Synergy between adaptations and resilience of livelihood from climate change vulnerability: A group-wise comparison of adapters and non-adapters

The similarities, differences, and contradictions regarding climate change adaptation and resilience by academics and practitioners have already been documented. It is the need of time to set new precedence by observing the adaptations and resilience as tools to respond to the climate variations. This study analyzed the influence of climate change adaptations and synergy between resilience from livelihood vulnerability and adaptations. A field survey of 489 farming households is conducted with the help of a well-structured questionnaire from four districts of the south part of Punjab province of Pakistan. This study uses the Endogenous Switching Regression model for the sake of analysis. The outcomes of the study reveal that age, education, family size, total land, and seed price have significant linkage with the adoption of adaptations. The synergistic effects of adaptation and resilience are also visible here as the adaptations factors are significantly contributing towards yield, per capita income, poverty, and poverty gap of the respondents. This study suggests the provision of proper education and smart technology to help in enhancing the adaptive capacity of farmers. More imperatively, adaptations to climate variations can be concluded as a remedial tool for resilient livelihood. It is believed that the present study can be considered as a guide for future research on other regions of Pakistan and neighboring countries.


Introduction
Climate change (CC) is one of the swiftly spread phenomena across the globe since last century and livelihood of residents of the planet is at risk [1]. One-third of the world population is directly or indirectly facing the heat of these variations [2]. The vulnerability of CC is exerting pressure on the livelihood of poor farming communities who are already on the brinks of more vulnerable to CC than the wheat crop concerning current and future farm production. [33] evaluated the effects of temperature and precipitation on wheat productivity and found the significant association among variation in temperature, precipitation, the area under cultivation, water, CO 2 and crop production. [34] estimated the impact of CC on wheat production of mountainous areas of Pakistan and found the negative influence of temperature on the productivity of the wheat crop.
CC is not only impacting the farm productivities but also exerting the pressure on the livelihood of rural populations [35]. Farmers' income is sensitive, as loss of income opens up a path for poverty and food insecurity [36]. Natural disasters also have negative impacts on rural households' welfare [4,37]. It can be concluded from the given literature that CC has negative impacts on farmers' welfare. In lieu of the above discussion, it feels necessary to understand the importance of the welling of farmers' livelihood. There is a need for the advancement in the literature to validate the performance of CC adaptations being practiced like the application of smartphone for better farming, resilience factors, and synergy existing between them. Evaluating the synergy between the said phenomena will help to understand the feedback mechanism of these strategies.
Adoption of adaptations against CC has an imperative role in compensating the farming returns [13,23,38,39]. Mitigation could not be meaningful in the case of developing economies, as these cannot go toward the reduction of carbon emission due to gradual expansion in the industrial sector and urbanization. Advancement in the technology in agriculture is often assumed as an adaptation in the era of CC [40]. This study takes farmers' responses toward predicted future CC risks and adjusting to predominant climate vulnerabilities through good practices (education, smartphone etc.) to have a resilient livelihood, as an adaptation [41]. Two types of adaptations techniques are being practiced globally, off-farm, and on-farm [42], and this study considers these both types. It is assumed that the choice of opting adaptations is based on regional and agro-ecological characteristics [43]. An adaptation measure is meaningful if it is cost-friendly for the resilient livelihood beyond climate variability [44]. According to an estimate, 300 dollars per household could be in surplus if a farmer adopts the adaptations [45]. Better crop revenues due to adaptations help in improving the material wellbeing and make farmers more resilient, and ultimately make them eligible to mitigate the risk associated with climate variations [46]. Diverse package of adaptation measures helps in attaining resilience to offset CC losses [47].
Resilience can be a prompt recovery [48] and in the present context, it is supporting farmers to recover from CC vulnerabilities. Here, in this study, we have taken farmers as the main stakeholders, who are vulnerable in the form of their yield and income losses, and worsening poverty situations. In the current scenario, farmers cannot get resilience without taking the remedial measures [49] and CC adaptations can help farmers to bounce back to their normal position.
A handsome amount of literature presents relevant contributions and most of these studies are problem and region-specific, and a research gap still exists. Several studies have focused just on the CC vulnerabilities like [3] or adaptation impacts like [11] those are conventional approaches. Although there is a long and multidisciplinary history of scientific research associated with adaptation and the definition of adaptation has varied by fields and practice [50]. This paper extends the existing literature by developing synergy between adaptation and resilience to respond properly to the variations in the context of agricultural vulnerability to climate. This study sets a new precedence by observing adaptations as a tool to combat livelihood vulnerability and developing synergy between CC adaptations and resilience through comparison of adapters and non-adapters groups. Estimation of association between adaptation and resilience will help to plan for and respond to, current and future climatic variability (see Fig 1 for details). Table 1 describes the brief review of available literature by discussing time-period, data type, the universe, methods used, and the main results of studies.
This study has tried to quantify the synergy existing between CC adaptations and resilience from the livelihood vulnerability. As it is clear from the above table that previous studies mostly discussed the CC adaptation based on households' perceptions or secondary data, and to date, no study has investigated farmers' adaptive behavior and allied synergy with CC resilience that could influence their choices. Further, this article has also attempted to answer the following research important questions, is there any significant variance between the farm incomes of adapters and non-adapters? Either, socio-economic indicators of adapters are getting better than the non-adapters, or not? Considering the cited research questions above, this study has two-fold objectives. In the first, the influence of CC adaptations on the wellbeing of rural households is quantified and secondly, we estimated the synergy between adaptations and resilience of rural households from climate variability. Outcomes from this study could contribute to an understanding of how to better coordinate the relative strengths and contributions of adaptations with the resilient behaviours to cope on-ground challenges of climate variations, as measures of increased resilience enhance adaptive capacity and decrease the vulnerability of farmers. After the essential part of the introduction and review of the literature, the remaining paper is arranged as follows: Section 2 elucidates the methodology and data used in the study. Section 3 and 4 are about assessed results and their discussion, and finally, section 5 concludes the problem under discussion based on outcomes and provides policy lessons for the solution of the issue.

Data
To ascertain the objectives of the study, the farm-level household survey is conducted in the rural areas of Punjab province. We selected Punjab as the study universe as it has significant importance for Pakistan's economy, particularly, its agricultural share in national Gross Domestic Product. Therefore, four districts namely Bhakkar, Khanewal, Multan, and Muzzafargarrh from the southern part of Punjab province of Pakistan are taken in the sample. The selected sampling units have diverse climate conditions and cropping patterns. Household heads (HHH) or representatives of farm households are interviewed for the data collection. As the population is heterogeneous, this is why multi-stage stratified random sampling technique is employed for the data gathering. In the first stage, four districts are selected and then in the second stage, two Tehsils from each district are taken. In the next stage, four villages from each tehsil are selected. Sixteen growers are randomly selected from each village. Cross-sectional data employed for the study were collected in June-July 2019 with the help of a team of enumerators who were trained before the survey. A pilot survey was conducted to remove the discrepancies and loopholes to produce a better quality of data. In actual, 512 farmers were selected for the interviews and out of this sample, 489 farmers were considered as valid respondents.
As this study is about the economic implications, it is why Ethical clearance is not taken. In Pakistan, ethical clearance body (National Bioethics Committee) issues clearance number to basic sciences experiments oftenly, containing human or animal tissues etc. However, authors and enumerators have taken verbal consent from the study participants before the conductance of the survey. Participants were briefed that their data would only be used for the study purpose and they agreed to give the required information. The authors declared that they did not have any conflict of interests.
However, we faced rejection from a few farmers but those were replaced. A brief and wellstructured questionnaire was developed and farmers were asked about; socio-economic and farm characteristics, CC awareness, and vulnerability according to their perspective, institutional facilities, and CC adaptations. Information on farm households, agricultural practices, production and costs, access to extension, social networking, overtime climate-related changes, and allied risks, adaptations to climate change, access to credit, farm and household assets, other income sources, and other were collected during the survey. The study employed both qualitative and quantitative data collection for a deep understanding of the topic. Datasets show that farmers are practicing more than one adaptation measures. Study has taken two types of the farmers, one group is using the adaptation measures to cope with the climaterelated issues and consider them as adapters, and another group of respondents is not using these techniques (non-adapters). However, the use of a mixture of on and off-farm adaptations is also reported. In the study area, the main adaptation techniques being practiced include crop diversification, use of the hybrid seed, and cultivation of supplementary crops to support the input cost of the main crop. The questionnaire used for the study purpose has been provided in the supporting information (S 1 Appendix. Questionnaire).

Conceptual framework
This study is primarily centered on the expected utility theory which states, a decision could be made based on the expected gain from participating adaptations in response to CC. Conceptual framework of this study given in Fig 2 that is based upon a further extension of Driving force, Pressure, State, Impact response (DPSIR) sustainability framework developed by [65].
Here, CC is a basic driver that exerts pressure on the livelihood of rural households; uneven rainfall is the state, which can have an impact on the livelihood of rural households in the form of a reduction in the farm yield, loss of income and poverty elevation. Response to this phenomenon is mitigation or adaptation to CC. Farmers take a decision, as they have the constraint to resources, and mitigation may not the appropriate strategy and adaptation becomes meaningful in this scenario. Adaptations are based on the expected utility, if utility from adoption is more than the utility from non-adoption, a farmer definitely would switch to an adaptation.
The actual amount of utility is hard to observe. So, a dummy for utility would be used with the value equal to 1, if a farmer adopts adaptation to CC, and its value would be 0 in the rest of the cases. Let us assume, the expected utility follows a latent variable A x 1i denotes the expected utility gained by adopting adaptation [49][50][51]. Whereas, A x 0i denotes the expected utility gain if he does not adopt. Therefore, the decision can be modeled in the index function as in the following Eq 1, Where A x i is a dummy for expected utility, Z x i a is a vector for the parameter of observed variable and � i is error term following the normal distribution. Here, the utility is expected to gain from adaptation in the form of an increase in the yield, income, reduction of poverty and resilience form vulnerability of climate. It is assumed that if a farmer adopts the adaptation then he may gain welfare, and resilience from vulnerability.

Endogenous Switching Regression (ESR)
ESR is a parametric approach, which addresses the problem of selective biased and missing unobserved characteristics, which influence the decision to adopt the adaptation to CC [66][67][68][69][70]. It follows the two-stage analysis. In the first stage, the model estimates the selection equation to quantify the impact of determinates of adaptation. In the second stage, the outcome variable is estimated including inverse mill ratios to address the selective bias. By using the ESR model, we face two different regimes (Eqs 2 and 3), Here Y i 1 is the outcome of household if he adopts, and Y i 2 is the outcome if he does not adopt, Z i is vectors of observed explanatory factors including socioeconomics characteristics of CC indicators, and institutional facility that influences the outcomes. β i is the parameter or coefficients to be estimated, ε 2 is the disturbance term. Following [38,66,71,72], we used full information maximum likelihood to estimate ESR and its selection equation for the outcome are given below.
In Eq 4, ω i is the optional weight of observationι, ρ 1 μ = σε 1 μ i /σε 1 σμ i is the correlation coefficient of error term of regime one, and selection equation, ρ 2 μ = σε 2 μ i /σε 2 σμ i is the correlation coefficient of error term of regime two and selection equation in addition to it if ρ 2 1 μ and ρ 2 2 μ have alternative signs.

Conditional expectations
Average treatment on treated (ATT), average treatment on untreated (ATU) and heterogeneity effect are estimated by conditional expectation [73] (Eqs 5 to 8): Here, E(Y 1i |A i = 1,Z i ) and E(Y 2i |A i = 0,Z i ) represent the farmers who adopt the adaptation and do not adopt, respectively, and these indicate the actual expectation. E(Y 1i |A i = 0,Z i ) and E (Y 2i |A i = 1,Z i ) represent the farmers who do adopt the adaptation if they do not adopt the adaptation and if those who do not adopt the adaptation do adopt the adaptation, and these represent the counterfactual expected outcome.
Hence, ATT can be derived as: Moreover, ATU can be specified as, Yield has been taken in the form of maund per acre. Per capita income is calculated by dividing total household income per day by household size. Poverty status has been computed with the help of income (PKR) threshold given globally. The poverty gap is calculated by following Foster-Greer-Thorbecke, 1984 methodology [74]. The poverty gap is used to estimate the difference of poverty line with an actual income of a member of the society. All statistical analyses are performed in STATA software.

Results
This section presents the results of the empirical estimation of determinants of CC adaptations, and trends of explanatory variables. Analysis of this article is divided into three parts. In the first, descriptive analysis is presented, the second part of the analysis is about the outcomes of ESR modeling and the last portion gives the conditional expectation investigations. Variables description of all explained and explanatory variables are given in Table 2. Whereas Table 3 depicts the summary statistics, and the mean difference between adapters and non-adapters are given in Table 4. Farmers' perceptions about CC vulnerability are given in S 2. Datasets used for the study are also provided in supplementary information as S 3. Table 5 shows the results of likelihood ESR for the impact of CC adaptation. This table has four subpanels that describe the results of four different analyses by taking yield, per capita income, poverty status and poverty gap as dependent variables, while the same explanatory variables have been used for all four analyses, respectively. The results for the synergy analysis are given in Table 5.

Adoption of adaptations
Results showed that several explanatory variables would increase the likelihood of farmers' decisions to adopt the adaptation to CC. Determining the overall effectiveness of adaptation solutions in agriculture is challenging because it is impossible to accurately enumerate and model all economically feasible options [60]. In general, adaptations have a significant association with age, education, farming experience, smartphone, total land, institutional facility, and farmers' knowledge about CC. Consequently, age and smartphone positively affect the decision to adopt with statistically significant values (see Table 5 for details). Similarly, education has a positive significant association with CC adaptation. In contrast to it, the coefficient of household family members depicted a negative association with the dependent variable. Moreover, the area of wheat cultivation shows a highly significant relationship with the dependent variable. Variable related to institutional facilities (i.e. access to credit and agricultural extension) also have a positive association with the CC adaptations.

Impact of CC adaptation on yield
The coefficient of age is not statistically different from zero for the case of CC adaptors. In contrast to it, age showed a negative significant impact on the yield of non-adapters. Similarly, the coefficient of time HHH spend on the farm has a positive sign for the adapters but negative in case of non-adapters. Results showed that seed price would have a significantly positive correlation with the adopting adaptation. Moreover, using a smartphone for farm adaptations showed positive and significant impact with the wheat yield of adapters, in contrast to it, its negative association has been observed for the case of non-adapters. Results showed that education could have a positive significant impact with yield in both cases of respondents. Furthermore, the numbers of livestock and total land showed significant association with the dependent variable. Hence, it can be concluded from the results of Table 6 that adaptation shares much in common with resilience in preventing the harmful impacts of climatic variations, as ATT is 5.25 Kg if a farmer chooses to adopt the CC adaptation for the yield and ATU would be 3.29 if he does not adopt.

Impact of CC adaptation on per capita income
Results of per capita analysis depicted that family size is negatively related to per capita income of a household at 1% level of significance, for both, adapters and non-adapters. Another   (Table 6) showed that ATT is 9.95 and ATU is -2.514.

Impact of CC adaptation on poverty
The finding of poverty analysis given in Table 5 implies that the age of HHH would negatively influence the poverty status of non-adapters. In addition to it, the family size would have a statistically significant and negative relationship with poverty. Another variable, time spent at the field, show a negative impact on poverty for both cases of adapters and non-adapters. Furthermore, both variables level of education and total land, depict a highly significant association with the dependent variable with negative signs. Impacts of synergistic effects of adaptation and resilience become visible when we make a comparison of ATT of adapters and nonadapter (Table 6), ATT is -0.324 in case a farmer who does not adopt the option of 'chose to adapt' and ATU is -0.286 in case if he chose the option 'not to adopt the CC adaptation'.

Impact of CC adaptation on the poverty gap
For the case of the poverty gap, findings showed that there is a negative association between the time given to farm by the HHH and the poverty gap. Total land and livestock are statistically significant at 1 and 5% level of significance for adapters and non-adapters, respectively. The coefficients of variables, education and availability of irrigation facility, are statistically different from zero in the case of non-adapters, with negative values. Moreover, farmers who

Discussion
This section provided the rationale and discussion of selected variables' outcomes. The results obtained appeared to be robust because most of the estimated parameter coefficients showed the expected signs. An in-depth analysis of the study explores that overall; our results are consistent with the previous literature. In the case of the coefficient of the education level of respondents, its value specifies its positive association with adaptation. Accordingly, an increase in the number of years of education of farmers would increase the willingness to accept the technological advancements in farming operations. Our results indicate that farmers who are more educated would be more productive and efficient, and eventually, they would generate more farm income and overcome their poverty as compared to the less educated ones, and this argument is in line with the findings of other relevant studies [e.g. 23,37,51,[54][55][56].
The negative signs of family size in case of CC adaptations, yield, and per capita income are probably due to an increase in the household members that would decrease the likelihood of adoption of adaptation and these results validated the outcomes of previous studies like [55,75]. Whereas, positive signs of the coefficients of family size in case of poverty and poverty gap indicate that increasing family size would increase the poverty and ultimately it would make difficult for the farmers with more family members to reduce the poverty gap relatively, and similar argument has been presented by [76].
Furthermore, results suggest that usage of smartphone for farming purpose increases the possibility to adapt to cope with CC. We hypothesized that farmers utilize this smart technology to explore the ongoing scientific advancements, information of changing climate and selfprojection of future climate risks. In addition to the above, smartphones provide updated to the farmers about several adaptation strategies being applied to the agriculture sector around the globe. Coefficient of smartphone use is statistically different from zero and has a positive sign in case of adapter but negative for non-adapters, showing that adapter uses smartphone to get an update on climate risk and farm-level adaptation measures. In contrast, it is assumed that non-adapters are conventional and proportionately less efficient, and does not much rely on such technologies. Further, the smartphone showed a positive significant impact on per capita income headcount in the case of adapters. Per capita income headcount has strong linkage with the crop productivity. Meticulously, it can be said that an increase in the use of smartphones can help in increasing the mean net farm returns, income, and poverty reduction, and results of [77] are consistent with our findings.
Moreover, the findings of the study showed that an increase in age would increase the chances to adopt the CC adaptations. Inversely, the negative sign of the coefficient for wheat's yield specifies that with the increasing age, the capacity to work also goes down and these arguments are maintained by [51,55,68,78], conversely, outcomes of [79] opposed this. Results show that total landholding and time spent by the HHH on-farm could have a positive and significant impact on the yield of adapters, implying an increase in the values of these both parameters would increase farm productivity overtime. Similar results have been observed for the case of total land for per capita income of headcount for both categories of farmers i.e. adapters and non-adapters, and through this finding our study validate the outcomes of a similar study by [37].
Results further showed that conditional expectation and ATT are statistically significant. The value of ATT indicates, those farmers are well off who have been treated by CC adaptation than those who are untreated. In conclusion, the study shows that overtime climatic variations are significantly exacerbating poverty and inequality. Thus, policymakers are well recommended to include the adaptations as potentially contributing factors, when designing policies to alleviate poverty and inequality. In general, findings of the study depicts that there is deep interaction between CC adaptation and resilience from livelihood vulnerability, as adapters are getting better farm outcomes and they are more well off than the non-adapters. The key determinant of individuals, households, or communities' adaptive capacity is to reduce risk, cope with, and adapt to increased risk level with farm resource portfolio. There are close linkages between vulnerability and resilience, and resilience is basically about the expanding and sustaining their farm resource [64,80]. It is noted that if farmers who are adopters of adaptation, had not adapted, would be more vulnerable to climate variabilities than their current economic state. Hence, adopting adaptation is a meaningful tool for resilience from livelihood vulnerability.

Conclusions
The present study analyzes the synergy between CC adaptation and resilience from livelihood vulnerability. The study used data from the four districts of the south part of the Punjab province of Pakistan. An intensive field survey of farming households is carried out. ESR model is employed for the analysis purpose. Results of the study reveal that education, usage of smartphone, and total land holding of the farmer are the main factors influencing the likelihood of adaptation to CC and above-said factors are also contributing positively towards yield, per capita, income and poverty indices found.
It can be concluded from the results of this article that education is the key factor in developing human capital. An increase in the level of education is indirectly helping the growers in enhancing farm net return through improved yield and income, and alleviation of poverty. Education builds the capacity to make a rational decision at the right time, hence they adopt the adaptation [81,82]. Consequently, educated farmers are more productive and better off in monetary terms than their counterparts. This study also concluded that family size is one of the dilemmas, and increasing family size worsens the welfare gains. Furthermore, the study concludes that landholding and using a smartphone for updated agriculture-related information have a notable role in building adaptive capacity and a farmers' resilient livelihood. It can be said from the study findings that smartphones, adaptive capacity, and farmer resilience are deeply interlinked with each other. Therefore, a farmer using a smartphone would have more knowledge about the new adaptation techniques and use them to moderate farm loses [61].
The most imperative conclusion of this study is the ATE established in the study. Results of ATE on treated showed that farmers who adopt are better off than the other due to potential gains from adaptation measures, and 14% of farm production of adapters is higher than the non-adapters that could be due to adaptations. The per capita income of adapters is also higher than the other groups. Adaptation contributes to the resilience of CC vulnerability [83,84] and our findings validated the results of other regional-level studies [3,[83][84][85]. The ATE reflects the effectiveness of adaptation strategies to abate CC vulnerabilities. An increase in the output of crops will help in increasing the living standard of the rural household through income increase and poverty reduction. The study shows that adaptations play a supportive role in recovering the vulnerability that substantiates the synergy between adaptation and resilience to CC vulnerability.
This study has successfully archived its stated objectives. Study yields a wide spectrum of policy options for policymaking, and practice-oriented solutions to cope with and adapt climate change (Fig 3). Firstly, the study suggested the need for proper education infrastructure in rural areas for farming communities to understand modern problems with modern solutions. Secondly, concerned stakeholders are suggested to provide the facility of smart technology in the rural areas to equip the farmers with timely updates about CC and new adaptation techniques. Policymakers are also suggested to provide their urgent concentration on the adaptive capacity building of the farmers.