Smallholder farmers ’ perception of climate change and adaptation strategy choices in Central Ethiopia

Purpose – Smallholder farmers have always been profoundly the ﬁ rst to be impacted by climate change, and therefore, farmers understanding of climate change and accessibility to alternative adaptation strategies are crucial for reducing the effect of climate change. The purpose of this study is to assess the perception of farmers to climate change, adaptation strategiesanddeterminantsof adaptation choicein centralEthiopia. Design/methodology/approach – The study used data from randomly selected 240 farm households. Descriptive statistics were used to describe farmers ’ perceptions of climate change and adaptation strategies. Also, a multivariate probit model was used to identify the major factors affecting farmers ’ choice of adaptation strategiesto climatechange incentral Ethiopia. Findings – Smallholder farmers perceive climate change in the past two decades in response; the majority (91.47%) of farmers used adaptation options. Improved crop varieties andinput intensity, crop diversi ﬁ cation, planting date adjustment, soil and water conservation activities and changing of the crop type were used as adaptation options in the study area. A few of these strategies were signi ﬁ cantly con ﬁ rmed a complementary and supplementary relationship. The study identi ﬁ ed sex, family size, agroecology, climate information, crop- fail history and formal extension service as signi ﬁ cant determinants for farmers ’ adaptation choices as these variablessigni ﬁ cantly affected more thantwofarmers ’ adaptation strategiessimultaneously. Research limitations/implications – Farmers ’ choice of adaptation was highly constrained by institutional factors and all these identi ﬁ ed factors can be possibly addressed through a better institutional service provision system. It is, therefore, recommended that local administrators should explore the institutional service provisionsystemforabetterfarm-leveladaptationwhileconsideringdemographiccharacteristicsaswell. Originality/value – This study identi ﬁ ed factors affecting farmers ’ several adaptation strategies at a time and provides information for the policymaker to make cost-effective interventions for better farm-level adaptation practices. Different studies have been conducted to provide information on farmers ’ perception toward climate change, adaptation level and factors affecting the adaptation level of smallholder farmers by using Probit, Logit, Tobit and Heckman econometric models to enhance their adaptation level in a different region of However, the majority of the in the ﬁ eld of climate change in Ethiopia including literature reviewed for this study are largely documented based on information from smallholders farmers adaptation habits in the Northern and Northwestern part of Ethiopia and only a few from the Eastern part of Oromia region. Smallholder farmer ’ s perception of climate change existed adaptation strategies used by farmers, factors affecting adaptation level of farmers and their choice among the available alternative adaptation strategies are not still investigated in a different part of Ethiopia, including Central province. Thus, this study aimed to narrow the existed information gap of literature by providing information about smallholder farmer ’ s adaptation status and adaptation strategies choices in Central Ethiopia. Also, this study provides information about the possible supplementary and complimentarily association among adaptation strategies by using the multivariate probit (MVP) model, which is still scantily documented in the literature.


HHs = households; HHHs
= household heads; NAP-ETH = National Adaptation Plan of Ethiopia; and SARI = South Ethiopia Agricultural and Research Institute.

Background
Climate change affects the economic growth of developing countries due to its multiple interdependent effects on different sectors including, agriculture, health, forestry, fisheries and tourism (Kahsay et al., 2019;Lawson et al., 2019;Gedefaw et al., 2018;Weldegerima et al., 2018;Kompas et al., 2018;Kebede and Gizachew, 2017). Currently, the unfolding change in climate with its effect on temperature and rainfall affects all aspects of the food system including, food production and availability, access, quality, utilization and stability of food systems (Collins et al., 2013;Bryan et al., 2013); and it is also true in Ethiopia (NAP-ETH [Ethiopia's National Adaptation Plan], 2019). Developing countries with high levels of hunger are highly vulnerable to climate change impacts and have a low capacity to adapt. Specifically, smallholder farmers are in danger due to the multiple effects of climate change as they are susceptible because of their socioeconomic situation (De Souza et al., 2015). Improvement of smallholder understanding of climate change, the availability of alternative adaptation strategies and look for determinants of farmers adaptation capacity are among the strategic means of reducing climate change effects and also ensuring food security issues in most developing countries (Berhanu and Beyene, 2015;Wondim et al., 2018). However, farmers' decision to use adaptation strategies, adaptation level and choice among the available adaptation options have been affected by demographic characteristics, agro-ecological setting, socioeconomic factors and other institutional constraints (Paulos and Belay, 2018). In this regard, numerous national and international research institutions and other stakeholders have been engaged in providing alternative adaptation strategies to enhance farm adaptation capacity. For instance, the Ethiopian government made numerous investments to enhance the generation and dissemination of new agricultural technologies (NAP-ETH [Ethiopia's National Adaptation Plan], 2019).
Also, location-specific information about each adaptation strategy is required for different stakeholders to make positive interventions in the adaptation process (Kahsay et al., 2019;Yamba et al., 2019). Different studies have been conducted to provide information on farmers' perception toward climate change, adaptation level and factors affecting the adaptation level of smallholder farmers by using Probit, Logit, Tobit and Heckman econometric models to enhance their adaptation level in a different region of Ethiopia (Gedefaw et al., 2018;Paulos and Belay, 2018;Birhane et al., 2020;Debela et al., 2015;Habtemariam et al., 2016;Abrha et al., 2017;Addisu et al., 2016;Beliyneh et al., 2013). However, the majority of the literature in the field of climate change in Ethiopia including literature reviewed for this study are largely documented based on information from smallholders farmers adaptation habits in the Northern and Northwestern part of Ethiopia and only a few from the Eastern part of Oromia region. Smallholder farmer's perception of climate change existed adaptation strategies used by farmers, factors affecting adaptation level of farmers and their choice among the available alternative adaptation strategies are not still investigated in a different part of Ethiopia, including Central province. Thus, this study aimed to narrow the existed information gap of literature by providing information about smallholder farmer's adaptation status and adaptation strategies choices in Central Ethiopia. Also, this study provides information about the possible supplementary and complimentarily association among adaptation strategies by using the multivariate probit (MVP) model, which is still scantily documented in the literature.

IJCCSM
The Central part of Ethiopia is among the potential area suitable for crop production and smallholder farmers are vulnerable to frequent climate change challenges (Befikadu et al., 2019;Eshetu et al., 2016). As a result, enhancing the adaptation capability of smallholder farmers in the region is the primary means of increasing income, livelihood situation and reducing food insecurity problems for millions of smallholders farmers in a changing climate. The present study was, therefore, planned to elucidate farmers' perception toward climate change, factors determining the choice of adaptation options and their best combinations of adaptation options in Central Ethiopia. The information so generated will be useful in optimizing the future adaptation strategies in the study area.

Methodology
Description of the study area This study was conducted in Gurage Zone; Central Ethiopia situated 188 km south of Addis Ababa and 214 km in the southwest part of the regional capital city of South Nation Nationalities and Peoples Region of Ethiopia (Hawassa). Its geographical location is between 7°44' 46" and 8°28' 29" N latitude and 37°27' 30" and 38°42' 42" E longitudes ( Figure 1). Gurage administrative zone has 15 districts with a land size of 593,200 hectares, of which about 52% of the land has been cultivated, 13 0.4% are grazing, 9.9% are covered natural and plantation forest. The remaining 7.3% is unproductive (GZoLR [Gurage Zone office of Agriculture Livelihood Report], 2019).
The agro-ecological category of the Gurage zone has split into lowland (3%), midland (62%) and the highland agro-ecology zone (35%). Rain-fed agriculture is the main economic activity in all agro-ecological categories of the zone. Crop and livestock mixed farming systems are practiced in the majority of areas of the zone. Stimulant cash crops such as Farmers' perception of climate change coffee and khat are cultivated in the zone in addition to cereal (tef, wheat, maize and barley), horticultural (inset, Irish potato and cabbage) and pulses (faba bean and field pea) (SARI [South Ethiopia Agricultural and Research institute], 2017).

Source of data and methods of collection
The study data were collected from both primary and secondary sources. Primary data were collected from 340 farm households through the cross-sectional households survey method. The primary data consisted of demographic, basic socioeconomic characteristics, institutional factors and perceived change in climate situations and adaptation practices in the past two decades. The related kinds of literature in the field of climate change and researchers' experience in the study site were used as an input to prepare questionnaires. The study also used secondary data from journals, bulletin and governmental offices to supplement data from primary sources.
Sampling procedure and sample size Gurage zone was selected purposively because this region is the most vulnerable to climate variability and change among others in Central Ethiopia (Zelalem and Belay, 2019). The survey populations for this study were smallholder farm households who own or/cultivate less than 2 ha of land and heads of households were the units of analysis. In this study, Yamane (1967) was applied to estimate the required sample size from the study population. Accordingly, a sample size of 340 household heads was used for the study.
where N is the population size and e is the level of precision. The sample for this study is calculated by assuming a 95% confidence level, the study used this formula for the population, in which N = 2,274 with 65% precision. In this study, for the selection of samples, multistage sampling procedures were used. In the first stage, 14 districts of the Gurage zone were stratified into 3 categories of agroecological zones such as lowland, midland and highland. In the second stage, six districts (two each from stratum), namely; Maraqo and Abeshgye from the lowland; Cheha and Mesqan from midland; and Sodo and Ezia from highland were randomly selected. In the third stage, from the selected districts, a total of 6 kebeles[1] (one each from the district) were randomly chosen in this study. Finally, on the basis of probability proportional to the size sampling method, 340 farm households were randomly drawn from the selected Kebeles.

Method of data analysis
Descriptive statistics such as frequency, mean, percentage and standard deviation were used to summarize the socio-economic, demographic and institutional characteristics of households. It was also used to examine and describe perceptions of climate change and adaptation strategies used by smallholder farm households in the study area. An MVP model was fitted to evaluate different factors affecting the choice of adaptation methods and possible interdependence among these chosen adaptations strategies used by smallholder farmers.

IJCCSM
Analytical framework Smallholder farmers' adaptation strategy choice is based on the theory of rational choice, which assumes that smallholder farmers are rational and can rank alternative adaptation strategies for utility maximization. This study used random utility maximization modeling to conceptualize smallholder farmers' adaptation strategy choice. This model is predominantly applicable for representing discrete choice decisions such as adaptation strategy choice based on the principle that the smallholder farmers chosen the strategies, which maximizes his/her utility (Ben-Akiva and Boccara, 1995).
On the basis of demographic, institutional factors, socio-economic characteristics and relevant issues, farm households will have an average utility level from each alternative adaptation strategies choices. In this case, smallholder farmers are expected to make a comparison on marginal profit and cost considering on the utility gained from using a combination of adaptation strategies. The utility from each adaptation strategies cannot be directly observed rather it can be indirectly obtained from the choice made by the farmer that revealed, which adaptation strategies provide the better utility subject to its determinants (Greene, 2012;Djalalou-Dine et al., 2015;Tarekegn et al., 2017). For this study case, smallholder farmers choose an alternative adaptation strategy if utility (U) of that alternative is greater than other alternatives, expressed as U in > U jn V j = i, where j are the different choices of adaptation strategies from the available choice set C n and smallholder farmers is labeled n. Also, the researcher cannot distinguish the complete aspects of smallholder utility function, and thereby a representative utility function V jn = V (x jn , S n ) is considered, with x nj V j the characteristic of the choice and S n , some characteristic of smallholder farmers (Kamba et al., 2015). Consequently, the utility is decomposed into deterministic (V ij ) and random (« ij ) component: where « jn indicates the factors that affect utility but are not observed by the researcher, and therefore, are not included in V jn . The probability that a smallholder farmer chooses a certain alternative adaptation strategy is expressed as follows.
P ijC n ¼ Pr U in ! U jn ; 8 2 C n ; À À (3) Empirical estimation of multivariate probit model Farm households possibly use more than one adaptation strategy simultaneously to confront climate change consequences. The adaptation strategies used by farmers share the same background and scenario and there is a possible correlation among these adaptation strategies as they share the same background and scenario. This interaction has its own role in policymaking but for the possible correspondence among different adaptation strategies have clearly been investigated. An MVP model is a qualified discrete choice model to estimate the possible multi-entity interactions among choices in a real-world situation (Chen et al., 2018). The model is efficient in investigating the effect of several independent variables on dependent variables simultaneously and allows the error terms to correlate. Given a set of explanatory variables used in MVP, the multivariate response assumes to have arisen from a multivariate normal distribution and falls within a certain interval (Ademe et al., 2019). Following Greene (2000), the MVP model assumes that the response variable has K distinct binary response. Let Y* refers to unobserved latent variable; i = 1 . . . n be the individual observation; K =1. . . K is the available options of binary response and X i is the list of Farmers' perception of climate change explanatory variables. Thus, Y ij = (Y i1, . . ., Y ij ) represent the K dimensional vector of observed binary response taking value either 0 or 1 on the i th households and Y* ij = (Y* i1, . . ., Y* ik ) ' represents a K variant normal vector of latent variables; b = (b 1, . . ., b k) represents a matrix of the regression coefficient, e I represents a vector of residual error distributed as a multivariate normal distribution with zero mean and variance of one; e I~N (0, R) where R represents the variance-covariance matrix and also the relationship between Y* ij = Y ij is given as follows: The likelihood of the households to adopt the given adaptation methods data is then obtained by integrating over the latent variables: Interpretation of MVP results is based on marginal effects rather than the parameter estimates. Thus, the marginal effects of independent variables on the propensity to adopt different adaptation options were calculated by the following equation (Charles et al., 2014): Accordingly, this study identified five main different adaptation strategies (ADS) choices used by smallholder crop-producing farmers. Following the specification in equation (2) and the information indicated above, the MVP model deployed for this study is estimated as follows: Y i2 refers improved varieties and input use intensity (8) Description of variables and sample statistics The dependent variable, farmers' adaptation strategies to climate change, was a binary choice dependent variable and five adaptation strategies were identified [equations (3)- (7)]. Explanatory variables used for the study were chosen by considering previous works of literature related to the field of climate change and the experience of the researchers in the study area. Thus, the description of explanatory variables used in the analysis and measurements expressed in the table below:-

Results and discussion
Description of variables and sample statistic Farmers' perceptions of climate change Smallholder farmers were not directly examined on what they perceive toward climate change, rather they were asked about what they had observed the trend of the main climate indicators, particularly rainfall, temperature and drought for the past two decades. The result confirmed changes in raining time and rainfall problem had observed in the past two decades, such as about 49.41% perceive rainfall amount had decreased, about 4.71% perceive as it had increased, about 18.53% had not observed the trend and about 27.35% perceive rainfall as it had remained the same. Furthermore, they revealed an onset rainfall problem (51.18%), rained after the perceived time (42.06%), cessation rainfall problem (71.18%) and end up before the perceived time (61.18%) (Figure 1). For this reason, smallholder farmers have been challenged by unpredictable rainfall nature as crop farming largely depends on rainfall. Thus, pre-production, pre-harvest and post-harvest production activities depend on rainfall characteristics and, in turn, determines overall farm-level productivity.
In the past two decades, smallholder farmers of Central Ethiopia perceived and observed temperature variability and change, such as about 71.98% of the respondents perceive as the temperature had increased, about 4.42% as it had decreased, about 27.35% revealed as it had unchanged and about 18.53% of respondents not observed the change. Also, the study considered drought as an important climatic indicator variable and about 69.41% of the Farmers' perception of climate change respondents confirmed the frequent occurrence of drought, but about 43.24% and 6.18% of the respondents perceived the frequency of drought occurrence as decreased and remained unchanged, respectively. The remaining 17.65% had no clear information about temperature change (Figure 2). The result indicated that the farming business in the study area is heavily influenced by the frequent occurrence of drought and increased temperature. In addition, flooding was another climatic indicator that smallholder crop-producing farmers perceived changes in the past two decades. However, about 7.94% of the respondents perceived the frequency of flood as increased, 43.24% decreased, 38.8% unchanged and the remaining 10% of the respondents had not observed the flooding pattern  Farmers' adaptation strategies to climate change Table 3 presents the adaptation options that were used by smallholder farmers across different agro-ecologies of the study area. In response to climate variability and change, the majority (91.47%) of surveyed households have practiced adaptation strategies to reduce the effect of climate change. The main adaptation strategies used by smallholder farmers in the study area were crop-diversification (51.47%), improved crop varieties and input use intensity (62.65%), adjusting planting date (45.59%), soil and water conservation (49.12%) and changing of the crop type (50.59%). Agro-ecologically, soil and water conservation activities were largely practiced by households in the midland agro-ecology as about 55.09% of the respondents used this as an adaptation measure. About 53.71%, 28.57% and 17.71% of the surveyed households in the  Farmers' perception of climate change midland, highland and lowland agro-ecology, respectively, adopted crop diversification strategy. Improved crop varieties and input use intensity are widely practiced by farm households in the midland and highland agro-ecology. Also, about 38.5% of households found both in midland and highland agro-ecology adopted improved crop varieties and input use intensity, compared to sample farm households in the lowland agro-ecology that was only about 23%.
Furthermore, about 40% and 35.48% of the farm households in the lowland and highland agro-ecology, respectively, used adjusting plant date adaptation strategy. Sample households exist in the midland agro-ecology widely used four from the identified five adaptation strategies, namely, the use of crop diversification, improved crop varieties and input intensity, soil and water conservation and changing the crop type. However, adjusting plant date adaptation strategy is not well-practiced in the midland agro-ecology. This could be due to the raining season in lowland agro-ecology is shorter than highland and midland agro-ecology and this might influence households' decision to practice the planting dates adjustment adaptation option. This result is in agreement with Abrham et al. (2017) in Ethiopia and Dasmani et al. (2020) in Ghana that revealed crop diversification is largely practiced as an adaptation by the majority of smallholder farmers but soil and water conservation practices and adjusting planting date adaptation strategy also practiced well (Table 2). Notes: Likelihood ratio test of rho21 = rho31 = rho41 = rho51 = rho32 = rho42 = rho52 = rho43 = rho53 = rho54 = 0: chi 2 (10) = 23.9732 Prob > chi2 = 0.0077 * and *** means statistically significant at 10 and 1% probability level, respectively IJCCSM Empirical model results Correlations among adaptation strategies. Table 3 presents the correlation coefficients of the climate change adaptation strategies from the MVP estimation in this study, which test either the adaptation strategies are independent or not. The likelihood ratio test of independence of the error terms of the different adaptation strategies used for this study is rejected at a 1% level of significance (chi 2 (10) = 23.97, p < 0.0077) ( Table 3). The result shows the presence of mutual interdependence among five adaptation strategies and supports the appropriateness of an MVP model for this study data set. Thus, smallholder farmers of Central Ethiopia practiced more than one adaptation strategy concurrently, and a few of these adaptations choice correlated significantly. Majority of the pairwise correlations among these adaptation strategies are positively correlated implying the presence of a complementarily relationship. The pairwise correlations are significant in 3 of the 10 pairwise correlation coefficients. For instance, soil and water conservation practices correlate positively with crop diversification strategies and showing that undertaking soil and water conservation practices increase the cultivated land for diversifying different crops. Similarly, the use of improved crop varieties and input intensity correlated with changing crop type positively, indicating that accesses to improved varieties and other input influence decisions on what and how much to produce. This result was true at a 1% level of significance. Also, farmers might replace the old crop varieties by new varieties or completely change the crop type for more additional benefit and production if they have accessed improved crop varieties and inputs. On the other hand, surprisingly, changing the crop type adaptation strategy negatively correlated with adjusting the planting date and the relationship was significant at 10% (p = 0.087) level of significance. This could be because of crop production in Ethiopia being largely rainfall dependant and seasonal; and the majority of crops are grown at the same period. This indicates farm households in Central Ethiopia used planting date adjustment and changing crop type as substitute adaptation options; as households who used changing crop type adaptation are less likely to use planting date adjustment as an adaptation option and vice versa.
The model results also revealed that the use of crop diversification, adjust planting date, improved crop varieties and input use intensity, soil and water conservation and changing of crop type adaptation strategies in the study area increases by 51.7%, 62.4%, 45.6%, 49.1% and 50.03%, respectively. Also, households are more likely to use all the adaptation strategies jointly than fail to use all these adaptation strategies. Thus, the likelihood of households using all the five adaptation strategies simultaneously was about 6.71% while the likelihood households failed to use all adaptation strategies were about 5.63% (Table 4).
Factors affecting farmer's adaptation strategies choice.    indicated that the explanatory power of independent variables on a dependent variable are satisfactory because the authors failed to accept the null hypothesis because the likelihood function from an MVP model was significant (Wald x 2 = 308.93 with P < 0·001). Among the hypothesized variables in determining the choice of adaptation strategies the sex of a household head, age category, family size, literacy status, farming experience, access to extension service, the crop failure history, accesses to credit, distance to research, distance to market, accesses to climate information, agro-ecological category significantly determine the choice of adaptation methods among farmers. Sex of household head showed a positive and significant effect on crop diversification (p < 0.1), the use of improved crop varieties and input use intensity (p< 0.05), and also adaptation by soil and water conservation (p < 0.05) adaptation strategies (Table 5). Based on the estimated marginal effect from an MVP model presented in Table 6, male-headed households were 9% more likely to diversify crops, 10% more likely to use improved crop varieties and input intensity and 11% more likely to conserve soil and water. This is associated with the fact that male-headed households have a better opportunity to exchange ideas with other farmers and development agents about climate change adaptation techniques. Also, male-headed households are more likely to engage in labor share work groups, which might influence their involvement in labor-intensive adaptation options such as soil and water conservation practices. This is in line with studies by Harvey et al. (2018) in Central America and Ademe et al. (2019) in Ethiopia as male-headed households are more likely to use crop diversification and using crop varieties adaptation strategy. Also, the finding that revealed male-headed households practiced adaptation methods requiring labor is in agreement with this study (Paulos and Belay, 2018).
The age category of the household head showed a positive and significant effect on improved crop varieties and input intensity at p < 0.05 (Table 5). As compared to older households, those households within the adult age group were 11% more likely to use improved crop varieties and deal with input intensity (Table 6). It might possibly because of adults are more energetic, self-motivated and actively ready to use new technology including improved crop varieties, fertilizers and, in turn, for securing their future lives. This result is consistent with earlier findings by Sisay et al. (2019) in Ethiopia and Kamba et al. (2015) revealed a significant relationship between household head age (not to be an older) with the use of improved crop varieties.
The available family labor within a household showed a positive and significant effect on households' decision to use crop diversification (p < 0.05) and soil and water conservation practices (p < 0.01) adaptation strategies (Table 5). The estimated marginal effect for the family size variable showed that one additional family labor in man equivalent increases the probability of diversifying different crops and using soil and water conservation practices by 3% and 4%, respectively (Table 6). Farmers' preference of adaptation measures might depend on the accessibility of labor for preparation, production and harvesting time. Different crops require similar action at a time and, in turn, failure to do so might result in a possible loss from their produce, and therefore, households with a better source of labor within a family were more likely to diversify crop production and conserve soil and water. This finding opposes the study by Ademe et al. (2019) in Ethiopia who revealed large family size decline the probability of using portfolio diversification but, in agreement with the finding by Birtukan and Abraham (2016), who revealed having large family size are more likely to use crop diversification and soil and water conservation practices.
The literacy status of a household head in formal education significantly determines the use of changing crop type (p < 0.01) as an adaptation strategy (Table 5). The result showed that households who accessed formal education were 19% more likely to change the crop   (Table 6). Literate farmers may possibly have better awareness about cost, productivity, market value, soil fertility, crop character and also agricultural uncertain events such as climate change. On the contrary, uneducated farm households might relatively passive and old-fashioned to take counteractive measures. This result is in consonance with the finding by Adégnandjou et al. (2018) in South Benin that revealed literate farmers reached with information and makes more competent choices among adaptation strategies. Farming experience showed a significant and positive effect on adaptation to the use of improved crop varieties and input intensity and changing crop type as adaptation strategies at p < 0.05 and p < 0.1, respectively ( Table 5). The probability of using improved crop varieties and input intensity and changing crop type adaptation measures increases by 8% and 6%, respectively, for each additional year of farming experience (Table 6). Observation, model farmers and formal training approach enhance farmers' knowledge of production or risk management strategies and the experience might enhance farmers' knowledge of the right adaptation strategies. The result showed the use of improved crop varieties and input use intensity, and more experienced farmers use changing crop type adaptation strategies. This result is in line with a study by Zerihun et al. (2018) in Ethiopia that revealed the positive correlation of farming experience with improved crop varieties use and input use intensity adaptation strategy, respectively. Access to extension service had a significant and positive influence on all adaptation strategies except crop diversification. The result showed that compared to these households heads who did not receive extension service, households who received extension service are more likely to use of improved crop varieties and input use intensity and adjusting planting date an adaptation strategy (both at p < 0.01). Similarly, it would also influence the choice of soil and water conservation practice and changing crop type adaptation strategy (both at p < 0.05) [ Table 5]. The availability of extension advise increases the probability of using improved crop varieties and other inputs by 14%, adjust planting date by 19%, soil and water conservation practices by 10%, changing crop type by 11% in southwest Ethiopia ( Table 6). The study result by Abrham et al. (2017) in Ethiopia and Charles et al. (2014) in South Africa revealed the significant relationship between extension service and farmers' choice of most adaptive measures, including the use of improved crop varieties; adjusting planting date, soil and water conservation and changing planting date adaptation measures are consistent with this finding.
The crop failure history of households showed a significant and positive correlation with the most identified adaptation choices, including crop diversification (p < 0.01), the use of improved crop varieties and input intensity (p < 0.05) and adjustment in crop date (p < 0.05) ( Table 5). The results showed that households who faced crop failure problems were 18% more likely to diversify crop production, 12% more likely to use improved crop varieties and inputs and 12% more likely to adjust planting dates (Table 6). This is because most farm households are relatively risk-averse and need to use measures that can protect or cut from the previous risk. The result by Zerihun et al. (2018) that revealed the positive association of crop failure history with changing crop date adaptation strategy is in line with this result. Surprisingly, households with crop failure history was expected to use soil and water conservation practices and change the crop type adaptation strategies, but the result showed insignificantly correlated and probably the problem for their crop failure might not be associated with flooding, drought and site selection problem.
Accesses to credit services showed a significant (p < 0.01) effect on the use of crop diversification strategies to lessen climate change effects (Table 5). The availability of credit Farmers' perception of climate change services increases the probability of using crop diversification by 13% (Table 6). Credit helps to strengthen households' purchasing power of agricultural inputs, and augments the likelihood of households to diversify their crop production. The study finding by Arun and Jun (2019) in Nepal revealed the positive correlation of credit service, and using improved crop varieties agrees with this result.
A distance of households residence from the nearest market showed a negative and significantly associated with the use of improved crop varieties and input use intensity (p > 0.01) ( Table 5). The computed marginal effect indicated that as the distance from the nearest input and output market increases by one kilometer, the probability of using improved crop varieties and other inputs decreases by 0.2% (Table 6). A village market is the most relevant source of information for farmers and those farmers found in the different communities probably share their experience in such a market. On the other hand, those farmers far away from the market might not receive timely farming information, including improved crop varieties and the transaction cost of improved crop varieties might exert influence on farmers' use of improved crop varieties. The result is in agreement with a study by Abrham et al. (2017) in Ethiopia and Diallo et al. (2020) in southern Mali revealed that market accesses increases the likelihood of changing the input use intensity as an adaptive measure.
Distance to a research center has a significant (p < 0.01) positive effect on the use of changing crop type adaptation strategies (Table 5). Increase in the households' residence from the research center by 1 km increases the probability of changing crop type adaptation by 0.3% (Table 6). Thus, households who are close to research service centers can use updated information on the productive crop in their locality and thereby more likely focus on these researched crops. However, those households who are far from a research center may not receive researched information and they would possibly change the crop type to combat climate change rather than simply stick to researched resistant varieties. The result is consistent with a finding by Adimassu and Kessler (2016) that revealed the negative relationship between distance from the farming service center and change crop type adaptation strategies.
The availability of climate information like an early warning and preparedness affects households' adaptation strategies choice significantly. Thus, households who took climate information more likely used crop diversification, adjust planting date and soil and water conservation practices at p < 0.01 as compared to counterparts (Table 5). Compared to household heads who did not receive climate information, household heads that got climate information were 12% more likely to diversify their crop production, 14% more likely to adjust planting date and 13% more likely to conserve soil and water (Table 6). Smallholder farmers might be limited by capital to run their farming business and prefer to choose adaptation strategies largely managed at the household level that did not require much cost of adoption. A study by Chalmers et al. (2017) in Malawi revealed the positive and significant effect of climate information on the choice of most adaptation strategies is in covenant with this result.
The agro-ecological category (midland) showed a positive and significant effect on the use of crop diversification and soil and water conservation practices but negatively correlated with adjusting the planting date at p < 0.01 (Table 5). The marginal effect result estimated from the MVP model showed that compared to households living in the lowland agro-ecological category, farm households in the midland agro-ecology were 37% more likely to diversify their production, 37% less likely to adjust planting date and 42% more likely to conserve soil and water. Also, households living in the highland agro-ecology were 11% more likely to diversify crop production, 15% more likely to use improved crop IJCCSM varieties and manage input use, 17% less likely to adjust planting date and 17% more likely to conserve soil and water (Table 6). Thus, households in different agro-ecology have different experiences of farming as they differ in the crop is grown, farming system and vulnerability to climate. For instance, households in the lowland agro-ecology were more likely to use adjust planting dates because crops in such agro-ecology acquire a short duration. However, such adaptation measure is risky to practices it in midland and highland agro-ecology. A study by Birtukan and Abraham (2016) that revealed the positive and significant relationship of the midland agro-ecological zone and the use of crop diversification support this finding.

Conclusions
The study concludes that the surveyed farm households in the Central part of Ethiopia observed changes in three climatic indicators, including rainfall, temperature and drought in the past two decades. In response, the majority (91.47%) were used adaptation methods to reduce the adverse effect of climate variability and change. The adaptation strategies used by crop-producing smallholders are interrelated to one another as the coefficient of the parameter significantly showed complementarities and substitutability relationships. The study indicated farmers of choice of adaptation were not largely affected by socio-economic factors rather demographic factors (sex, family labor, age groups, farming experience and literacy status) and institutional factors (access to credit, access to extension service and distance from market and research center) were largely determining farmers' choice of adaptation.
The policy should aim at looking for farm financing, strengthen extension service (climate change inclusive, considering the technical language usage and location-specific), ensuring farmer's accessibility to market infrastructure (road accesses, market accesses and transport service) and developing farmer's experience (training, model farmer's field visit and guest speaker). Moreover, developing socio-economic, resource and farm management strategies to farmers (family labor management, farm decision-making, use of adopting technology and a potential loss from exaggerated social life and expenses), encouraging women involvement in the affairs of the community and continuous awareness to farmers on the existed and updated farming technology would be grateful in enhancing adaptation practices in the Central part of Ethiopia.