1. Introduction
The interaction and exchange between the natural environment and human activities affect each other [
1] and outcomes from this interaction are important for both sides in terms of sustainability. Global warming and climate change are occupying the world’s agenda as important problems. Many areas of the world experience variability as a part of their normal climate over both short and long periods [
2]. One of the biggest challenges is dealing with uncertainties in the future climate and it is critical for nations, economies, and societies. This situation requires a new way of planning [
3] and new approaches. Climate change is predicted to make the existing problems worse in many regions [
4], and it will change features of the environment and ecology [
5]. This results in the deterioration of resources, which negatively affects the welfare of societies. Current management practices are increasingly being assessed in terms of the ability to overcome climate change and uncertainty [
6]. Agriculture plays a substantial role in food security. It is known that climate change and water-based problems will negatively affect agricultural production, which leads to significant yield losses [
7], especially in dry farming areas [
8,
9] and semi-arid and arid areas, with significant impacts on the livelihood of rural communities, especially smallholder farmers [
10]. This will be more important for societies where agriculture is predominant in the economy and provides the most employment [
11]. Are the farmers aware of climate change and the associated risks? What are the factors that influence this awareness and farmers’ perceptions? How will they adopt? What will be the adaptive practices? Are they willing to pay for adaptation? As adaptation becomes more tightly integrated into the range of responses due to climate change, understanding how knowledge of climate change impacts on farmers and how vulnerabilities can effectively be used is necessary both to direct research and to support action [
12]. It may also have long-term impacts on the adaptation capacity of a society in terms of livelihood development and by resulting in a reversed development [
13]. Information needs of policy-makers for adaptation evaluations are being reversed in beginning with the adaptation problem in its decision context rather than with climate projections [
14].
Attitudes and expectations play a noteworthy role in society, development, environment, and resource economics. Risk is an unlikeable and subjective concept that may potentially result in individuals’ loss of values, and risk perception differs based on many factors. These factors contribute to an understanding of the overall risk perception, although they do not explain it all [
15]. Human risk perception varies across individuals, households, and societies [
16]. Preferences with respect to risk, time, and the environment shape the decision-making processes of individuals and can be used as an important source of information for policy design development [
17]. When risk perception increases, it may result in increasing the conservative attitudes of individuals due to livelihood considerations.
Numerous recent studies indicate that there is an increasing tendency of climate instability in Turkey [
18,
19]. A precipitation feature of Turkey is represented by a complex pattern of spatial, seasonal and inter-annual variability that is subject to extreme climatic variations, resulting in recurrent droughts and floods that negatively affect resources [
20] and the degradation of water-related ecosystems, causing serious problems for agriculture. These lead to loss of crop yield, shorter growing seasons and salinity. Turkey will be at a higher risk of experiencing meteorological droughts as well as these droughts turning into agricultural and hydrological droughts in the near future [
21,
22]. Seasonal variability of precipitation characteristics in Turkey is relatively higher than the annual average, which is around 5%–65% in the southern part of the country [
23]. Şanlıurfa is a city where agriculture is extensively carried out and it is located in the southeastern part of Turkey. Agriculture is the main base of the economy of Şanlıurfa [
24]. The number of abnormal to exceptionally dry years was 17 in Şanlıurfa, and that is based on 64 years of data from 1951 to 2015. Five of the last 10 years saw drought [
25]. These figures indicate that there is a causal relationship between climate change and dry farming, and thus precipitation, in another word, water.
Beliefs, attitudes and cultural influences affect farmers’ behaviors [
26,
27], and an understanding of farmers' beliefs, concerns and willingness to adapt is necessary for political, economic and social action towards risks arising from climate change [
28]. Most of the literature on climate change in Turkey deals with impact studies [
18,
19,
20,
21,
22,
23,
25] and none of the studies focused on the risk perceptions from the farmers’ perspective in the GAP (The Southeastern Anatolian Project) Region and Şanlıurfa. This study is aimed at determining farmers’ risk perception towards climate change in the dry farming areas of Şanlıurfa and their willingness to pay (WTP) for adaptation practices and factors that contribute to the understanding of them.
3. Results and Discussions
All questionnaires were conducted with male farmers due to the rural family structure. The average age of the farmers was calculated as 43.45 years, and the average amount of land owned was 17.11 ha together with 22.11 years of farming experience. The descriptive statistics of the model are given in
Table 1.
The farmers who have a risk perception towards climate change, a dependent variable, accounted for 53%, and those WTP for adaptation practices to reduce the effects of climate change numbered 61.6%. Then it was asked how much they want to pay per hectare. The initial value was selected as 120 TL/ha which was the minimum fee for cereals in the irrigation area of Şanlıurfa at the survey time. Then, starting from 10% and increasing with multiples of 10%, up to 100% of the increased current price was randomly drawn for each farmer and he was asked to accept to pay or not this amount for adaptation practices. The cross-tabulation table of risk perception and WTP is given in
Table 2.
The Chi-square test statistics of the cross-tabulation are given in
Table 3. The results indicated that a significant relationship exists between risk perception and WTP by
p < 1%.
There are 8.6% (40 people) of the farmers who do not have a risk perception but tend to have WTP for adaptation practices. The average WTP of 61.6% of the farmers was calculated as 164.4 TL/ha, that was 15.8% of their income derived from agriculture. These results indicate that more than half of the farmers have a risk perception and 8.6% more of the farmers tended to pay without having a risk perception. This outcome could be used by policy-makers for adaptation practices. All the respondents were classified as “yes” by the model and the percentage of verification was 53% in the classification. Then the next step was run and the model coefficients of the Omnibus tests are given in
Table 4. The results indicate that the independent variables were making meaningful contributions to the model.
The Omnibus test showed the existence of statistical significance between dependent variables and independent variables at the level of
p < 1% in the model. The model summary is given in
Table 5.
The Cox and Snell and Nagelkerke R
2 values indicate the amount of variance explained by the model. The variance of risk perception towards climate change was explained to be 49.3% by Cox and Snell and 65.9% by Nagelkerke R
2. The test result of Hosmer and Lemeshow is presented in
Table 6.
The test assessed the compliance of the logistic regression model as a whole and the result indicated that it was insignificant (
p > 0.10); it means that a sufficient level of model-data fit existed. The classification table that was obtained from the logistic regression is given in
Table 7.
The percentage of verification was increased from 53% to 85.4%, which indicates that variables were making significant contributions to the model. Initially, more variables were used in the model these were given in the descriptive statistics, and significant ones were selected after the first run in terms of contribution to the model. This is because of the increased demand on the robustness and the reliability of the estimates [
49]. The marital status and additional income besides farming variables were found as insignificant and omitted from the model. In fact, 90% of the respondents were married and 37% have an additional income that was up to 20% of their farming revenue. The most appropriate multivariable, which are age, education, farming experience, agricultural income, land amount, agricultural credit usage, household, non-agricultural manpower and water perception, were selected as variables in the model. The model was run for the second time and results are given in
Table 8.
The results are obtained by a logistic regression analysis that is based on odds ratios which, presented for dummy variables, are interpretable in relation to their respective reference group where less than one unit change indicates the existence of a negative relationship, which means having less risk perception towards climate change. A more than one unit change shows the existence of a positive relationship, which means having greater risk perception than their respective reference group. Interpretations can be done according to the coefficient of the exponential logistic regression as well. In this case, the formula (exp
b −1) × 100 could be used in order to understand how an increase or decrease will happen based on the superiority rate of a dependent variable based on a unit change in the independent variables [
50,
51,
52]. The interpretation of this research is done according to the general literature surveys. There is a relationship between age and risk perception (
p < 1%). There are many studies showing that age is one of the important indicators and influencers of risk perception by individuals [
10]. The youngest ones (Gr.1) have 28.5 times greater risk perception regarding climate change as compared to the reference group. The middle-aged (Gr.2) have seven times and the upper-middle-aged (Gr.3) have 3.1 times greater perception in a positive way. Age has an effect on the attitude of individuals. The preferences and behaviors are related to expectations and experiences that influence the formation of different value priorities. Environmental concerns and value priorities tend to become increasingly important as the younger generation replaces the older generation [
42] and this can be explained by the level of education and awareness.
There is a relationship between the income level derived from farming and risk perception (p < 1%). Risk perception is increasing with the increasing income level of the farmers from agriculture, and vice versa. The Gr.1 and Gr.2 have an 8.1 and 8.2 times greater risk perception as compared to the reference group (lowest income derived from farming), respectively, in a positive way. Higher income groups are more sensitive to climate change due to their income, which is mainly based on climate in the dry farming areas. It is statistically significant (p < 1%). There is a relationship between the amount of land owned and risk perception (p < 1%). Farmers’ incomes are dependent on the amount of land owned. A bigger amount of land means more revenue for the farmer. The Gr.1 and Gr.2 have greater risk perception as compared to the reference group by 5.4 and 7.7 times, respectively, in a positive way. The result is significant (p < 1%).
There is a relationship between farming experience and risk perception (
p < 1%). Groups with less farming experience which are the youngest ones (Gr.1) have greater risk perception as compared to the reference group by 21.5 times in a positive way. It was unexpectedly high. Gr.2 has a 2.8 times greater perception in a positive way as well. The outcomes are consistent with age-related results and it is statistically significant (
p < 1%). The younger ones have less experience about how to deal with and overcome the effects of climate change. The older farmers are more experienced about agriculture, farming and the potential effect of climate change and know how to overcome this situation. They are also more conservative as opposed to younger farmers and may consider climate change religiously rather than by evidence. The level of suspicion is dependent on demographics, personal experience and values, among other factors [
41].
There is a positive relationship between agricultural credit usage and risk perception (
p < 10%). Agricultural credit users have 1.9 times greater perception as compared to the non-credit users. Generally, farmers pay their debts at harvest time. Thus, climate change directly affects the income of these farmers. Therefore, they have more concerns about risk. The result is statistically significant. There is a positive relationship between household numbers and risk perception (
p < 10%). When the number of households is increasing, the perception of risk increases by 1.1 times. This result arises mainly from livelihood and household members are considered as manpower in agriculture. Crowded families will be more affected. There is a negative relationship between non-agriculture manpower and risk perception (
p < 5%). It means that a one-unit increase of non-agricultural power decreases the odds of having risk perception by 60.3%. It is hypothesized that people’s perception of risk is directly dependent on environmental conditions and the employment status of the household [
43]. These farmers have income besides agricultural activities, so they will be less affected by the loss of prosperity due to climate change as opposed to loss of manpower in agriculture.
There is a relationship between education levels and risk perception (
p < 1%). When the education level decreases, there is a decrease in risk perception. The opposite is true, too. Education is an important indicator of life quality and attitudes. There are some studies showing that different cultural groups based on different socioeconomic and education levels differ in their risk perception considerably [
44,
45,
46,
47]. Gr.1, Gr.2 and Gr.4 have 2% odds of having risk perception as compared to the reference group, in a negative way. Gr.3 has 3% odds of having risk perception as compared to the reference group, in a negative way, as well. The results of Gr.1 and Gr.2 were understandable. These farmers were conservative, the least educated and older as compared to the other groups. This point can be explained with beliefs. The results of Gr.3 and Gr.4 were unexpectedly lower. This can be explained by the rural culture and poor quality of education in the rural areas.
Water perception is defined as the availability of more or less water than is needed in the field by the farmers because the farmers are adversely affected in both cases. The flood, drought, and other measures of climate variability are perceived as influential, typically negatively, to livelihood strategies [
40]. There is a relationship between the understanding of water and risk perception at the significance level of
p < 1%. This is a predictable result but the values are lower than expected because risk is a function of water availability for the farmers in the dry farming areas. The strongly agree group selected as the reference group is mainly composed of the young and those less experienced in farming, with greater income and education levels. Gr.1 and Gr.2 have 5% and 7% odds of having risk perception as compared to the reference group in a negative way, respectively. The results were significant. Water is one of the most important inputs of agriculture, not only for dry farming but also in irrigated areas. Water-related issues always receive high priority among the farmers in the GAP Region. The studies conducted in GAP-Harran Plain have shown that age, land, farming experience, income and education levels are significant in water management [
48]; WTP for safe irrigation water is found as 71.69% more than the existing price [
53], 2.23-fold greater than current price under shortages [
54], and they believe in the necessity of irrigation training for sustainable usage and accept payment by 59% [
9] in GAP, the Harran Plain, and Şanlıurfa.