Perception of Climatic Change and Farmers’ Decision to Adapt in the Sudano-sahelian Zone in Cameroon

William Nemkenang Koguia ✉ Fidoline Ngo Nonga, Ali Madi, Antoine Leblois and Mabah Tene Gwladys Laure Faculty of Economics and Management, University of Yaoundé II-Sao, Cameroon. Faculty of Economics and Management, University of Douala, Douala, Cameroon Faculty of Economics and Management, University of Maroua, Maroua, Cameroon INRA, Institut National de la Recherche Agronomique. Center for Environmental Economics Montpellier IRAD, Institute for Agricultural Research and Development, Yaoundé, Cameroon ✉ Corresponding Author: William Nemkenang Koguia, E-mail: nemkes2001william@gmail.com

Research works (Molua, 2002(Molua, , 2006(Molua, , 2008, show that the agriculture in Cameroon shall be highly affected by climatic change. These results are in line with those of Ouedraogo (2012) et Jeder et al. (2013) in Burkina Faso, Da Silva (2009) in Canada, Mendelsohn et al. (1994) in the US.
This situation obliges farmers to develop adaptation strategies in order to preserve their means of subsistence. The majority of research on the determinants of adaptation to climatic change lay emphasis only on the demographic and socio-economic characteristics of farmers. However, according to Maddison (2007), Gbetibouo (2009), and Yegbemey et al., (2014), adaptation is a result of both perceptions of the evolution of climate and demographic and socio-economic characteristics. It is important that research should also take into consideration the perception of producers of the said climatic change. This article has as objective to describe simultaneously the perception and adaptation of farmers to climate change in the sudano-sahelian zone of Cameroon. This article is organized as follows: Section I is the introduction; section II examines the literature review on perception and the adaptation of farmers to climatic changes; section III presents the methodology and empirical model; section IV presents the results and Section V presents the conclusions and the implications in terms of public policy.

Literature Review
The notion of risk perception emerged in the late 60s during the anti-nuclear protest movements in the United States between the « objectives » risks of experts and the « perceived or subjective » risks of laymen (Finucane et al., 2000;Slovic, 1992;Slovic et al., 1979). Experts used to address risk through potential consequences and the probability of occurrence (Marris, 2001). For the laymen, qualitative dimensions varied their perception, their vision composed of believes, value scales, etc. (Rowe et Wright, 2001). Then the dichotomy between the risk perceptions of experts and laymen fades (Starr, (1969) cited by (Heitz, 2009)) with the evolution of research on perception, notably through the integration of methods and concepts from the social sciences. This made it possible to integrate risk managers the integration of risk managers (an important link in the communication and information between the "experts" and "laymen") and all the actors (individuals, communities) directly related to any risk into research investigations. Perceptions were no longer seen as distortions but as representations of a reality influenced by many factors inherent to individuals (Kouabenan et al., 2007;Marris, 2001), regardless of their field of action. We then find the notions of belief (Douglas, 1967;Douglas et Wildavsky, 1983;Kunreuther et al., 1988) and responsibility used in the cultural or psychometric theories of classification of perceptions. In many studies, perception links individual judgment of risk and action (O'Riordan (1986), in others, perception goes beyond the individual, it is a social and cultural construction that integrates values, symbols, history and ideology (Sjöberg et al., 2004).
According to Van Den Ban and Hawkins (1996), perception is a process by which an individual receives information and stimuli from his environment and transforms them into conscious psychological acts. Two types of perception are distinguished in the human perception model: psychic perception related to the psychological situation of the individual and the sensory perception that is related to the senses (Yegbemey et al., 2014). Psychological perception is a function of functional factors such as experiences, notions of values, expectations, needs, opinions and socio-cultural norms. (Van Den Ban et al., 1994). Sensory perception depends on structural factors which our five senses (sight, taste, smell, hear and touch).
This model of analysis is applied as well to climate change, given that farmers do not adapt directly to the change in question but according to the manner in which they conceived and perceived. Climatic change is just a stimuli where the visible response is adaptation. Adaptation is a response to observable change responding to forces or perturbations such as climatic change (Smithers et Smit, 1997), an increase in the prices of fertilizers etc. In this perspective, the ultimate goal of adaptation to climatic change will be to moderate the negative impact of this change, to confront its consequences and eventually benefit from its new opportunities (Adger et al., 2003).
With the perception of climatic change being an observed stimulus, the adaptation decision of farmers will be a reasoned process. This leads us to the theory of reasoned action developed by Fishbein et Ajzen (1975). According to this theory, the strategy adopted by the producer will be determined by his behavioural intention to adopt.
With these theoretical considerations, the main hypothesis according to which the adaptation of the producers given the stimuli such as climatic change is only coherent regarding their conception and perception

Study area and data
The data was obtained from an interview using a sample size of 721 agricultural exploitations in the North and Far North regions of Cameroon. This survey was sponsored by the Institute of Agricultural Research for Development (IRAD) of Garoua, under "C2D agro-system's" project, and carried out from July to September 2014. It includes a questionnaire composed of open and closed questions which allows us to categorize farms according to socio-economic characteristics (age, level of education, the main activity of the family head, family size, surface area, etc.); production systems, (agricultural inputs, equipment, and labor force used); socio-institutional environment of the exploitation (access to loans, access to vulgarization, etc.); the perception of the exploitation leaders of climatic change and finally the different adaptation strategies to climatic risk.
We are now moving to a agro-ecological zone realized by Dugué et al. (1994). This zoning took into consideration the physical characteristics (climate and soil), demographical (density and population movement), the potentials and constraints, level of agricultural equipment, cultures in the regions, type of soils, and economic criteria (presence of markets, accessibility). This zoning culminated in seven (07) sub zones. In function of security dispositions and action zones of the project, four out of the seven zones were retained. In each of the chosen zone, a sample of 200 farms was obtained by random draw.
The reject rates for questionnaires are 7, 5% and 10, 5% 17% and 4, 5% respectively in the MOWO, SIRLAWE, ZERA and SANGUERE territories. It involves incomplete questionnaires, doubtful answers given by interviewees and finally the farmers not having as the main agricultural activity. The second one constituted only 5.2% of the sample and was thus rejected. Source : author's calculations using interviews ran in 2014

Treatment of perception data: Calculation of the perception index of climatic change in the sudano-sahelian zone
The calculation of the risk perception indices has been under many research studies (Glatron et Beck, 2008;Kunreuther et al., 1988;Lindell et Barnes, 1986;Slovic et Peters, 2006). We have been inspired by the psychometric paradigm which defines important factors to be integrated in estimation of indices. This method was also chosen because it provides a good translation of the representations that are qualitative data into quantitative data. The choice of our variables was based on the works of Kahneman et Tversky (1979). According to the psychometric paradigm, the evaluation of the perception was made thanks to notation scales fixed around nine dimensions permitting quantitatively translate the representation of a risk by the sampled individuals. Therefore, we have a voluntary character of risk-taking, the immediate effect of risk, the knowledge of risk, its controllable or reducible character, its recent character, its chronic or permanent character, the threat it represents, and its potentially catastrophic character (Sjöberg et al., 2004;Slovic, 1992).
The data collected does not permit us to consider the set of factors presented in the literature. In addition, the nature of risk requires us to ignore certain factors such as: the voluntary or suffered nature of the risk and the recent nature of the risk. Indeed, the climatic phenomenon is general and automatically applies to all people living in the same agro-ecological zone. Moreover, climate change is continuous and therefore cannot be analyzed as one-off phenomenon. However, on the basis of this literature, we identified two major groups of factors which seem necessary in the realization of our objectives. So we have:  The knowledge of risk: evaluated amongst others by the level of information on the evolution of the climatic phenomenon.
The question «according to you, what are the major changes that in the last 10 years in the climate of the zone?», permits the surveyed to directly express their knowledge of the climatic parameters of the zone. This is an open question and the answers are debatable and requiring re-coding.
We do not wish to highly intervene in the adaptation of the discussion so as to minimize any bias with respect to re-coding.
We chose to use a semantic grouping because individuals do not always use the same terms to express the same idea. This question enables the surveyed to reveal his optimistic or pessimistic vision of climate. For example, the recognition of an increase in temperature indicates a state of mind where surveyed sees the future climate. Besides, many other questions permitted to measure the knowledge on the variability of climate through the beginning and end of the rainy season as well as a Spatiotemporal variability of precipitations.

 The threat it represents and its chronic character
It measured among others by the apprehension that the population has on the consequences of the climatic phenomenon. The answers of the farmers with respect to the consequences of climatic change materializes the supportable or not-supportable character of climatic risks. The threat of climatic change necessitates the examination of the questions related to the consequences of climatic change on the social activities. Also, the "climate pessimists" hold that climatic change negatively influences activities. However, all responses which indicate a negative influence on the farmer's activities indicate the pessimism of the investigation given the climatic change.
Based on the study of Kahneman et Tversky (1979), we will define the necessary variables in the construction of our index. The methodology of construction of the composite indicator of perception of climatic change is based on the inertia approach with the help of multidimensional analyses. The choice of this technique is due to the fact that it permits to eliminate as much as possible arbitrary in the calculation of a composite indicator. The technique of factorial analysis which is more adapted in our case, is that of multiple correspondence analysis since all the variables are qualitative.
In order to generate the indexes of perception of climatic changes, we will proceed to the realization of ACM on the set of available and pertinent variables which characterize climatic changes as well as their consequences on the zone of study. To ameliorate our indicator, we shall proceed to the reduction of the number of variables resulting from the first ACM following a certain number of criteria. The principal criteria is that of ordinal consistence on the first factorial axe (COPA). This property consists of a partial indicator having its ordinal structure respected by the coordinates of its modalities on the first factorial axle. The realization of a second ACM with the remaining variables permitted to ameliorate the explicative power of the first factorial axle for the COPA principle was respected. Finally, we constructed from the results of the second ACM the composite indicator of perception of climatic changes with respect to the inertia approach.

Processing of adaptation data
After the questions related to perception, a series of two questions were used to define the variable « adaptation to climatic change ». The first question was: « Have you taken some measures to reduce the negative effects of climatic change on your activities? ». Afterwards, the strategies developed in the framework of adaptation to climatic change were reviewed through the question: "If yes, did you adopt one of the following measures?"

Choice of model
The Logit and Probit models are commonly used to analyze the choice of determinants or the decision of the producers to carry out adjustments or modifications (Gbetibouo, 2009;Hassan et Nhemachena, 2008;Maddison, 2007). According to the nature of the dependent variable, the multinomial models are equally used. These models are presented in the general form as follows: (1) = Where and Represent the decision of adaptation of farmer i and a set of demographic and socio-economic characteristics of the same farmer i respectively. While considering the hypothesis of the relationship perception-adaptation, the simplest manner of integrating the perception of producer (P) in the previous model is expressed as: However, still on the theoretical framework, perception appears as an endogenous variable (depends on a certain number of individual characteristics). The estimation of the second equation presents endogenous bias. In these conditions, the specification of two distinct models, that is to say an adaptation model (equation 3) and perception model (equation 4) appear as an alternative that will reduce the estimation bias: Where represents a set of demographic and socio-economic characteristics of the same farmer i; who could be identical or different from .
This new formulation which reduces the endogeneity device related to perception does not take into account the hypothesis according to which adaptation of farmers face with stimuli such as climatic change. According to Maddison (2007), perception is a prerequisite to adaptation. In other words, you need to perceive before adapting. The problem of endogeneity is no longer posed rather the problem of selection. Also as proposed Maddison (2007) The defined form is based on two sub-models: output or adaptation model where the dependent variable is adaptation (A) and the selection model where the dependent variable is perception (P). Considering j, the demographic and socio-economic characteristics related to farmer capable of determining its adaptation decision (noted characteristics ) on one hand and j' demographic and socio-economic characteristics related to the same farmer I and is susceptible of determining its perception (noted characteristics ) on the other hand, the resulting economic model is: In this model, the adaptation decision (1 = adapt ; 0 = do not adapt) of farmer i and its perception defined as a silent dichotomy variable (1 = perceive ; 0 = do not perceive) ; α and β are variables to be estimated; finally, u and v are the errors terms. Equation (6) becomes: where Z is a j-vector of the demographic and socio-economic characteristics which could influence the adaptation decision, Y is a j-vector of the demographic and socio-economic characteristics which could determine perception, U and V are the error terms following the normal distribution independently from X and Y, and A and P are linked by the selectivity link A if P > 0. Also, the dependent variable A is defined as: From these observations and specifications, the farmers' characteristics such as the possession of a secondary activity, the number of agricultural assets per household, experience in agriculture, ownership right over the exploited land, access to loans, membership in an organization and contact were introduced in the adaptation model, and experience in agriculture, sex, educational level and membership in a farmers organization were introduced in the perception model. In our empirical model, the dependent variable of the selection model is a binary variable and indicates if the farmer has a pessimistic or optimistic perception of climatic parameters in the zone. In order to render the scores in the dichotomy form, we will start from non-normalized scores. The dependent variable of the result equation is also binary and indicates if the farmer decides to adapt or not after perception.

Perception of climatic change  The knowledge of risk : evaluated amongst others by the level of information on the evolution of the climatic phenomenon.
This permits us to realized that climatic change is not only observed by the researchers because for more than two decades now the population has noticed a certain « deregulation » in the normal order of things as it was in the past. About 98.6% of the research carried out in the zone confirm to have noticed a considerable modification in the climatic conditions of the zone. These changes are seen locally through the indicators conceived by the farmers so as to establish a reference framework as presented in the following table. The 69.8% of the agricultural areas we interviewed had perceived changes through a reduction in the number of days rainfalls throughout the rainy season.
For some who haven perceived climate changes, these modifications are at the level of the raining season by: I) a late onset of the rainy season (71.71%) and an early end (55.89%). 54.09% of farmers think that the rain is poorly distributed during the rainy season leading to dryness. Equally, 46.19% of the investigations declares a bad spatial distribution of rain in the land. Generally, the analysis of these results permits us to see that in the study zone, there exist a deregulation in the starts and end of seasons having as consequences the reduction of the raining season and the extension of the dry season.

 The threat it represents and its chronic character.
Given the heterogeneity of the physical environment, the incidences differ from one topography to the other and the vulnerability of different units of the landscape depends on the degree of their exposure to climate. This heterogeneity of the physical environment as well-conditioned the distribution of the resulting effects and affected persons as we present in the following table. The analyses of scattered points were obtained in ACM. This axe explains 73% of the total scattered variables while the second axe has weak explanatory power (6.6%). This distinction of the first axe puts a particular phenomenon which climatic change. The analysis of ACM is mainly circumscribed to this axe describing the phenomenon of climatic change. In observing the first factorial plan, we observed that the variables translating a critical and undoubtedly state of the climate are found on the left and that indicating a normal or advantageous climate is found on the right. The higher the perception index of climatic changes, the more the individual is skeptical of changes and negative climatic effects and consequently an optimistic vision of climate. While a weak level of perception index means the individual has an understanding of negative changes with respect to climatic parameters as well as the negative effects of these changes in the zone. The individual in question, however, has a pessimistic conception of climate. The choice of the analysis in terms of the pessimistic or optimistic individual with respect to the risk of climatic changes is guided by the fact that the evaluation of risks is influenced by a certain number of beliefs on risks. These beliefs can be qualified as positive in the sense where it reflects the tendency of the individual to manifest optimism given their vulnerability to risks. On the other hand, these beliefs can be qualified as pessimistic when it entails a tendency of the individual to manifest fatalism with respect to their predisposition to succumb to risks.
In the study zone, these perception scores are found between -1, 21, and 2, 64.
In accordance with the different orientations of the variables on the factorial axes we posed the postulate according to which an individual has a pessimistic perception when his perception score is less than zero and found between -1, 21, and 0; contrarily, an individual will have an optimistic perception if his perception score is greater than zero and found between 0 and 2, 64.

Adaptation to climatic changes
In spite of a higher degree of intensity and/or the appearance of the new climatic constraints, the farmers have the attitude to composing with a certain climatic risk level using the adjustment means they have. 88.9% of farmers adopted at least a strategy to face the climatic mutations in the zone.

Perception and adaptation determinants of climatic change
The estimation results show that the model is globally significant at the limit of 1%, car Prob > chi2 < 0, 01. Rho is negative and significant at 5% which shows that an optimistic perception of climate in the region negatively influences the adaptation decision of the farmers. According to the perception paradigm (Adesina et Zinnah, 1993), this study shows the existence of a significant association between the perception of extreme climatic events and the adaptation decision of the smallscale farmers in the sudano-sahelian zone. The latter are consequently rational and integrate the evolution of their environment in the decision-making. In conformity with the theory, a producer in a risky environment will first proceed by evaluating the risk. Then he will choose protective measures Villa et Bélanger (2012) or he will choose to be protected against this risk if he thinks he can confront it. These results corroborate with those of de Gbetibouo (2009) in the Limpopo basin in South Africa and those of Kansiime et al. (2014) in Uganda which also show that adaptation to climatic change is an appropriate response to the different perceptions which the producers have on the evolution of their environment.
sensitized to the actual variability of climate as well as on the present and future consequences on agricultural production and on the immediate environment of man. To these sources of apprenticeship is added the relations existing between the producers and which serve as canals of sharing of experiences which can initiate the common adaptation initiative. Membership to an organization can then facilitate access to information and technical practices where positive correlations are observed. This result corroborates with that of Yegbemey et al. (2014) au Benin.

Carrying out a secondary activity
Carrying out a secondary activity is also positively and significantly correlated with adaptation decision to climatic change at the threshold of 5 % (0,01 < probability < 0,05). A secondary activity constitutes another source of revenue for the producers. The revenue from the secondary activity can as well be used for investment such as manpower, fertilizers, and pesticide, new varieties, etc. The expected effect, in this case, is similar to that of access to loans. Besides, the diversification of activities is also an adaptation strategy to climatic change (Gnanglè et al., 2012;Maddison, 2007;Yegbemey et al., 2013) which permits minimizing the risk in case of bad harvest and hence loss of revenue. In this sense, the producers who already have a secondary activity have a high probability of adapting to climatic change. 

Level of wealth
In spite of its low rate, the level of wealth is negatively correlated with the producer's decision to adapt to climatic change at the statistic threshold of 10 % (0, 05 < probability < 0, 10). This result which seems paradoxical to expectations can be justified by the fact that poor farmers are most vulnerable to shocks that affect their revenue than rich farmers. Also, faced with chaos like floods, some farmers adopt autonomous strategies (Smit et al., 2001), while striving to stabilize their revenue than its maximization. . According to this theory, the evaluation of risk shall be affected by the emotional state of the subject perceiving as well as by the emotional tone (positive or negative valence) associated with the event. This result corroborates with many other research works in the literature. According to Slovic (1999), many research works have shown that men try to judge risks in a weak and least problematic manner than women.

Zone effect
The farmers in the sub-humid zone are more likely to adapt than those in per urban; the coefficient of this zone is positive and Significantly linked to adaption choices, with a 1% chance of error.
In spite of the degree of intensity and/or the appearance of new climatic constraints, the adaptation decision depends on the significant manner of perception that farmers have on such changes. The farmers do have the attitude to be content with a certain level of climatic risk in the function of the way they use the adjustment means at their disposal. However, the latter generally face many strategies which they combine at their convenience to attend to their objectives.

Conclusion
The agricultural producers in the sudano-sahelian zone in Cameroon perceive climatic changes and develop diverse strategies to adapt. These changes are seen through rainfall and thermal disturbances, as well as by an increase in extreme events. The diversification of sources of activities, adjustment of cultural practices, and agricultural calendar is the principal adaptation strategies developed as a response to perceived climatic changes. Sex, membership in an organization, religious beliefs, age, and information sources are the main determinants of perception and adaptation decision of producers to climatic change. Particular attention was to be given to these different determinants in the context of the implementation of policies to promote agriculture in the context of climatic changes. Also, regular sessions to raise awareness, training, exchange, and sharing of knowledge related to future climatic conditions and the adaptation to climatic changes could constitute potential intervention tools, oriented towards producer's organizations through popularization services.