iii) population growth argument with the lack of transparency in forest management and the so called 15 years period that they had agreed for the CAF's implementation. Through the Mayor and customary chief's mediation, the forest managers and the Ministry of Environment were compelled to remove the contested 1,500 ha from the forest for the people of that neighbourhood. In contrast to the case in the previous V village, here the calls for change came from local leaders before trickling down to the people. In addition, the claims over forestland were formalized, coordinated and resulted in an official removal of forestland for the claimants.

Village L: Reinterpreting and rejecting the CAF institutions "If a wooden plate cracks and it is not mended, it breaks into pieces so it becomes difficult to fix and
can only be thrown away. The forest managers have let the plate break completely and now the pieces cannot be put back together again". Acting customary chief (D61:11, 21-07-2017) In village L, a perceived ineffective enforcement of forest management rules and regulations has led to local peoples' discontent with the forest managers, the readjustment, reinterpretation and eventually the rejection of the CAF. As in the first village V, it all began when villages D and K, two neighbouring villages also involved in the CAF, trespassed their original customary forest borders to convert forestlands belonging to L. When asked, the male autochthons in D village justified their actions by the irregular forest maintenance activities that caused vegetation to cover up the natural forest boundaries and confused them about their customary land limits. They also argued that the mismanagement of their village development funds had hampered their relationships with the forest managers and led them to convert the forest in retaliation. The people in L rejected these explanations, arguing that the D villagers seemingly ignored their own customary land limits because they wanted to take advantage of the weak forest management to expand their territory and grap financial benefits. As in V, the village chief in L failed to get the CAF managers' mediation to solve the land dispute. Meanwhile, fuelwood resources declined in the forest along with the village development funds and people's overall incentives to keep the forest standing. In such a volatile context, Burkinabè citizens (not the autochthons' relatives) fleeing the political crisis in Ivory Coast continued to head toward the village seeking farmlands. The autochthons gave them farmlands inside the CAF as a strategy to see what would happen if the village also began the forest conversion. These autochthons put forward various arguments to justify their decision: They evoked first a state's call for solidarity with the migrants and second, the end of the CAF based on the argument of the 15 years deadline agreed for the implementation of the CAF. Third, they promised that no other farmers would be given land inside the CAF and that those that were already there would not extend their current farmlands. With no follow up, the autochthons continued to not only lend lands to more people inside the forest, but they also literally began selling those lands. This is how agribusiness (rich and influential city dwellers seeking to buy rural lands) entered the scene.
These actors forged networks through which they offered cash and promised to invest in the village. The autochthons saw in this an opportunity to express their discontent against the CAF managers and get rid of the CAF that had become abhorrent to them. However, they needed to align the public opinion. Therefore, they constructed rationales using the land dispute argument and blamed unsatisfactory forest management to justify their actions. The male autochthons explained: The forest officers, the director and the CAF managers all know what pushed us also to convert the forest. We are not bad people. If they had listened to us and stopped the forest conversion and sale [by the neighbouring villages] from the beginning, none of this would have happened. They abandoned us instead. Therefore, when we get someone seeking to buy land, we cut the forest and sell it. They pushed us into this position. (FGD, 51:15, 05-07-2017) Although motivated by frustrations, the autochthons engaged in the forest conversion and land sale strategically as they knew that selling the forestland was forbidden and that if not the CAF managers, the government could sue them and re-establish the forest. However, if they could gain money in the land transactions, the agribusinesses would be the new land owners and therefore, the ones to respond to the government if any complaints were launched in the future. What these autochthons did not anticipate is that they were pursuing immediate financial gratifications while depriving their own people and the next generations of farmlands. Hence, voices raised to oppose the autochthons' decisions to convert and sell the forest from the village's oldest resident (seeking to preserve the village lands as per the customs and for future generations). He was supported by women and the Fulani migrant herders (interested in protecting the trees for their livelihoods' values).
In these processes, the exercise of power based on social characteristics including the status of residence (autochthons versus migrants), social status and gender were prominent to determine whose voice was heard or not. Women's opinions were not considered as this autochthon woman affirmed: "The men argued that they were the owners of the forest and the village and that they owed us women no account" (D55:1, 06-07-2017). Likewise, the Moose migrants failed to make their opposition heard, arguing that because of their status of migrant they had no say in land matters and that they could be expelled from the village if they confronted the autochthons. For example, a migrant Fulani herder confirmed that the autochthons rebuked them, arguing that: "The same way you have the right to sell your cattle, we also have the right to sell our lands and forests" (D102:7, 06-07-2017). Consequently, the forest in L village was turned exclusively into private farmlands, which resulted in a scarcity of forest products and other ecosystem services specifically for women and herders. Finally, the male autochthons concluded: Since the forest is over, the CAF people have nothing to do here anymore. They know that the forest has been converted into farmlands. Nowadays whoever wants to sell fuelwood does as he pleases. There is no ticket, no village development funds, or forest management group to monitor who is logging fresh wood. We have no more contact with the CAF managers. Only the foresters come here often. Since they do not want to listen to us, we do not listen to what they tell us either. (FGD, 51:23, 05-07-2017) The CAF's technical director confirmed that the village L had sold its entire forest and that the customary chief said that they do not want to stay in the CAF anymore. Thus, the position made dominant in the village was the forest conversion and sale and accordingly the termination of the CAF. It is worth mentioning that the number 15 evoked in all the villages corresponded first to the number of forest plots, and second to a full rotating cycle of the CAF, after which the forest managers should revise the forest management plan and consult each village on their commitment to continue the CAF. Such consultation did not happen, therefore, people have interpreted the 15 years as the deadline that they agreed for the implementation of the CAF.

The drivers of institutional change
The CAF did not work as intended. The drivers of change identified by the informants can be classified into two larger categories: internal and external to the CAF. The internal drivers pertained to unsatisfactory forest management that nurtured actors' discontent and pushed them to pursue change. Amplifying these effects, informants pointed to external drivers including the commodification of land and population growth that have increased land pressure and local people's incentives to change the forestland uses. Population dynamic was frequently mentioned in the context of migration movements whereas the first driver referred to the emerging land markets and higher demand for land beyond local population and land use dynamics. Agribusiness actors were identified as main agents here, because of their financial and authoritative powers to to demand (and pay) for land ( Table 1). While the drivers differed in their frequency of mention, they were identical in all villages except two that were only mentioned once in village C: the national political context giving strength to popular contestations and the recent rural land law that pushed for a land commodification agenda. Moreover, the top three drivers of change-unclear forestland boundaries, weak management and enforcement and commodification of land-were the same in villages V and L yet the change trajectory was different as will be argued in the next section.

Internal/external categories of drivers for change Drivers Effects
Internal factors to the CAF:

Unclear forestland boundaries
At the creation of the CAF, the project's failure to recognize customary territory limits led to land conflicts among the villages. The CAF managers' inability to solve those disputes generated frustrations and led villagers to question the effectiveness of the CAF's institutions; hence, they took action to bring about change.

Weak management & enforcement
The ineffective surveillance of the forest, communication with the people and enforcement of the CAF rules and regulations (punishing offenses) has led to mistrust between forest managers and the people and eventually people' discouragement and disengagement.

Unattractive fuelwood market
The irregular silvicultural activities contributed to the decline of fuelwood stock inside the forest (the main financial source for the CAF) and the low wood price decreased local incomes and incentives and the overall attractiveness of the CAF.

Unfair benefit sharing
The exclusion from the forest management and benefit sharing schemes of actors such as the customary chiefs (custodians of the forestland) and the mayor nurtured revolts against the current system. Forest managers were perceived as the main beneficiaries of the CAF at the expense of the people.
Contested accountability The CAF managers' reluctance to be accountable to local leaders for the forest management created mistrust and led the people to question their authority and their ability to represent the collective interests.
External factors to the CAF:

Land pressure
Commodification of land People's frustrations and disappointments with the CAF motivated them to seek alternative usages of the forestland. The increase in land sales although culturally forbidden was perceived as a more attractive option for them.

Population growth
Population growth in addition to internal and external migration increased the need for more farmland; hence, pressure on the forest was intensified.

Unpacking the drivers of institutional change
The three villages share similar socio-political and cultural contexts. They also experienced the same procedures for the creation and implementation of the CAF, yet they displayed three different institutional change trajectories and outcomes for the forest under management (Figure 1). Among others, land tenure issues, poor design and implementation of the CAF project, ineffective forest management and unfair benefit sharing decisively drove the institutional change process. These drivers were identical in all three cases; however, they produced different pathways of change because each village perceived and dealt with them in its own way. Thus, highly diverse processes were at play with the actors in the different villages constructing and exchanging their perceptions of the same events through various social networks. They also expressed their agency through diverse strategic and interest-oriented actions and mobilized power and resources to influence other parties; hence, they crafted the change trajectory. On one extreme, in village V the actors acting on their frustrations and discontent with the overall CAF management, argued among themselves to (dis)continue their commitments to the CAF. Finally, they converged towards maintaining the forest standing and reconfirming the CAF. On the other extreme, in village L the actors also revolted against the perceived unsatisfactory forest institutions. However to retaliate, they choose to convert and sell their forest to private actors including the agribusiness. In between, in village C, the actors changed their failed attempt to discard the bureaucracy of the CAF into claims over forestland. Furthermore, they activated their networks and constructed narratives to argue that they would not have challenged and bent the forest institutions if the benefits were shared equitably. Consequently, we can infer that the determinants of the differentiated trajectories of change lie within the workings of the networks, agency and power. Despite its core idea of the unpredictability of institutional outcomes, critical institutionalism scholarship can still inform policy and the governability of the commons. A promising avenue would be to discover and act upon the conditions that cause changes in the rules in use instead of attempting to change the actors' behaviours themselves. Indeed, our case showed how impossible it was to anticipate or control people's strategies and actions because of the complex web of logic, motivations and meanings at play. In contrast, the forces driving these behaviour changes such as the ineffective forest management and land tenure issues were relatively straightforward and manageable.

Drivers of change and bricolage
At the establishment of the CAF, the pre-existing institutional actors "switched hats" to allow the operation of the CAF's formal structures. Eventually, ineffective forest institutions including weak management and enforcement, land tenure issues and perceived unfair benefit sharing provoked local actors' discontent with the CAF while reducing their incentives to keep the trees standing. In addition, land pressure due  to both population growth and increasing commodification of land in the area further drove the actors to make changes. In this context, they formed various coalitions with distinct priorities to either sustain or challenge the newly formed bureaucratic institutions. These findings are supported by the research of Weiland and Dedeurwaerdere (2010) as well as Carstensen (2011), who also argued that actors are likely to challenge and delegitimize current institutions and contrive new arrangements to replace them if the current ones prove to be ineffective. In this process, the bricoleurs of institutions made use of a range of suitable arguments to achieve their goals. The success of the actors in village C to reclaim their land from the CAF reinforces Cleaver's (2002) argument that, pressure from pre-existing institutions is needed to force bureaucratic institutions, such as the CAF in our case, to adapt to people's needs and interests. Consequently, in C village, the customary institutions rose again and mixed with bureaucratic rules. In contrast, in village L, a new type of institutional setting emerged that is neither entirely customary nor fully bureaucratic. It can be described as a sorte of "fine mess" of varied governance islands (Ingram, Ros-Tonen, and Dietz 2015). However, the likelihood of the new induced institutions to be more efficient depends on the convergence of those institutions with the broader society's interests (Agrawal 2005). In our case, the actors were heterogeneous and their interests at times contradictory.

How agency played out in the process of institutional bricolage
The operation of agency enabled the actors to give specific meanings to selected practices. Based on those meanings, they constructed preferred narratives to justify their actions. In line with Cleaver (2002) and Long (2001), in our study we observed that the public was guided by emotions of revolt, anger, frustration and disappointment. Moreover, the specific context in each village and its local leadership shaped the spaces for exercising and legitimizing agency. While the CAF was officially governed by bureaucratic institutions in all three villages, the strong agency of the autochthonous land rights owners enabled them to impose their preferred strategies (i.e., challenging the CAF in C, leaving it in L, and reconfirming it in V). Likewise, several other authors (Agrawal and Chhartre 2006;Gutu, Wong and Kinati 2014) have also argued that in a setting of consensual institutions, the exercise and outcomes of agency can be hijacked by social characteristics and captured by local elites. Our study reinforces these findings by further demonstrating that the local elites, in our case, the group of the autochthons, are not a homogenous set. Neither do they always exercise the same agency, nor do they necessarily follow the same interests.

How networks played out in the bricolage process
Acting upon the drivers, the social networks enabled the actors to draw up a menu of potential institutional solutions and exchange them. For instance, the forest conversion started in one village and then spread gradually to the other ones. Similarly, land sale that used to be forbidden by customary law, occured first in one village and then in neighbouring ones, until it was gradually normalized. We also observed the role of structural factors such as ethnicity, status of residence and the corresponding interests in influencing support or rejection of a certain strategy across all villages. Aligning to these results, Giddens (1984) also showed that social structures are dynamic and circulate opinions and norms that shape the actors' representations and behaviours; conversely, these are shaped by the actors' collective meaning making. Our results also confirm the different institutional bricolage practices defined and described by de Koning (2014) and Faggin and Behagel (2018): We found alteration in village L, aggregation in C and articulation in V. Further, these nuance the idea of systematic institutional path-dependencies (Heinmiller 2009) and the argument that being exposed to new elements is a crucial determinant of the magnitude of institutional change (Campbell 2005; Carstensen 2011).

How power played out in the bricolage process
In all three villages, the actors exercised power through control over the forest, land and social structures. Identifying who had this power and how it was used helped to explain how and the reasons why each village produced a certain outcome instead of another. The actors who could exercise more authority over resources could shape the direction of the bricolage process. In all three villages, the autochthons had more power than the migrants and women who could not manage to achieve their interests. These findings confirm the observations of Gutu, Wong and Kinati (2014), who showed, in a case from Ethiopia, that more authoritative actors curbed the bricolage outcomes. Likewise, reinforcing the results of Funder and Marani (2015), our study also illustrated that the bricoleurs face diverse constraints that limit their actions. Finally, our analysis supports the findings of Balkin (1994), Cleaver (2007) and de Koning (2014) who argued that bricolage as a political process can yield unanticipated results. As we have seen, in village L even the autochthons who converted and sold the forest land were facing forest products and land scarcity. Nevertheless, the cases of C and V villages illustrated that institutional bricolage processes can achieve the intended goal.

Conclusion
In this paper, we used the perspective of critical institutionalism and the concept of institutional bricolage to examine why and how institutions of the commons change over time. We focused on three neighbouring villages in Burkina Faso that shared similar conditions, including the introduction of a large forest management project, yet displayed diverse trajectories of institutional change with different outcomes for the forest and people. The results showed that the community forest management project created new bureaucracies to manage the forest to replace the previous customs-based systems. Poor design of bureaucratic institutions coupled with ineffective forest management, land tenure issues and land pressure led to the actors' discontent and their engagement in action for change. Although these challenges were identical in all cases, the three villages perceived and (re)interpreted them differently. Opposing factions and coalitions (for or against the current institutional establishment) emerged within and among the villages to push for their agendas and interests. The social networks, agency and power mediated these processes and determined the institutional change pathways, hence explaining why these differentiated changes trajectories occurred. The networks conveyed the manufactured strategies for the actors to contemplate while the exercise of agency and power sieved those repertoires to make them suitable to the actors' contexts and eventually curbed the direction of the change. Consequently, village L actors rejected the CAF's institutions. In C, they challenged and bent those institutions to their interests, while in village V they reconfirmed them. Each change trajectory had intended and unintended implications. These findings reiterate the necessity to rethink local actors' role in driving institutional change as well as the opportunities for policy makers to induce change for sustainable resources management.
These findings have implications for future research and for any forest conservation effort. First, it is crucial to design research and development projects in a way that is less linear so that disrupted processes of change are anticipated; an analysis of agency, networks and power is essential. In our case, applying the concept of institutional bricolage enabled us to capture and scrutinize both details and the grand scheme of the institutional dynamic and outcomes. Second, when possible, we encourage researchers to take a comparative perspective, which allowed us to unpack the processes of bricolage as a response from below and to investigate the construction of narratives, their translation into institutional practice and the actual outcomes of such processes. The findings demonstrate the importance for any forest conservation effort to ensure effective and satisfactory forest management responsive to both bureaucracies and socially embedded practices.