Socio-demographic determinants of farmers’ beliefs about climate change cause in the Sudanian zone of Benin

: Understanding farmers’ beliefs on climate change is crucial as it drives the adaptation strategies that they might adopt. This paper investigates farmers’ beliefs on climate change in the Sudan Savannah Zone of Benin, a region heavily reliant on rain-fed agriculture. The multinomial logit model is applied to cross-sectional data collected through a survey of 60 randomly selected farm households. The findings suggested that 33.33%, 31.67%, 21.67%, and 13.33% of the farm households believe that climate change is due to human activities, to natural changes in the environment, gods anger, and to both human activities and natural changes in the environment, respectively. Moreover, the estimation results of the determinants of climate change cause indicate that the gender of the household head, the ethnic group, and household size influence significantly climate change beliefs. Based on the findings, information on the fact that climate change is not only due to natural changes in the environment, but is also due to anthropogenic greenhouse gases should be provided to farmers. This paper contributes to the literature by analyzing what farmers believe as causes of climate change which is beyond climate change perception. Moreover, the variable ethnic group and household size are found for the first time to our knowledge to determine climate change beliefs.


Introduction
Climate change constitutes one of the major threats to economic activities across the world (Pörtner et al., 2022). Empirical evidence confirms that climate change affects agriculture in non-industrialized countries due to its rain-fed nature and it will still be affecting this sector if relevant adaptation measures are not taken (Lokonon et al., 2015;Di Falco & Veronesi, 2013Di Falco et al., 2011). The agricultural sector is the mainstay of the economy of non-industrialized countries. For instance, in Benin, the agricultural sector; characterized by the predominance of smallholder farm households, is the source of employment for about 70% of the active population, contributes 33% of the gross domestic product (GDP) and about 75% of export revenues (MAEP, 2017).
As climate change casts a shadow over the future of agriculture, the impacts of climate change in Benin are diverse (Adjacou, et al. 2022). The IPCC projects a decline in agricultural yields in West Africa, which could be between 5 and 20% in Benin (Pörtner et al., 2022). The most predominant impact indicator declared by the populations in North is food insecurity (22%), which is the consequence of the disruption of sowing dates (16%) and the drop in yield (15%). Poverty (14%) is seen as exacerbating and persistent over the years (Vodounou & Onibon Doubogan, 2016;Sodjinou & Hounkponou, 2019). Another study also shows a decline in the availability of fish resources, fishing success and water level in North Benin (Nago et al., 2019). Based on the Ricardian analysis developed by Mendelsohn et al (1993), another study shows that the impact of climate change on agriculture in Benin is likely to be negative in the long term (Agossou, 2012). The details from this paper regarding the horizon 2100 provide that the impact of climate change on the agricultural value added per hectare and indirectly on the agricultural income per hectare is estimated at -$365.53 or -21.87% compared to its 2008 value. Consequently, the whole economy will be affected by climate change. It is therefore urgent to put in place adequate policies and strategies that can facilitate adaptation to climate change. It should be noted that some areas in Benin and crops may benefit from climate change. The economic impacts of climate change on agriculture have shown that maize and yam production will increase by 2050 (Satoguina, 2016). Therefore, it is urgent to undertake actions for climate change adaptation.
The consideration of adaptation measures in the agricultural sector, to mitigate climate change impacts and/or to seize the opportunities, depends on the perception of climate change and the beliefs regarding climate change (Lokonon & Mbaye, 2018;Wheeler et al., 2013;Deressa et al., 2011;Maddison, 2007).
Climate change perception is a complex process that encompasses a range of psychological constructs such as knowledge, beliefs, attitudes and concerns about if and how the climate is changing (Whitmars & Capstick, 2018). The evolution of risk perception studies has been research on environmental beliefs, attitudes, and values. Although the literature on public opinion often uses the terms ''values,'' ''beliefs,'' ''attitudes,'' and even ''paradigms'' somewhat interchangeably, we use the term ''climate change beliefs'' to refer to non-issuespecific cognitive orientations. A reasonable assumption characterizing this research tradition is that environmental cognitions are the bedrock of support for environmentally friendly or hostile behaviors and are the basis of environmental risk perceptions (Dunlap & Scarce, 1991). From this perspective, risk perceptions are an integral by-product of environmental beliefs and not independent causes of behavior. Perception is an important link between beliefs and adaptation (Jooste et al., 2018).
Understanding climate policy risk responses and other social, economic and policy perspectives is a vital component of understanding climate change beliefs, risks and behaviors (Haden et al., 2012). Azadi (2019) found a complex relationship between overall climate change belief, risk perception, psychological distance, trust and risk salience, and farmers' adaptation behaviors. So, it is of paramount importance for policymakers to understand the beliefs of farmers on climate change in order to design policies towards boosting agricultural production and dealing with food insecurity. Heath et al. (2006) investigated three beliefs concerned with global climate change: the likelihood that it exists, whether it has human or natural causes, and whether its consequences are negative or positive. In the case of this paper, we assess the beliefs on climate change as the cause of climate change according to the beliefs of farmers.
Currently, most of the literature is relative to climate change perception. Few papers which deal with the beliefs on climate change lack the determinants of these beliefs (Chisale et al., 2022;Lee et al. 2015;Debela et al. 2015;Saleh Safi, et al. 2012& McCright, 2010. So, this paper includes what farmers believe as causes of climate change which is beyond climate change perception as well as their determinants. Actually, the nature of measures to be taken to mitigate the adverse effects of climate change on farming activities depends on what farmers think as causes of changes in climate (Arbuckle et al., 2013;Wheeler et al., 2013;Haden et al., 2012). In fact, farmers may think that climate change is due to the fact that they have offended God and/or gods so that he (they) is (are) angry. Only a few studies to our knowledge have considered farmer's beliefs. Menapace et al. (2015) analysed farmers' shortand long-run perceptions of agricultural risks related to climate change. They identified climate change beliefs as a critical factor explaining the short-versus long-run difference in risk perceptions: those who believe in climate change project larger future crop losses. Arbuckle et al. (2013) analysed, among others, farmers' beliefs about climate change and attitudes toward adaptation and mitigation. In reality, adaptation strategies could be linked to farmers' beliefs because for example, if farmers believe that climate change is a phenomenon due to God or gods anger, they will decide just to pray or to offer sacrifices to gods, to ask them to let rainfall coming back (Yegbemey, et al. 2020). As for Wheeler et al. (2013), they explored the influences associated with farm adjustment strategies, and particularly the role that climate beliefs play. Wheeler et al. (2013) found that the relationship between climate change belief and adopting various adaptive strategies appears to be often endogenous, especially for accommodating strategies. The presence of believers in the cultural and spiritual causes of climate change in a particular area presents a challenge to mobilize them toward implementation of climate intervention measures (Chisale et al., 2022). McCright (2010) found that a greater percentage of women (64%) than men believe that global warming is primarily caused by human activities. Moreover, older people tended to believe that the causes of global climate change are natural (Heath et al. 2006). So, the determinants of farmers' beliefs on the causes of climate change are gender and age. One can see that in the literature it is difficult to find out more determinants of farmers' beliefs on the causes of climate change. This study is an attemps to fill in this gap by checking if the variables determining the perceptions of farmers on the causes of climate change are the same in the case of the beliefs.
In the light of the literature on the determinants of farmers' perception on the causes of climate change, the independent variables included in the estimations (e.g., Chisale et al., 2022;Lokonon & Mbaye, 2018;Debela et al., 2015;Fosu-Mensah et al., 2012;Silvestri et al., 2012;Deressa et al., 2011;Maddison, 2007) are: the gender of the household head, the age of the household head, the ethnic group of the household head, the size of the household, the experience of the household head in agriculture, farm size, and having agriculture as the sole source of revenue or not. Experience in agricultural activities is positively associated to climate change perception (Chisale et al., 2022;Lokonon & Mbaye, 2018). Lokonon & Mbaye (2018) found that climate change perception is negatively related to household size. As for the formal education level, it is found playing a paramount importance in climate change perception (Chisale et al., 2022;Fosu-Mensah et al., 2012;Silvestri et al., 2012). The gender of the household head can influence the likelihood of perceiving climate change (Chisale et al., 2022;Fosu-Mensah et al., 2012;Maddison, 2007). Farm size is also hypothesized to affect climate change beliefs. Indeed, assets and wealth may influence climate change perception (Silvestri et al., 2012;Fosu-Mensah et al., 2012;Deressa et al., 2011).
This study aims to investigate farmers' beliefs on the causes of climate change. Specifically, this paper analyses (i) farmers' beliefs on the causes of climate change and (ii) the determinants of these beliefs in the context of the Sudan Savannah Zone of Benin.

Study area
This study focuses on the Sudan Savannah Zone of Benin located in the Northern part of the country. The primary data used in this paper were collected in three communes (Boukoumbe, Ouake and Banikoara). The area is characterised by a rainy season and dry season. Generally, the rainy season is from May to November in the study area. The annual rainfall in the study area is about 780 mm and the number of annual rainy days is about 45 days. The three communes received seven months of rainfall with a peak period in June. The temperature is generally warm in these communes ranging between 28º C and 34º C with a relative humidity between 65% and 94%.
The survey covers six communities within the communes, namely Dikouteni, Koumagou, Gomparou Peulh, Sampeto, Kakpala and Kawado ( Figure 1). The selection of the communes was based on the following criteria: (i) level of vulnerability to climate change, (ii) perceived levels of demonstration/application/use of indigenous knowledge, and this is due to farming methods and farming systems practiced (mechanization and irrigation are not available), (iii) accessibility for the survey. According to MCVDD (2014), West Atacora Zone (Boukoumbé) and North Donga (Ouaké) are vulnerable to climate change. Alibori (Banikoara) is less vulnerable to climate change.

Research units and sampling
The sample consisted of 60 smallholder farmers selected randomly in six villages (10 smallholder farmers each). This helped also to record the entire interview and take pictures.

Data and variables
Data was collected from 28 th August to 4 th September 2017. It was a digital survey conducted with KoboCollect. For the administration of semi-structured questionnaires, we used two enumerators who are agricultural economists. Sixty semi-structured questionnaires were administered to smallholder farmers. The collected data included socioeconomic characteristics such as age, gender, household size and questions on climate change beliefs: What do you think about climate change in regard to natural changes, human activities and the anger of God or gods? The response of the farmer is among the list: climate change is occurring, and is caused mostly by human activities; climate change is occurring, and is caused equally by natural changes in the environment and human activities; climate change is occurring, and it is caused mostly by natural changes in the environment; there is not sufficient evidence to know with certainty whether climate change is occurring or not; climate change is occurring and is due to God or gods anger and; climate change is not occurring; others (specify).
The dependent variable contains four categories characterising farmers' beliefs on climate change (Figure 2). These are Gods anger, Natural changes in the environment, Human activities and Natural changes in the environment as well as to human activities. This paper captures assets and wealth through farm size and having agriculture as the sole source of revenue or not. The age and the ethnic group of the household head are included in the set of explanatory variables as they have the potential to influence beliefs on climate change. For instance, experience in life (captured by age) may shape farmers' beliefs on climate change and the customs related to land tenure vary across ethnic groups and therefore, ethnic group can influence the formation of beliefs on climate change.

Model
In this paper we are interested in estimating the likelihood that a farm household will belong to a certain category of beliefs on climate change. Climate change beliefs were measured through a four-category question that measures belief about the causes of climate change.
Being a female leads one to believe that global warming is primarily caused by human activities as opposed to men (McCright, 2010). Moreover, older people tended to believe that the causes of global climate change are natural (Heath et al. 2006). So, the determinants of farmers' beliefs on the causes of climate change are gender and age. One can see that in the literature it is difficult to find out more determinants of farmers' beliefs on the causes of climate change. This study is an attempt to fill in this gap by checking if the variables determining the perceptions of farmers on the causes of climate change are the same in the case of the beliefs.
In the light of the literature on the determinants of farmers' perception on the causes of climate change , the independent variables included in the estimations (e.g., Chisale et al., 2022;Lokonon & Mbaye, 2018;Debela et al., 2015;Fosu-Mensah et al., 2012;Silvestri et al., 2012;Deressa et al., 2011;Maddison, 2007) are: the gender of the household head, the age of the household head, the ethnic group of the household head, the size of the household, the experience of the household head in agriculture, farm size, and source of revenue.
Experience in agricultural activities is positively associated to climate change perception (Chisale et al., 2022;Lokonon & Mbaye, 2018). Lokonon & Mbaye (2018) found that climate change perception is negatively related to household size. In contrario, large households may be forced to divert part of the labour force to off-farm activities in an attempt to earn income in order to ease the consumption pressure imposed by large family size (Yirga 2007). Then household size is positively related to the perception of adapting to climate change.
The variable education level of the household is a determinant of climate change perception (Chisale et al., 2022;Fosu-Mensah et al., 2012;Silvestri et al., 2012). The gender of the household head influences the likelihood of perceiving the cause of climate change (McCright, 2010;Fosu-Mensah et al., 2012;Maddison, 2007). Farm size is also hypothesized to affect climate change beliefs (Wheeler et al., 2013). Indeed, assets and wealth may influence climate change perception (Silvestri et al., 2012;Fosu-Mensah et al., 2012;Deressa et al., 2011). These positive and negative associations do not provide conclusive evidence in the context of the determinants of beliefs on climate change causes. They merely highlight the possible association between these variables and beliefs on climate change causes.
Climate change beliefs is of paramount importance in conducting agricultural activities owing to the rainfed nature of agriculture in Benin, including the Sudan Savannah Zone of Benin. Indeed, farmers will take appropriate adaptation measures with regards to their beliefs on climate change (Arbuckle et al., 2013;Wheeler et al., 2013;Haden et al., 2012). A rational farm household will conceptualize her/his beliefs on climate change with respect to her/his characteristics, X . These characteristics include those that are internal to the household such as the gender, the age, the education level of the household head, the size of the household, farm size, the ethnic group, and source of revenue. In this paper, four types of beliefs about climate change are considered: (i) climate change is due to natural changes in the environment, (ii) climate change is due to natural changes in the environment as well as to human activities, (iii) climate change is due to God or gods anger, (iv) climate change is due to human activities. Therefore, the dependent variable contains four categories, and there is no natural order between these four categories of beliefs. Thus, it is not possible to rank these four categories of beliefs.
Thus, the empirical model to be estimated is specified as follows: Natural changes in the environment 2 Natural changes in the environment as well as to human activities 3 gods anger 4 Human activities (2) is the error term which is independently and identically distributed. Because the dependent variable is categorical and there is no natural order between the categories, (1) should be estimated by a multinomial probability model (e.g., multinomial logit and multinomial probit models). The paper opts for a multinomial logit model which is an alternative to the multinomial probit model. The multinomial logit model relaxes the independence restrictions built into the multinomial probit model (Greene, 2012). Table 1 presents a detailed description of the independent variables and their descriptive statistics. Figure  2 presents farmers' beliefs on climate change. The results indicate that 33.33% of the farm households believed that climate change is due to human activities. This result confirms the findings that the belief in anthropogenic causes of climate change is a major determinant of risk perception (Lee et al. 2015;Saleh Safi, et al. 2012;McCright, 2010).

Farmers' beliefs on climate change
About one-third (31.67%) of farm households also considered climate change as a natural phenomenon. More than one-fifth (21.67%) of them believed that climate change is due to God or gods anger, while the remaining 13.33% attributed climate change to both human activities and to natural changes in the environment. Therefore, the surveyed farm households have beliefs that are somewhat consistent with scientific knowledge on climate change. However, these findings suggest only 13.33% of them have beliefs which are in line with scientific knowledge on climate change; as mainly due to anthropogenic GHGs (Pörtner et al., 2022), while those believing that it is due to human activities are not far from this scientific knowledge.
It is striking that more than one-fifth of the farm households attributed climate change to God or gods' anger, which means that the actions they will undertake to mitigate the adverse effects of changes in climate may differ from those of their counterparts (O'Neil, et al. 2022;Yegbemey, et al. 2020).

Determinants of farmers' beliefs on climate change
The estimation results of the multinomial logit model of the determinants of farmers' beliefs are presented in Table 2 with "climate change due to human activities" as the reference category. The model is overall significant and passes all the diagnostic tests. The findings reveal that the gender of the household head, the ethnic group of the household head, and the household size are significantly related to climate change beliefs. The likelihood of believing that climate change is attributable to natural changes compared with believing that climate change is due to human activities decreases with household size. The relative risk ratio was 0.88 and was significant at a threshold of 0.10. Keeping all other variables constant, if the household size increases one unit, the head of household is 0.88 times more likely to believe that climate change is Natural changes in the environment as compared to the belief that climate change is due to human activities (a 12% lower risk ratio or odds). Similarly, the likelihood to believe that climate change is attributable to God or gods anger compared with believing that climate change is due to human activities decreases with household size. This result is in line with that of Lokonon & Mbaye (2018) relative to the association between household size and climate change perception. In the case of climate change beliefs, this is a new relationship that has not been discussed in previous literature: the relationship between household size and beliefs in climate change.
Having a male head in the household increases the likelihood of believing that climate change is caused by natural changes in the environment compared to conceptualizing climate change as due to human activities. Similarly, the gender of the household head (having a male household head) increases the probability of believing that climate change is caused by God or gods anger compared to thinking that it is due to human activities. This finding is not in line with that of McCright (2010) who found that a greater percentage of women (64%) than men believe that global warming is primarily caused by human activities.
Farmers' beliefs about climate change are also related to the ethnic group of the household head. For example, individuals belonging to the Lokpa ethnic group, as compared to the Bariba ethnic group, are more likely to attribute climate change to natural changes in the environment compared to believing that it is due to human activities, with the relative-risk ratios being significant at the 5% level of significance. This insight aligns with previous studies that have emphasized the influence of spiritual beliefs on climate change perceptions (Chisale et al., 2022;Saleh Safi, et al. 2012). According to Chisale (2022), the presence of the believers in the cultural and spiritual causes of climate change presents a challenge to mobilize them toward implementation of climate intervention measures and forest management in Malawi. Saleh Safi (2012) assessed rural Nevada and climate change perception and found ethnic group as an important factor of individual socioeconomic vulnerability. In the case of climate change beliefs, this is a new relationship that has not been discussed in previous literature: relationship between ethnic group of the household head and beliefs in climate change. The remaining variables such as experience, farm size, having agriculture as the only source of income and age do not significantly affect farmers' beliefs on climate change. The result of the last variable is surprising, as Heath (2006) found that older people tended to believe that the causes of global climate change are natural using the correlation method. One limitation of this paper is the lack of inclusion of religion as an independent variable. However, Haluza-DeLay, (2014) suggests the potential utility of religions engaging the issue of human-induced climate change can be compromised by the perception that religious beliefs are often in conflict with scientific understanding. Those following modern religion believe on climate change, because both climate change and the religions modern concepts, but animists are expected to not believe or to think that it is God made, because of their traditional knowledge. Note: Robust standard errors in parentheses. *, **, *** indicate statistical significance at the 10%, 5%, and 1%, respectively.

Conclusion and policy implications
Due to the importance of climate change beliefs in shaping adaptive measures, this paper investigated the extent of these beliefs as well as the factors associated to them in the Sudan Savannah Zone of Benin, a region heavily reliant on rain-fed agriculture, using a multinomial logit regression. The findings paint a diverse picture: 33.33%, 31.67%, 21.67%, and 13.33% of the farm households believed that climate change is due to human activities, to natural changes in the environment, to God or gods anger, and to both human activities and natural changes in the environment, respectively.
The analysis unveiled intriguing associations: climate change beliefs are related to the gender and the ethnic group of the household head, and household size. In a striking contrast, male-headed households are more likely to attribute climate change to natural phenomena or divine intervention compared to their female-headed counterparts. Similarly, the Lokpa ethnic group and others ethnic groups, exhibited different beliefs compared to the Bariba ethnic group. However, household size was found to decrease the likelihood to believe that climate change is due to natural or divine causes compared to believing that it is due to human activities.
Armed with this knowledge, it becomes imperative to disseminate information that climate change is largely a consequence of anthropogenic greenhouse gases (from human activities). This can enable those that believed that climate change is attributable to natural changes in the environment to be consistent with the scientific knowledge on the causes of climate change as found as the IPCC (Pörtner et al., 2022). This crucial information should be tailored to reach those who attribute climate change to natural changes or divine will, as they might otherwise rely solely on prayers and sacrifices as adaptation measures, overlooking practical and scientifically-supported strategies.
Moreover, special attention must be given to raising awareness among male-headed households, the Lokpa and other ethnic groups regarding the actual causes of climate change, fostering a deeper understanding and potentially transforming their adaptive approaches. Future research may probe deeper into the link between farmers' beliefs on climate change and adaptation decisions, as well as exploring innovative methods for communicating climate science to diverse audiences.

ACKNOWLEDGEMENT
We would like to thank BIARI ALUMNI RE-SEARCH TEAM 2016-17. The authors would like to thank their peers from seven nations across Africa, Asia, South America, and Europe for giving them the opportunity to collect this extra data during their participative work. They are: Chinwe Maryann ONUEGBU, Cristobal Delgado MATAS, Efosa ADAGBASA, Eustace UZOR, Mariaelena HUAMBACHANO, Portia ADADE WILLIAMS, Siri VELAND, Solomon ADESOJI, Tyseer EL HADI OMER and Wang HUAIYU. Moreover, the special thanks and heartfelt gratitude go to the farmers in Benin who participated in the survey for their time and contributions as well as their efforts in sharing their experiences during the field work which contributed to the success of this study. We will acknowledge BIARI (Brown University) which has founded this research. We would like to extend our gratitude to the journal editor and reviewers for a number of useful comments. We also extend our gratitude to the enumerators: Ken Kounouewa and Morel Kogbevi and the farmers who participate to this survey.