Assessing Livelihood Adaptation Indices and the Sustainability of Rice Farmers in Bangladesh's Northwestern Region

Climate change alters the impact of the environment on economies, and sustainable adaptation which may improve food security is embedded in the actualization of the 2030 United Nations Sustainable Development Goals (SDGs) 2, 13, and 14. Hence, this study contributes to the debate by measuring how farmers in the Northwestern region of Bangladesh adapt to climate change. A cross-sectional multistage random sampling is used to collect 500 data points by the face-to-face interview method. For robustness, the study demonstrates climate change adaptations, adaptation indices, and the sustainability indicators in social, economic, and environmental concepts using the composite indicator method. Also, Rasch analysis and marginal contribution were used to explain the adaptation indices. Finally, a trivariate Tobit regression is used to examine sustainability analyses of climate change adaptation strategies and explain how the climate change adaptations affect different dimensions of sustainability. The results showed that dominant male households, extended family, skilled farmers and food security influenced adaptation index as well as adaptation strategies like organic manure, changing planting dates, diseases tolerant varieties and irrigation. Though, most dominant strategies like irrigation and applying fertilizer are not sustainable. The study also found that farm size, credit access, and extension contact significantly affects sustainability. Moreover, off-farm activities, crop diversification, and using high-yield varieties are more sustainable adaptation strategies. Policy should be implemented on the basis of region and sustainable manner.


INTRODUCTION
The agricultural sector is burdened with the responsibility to tackle food security at the national and global levels (Schmidhuber and Tubiello, 2007;Barrett, 2010;FAOSTAT, 2015;Alam, 2016;Pandey et al., 2016;Adeleye et al., 2020;Osabohien et al., 2020). But food sustainability and security are hampered by the incidence of climate change. Though agriculture contributes the least to climate change, its adverse effect is felt due to the increase in the severity of the weather (Herring et al., 2014;Hulme, 2014;Adeleye et al., 2021), which adversely affects climate-related food productivity and accessibility (IPCC, 2001;UNFCCC, 2009;Dodman and Miltan, 2011;Wilson et al., 2014). Given that agricultural integration is crucial for sustainable socio-economic development, when agriculture is adversely affected by climate change it causes a decline in economic growth, worsens poverty levels, and further generates new poverty traps (Gautam, 2008;IPCC, 2014;Alemayehu and Bewket, 2016).
Adapting to climate change requires the need to minimize susceptibility and attendant risks (Barnett, 2010). Given this, several adaptation measures have been implemented in the bid to reduce the harmful effects of climate change (Donner et al., 2016;Dinesh et al., 2017;Weiler et al., 2018). Notably, existing adaptation choices are tied to resource constraints, human cognition, and socio-economic factors but ignore the adverse effects of adaptation policies and sustainable practices (Grothmann and Patt, 2005;UNFCC, 2009;Brown, 2011;Eriksen and Brown, 2011;Yegbemey et al., 2017).
The focus on Bangladesh is germane. The country's agricultural sector contributes about 14.74% to GDP and employs about 40.15% of the population (BBS, 2018), and strives toward the actualization of the SDG associated with food security and climate change. The Fifth International Rice Congress (IRC) 2018 marked Bangladesh as the fourth largest consumer of rice and branded the "country of rice" for contributing 7.5% of the total volume of rice in the whole world. Despite this, the Bangladesh government imported 10 million tons of food cereals to which rice imports amounted to six million tons in order to cushion supply shocks. This is the highest volume of rice imported in the last 32 years (Food Ministry and Daily Star 1 ). This shortfall in rice production is due to unusual climatic events (Wassmann et al., 2009;Hossain and Silva, 2013). Therefore, to address the potential threats to rice productivity, finding alternative and adaptive measures to climate change became the priority of farmers and stakeholders (Kirrane et al., 2010;. Identifying successful climate change adaptive measures proved vital in coping with the menace. Given this, it is critical to examine the adaptation strategies employed by smallholder rice farmers in Bangladesh. Several studies have identified strategies such as shifting sowing dates, using early maturing and drought-resistant rice varieties, using artificial fertilizers, farming near water bodies, crop rotation, better irrigation, establishment of deep pond tubes, construction of embankments, introduction of trees in rice farms, mixed cropping, selection of short-term crops, and preserving rain waters. These documented measures have enhanced agroproductivities, efficiency, and profits of producing units (Kaiser et al., 2010;Ghosh et al., 2015Ghosh et al., , 2016Eyasmin et al., 2017). To the best of our knowledge, no study has investigated the challenges faced by rice farmers in this region as a result of the effects of climate change. Moreover, this study contributes to the literature by assessing the weighted climate change adaptation strategies of rice farmers in the Northwestern region in Bangladesh. 1 https://www.thedailystar.net/world/south-asia/bangladesh/bangladeshgovernment-import-rice-food-staple-price-production-agricultural-1403068

Study Area, Description of Variables, Summary Statistics
During the 2017-18 production year, this study performed a cross-sectional pilot survey through a face-to-face interview process. The survey had 500 sample households of rice farmers to exclude missing data in the northwestern region of Bangladesh. The Northwestern region of Bangladesh is the most climatesensitive and poorest area of Bangladesh (BBS, 2019). The region has experienced groundwater depletion and severe droughts (Alauddin and Sarker, 2014;Kabir et al., 2017). Two upazilas (sub-districts), Ishwardi in Pabna and Lalpur in Natore, were selected and assumed to be representative of the Northwestern region of Bangladesh. Figure 1 shows the study area, in which 79.85% of the people in the survey have their livelihood on agriculture. Moreover, Ishwardi Upazila and Lalpur Upazila reach their peak vulnerability from the intensity and frequent drought (Alauddin and Sarker, 2014;Eyasmin et al., 2017). The survey questionnaire included the 14 adaptation strategies such as irrigation, chemical fertilizer application, organic manure application, restructure cash flow and debt, crop rotation, offfarm activities, diseases tolerant varieties, changing planting dates, drought tolerance rice varieties, and crop diversification, which were identified in different studies in the drought-prone area (Ghosh et al., 2016;Eyasmin et al., 2017;Kabir et al., 2017). Table 1 shows the description of the variables.

Rasch Analysis
Assessing all the adaptation strategies is quite difficult due to their heterogeneous characteristics (Wheeler et al., 2013). To identify the most prioritized strategies by transitive rank, Rasch analysis was used. The model can be explained as, where, A ni = the Adaptation strategies A ∈ (o, 1) β n = parameters of nth households δ i = Parameters of ith region at same continuum.
And γ ni = 1 + exp (β n − δ i ) indicates the normalizing factor to ensure that the sum of two probabilities in Equation (1) is equal to zero.

Livelihood-Based Adaptation Index
The livelihood-based adaptation index (LAI) was developed to evaluate the determinants of adaptation by assessing the weights of adaptation strategies on the basis of their marginal contribution to household livelihood. Ordinary least square (OLS) is used to measure the index which contributes to aggravate poverty: where, L i = Livelihood based adaptation index of ith household which is weighted by their marginal contribution  A ij = jth adaptation strategy are undertaken by ith household β j = parameter explained as poverty line unit, a more obvious interpretation of the result.
However, LAI < 1 indicates the household below the poverty line and otherwise.

Model of Sustainability Analysis
The sustainability of climate change adaptation strategies is measured by the socio-economic characteristics and the adjustment practices as well.
This assumption can be expressed as follows: where, S = household's sustainability level, Z = household's socio-economic characteristics A = Adaptation strategies prioritize by the Rasch analysis.
The level of sustainability is estimated in economic, environmental, and social dimensions. As such, sustainability S is assessed by the nth household's socio-economic characteristics and mth adaptation strategies. So, Equation (3) may be written as: As there are only three dimensions of sustainability, j = 3. The extension of Equation (4) is as follows: where, α = constant terms, λ and ρ are the parameters, and u is the error term.
Due to the barrier of implementing OLS to multidimensional variables, a multivariate Tobit model is applied to sustainability which is limited between 0 and 100. Moreover, the Tobit model is advanced with its feature estimation of M equations together. Their sustainability is derived from their coefficient and level of significance. The specification of the sustainability Equation (5) is derived as follows: The sustainability score measures from 0 to 100 in scale. All sustainability indicators and components are scaled by the agriculture extension officer and farmers (Yegbemey et al., 2017).

RESULTS AND DISCUSSIONS
Descriptive Statistics of Households in the Study Area Table 1 shows the summary statistics of the variables included in the study. The variables included dependent and independent variables such as sustainability in three dimensions, age, sex, farming experience, education, farm size, access to extension services, credit accessibility, and 14 adaptation strategies.

Rasch Analysis Results
The goodness-of-fit of the model is assessed by using strategies and individual fits residuals to rank the adaptation strategies from easiest to challenging (Bond and Fox, 2001). The selected range is between ±2.5. All other outer practices from the range are dropped down for further analysis (Bond and Fox, 2001). Differential item functioning (DIF) further found that each strategy is unbiased at a 10% significance level. Table 2 shows the results from the Rasch analysis. Table 2 shows that the fitted residuals are between −1.5 and 1.2, whereas person's residuals are between −2.103 and 1.231, both are fallen within the suggested range of ±2.5. That's both the fitted and person's residuals are well-fitted according to Rasch model. Besides, the reliability of the Rasch model is tested by the person separation index: 0.66 (Bond and Fox, 2001). From the Rasch analysis, the strategies of chemical fertilizer, irrigation, organic measure, redistribution of credit accessibility, and crop rotation have lower negative values that indicate that they are easier to implement. The other strategies, such as implementing edge of field conservation practices, intensifying or expanding the cultivated area, implementing field conservation practices, and testing new crops, might be challenging to engage in cultivating (Shikuku et al., 2017).

Livelihood-Based Adaptation Index
Tables 3, 4 show the influence of adaptability to climate change through household features via the OLS method and the marginal contribution of each adaptation strategy, i.e., the adaptation index. Before interpreting the results, the diagnostics reveal that the errors are normally distributed, and the White test for heteroscedasticity cannot reject the null hypothesis of homoscedasticity. For multicollinearity, the variance inflation factor (VIF) ranges between 1.87 and 5.23; indicating the absence of multicollinearity. Lastly, the Durbin Watson statistic indicates no autocorrelation. Table 3 shows a higher adaptation index for a male head of household with a larger family and more rice farming experience. Food accessibility and credit accessibility influence the adaptation index positively. Households with a male head and larger families are more likely to adjust to climate change by applying organic manure, chemical fertilizer, irrigation, and changing sowing dates.
A positive relationship for the size of the family, and the tendency to implement irrigation, chemical fertilizer, and changing planting dates have been established by Eyasmin et al. (2017) and Yasin and Ghosh (2019).
Food security lifted the adaptation index, which stimulates the probability of adjusting through organic manure application, changing planting dates, and disease-resistant varieties. This revelation proposes that farmers' hunger and struggle to adapt become obstacles to adjust effective adaptation and decisive adjustment for improving the sustenance food security. Kristjanson et al. (2012) have found that families with food stressed are more inactive to adaptive systems.
It was more likely that highly skilled farmers would apply chemical fertilizers, crop rotation, and shifting planting dates which promote the adaptation index. Shikuku et al. (2017) noted that the chance of using chemical fertilizer, crop rotation, and changing planting dates was more likely higher among experienced farmers. However, the continuing climate change demotivates the farmer from applying adjustment practices that force the adaptation index to be lower following the study (Bryan et al., 2009;Deressa et al., 2009Deressa et al., , 2011.

Sustainability Analysis
The analysis of the multivariate Tobit model of sustainability is displayed in Table 5. The results reveal that the overall sustainability is 45. The results indicated that about 50% of farmers' farming systems and adaptation strategies were sustainable in the social, economic, and environmental dimensions. The likelihood ratio of sustainability of 167.0234 indicates a significant level at 1%. Moreover, the likelihood ratio statistic of 173.71, 176.24, and 174.74 in the economic, environmental, and social aspects of sustainability are statistically significant at the 1% level. The results reveal the connection of the equations in the model and the significance of at least one covariance in the error terms. Hence, the estimated coefficients in the model present major driving forces on the sustainability dimension through adequate goodness of fit. Additionally, the corresponding coefficients indicated how rice farmers' socio-economic characteristics and climate change adaptation strategies related to social, economic, and environmental sustainability.

Sustainability of Socio-Economic Characteristics
Socio-economic characteristics such as level of education, farm size, credit access, and extension service access significantly impact sustainability dimensions. Education has a positive and significant effect that stimulates adaption with climatefriendly strategies as well as sustainable farming. However, the results show it to be an insignificant tool (p > 0.10) in both economic and social sustainability but a significant (p < 0.10) tool in environmental sustainability. However, significant environmental sustainability has a negative impact with a level of significance of 10%. It is shown to be an unsuitable tool to promote the sustainability of rice farming. Farm size is statistically significant and positively linked with sustainability, similar to other studies (Sarker et al., 2013;Ren et al., 2019). Therefore, farm size positively affects economic and environmental sustainability as expected at 10% probability but 1% at social sustainability. This reflects that land fragmentations reduce rice productivity and affect economic development, urbanization, and technological development as shown in Hung et al. (2007) and Adamopoulos and Restuccia (2014). Other variables such as accessing credit positively affect economic, environmental, and social sustainability at 5, 1, and 1% probability levels. Farmers who can access more credit adopt a more sustainable farming culture than those who access less. Sarker et al. (2013), Eyasmin et al. (2017), Linh et al. (2019), and Adeleye et al. (2020) also found a positive and statistically significant relationship between accessing credit and sustainable agricultural production. In all three dimensions of economic, environmental, and social sustainability, the relation between extended contacts and production sustainability is highly significant at a 1% significance level, though receiving higher extension services will result in higher social and environmental sustainability than economic sustainability as extension contacts become environmentally viable such as soil conservation, minimizing pesticides use, and organic farming. However, the traditional farming technique negatively affects sustainability and extension contact in the study area.

Sustainability Analysis of Climate Change Adaptations
Crop diversification, off-farm activities, and hybrid varieties for rice production significantly correlate with all three sustainability dimensions. Sustainability is highly related to off-farm activities at the 5, 1, and 10% significance levels in the economic, environmental, and social dimensions. Since it is challenging to manage the high cost of rice farming, farmers tend to do other off-farm activities like rickshaw pulling, plowing by machine, shopkeeping, day labor on other farms, and selling storage rice production off-season (Al-Amin and Hossain, 2019). The positive coefficients with significant probability indicate more sustainability of another source of income. Several studies (Coelli and Fleming, 2004;Chavas et al., 2005;Gonza'lez and Lo'pez, 2007) have concluded the positive relationship between off-farm income and higher agricultural production. Other sources of income might boost investment in rice production and help cope with the impact of climate change in three dimensions of sustainability (Zhang et al., 2014;Alamgir et al., 2018;Chen and Mueller, 2018;Rahman et al., 2019). Crop diversification helps to minimize mono-crop losses and enhances sustainability at a 1% probability level in both economic and environment. Therefore, farmers have tried enhancing economic performance to lead to better social status. However, a negative and insignificant (p > 0.10) relation was shown in terms of social sustainability. High-yield seeds help a farmer gain higher economic return (i.e., higher yield, income, and maybe higher technical efficiency) and are more environmentally viable (i.e., remove biological constraints). In the study area, high yielding seeds (HYV) significantly affect rice productivity in the economic (p < 0.05), environmental (p < 0.01), and social (p < 0.05) dimensions. The most common and dominating adaptation strategies are using chemical fertilizer and irrigation, which has no significant impact on sustainability with probability p > 0.10. The efficiency of irrigation depends on the availability of fresh water on both ground and surface, and chemical fertilizer limits the quality of the soil. Depleting groundwater and soil degradation might cause being not sustainable (Selvaraju et al., 2006).

CONCLUSION AND POLICY IMPLICATIONS
This study aligns with the 2030 United Nations Sustainable Development Goals (SDGs) 2, 13, and 14 to evaluate the relationship between climate change strategies and rice farming sustainability in the northwestern region of Bangladesh. The study focuses on the leading climate change adaptation strategies and their implementation through the adaptation index and sustainability in economic, social, and environmental dimensions. Significant findings reveal that using adaptation strategies was not sustainable in social, economic, and environmental dimensions. The study showed a greater tendency to implement irrigation, chemical fertilizer, and changing planting dates in larger families, while organic manure application, changing planting dates, and diseasesresistant varieties were related to food security and rice farming experience. In the case of sustainability analysis, the study found that farm size, credit access, and extension contact significantly affect sustainability. Moreover, off-farm activities, crop diversification, and using high-yielding varieties are more sustainable adaptation strategies. Policymakers should try to organize agricultural training on crop diversification and high-yield rice varieties with a promotion that improves food security and the supply chain. Since education has a significant relation to sustainable agriculture, existing education, vocational training, and vocational education should be linked to sustainable agriculture by helping them preserve rainwater, limit the use of chemical fertilizer, and utilize adaptation strategies. Extension services should be more explored so that farmers can gather knowledge about undertaking climate change adaptation and the harmful impact of their overuse. New technologies should be developed and made available and affordable so that farmers can adapt easily. As the different regions have different weather conditions, regional policy to strengthen qualitative research needs to be pursued, and focus on the farmers who have decided to implement or expand the utilization of more adaption strategies.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.