How Endogenous Cultural Institutions May (Not) Shape Farmers’ Climate Adaptation Practices: Learning from Rural Cameroon

Abstract Climate change is high on the science and policy agenda of sub-Saharan Africa (SSA), requiring context-specific adaptation. Studies on endogenous cultural institutions (ECIs) in climate adaptation are fragmented, impeding policy orientation. This suggests the need for complementary evidence on ECIs to inform adaptation practice. Through a sample of 158 households in rural Cameroon, we contribute to bridge this gap, by: (1) exploring socio-political factors shaping peasant farmers’ compliance with agro-based ECIs, and (2) estimating the effect of ECI compliance on farmers’ adaptation practices. Our estimation revealed the following: While political factors do not sufficiently explain compliance, state influence, mirrored through the application of formal rules negatively affects ECI compliance. Traditional rain forecast significantly explains climate adaptation; however, the pouring of libation does not. The age of peasant farmers also positively affects compliance with ECIs. We conclude that compliance with ECIs manifest in climate adaptation in varying proportions.


Introduction
Dealing with climate change effects is perhaps one of the most daunting challenges to sustainable development today. This holds especially true for the agricultural sector, which is earmarked as one of the most significantly affected (Borras 2009;Wood et al. 2021). Estimates show that rising food insecurity linked to climate change will have to be addressed through adaptation practices at all scaleswith an estimated 60 percent increase in agricultural production required by 2050 to feed the growing human population (FAO 2009, Campbell et al. 2016). Yet, 22 percent of cultivable land across the globe is projected to experience negative impacts from climate change. Standing at 56 percent, this figure is even higher for Sub-Saharan Africa (SSA) (Campbell et al. 2016) where more than 40 percent of the population fall below the poverty line. 1 Addressing rural poverty for farmers therefore remains a highly relevant objective in SSA, given its negative short, medium, and long-term welfare effects (Clay and King 2019;Nchu et al. 2019;Kimengsi et al. 2020).
It is plausible to indicate that agro-based adaptation to climate change remains high on the science and policy agenda of SSA as more than 60 percent of SSAs' population relies on agriculture. Agriculture in the context of SSA is predominantly rainfed. Rainfed agriculture is considered to be highly sensitive to climate change as compared to non-rainfed agriculture (Kurukulasuriya et al. 2006;Vanschoenwinkel and Van Passel 2018). However, even within non-rainfed agricultural systems, marginal changes in climatic conditions with respect to temperature and precipitation have heterogenous impacts all over the world. This is explained by the variations in the levels of technology, regional differences, and crop choice (Vanschoenwinkel et al. 2016). Consequently, climate response approaches vary between different non-rainfed agriculture systems but, more importantly, they are very different for rainfed agricultural systems such as in SSA and non-rainfed systems (Vanschoenwinkel and Van Passel 2018).
Although concurrently several climate adaptation interventions are applied in diverse settings in SSA, they have been criticized for shifting the burden of responsibility and the costs to the poora majority of whom are farmers (Ribot 2014). For instance, interventions to promote sustainable agriculture places the burden on farmers to implement and produce results, despite their limited resources and capacities . This mainly results from the fact that most climate adaptation practices (CAPs) are exogenously rooted. 2 Although some of them strive to align with endogenously shaped adaptation systems, a significant proportion of these interventions are still in contrast with Endogenous Cultural Institutions (ECIs) 3 (Whitfield 2015; Kimengsi and Balgah 2015). For instance, in recent attempts to shape adaptation practices in the agricultural sector, the climate-smart agriculture (CSA) framework of the World Bank (exogenously rooted), has been used in several contexts (World Bank 2005;Collins 2018). However, these interventions significantly give rise to maladaptive practices by inadvertently reinforcing, redistributing or creating new sources of vulnerability. For instance, the adoption of crop substitution techniques with regards to places resistant to pest and hazards proved unsuccessful in many instances (Eriksen et al. 2021). This suggests the need for adaptation to draw endogenous capacities and aspirations for selfdetermination and cultural continuity (Johnson and Parsons 2021).
Endogenously shaped adaptation systems are of particular relevance for SSA because of the diversity in the continent's agro-ecological systems, which demand context-specific local adaptation practices. Furthermore, as the most ethnically diverse region in the world, SSA inherently hosts an abundance of endogenous cultural institutions (ECIs)although their potential role in shaping local adaptation practices of farmers is still to be sufficiently uncovered. Several studies (largely qualitative) on institutions report that ECIs continually countervail exogenous ones (Yami, Vogl, and Hauser 2009;Ribot 2005); this represents a yet-to-be resolved conundrum in agriculture and natural resource management in SSA (Ntuli et al. 2022). Complementary quantitative evidence provides a good opportunity to enhance generalizations in the spectrum of climate and institutions. Even recent studies on climate adaptation in Cameroon draw more from qualitative evidence (Azong 2021;Chia, Kankeu, and Hubert 2019). The foregoing assertion suggests that there is still limited quantitative empirical evidence on the role of ECIs in shaping CAPs for peasant populations (Ford et al. 2015;Feola et al. 2015;Clay). This concerns SSA in general and Cameroon in particular. Additionally, due to the latent nature of ECIs, very few studies have approached the subject using a quantitative approach (Acemoglu and Johnson 2005). This paper advances the following central topic: While ECIs are relatively intact in the Western Highlands, when compared to other parts of Cameroon Kimengsi, Abam, and Forje 2021) there is lack of clarity on their role in shaping agrobased climate adaptation. This knowledge is needed to enhance comprehension on ECIs and further inform climate adaptation practices. To clarify this, this paper seeks to: (1) explore the household socio-economic attributes that determine peasant farmers' compliance with agro-based ECIs (2) investigate the socio-political elements that shape compliance with ECIs, and (3) estimate the relationship between ECI parameters and farmers' adaptation choices. The results will potentially contribute to bridge knowledge gaps on this subject, and will equally support the design of more inclusive, pro-poor and efficient institutional arrangements for climate adaptation for peasant farmers.

Analytical Framework on Endogenous Cultural Institutions and Their Determinants
Endogenous institutions refer to structures, rules and norms, which emanate from the community itself. Exogenous institutions relate mainly to the formal rules emanating from the state (e.g. national laws) or international processes (e.g. international regulations and conventions) which are codified through law, policies and regulations (Osei-Tutu, Pregernig, and Pokorny 2015;. They are generally introduced from outside the community. It is sufficiently documented that exogenous institutions predominate natural resource management settings in SSA (Ntuli et al. 2022). For instance, farmers in rural Ghana significantly made use of formal interventions in climate adaptation over informal ones (Abass, Mensah, and Fosu-Mensah 2018), while in Ethiopia, studies show that both ECIs and exogenous institutions have not addressed the deeper systemic problems of agriculture and food security. Additionally, farmers and environmental resource users in Mozambique are constrained by formal institutions (Ntuli et al. 2022).
ECIs compliance varies across communities. In our study, the context-specific parameters include norms (traditional rain forecasts, and the traditional launch of planting seasons), customs (pouring of libation before planting and adherence to spiritual guidance in farming), values (respect for spiritual days) and taboos (encroachment into spiritual sites) (see details in Table 1 and Figure 1). Traditional rain forecast as an ECI guides decision-making in cropping and livestock raising in East Africa (Ethiopia, Tanzania and Uganda), and demonstrates both beneficial and detrimental outcomes in agro-based climate adaptation (Radeny et al. 2019). In Botswana, farmers adhere to the Letsemathe practice of waiting for the traditional ruler to grant permission to begin planting or harvesting (Selato 2017;Davies et al. 2019). In the context of our study area, traditional rain forecast guides decision-making as farmers draw from a range of  forecast norms to signal the onset of rainfall. This ECI informs the traditional launching of the planting season to guarantee a good farming season. The flipside of such a practice is that, even if signals of planting are evident, farmers have to wait for the pronouncement of the traditional ruler. Traditional values such as the respect for spiritual days and taboos (non-encroachment into sacred sites) prohibits planting, harvesting and other natural resource management activities on certain days and places within communities (Schwartz 1992;).

Study Area
Cameroon represents a classical setting in SSA to analyze the dynamics around institutions. Its vast agro-ecological diversity gives a semblance of the African agricultural spectrum. Cameroon has a diverse cultural settingwith over 250 cultural groups (Yenshu 2011). Attempts to build resilience for rural farmers in the face of climate change and rising poverty culminated in the introduction of several exogenous agrobased schemes (Balgah 2018;Muluh, Kimengsi, and Azibo 2019;Kimengsi et al. 2020), leading to the implementation of several agribusiness projects as adaptation choices (Kimengsi et al. 2020). The Western Highlands of Cameroon ( Figure 2) is one of Cameroon's four agro-ecological zones.
It is one of the most densely populated areas in Cameron (128.5 inhabitants per km 2 ). This region has several household and community-based shrines, which are used to perform traditional rites as well as to pour libation. The co-existence of traditional and religious practices makes this area a good landscape to understand the dynamics around ECIs. For the past five years, the landscape has received significant waves of internal migrants fleeing from the conflict in the North West Region of Cameroon (Bang and Balgah 2022). This new wave of migrants are significantly engaged in farming activities. The south eastern part of the Western Highland hosts the Bamilekes who are mainly engaged in crop production for daily subsistence, with a little portion being sold at the local markets to fetch some income for the household. Typical farm systems here are characterized by agroforestry plots (cropping and the growing of fruit trees), and mixed cropping. The major crops grown in this area include Irish potatoes, beans, corn, tomatoes, soya beans, vegetables, cocoyams, pumpkins, groundnuts and cassava. The region is equally chosen for the study because, it still shows some relative stability in terms of the nature of its ECIs when compared to other parts of Cameroon (Lambi and Ndenecho 2009;Jiotsa, Okia, and Yambene 2015). Each community has a traditional ruler who ascends the throne based on inheritance. The traditional ruler makes decisions, in consultation with village elders. In terms of land access, while all lands belong to the state, traditional rulers oversee the governing of lands. Furthermore, family land ownership is predominant.

Data Collection
This paper is a product of an on-going mixed methods collaborative research on institutions and natural resource management in rural Cameroon. We collected data for this study between March and May 2019. A key informant interview guide (6 items) was developed and used to identify the key ECIs and CAPs in the study sites. The key informant interview (KII) guide emphasized on questions linked to the respective farming practices, knowledge regarding climate adaptation, the different ECIs (beliefs, norms, taboos) which have determined farming activities in the context of climate change. In this regard, we interviewed 8 key informants; 3 notables, 2 women, 2 youths and 1 NGO leader. The interviews lasted between 35 and 50 minutes and informed the design of the semi-structured questionnaire (30 items). 4 The semi-structured questionnaire took into consideration compliance with endogenous cultural institutions, i.e. norms, customs, values and taboos, the socio-political determinants of institutional compliance and peasant farmers' climate change adaptation choices. Household socio-economic characteristics, i.e. age, sex, income, household size, number of working household members, education of household head, and access to climate related training, were equally considered. The instruments (key informant interview guide and semi-structured questionnaire) used were translated into French language prior to data collection.
Furthermore, three enumerators were recruited and trained, to support the data collection. The questionnaire was pretested (N ¼ 10) with farmers in a similar locality, and further refined to capture relevant CAPs. Five villages which are described as agricultural hot-beds in the Western Highlands were purposively selected for the study (Table  2). With their high agricultural production potential, characterized by the growing of short and long cycle crops, they mirror the realities of farmers in the region. Furthermore, each of these villages exhibits unique cultural traits with regards to their traditional belief systems, the belief in the pouring of libation to enhance farm performance, the respect for spiritual days (e.g. contri Sundays) and their strong attachment to traditional institutions. These provide a comprehensive picture of the endogenous cultural institutional landscape of the Western Highlands. By farming households, we refer to human groups, which share the same roof and resources (Mphande 2016) and are engaged in a range of farming activities involving mixed-cropping, agroforestry and mono-cropping. These household members may equally be engaged in a range of non-farm and off-farm activities.
The target villages are linear in nature, with the assistance of a village members, we proceeded to assign numbers to households during a quasi-village transect walk. From the five villages, we chose a random sample of farming household representatives (male and female-headed households). After counting the total number of households, we proceeded to repeatedly conduct raffle draws to randomly select household numbers. Guided by the selected household numbers, we located the exact household and administered our semi-structured questionnaire. In total, we collected data from 158 representative farming households. A more than 5 percent sampling fraction was assured for each target village. Following Angelsen et al. (2011), a smaller sample size could be used especially when the study population exhibits homogeneity. In our case, homogeneity is exhibited on the basis of the traditional structures (e.g. all villages have a traditional ruler) and on the basis of the climate adaptation practices.
Despite having conducted 158 interviews in total, a significant number of those was incomplete and did not allow for the inclusion into the regressionstherefore 105 of the questionnaires were retained for the analysis. 5 The large number of missing variables can easily be explained by the fact that cultural issues are quite sensitive, and people sometimes did not feel comfortable responding. For example, some respondents were not comfortable answering questions linked to the pouring of libation and other specific ritual and traditional practices. They therefore skipped these questions. Equally, questions linked to political affiliations raised skepticism. Regarding ethical consideration, the free, prior informed consent approach was employed; the consent of all respondents was sought before proceeding with the interviews. The research team explained the motive behind the research and sought the approval of every respondent. In cases where respondents were hesitant to provide further details, the research team did not insist. Some respondents raised skepticism when asked about their specific farming practices (e.g. bush burning) and its effect on the environment. This might have been the determining factor to some non-responses.

Data Description
Our data (Table 1) consists of a set of household socio-economic variables (e.g. age, gender, number of household members, farm income). These attributes explain compliance with ECIs on the one hand, and the agro-based CAPs on the other hand. Besides household level determinants, community-specific factors which are captured as political (e.g. dominance of state rules, political leanings and elite phenomenon) and social (education, migration and modernization) determinants were used to explain compliance with ECIs. Furthermore, the compliance with ECIs considered norms (e.g. traditional rain forecasts), customs (e.g. pouring of libation), values (e.g. respect for spiritual days), taboos (e.g. encroachment into spiritual sites). The responses and the resulting indicators are binary in nature (e.g. one if there is compliance, and zero otherwise). Finally, CAPs 6 were framed under adjustment of dates (e.g. alteration of planting dates), and farm-based practices (e.g. expansion of farms, local irrigation).
Traditional rain forecasts represent the process of peasant farmers relying on certain traditional signs or from diviners 7 to determine whether it will rain or not, to enable to them cultivate their crops. Closely followed by this is the traditional launch of planting seasons. Usually, before the start of the farming season, some rituals (e.g. the sprinkling of salt and palm oil) are performed and the village or family heads traditionally opens the farming season. This organized approach is to ensure that people are well informed and prepared to avoid losses linked to weather failures. To enhance crop yields and to avoid the effect of pests and diseases, village or family heads usually pour libation to invite the "gods" to visit and bless the farms with good harvest. During the farming periods, family heads usually visit some native doctors to know the mind of the "gods" regarding the state of their farms. This practice relates to the adherence to spiritual guidance. Certain days are reserved for the gods to visit and bless the farms. If they respect spiritual days, peasant farmers are not supposed to work or to visit the farms. While farming is allowed, certain spiritual sites (reserved for sacrificesalso known as sacred sites) are prohibited from encroachment. 8 This practice is captured by the variable nonencroachment into spiritual areas. In principle, farming households are expected to adhere to these ECIs; however, adherence to specific ECIs varies.
The first set of explanatory variables is related to the socio-economic attributes of household representatives. Age is as a continuous variable that captures the age of the respondent. With respect to adhering to and upholding ECIs, older community members are generally more involved than the younger generation as younger members are witnessing growing disengagement with endogenous cultural institutions (Haller 2001). This has also been established for the Western Highlands ). Consequently, we expect older respondents to adhere to ECIs to a greater degree than younger ones. Male is a binary variable that equals one for male and zero otherwise. Additionally, we have collected a number of continuous variables. Household members denotes the number of members in the household. Working members denotes the number of working members in the household. Farm income is the monthly income 9 (measured in FCFA) that respondents receive from their farming activities. Schooling represents the number of years that the respondent attended school. 10 The correlation between the ECI and the socio-economic attributes is shown in Supplementary  Figure 1.
Socio-political forces were considered as crucial determinants of compliance with ECIs. Therefore, we sub-categorized determinants of compliance into political and social determinants. Political indicators include political leanings of respondent, state dominance, elite phenomenon. In terms of categorization, political leaning is assigned the value one, to denote that the respondent belongs to a political party, and zero otherwise. For state dominance, the value one was assigned when respondents noted the strong influence of the state in climate adaptation interventions, and zero otherwise. Furthermore, for the elite phenomenon, which is defined as a situation in which elites do not feel committed to respect traditions, due to their socio-economic and political strength, the value one was assigned if they reported the elite phenomenon, and zero otherwise. For social determinants, communication/social media, modernization, education and migration were considered. In terms of communication/social media, the value one is assigned if the respondent blamed some of the changes in the respect for cultural values to the strong influence of communication/social media channels, and zero otherwise. Furthermore, modernization, linked to the adoption of new and seemingly "modern" lifestyles were attributed one, and zero otherwise. Migration is one, if the respondent blamed changes on migration and zero otherwise. Education as a social factor is a binary variable, denoting one if the respondents noted education as a factor of change in ECIs and zero otherwise.
Finally, our second dependent variable refers to CAPs. There exist various CAPs but there is consensus on the need to appreciate what is actually practiced in different agroecological zones. In the context of this study, the main category "CAPs" is further subcategorized into three sub-components; the first component includes mixed cropping and expansion of farms. The second sub-component involves the alteration of planting dates and new harvesting dates, while the third sub-component includes local irrigation and local pest control. The ECI variables were not significantly interlinked when compared with CAPs. For instance, rain forecast as an ECI and the pouring of libation are not interlinked. However, CAPs such as mixed farming and the expansion of farmlands are compatible and could easily be combined for a PCA. This also applies to the alternation of planting and harvesting dates. These parameters and their indicators were also binary in nature (e.g., one if the respondents altered planting dates, and zero otherwise). Their correlation with our socio-economic household attributes can be seen in Supplementary Figure 2.

Empirical Model
The objective of our empirical analysis is twofold. First, we explore individual respondent and household socio-economic attributes and general social and-political elements that determine peasant farmer's compliance with agro-based endogenous cultural institutions. Second, we estimate the relationship between ECI parameters and farmers' adaptation choices. The first model is presented as follows: where ECI is a binary variable that equals one if the farmer i complies with a particular ECI and is zero otherwise. HHSEP is a vector that captures several household socio-economic attributes. To capture social and political determinants that might affect compliance with endogenous cultural institutions we include the vectors SOC and POL. The variable denotes the error term.: Given the binary nature of the dependent variable, we employ a logit model. 11 In a second step, to estimate the relationship between endogenous cultural institutions and farmers' adaptation choices, we employ the following model: Since some CAPs are strongly interlinked, we employ Principal Component Analysis (PCA) to group them into three indexes. More specifically, we join Mixed Cropping and Expansion of Farms (Model 1), Alteration of Planting Dates and New Harvesting Dates (Model 2), and Local Irrigation and Local Pest Control (Model 3) as our dependent variables of interest. 12 In Equation (3), ECI is a binary explanatory variable that captures whether the farmer i complies with a particular ECI. We then proceed to use Ordinary Least Squares (OLS) to estimate the relationship between ECI and CAP, controlling for individual features of the respondent (i.e. age, sex) as well as their households' characteristics (i.e. household income, household members) in the vector HHSEP.

Results
In line with our objectives, we will first present the determinants of ECIs before turning toward ECIs as possible determinants of different CAPs. In estimating the determinants of ECIs we follow a stepwise approach and first estimate socio-economic determinants individually and then look at political and social determinants of ECIs.

Socio-Economic Determinants of Endogenous Cultural Institutional Compliance among Peasant Farmers
The results (Supplementary Table 1) indicate that most of the respondents and household characteristics do not contribute much to explaining different ECIs, as can also be seen by the relatively low R 2 throughout the regressions. It is, however, important to note that this can also be a result of our rather small sample, which bears the problem of relatively low variation in the variables under investigation, a problem we are facing throughout all our estimations owed to low response rates. Nevertheless, of all the ECIs under investigation, Pouring Libation seems to be best explained by our respondent and household characteristics.
In general, increased Age seems to be associated with a stronger use of ECIs that is statistically significant for three of the six ECIs (Launch of Planting Seasons, Pouring Libation, Adherence to Spiritual Guidance). An increase in Age by one year is associated with an increased probability of complying with the relevant ECIs by one to two percentage points. 13 Older respondents are more likely to be familiar with ECIs than younger and possibly less familiar with more "modern" ways of farming. Males seem to be more likely to comply with the practice of Pouring Libation with a difference of about 30 percentage points between males and females, while for the rest of the ECIs, we do not observe any statistically significant effect of Male. Similarly, the number of Household Members does not seem relevant for compliance with different ECIs. The number of Working Members within the household, however, seems to be associated with a stronger use of Rain Forecasts by over eight percentage points, while Farm Income has a small but negative association with the Respect for Spiritual Days. Spiritual days are usually dedicated days for the performing of rites, and also the days when ancestors are believed to be visiting and parading through the community. There is an inverse relationship between Schooling and Pouring Libation, and Adherence to Spiritual Guidance. 14 Also interesting to note is the fact that Trainingan exogenously-rooted processhas no statistically significant association with either one of the ECIs.

Social and Political Effects on the Compliance of ECIs among Peasant Farmers
In Supplementary Table 2, we present the logit estimates of the political forces that shape compliance with ECIs among peasant farmers. 15 The socio-economic respondent and household characteristics are mostly in line with the results of Supplementary Table  1. Most of the political variables do not seem to significantly account for peasant farmers' compliance with ECIs. However, State Dominance in terms of the introduction and the application of formal rules is associated with a lower probability compliance of traditional Rain Forecasts as an ECI. Farmers who noted the strong influence of the state in climate adaptation interventions have probability of compliance about 24 percentage points lower than the probability of compliance of the farmers who did not note the influence. The probability of compliance with Non-Encroachment into Spiritual Area for those farmers, however, is about 23 percentage points higher. Peasant farmers with Political Leanings have a higher probability to Adhere to Spiritual Guidance by about 24 percentage points, while the role of Powerful Elites was surprisingly found to be completely insignificant.
The logit estimates of social factors that shape compliance with ECIs are presented in Supplementary Table 3. 16 Once again, the respondent and household parameters are much in line with the previous results, with two notable exceptions. Once we include the set of social determinants, Schooling becomes a statistically significant predictor of Non-Encroachment into Spiritual Area, with a positive relationship. On the other hand, peasant farmers who have received Training were more likely to adhere to the Traditional Launch of the Planting Season. Regarding Education as a social factor, respondents who noted education as a factor of change in ECIs showed a higher probability of compliance with the ECI of Non-Encroachment into Spiritual Area for farming by almost 52 percentage points, while they were less likely to observe Traditional Rain Forecast, Traditional Launch of the Planting Seasons and (Adhere to) Spiritual Guidance by a lower amount. While the aspect of increasing Communication and Modernization did not provide a meaningful explanation on the compliance with ECIs, respondents who mentioned Migration showed a higher chance of noncompliance with the rules regarding Non-Encroachment into Spiritual Areas.

Relationship between ECI Parameters and Peasant Farmers' Climate Adaptation Choices
Taking into consideration relevant and compatible CAPs, we obtained three indices through PCA. 17 The results are presented in Supplementary Table 4. Most of the socio-demographic variables do not significantly explain the adoption of CAP. Respondents who participated in Training, which can be considered as exogenously influenced, are less likely to employ the practices captured in Model 3. This is not surprising, as they are more likely to lean toward exogenous institutions as opposed to their counterparts who did not receive training. Most exogenously shaped trainings are associated with extension services and the provision of support through farm inputs. This incentive facilitates the shift in perspective away from endogenously shaped practices. Model 1 shows that farmers who complied with the traditional Launch of Planting Seasons as an ECI had a significantly higher chance of undertaking mixed cropping and expanding their farms as a CAP as reflected in Model 1. The alternation of planting dates and setting of new harvesting dates (Model 2) is not statistically significantly associated with any of the endogenous cultural institutional parameters. In Model 3, peasant farmers who complied with Traditional Rain Forecast and Traditional Launch of Planting Seasons were more likely to engage in local irrigation and local pest control as an adaptation index than their counterparts who did not. On the other hand, those who adhered to the Pouring of Libation were less likely to engage in the CAP captured in Model 3.

Discussion
The debate on the effectiveness and compliance with institutions continues to attract much attention. In this paper, we attempt to shed light on the determinants of farmers' compliance with agro-based ECIs and their relationship with different CAPs.

Household Socio-Economic Determinants of Compliance with ECIs among Peasant Farmers
Age was found to increase the chance of compliance with several ECIs. Furthermore, higher farm income earning households hardly complied with the respect for spiritually assigned dates. This is consistent with findings from other contexts: for example, in India, farmers were very much attached to informal arrangements (ECIs) as opposed to rich non-farmers (Birthal et al. 2017); in Kenya, farmers patronized informal systems in seed acquisition which were judged to have greater potential to meet food security needs (Croft et al. 2018). In rural Ghana, the practice of exogenous-driven climate adaptation interventions was influenced by age, gender and the physical location of the farmer (Wood et al., 2021). However, evidence from farmers in the Northern Region of Cameroon point to a contrary view that gaining income does not completely eliminate ECIs. That is, ECIs which are beneficial to a few actors are usually sustained (Haller 2001). Surprisingly, being male and years of schooling all decreased the likelihood of compliance with this practice. Furthermore, the higher the number of years of schooling, the more the chance of peasant farmers not adhering to spiritual guidance during farming periods.

Social and Political Effects on Compliance with ECIs among Peasant Farmers
Respondents argue that the propagation of state institutions and their interventions, including the influence of political elites (government officials and elected representatives such as senators, parliamentarians and mayors) contributed to a decline in compliance with several ECIs, including traditional rain forecast, and non-encroachment into spiritual areas for farming. State institutions generally promote more formal institutional arrangements in the form of laws and policies which are exogenous to communities. Abass, Mensah, and Fosu-Mensah (2018) showed that in rural Ghana, farmers tend to lean more toward exogenous institutions, interventions and resources to adapt to climate change. This particularly occurs in the sphere of local capacity building (void of innovations). However, farmers who were militants of political parties complied with spiritual guidance. Studies in diverse settings of SSA show that the interventions ended up shifting the burden of responsibility to peasant farmers (Ribot 2014). As revealed during key informant interviews, this case could be further explained by the links several political elites have with traditional practitioners in communitiesthey seek protection and guidance to attain their ambitions, which indirectly motivates them to do so, even for their economic endeavors. Using data from more than 1400 ethnic groups, Baldwin and Holzinger (2019) showed that contrary to a widely held position that traditional systems do not have clear and strong accountability mechanisms, they in effect do. These systems regulate the behavior and actions of not only traditional rulers, but also subjects, including politicians. Political elites in the Western Highlands of Cameroon seek to adhere to existing ECIs for their political gains; by interacting and respecting these institutions, they indirectly convince the local population and gain votes during election periods. Education as a factor of change reduces compliance with traditional rain forecast and traditional launch of the planting seasons. Migration hardly explained compliance with ECIs, but a shift toward modernity reduced compliance to some ECIs (e.g. non-respect for traditional rain forecast). These positions have been echoed by several authors showing that migration contributes to a change in the degree of compliance with existing ECIs (Ensminger 1992;Haller 2001).

Relationship between ECI Parameters and Peasant Farmers' Climate Adaptation Choices
Household socio-economic factors, together with ECIs, differentially interacted with climate adaptation choices. Most of the socio-demographic variables do not significantly explain the adoption of climate adaptation choices. We find particularly surprising that education does not affect farmers' adaptation choices. In line with related literature, we would have expected the more educated to appreciate changes in climate and adjust their farm practices. In rural Togo, Ali, Awade, and Abdoulaye (2020) showed that education did not significantly determine the decision to embrace climate adaptation strategies in the agrarian sector. A similar situation was reported for rural Ghana, in which gender (and not education) was a key determinant to climate adaptation decisionmaking (Wood et al. 2021).
With respect to ECIs as determinants of climate adaptation, we find that farmers who complied with the traditional launch of planting seasons were likely to adopt mixed cropping and farm expansion as a climate adaptation strategy. This concurs with previous findings: for instance, in Zimbabwe, peasant farmers relied very much on indigenous customs to make major strategic, tactical and operational crop production decisions (Mapfumo et al., 2016). In the case of Uganda and Tanzania, farmers employed customary seed and farm management approaches to cope with increasing dryness, linked to climate variability (Namanda et al. 2013). Kihila (2018) further showed that peasant farmers in Tanzania employed a wide range of climate adaptation farming practices such as mixed cropping, terracing, tree planting, and local irrigation (through locally based water reservoirs). This shows that future interventions to climate-proof the agrarian sector need to increasingly consider attachments toward ECIs (Dow et al. 2013). During the launch of the farming season, certain pronouncements are made to bless the land, while small village celebrations to mark the planting season also characterizes some communities. The confidence that develops during the process explains why farmers who complied with it will aim to undertake mixed-cropping and to expand the size of their farms, in the belief that it will assure greater yields. What remains to be uncovered is the performance of these farms in the face of a changing climate and their applied adaptation strategies.
Further studies should therefore shed light on the performance of peasant farmers (in terms of crop yields), by comparatively studying groups that adhere to ECIs and those that do not. This will provide a holistic information on the dynamics linked to the practice of ECIs and their outcomes in terms of farmers' climate adaptation choices. Farmers who complied with the practice of traditional rain forecast and traditional launch of planting season engaged more in local irrigation and local pest control as adaptation choice. However, compliance with ECIs does not significantly explain the choices of alternating planting dates and setting up new harvesting dates. In different agro-ecological zones in Asia, Karki, Burton, and Mackey (2020) showed how farmers employed several strategies including the change of crop types and varieties, the application of new technologies, and other diversification strategies. In rural Ghana, Awuah-Nyamekye (2013) explained how people employed a range of indigenous and conventional methods in agriculture and natural resource management, although the degree of dependence on both systems was not separately analyzed. This suggests that farmers do not only rely on what formal systems provide, they also combine what both formal and informal systems provide, to navigate through the process of adapting to climate change in the agricultural sector (Lebel 2013;Karki, Burton, and Mackey 2020).

Conclusion
ECIs are a part of rural life, and continue to define several activities, including agriculture. This is largely defined by the degree of compliance. From the foregoing analysis, the following conclusions can be drawn: Political factors enhance ECIs compliance more than they constrain it, in the context of climate adaptation. Social forces do not significantly explain compliance with ECIs; however, education significantly accounts for growing noncompliance with ECIs. ECIs relationship with CAPs remains mixed; while some ECIs are strongly linked with several CAPs, others are not. A key message derived from this study rests on the fact that the solo functioning of ECIs is insufficient to drive effective climate adaptation for farmers. It therefore needs to be complemented by exogenous institutional arrangements to drive effective climate adaptation. A twoway acceptance is expected in the Western Highlands; actors driving exogenous institutions should accommodate endogenous cultural institutional arrangements, and vice versa, in the context of climate adaptation. It is clear that in the wake of renewed efforts to combat climate change, ECIs should be further repositioned to play significant and complementary roles toward climate adaptation. In this light, leveraging ECIs in climate adaptation would require effective documentation and capturing in formal interventions toward climate change adaptation.
This study is limited by the fact that a relatively small sample size was employed to analyze ECIs and climate adaptation practices. Therefore, future studies should consider the use of largely samples and data sets to analyze these issues. Additionally, as culture is shared, further studies need to embrace more community level approaches to understand how ECI elements shape the climate adaptation behavior of farmers. As a limitation, this study did not compare the climate adaptation practices as a function of proximity to the city. Future studies should consider these aspects and complement them. Furthermore, while migration is analyzed, this paper did not consider the origin of migrants which could potentially define the degree of adherence to certain ECIs. Further studies should therefore consider these questions. In analyzing ECIs, future studies should their source and change pathways in the context of climate adaptation. This research contributes to address the existing lacunae on the one hand, and could inform theoretical frameworks on institutions and climate adaptation in SSA.

Notes
1. The poverty rate is $1.9/day. The world's poverty average is estimated at 10 percent (United Nations 2015). 2. Exogenous institutions here relate to interventions in communities, which are externally induced; they might emanate from the state or international actors. Details are provided in section "Analytical framework on endogenous cultural institutions and their determinants". 3. By endogenous cultural institutions, we refer to informal rules which are developed within communities and transferred from one generation to the other, through norms, values, beliefs and taboos (Colding, Folke, and Thomas 2003). 4. Apart from the key informant interviews, the design of the survey instrument was also informed by previous studies on endogenous cultural institutions in the Western Highlands of Cameroon (see Kimengsi, Abam, and Forje 2021 and Kimengsi and Mukong 2022 among others). 5. The sample size in our regressions ranges from 95 to 105, depending on the specific variables we are using. 6. Descriptive statistics are in Supplementary Appendix 1. 7. These are also known as soothsayersthey are believed to predict future occurrences in people's daily lives, including rainfall. 8. There are sites reserved as spiritual sites where human activity is prohibited. Nonencroachment here refers to the tendency of the respondent to respect this cultural norm. 9. The monthly income was derived by calculating the average incomes realized during the major cropping seasons, including the low seasons. 10. We have also captured degree in our data but due to the high correlation with schooling, we chose the latter as our primary variable. The results for degree are comparable and can be obtained from the authors. 11. For robustness, we also computed the same regressions using a probit model and obtained similar results that are available from the authors.
12. Mixed cropping usually encourages farmers to expand their farms. Also, the alteration of planting dates almost invariably causes farmers to adopt new harvesting dates. Finally, local irrigation practices influence the timing of pest control to avoid wash offs from pests. 13. The marginal effects for Supplementary Table 1 can be found in Appendix 2. 14. An additional year of Schooling is associated with a reduced probability of ECI compliance by roughly two to four percentage points. 15. The marginal effects for Supplementary 17. We are well aware of the drawbacks when employing PCA on binary variables, in particular the fact that linear functions of binary variables are not readily interpretable (Jolliffe 2002). Our main objective, however, is to summarize much of the variation of our original dataset into a smaller set of variables and if all included variables are binary the PCA indeed provides a plausible low-dimensional solution (Gower 1966). To test the robustness of our results we also test a manually generated variable that joins the components of our models. The results are comparable across the different variable specifications and can be obtained from the authors.