Determinants of Rural Women’s Adaptation Measures to Climate Change in Yeki Woreda, Sheka Zone of South Western Ethiopia

: Climate change is a global challenge that burdens all of humanity. However, the world’s poor, the majority of whom are women, are encumbered disproportionately. Adaptation is, therefore, considered as an important response, especially for those groups who are most exposed to and affected by its adverse effect. Hence, this study assessed the determinants of women’s choices of adaptation strategies to climate change in Yeki Woreda, Sheka Zone of South Western Ethiopia using a multinomial logit model. A three-stage sampling procedure was followed in selecting 150 women farmers from three kebeles. The result revealed that productive labor force, farm size, farming experience, wealth status, access to climate information, credit service usage and extension visit were found to influence women’s choice of adaptation strategies positively and significantly (P ≤ 10%). Therefore, development interventions, policies and supportive services should be designed at a different level to ensure effective access to climate information and affordable credit schemes for women households to enable them to plan and adapt in the face of the changing climate.


Methodology 2.1. Description of the study area
Geographically, the study district is located between 70 3' N latitude and 350, 0' E longitude in the south-western parts of Ethiopia. The maximum and minimum temperature of the area is 30 °C and 15 °C respectively, with an altitude of 1200 m ASL. The mean annual rainfall of the area is 1591 mm which extends from April to December and the area is located under hot to warm humid lowland agroecology (TNSRC, 2015).
Yeki Woreda has a total population of 134,519, of whom 68,895 are men and 65,624 women; (CSA, 2007). The livelihood of the rural population is based on growing cereals (like maize, sorghum, rice, etc), enset, coffee, and spices (such as Turmeric, Black pepper, and cardamom). The land cover composition of the Woreda includes perennial agriculture (42.4%), annual agriculture (34.3%), and forests (22.5%) (TCPE, 2010). Landholding sizes range from 0.5 ha to 5.5 ha, with a mean size of 1.3 ha of cropland, 0.6 ha coffee, 0.3 ha spices, 0.4 ha grazing and 0.3 ha plantations (Yeki Woreda Agricultural and Natural Resource, 2010). Forest cover in Yeki Woreda declined rapidly from to 73% by 1973 to 32% by 2010 forest as a result of the expansion of agricultural lands and settlements, and coffee farms. Consequently, the landscape changed from a forest dominated to non-forest dominated during the last four decades (Ibid).
Source: Own construction, 2017 Figure 1. Map of the study area

Sampling techniques
Three stage sampling procedures were followed to identify the study kebeles and sample respondents. In the first stage, out of the total 22 rural kebeles in Yeki Woreda, three (Bechi, Beko and Ermich kebeles) were selected randomly because of the biophysical, socio-economic, and socio-cultural homogeneity in each kebeles. Prior to selecting the sample respondents, the list of households (sampling frame) was collected in collaboration with kebele administrators of the respective kebeles. In the second stage, male-headed households (MHH) and femaleheaded households (FHH) were identified from the sampling frame. In the third stage, a total of 150 women households were selected from the three kebeles based on systematic random sampling procedure proportional to the size of the population in each group using the following simplified formula (Cochran, 1977). * . * . . . Source: Cochran's (1977) Where, t is the value for alpha level = 1.96, (p) (q) are the estimates of variance = 0.25 and d is the acceptable margin of error= 0.08.

Type and Methods of Data collection
For this study, both primary and secondary data were employed, which are quantitative and qualitative in nature. Accordingly, primary data was gathered from carefully selected women households using household survey, focus group discussions (FGDs) and Key informant interviews (KIIs). Prior to the actual survey, a pre-test was conducted on 15 non-sample respondents and necessary modifications were made on the structure of the questionnaire. On the other hand, secondary data were obtained from governmental offices and a review of published and unpublished literature relevant to the study. The type of data collected pertains to personal characteristics such as demographic, socio-economic and institutional factors of respondents, the type and choices of adaptation strategies employed to climate change, particularly on the change in rainfall and temperature of the area.

Multinomial logit model specification
When there is a dependent variable with more than two alternatives among which the decision maker has to choose (i.e., unordered qualitative or polytomous variables), the appropriate econometric model would be either multinomial logit or multinomial probit regression model. Regarding estimation, both of them estimate the effect of explanatory variables on a dependent variable involving multiple choices with unordered response categories (Greene, 2000).
However, MNP is rarely used in empirical studies due to estimation difficulty imposed by the need to solve multiple integrations related to the multivariate normal distribution (Chilot, 2007). Moreover, the MNL model was selected not only because of the computational ease, but also MNL analysis exhibits a superior ability to predict adaptation strategies and pick up the differences between adaptation strategies. In this study, therefore, a multinomial logit model specification was employed. This model makes it possible to analyse factors influencing women's households' choices of climate change adaptation strategies in the context of multiple choices. Following Green (2003), the MNL model for a multiple choice problem is specified as follows: Where pij is the probability representing the ith respondent's chance of falling into category j or (it is the probability of household i s' choice of adaptation strategies from category j), xi is predictors of response probabilities; e is the natural base of logarithms; and is the parameters to be estimated by maximum likelihood estimator (MLE). The estimated equations provide a set of probabilities for the j +1 choice for a decision maker with characteristics. For the identification of the model, we need to conveniently normalize by assuming =0 (Greene, 2003). Therefore, the probabilities are given by: The marginal effects (δij) of the characteristics on the probabilities are specified as

Description of variables and Research hypothesis
In this study, the dependent variable was a women's choice of climate change adaptation strategy. It was identified by categorizing the sample households into adaptation strategy groups based on their choice; improved crop varieties (A), soil and water conservation practices (B) and crop diversification practices (C). However, farmers usually adopt more than one adaptation strategy at a time (Tessema et al., 2013;Kidanu et al., 2016

Women's Adaptation Strategies to Climate Change
Based on the findings of the study, women households employed different farm-level adaptation measures in response to climate change in the study area over the last ten years. Accordingly, the majority of woman households (72.1%) practised one or more adaptation strategies. The major strategies include the use of improved crop variety like early maturity and drought-resistant crops (12.7%), use of improved crop variety and soil and water conservation practices (12.7%), use of soil and water conservation practices and crop diversification practices (10.7%) and all adaptation strategies (19.3%) ( Table 9). The findings of the study were also in agreement with the response of focus group discussants as they pointed out that:  (turmeric, rice, coffee, enset, etc.) to diversify the risks associated with the changing climate (Woman farmer in FGD, 11.03.2009 E.C). In addition, women households practised both biological and physical methods of soil and water conservation practices. Among the biological methods, planting banana in the border of farms, trenches to divert runoff, mulching, tree planting (Gravellia, Korch, Sesa, Wanza etc.) were common in the area. On the other hand, women households practice soil bund and faniagou as a physical soil and water conservation strategies. Thus, the result of the study was found inconsistent with the responses of the key informant interviewee as they gave evidence that multiple adaptation strategies to climate change are commonly employed by women households in the study areas.

Determinants of women's choices of adaptation strategies to climate change
The results of the MNL model indicated that different demographic, socio-economic and institutional characteristics influenced women's choice of adaptation strategies to climate change. The result further indicated that among fourteen (14) hypothesized explanatory variables, seven (7) variables were found significantly influenced women's choice of adaptation strategy to climate change in the study area. These were; productive labor force, farm size, farming experience, wealth status, access to climate information, credit service usage and extension visit. Productive labor force: The finding indicates that as the number of the productive labor force within a household increased by one individual, the probability of implementing all adaptation strategies increases by 9.1% at less than 10% probability level. Some adaptation options are labor intensive and hence, women households with large productive labor force are more likely to share labor and as a result, reduce labor constraints to undertake each adaptation measures. This is inconsistency with the findings reported by Urgessa et al. (2014). Farm size: The model result depicts that a one unit increment in farm size of the household results in an increase in the probability of adapting soil and water conservation practices by 0.2% and all adaptation practices by 43.09% at less than 5%, and soil and water conservation practices + crop diversification practices by 11.6% at less than 10% probability level over doing nothing. Large farm size offers women households more flexibility in their decision-making process, more opportunity to take up new practices on a trial basis, and more ability to deal with risk. This result is in agreement with the results of Gebrehiwot and Van der Veen (2013).
Farming experience: Farming experience of women household had a positive and significant influence on adopting improved variety + crop diversification strategies and soil and water conservation practices + crop diversification practices at less than 5% probability level. The possible reason is that experienced women households are more likely to have more information and knowledge on the changes in climatic conditions and they can easily adjust themselves to climate change stresses. This result confirms the findings of Gbetibouo (2009), Maddison (2006 and Esiobu and Onubuogu (2014).
Wealth status: The probability of resource medium women households in adapting soil and water conservation practices is greater by 0.1% and all adaptation practices by 44.7% at less than 5% probability level than resource-poor households. Adaptation is a costly activity and those women households with better resources are able to buy new irrigation technologies and other important inputs they may need to change their practices to suit the forecasted and prevailing climatic conditions. This result confirms the findings reported by (Dasgupta et al, 2010).
Access to climate information: Women households' access to climate information increases the likelihood of using crop diversification strategy by 0.7% at less than 5%, improved variety + soil and water conservation practices by 8.8%, soil and water conservation practices + crop diversification practices by 19.4%, and all adaptation practices by 35.7% at less than 1% significance level. Women households who had access to climate information will be conscious of the changes in climatic conditions and have higher chances of taking adaptive measures. The finding of this study is in line with the arguments conveyed by Maddison (2006) and Onubuogu and Esiobu (2014).
Credit service usage: It had positively and significantly influenced women households' choice of climate change adaptation strategies. As woman households are a customer of credit service, the probability of adapting improved variety + soil and water conservation practices increases by 49.6% at less than 1% probability level. The probable reason could be with more financial resources at their hand, women households are able to change their management practices in response to the changing climatic condition. This result is in agreement with the findings of Hassan and Nhemachena (2008) and Fosu-Mensah et al. (2010).
Extension visit: Increasing the frequency of extension visit increases the probability of women households to adapt improved variety by 27.9% and soil and water conservation practices by 0.1% at less than 5% and improved variety + soil and water conservation practices by 21.3% at less than 10% probability level. The possible reason could be extension service assist women households to take climate change adaptation strategies through advice and the provision of information on how to deal with climate variability and change. This observation is in line with Ndambiri et al. (2012) and Maddison (2006). Table 4. Parameter estimates of MNL models of women' choice of adaptation strategies.

Conclusions And Recommendations
For rural women households in Yeki woreda, farming is not only a way of life, but it is also a means of livelihood. However, they have been affected by the impacts of climate change through the reduction of yield, the occurrence of human diseases and the reduction of water sources. In order to improve the ability of women households to adjust to the ongoing and future climate change, the need to understand their adaptation strategies at the local level is important. From the findings of this study, that the majority of women households employed a variety of farmlevel adaptation practices. Among these, the use of improved varieties like early maturing and drought resistant crops, a combination of the use of improved varieties and soil and water conservation practices (both biological and physical conservation practices), diversification of crops in the form of intercropping and growing different crops at different farmlands.
A range of factors influences women households choice of climate change adaptation strategies. Among these, farming experience, farm size, wealth status, access to climate information, credit service usage and extension visit influence choice of adaptation strategies of women significantly and positively in the study area. In some cases, a wide range of institutional and technological support enhanced the local adaptation strategies of women households in the study area. Labor saving technologies, for example, played an important role to reduce their labor constraints and thereby respond to the consequences of climate change. Moreover, ensuring affordable credit schemes for women households is important as this helps them increase their ability and flexibility to different adaptation strategies in response to the changing climate. On the other hand, local agro-meteorological stations should be established to the level that they are efficient in providing reliable, trustworthy weather forecasts pertinent for women households. The role of government, NGOs and other responsible bodies is, therefore, imperative in improving women households' level of awareness on adaptation strategies through training and making the extension services (climate-related information and advice) more accessible to women households.