Livelihood vulnerability assessments and adaptation strategies to climate change: a case study in Tanguar haor, Sylhet

Bangladesh is very prone to climate-driven hazards due to its unique geographical location, high population density, and low climate resilience. Therefore, this study addresses the livelihood vulnerabilities and adaptation strategies in response to the climate change issues based on empirical evidence from wetland communities using the livelihood vulnerability index (LVI) in Tanguar haor, located in Sylhet, Bangladesh. The ﬁ nding shows that people of the study area are switching their formal livelihood activities into alternative sources. Crop diversi ﬁ cation, homestead gardening, erosion proo ﬁ ng, and cage aquaculture are commonly used by the studied communities to adapt to climate change impacts. The main driver of vulnerabilities was found to be lack of education, underdeveloped infrastructures, lack of advanced health facilities, and lack of alternative livelihoods during extreme events. Therefore, people living in and around the haor region need special consideration. Overall, the study suggests that policy or institutional arrangements are required to promote the better governance system for local communities to improve their living standard so that they can achieve resilience to the impacts of climate variability and change in this area.


INTRODUCTION
Climate change distresses the whole world. But worst triggers will be those who are already vulnerable. Climate change may affect people differently based on their livelihoods and socioeconomic status (Kabir & Serrao-Neumann 2020). As with other developing countries in South Asia, Bangladesh is the most susceptible country to climate change due to its geographical position, high population density, and poverty, low capacity for adaptation and mitigation (Mirza 2002;Dastagir 2015;Alam et al. 2018). According to the World Risk Report 2012, Bangladesh ranked fifth out of 173 countries to natural hazards and the vulnerability of societies (Joarder & Miller 2013). A lot of studies have been carried out about climate change and its impact on our society and environment. Researcher claimed that frequent natural calamities are the most common issues and affected many districts of Bangladesh (Islam et al. 2015). Mostly the riverine households of Bangladesh are at high risk of negative impacts of climate change and experience extreme events such as drought, flood, tropical cyclones, and storm surges every year (Alam et al. 2017). These events have killed thousands of people and destroyed homes and livelihoods. The prolonged flood events push thousands of people to become homeless and they lose their homestead and croplands (Saha 2017). Also, in riverine areas of Bangladesh, frequent floods pose a different threat to food security by rice crop damage which ultimately impacts both the country's food security and economy. Due to flash floods, crops are destroyed every year in the northeastern haor region. This situation leads to inflated prices for available food. These price spikes leave the poorest households most vulnerable. As the food systems are dependent on one another, this means that more frequent and more extreme events in one region could disrupt a cluster of food systemseven the country. Although the riverine haor areas are much more fertile land for agricultural production, these areas are relatively poor income compared to the rest of the country. There are many reasons attributed to it such as the production of the single crop throughout the year, seasonal unemployment, flood, other natural disasters, lack of communication and other infrastructure facilities, employment opportunities, lack of due attention on the part of the government, and other service providers (Nowreen et al. 2013).
People of Northern riverine areas are highly exposed to the impacts of climate-driven hazards through rainfall variability and changes in the river flow. Due to the monsoon flooding and deltaic topographical condition, people in the riverine areas are facing riverbank erosion which is the most common destructive disaster in terms of financial loss (Alam 2017). Riverbank erosion adversely affected the agricultural sector by changing the local land configuration, soil texture and structure, drainage, and surface conditions and makes the riverine people vulnerable by changing their income and consumption pattern. So, the vulnerable people who are affected by these types of natural hazards persuasively displace the community and thus the occupation and livelihoods (Alam 2018). The future of the riverine haor people of Bangladesh relies upon the improvement and disaster management policies of the responsible government and on the regional cooperation and international effort fighting climatic dangers. Therefore, climate change is followed by a significant unpredictability of the new local conditions and vital that initiatives for better climate adaptation.
A significant number of adaptation plans have been unsuccessful basically, as they were not effective to accurately identify the major facets and enormity of susceptibility of the community where projects were to be announced. If we can find the prime aspects of vulnerability and its enormity for a susceptible community, it will be beneficial to measure the degree of sustainability of livelihoods that the community possesses. Scholars revealed that the affected people who are living in vulnerable areas have not enough knowledge to deal with the likely impacts of climate change. But nowadays, most of the households have now become accustomed by taking some steps in different events like floods and heavy rainfall. Therefore, exploring adaptations associated with local-level knowledge is regarded as an important adaptation strategy for people to deal with an extreme climatic event. So far only a few studies have been conducted to identify the impacts of climatic events, the reason of happening such events, adaptation strategies especially adopted by local people, and how it can be mitigated through building local awareness. To date, there have been only a few in-depth studies of livelihood vulnerability assessment. Without recognizing local level vulnerability patterns, it is likely to be least effective to recommend any copying or adaptation mechanism. Through implementing a cohesive and participatory action research approach, this project explores how rural communities living in disaster-prone areas in Bangladesh respond to increasing incidences of extreme events. Therefore, the objective of this research is to assess the meteorological conditions, people's perceptions regarding vulnerabilities to natural hazards, and their adaptation strategies based on their socioeconomic perspective in Tanguar haor, Sylhet, Bangladesh.

Study area
Tanguar haor is located at the northern outskirt of riverine Bangladesh covering 9,727 ha. It is adjacent to the Indian border and one of the largest haor (river basin) which is part of a wetland of the Surma-Kushiyara river basins (see Figure 1). It is situated in the northeast region relative to natural state between 25.1615°N and 91.0778°E. The ecosystem is most diversified in Tanguar haor. It is not merely ecologically but has also socioeconomic importance for supportive livelihood to around 70,000 people. Around 50% of the area of Tanguar haor is placed in water bodies, followed by 31% of crop land (Islam et al. 2014). It comprises 46 villages and around 120 beels (Hossain et al. 2017). The swamp forest land of the haor is one more exceptional ecological feature of the haor ecology which acting as a significant role in fish production and functions as a 'mother fishery' for the country (Chowdhury 2010). Since its ecological importance, it has been acknowledged as the second Ramsar site of Bangladesh in 2000 (Sun et al. 2017). structured questionnaire was used to collect data on each topic from the head of the households. The data were collected on the livelihood practices, agricultural system, monthly income-expenditure, local awareness, perception of climate change, and natural disasters. They were asked about their socioeconomic condition such as number of family member, their age, gender distribution, educational background, housing condition, occupation, and their income level per month. The impacts of natural hazards on shelter, water supply system, sanitation system, health condition, food the change of intensity of livelihood vulnerability, and percentage of occurrence of various natural disasters were also asked. Through detailed interviews of the respondent, data were collected by using a semi-structured questionnaire. During the field visit, several questions related to climate change adaptation were also asked to the respondents.

Selection of villages
Tanguar haor, covering an area of 9,727 ha comprises two Upazilas namely Tahirpur and Dharmapasha under Sunamganj district located in the northeast part of Bangladesh. It is very tough, expensive, and time-consuming to carry out research work in two of these Upazilas. In this context, the selection of villages was made because of secondary information based on the level of vulnerability to flooding condition and waterlogging. Areas of study were chosen to recognize the impact of climatic catastrophe and the mechanism of adaptation to these various dangers. Considering all the criteria, Tahirpur and Jamalganj Upazilas were selected purposively because the areas are situated in the 'haor basin'. Then a list of villages was fixed afterward. Then 10 villages were selected randomly from the list. Five villages were selected from each village. Manik Khila, Dumal, Santospur, Ramshinghpur, and Janjail villages were selected from Tahirpur Upazila, and Fenarbag (central), Fenarbag (west), Vobanipur, Santipur, and Badarpur were selected from Jamalganj Upazila.

Selection of respondents
Prior to individual respondent data collection, study villages were visited, and an FGD was done to know about the local people condition, locality, and farming system in order to make an appropriate sampling approach. In each FGD, a variety of respondents like local Upazila members, government officials, and development workers were included. A total of four (two from Tahirpur Upazila and two from Jamalganj Upazila) FGDs were done in 10 study villages, and a number of respondents were fixed during the FGDs. Then, data were collected from a total of 120 respondents of the 10 study villages 10 from each village. The male and female respondents were selected who are young or old age for interview because they are experienced and head of the family.

Livelihood vulnerability index analysis
The livelihood vulnerability index (LVI) was developed by Hahn et al. (2009) for vulnerability assessment of climate change. It has been designed to provide information to the development organizations, policymakers, and public health practitioners to understand climate vulnerability and its related component such as demographic, social, and health (Hahn et al. 2009). The LVI uses an evenly balanced weighted approach. This weighting scheme could be adjusted by future users if it is needed. Because each of the subcomponents is measured on a different scale, it was first necessary to standardize each as an index. The LVI includes five major components: sociodemographic profile, livelihood strategies, social networks, climate sensitivity of households (food, water, health, housing, and land tenure), and natural disasters and climate variability. Each major component is comprised of several subcomponents (see Tables 1 and 2). As this method has flexibility, we used more relevant subcomponents in this study based on the literature review and expert opinion. To analyze the index, we have followed four main steps, namely: Step 1. Transforming measurement units: All the raw data of each subcomponent which was collected through a questionnaire survey was transformed into appropriate standardized measurement units, such as percentages, ratios, and indices.
Step 2. Standardization of subcomponents: It was first necessary to standardize each subcomponent as an index because these were measured on different scales. The equation (Equation (1)) is used for the conversion, which was adapted from that used in the Human Development Index to calculate the life expectancy index, which is the ratio of the difference of the actual life expectancy and a pre-selected minimum, and the range of pre-determined maximum and minimum life expectancy (UNDP 2010): where S d is the average value of the subcomponent or indicator value of village d, S max is the maximum value of subcomponent, and S min is the minimum value of subcomponent.
Step 3. Averaging of Subcomponents: The standardized scores of each subcomponent were averaged by using the following formula: where M d is one of the seven major components for district d [sociodemographic profile (SDP), livelihood strategies (LS), social networks (SN), health (H), food (F), water (W), housing and land tenure (HLT), or natural disasters and climate variability (NDCV)]. Index S d i is the ith subcomponent value, belonging to major component M d for village d. n is the number of subcomponents in each major component.
Step 4. Calculating final LVI score: When values for each of the major components were calculated, then they were averaged using Equation (3) to obtain the final LVI score. The average of each major component was calculated from the number Percentage of households drinking unsafe water (river, pond, or canal water) Higher percentage indicates less adaptive capacity since the water of pond/river is polluted.

Natural disasters Percentage of households of which any member injured in the flood
The households are less vulnerable whose percentage is low. Average number of flood affected them.
Higher value reflects more vulnerability. Average number of flood occurred in the last 5 years More flood events reflect higher exposure.
Average time of last flood stayed/lagged at this area Long time reflects higher exposure.
Journal of Water and Climate Change Vol 00 No 0, 5 Uncorrected Proof of subcomponents of which it was comprised. The LVI is scaled from 0 (least vulnerable) to 0.5 (most vulnerable).
where LVI d is the livelihood vulnerability index for district d, M d i is the major component for community d indexed by i, and W M i is the weight of each major component.

LVI-IPCC calculation
The LVI-IPCC index is an alternative to evaluate LVI by merging the definition of vulnerability corresponding to IPCC. Vulnerability is an action of a process that exposes and susceptible to environmental stimuli and their adaptability to negative impacts (Shah et al. 2013). For this analysis, the major components of LVI data were categorized into three groups, these are adaptive capacity, sensitivity, and climate exposure. Table 2 describes the processes of calculating LVI-IPCC. Exposure represents the nature and degree to which a system is subjected to large climate variations. The exposure of this research population is measured by the average number of natural disasters (flood, flash flood, drought, and storm surges) that occurred and reported by the respondent in the last 5 years. Adaptive capacity denotes the ability of a group or individual to deal with the challenges through local knowledge and strategies to make an adjustment. Sensitivity is measured using the facilities of available food, health, and water in the study area. Understanding these components can assist to assess the nature and magnitude of the climate change risk together with the recognition of substantial sources of vulnerability. The LVI-IPCC was ranged from À1 (least vulnerable) to þ1 (most vulnerable) (Hahn et al. 2009). The LVI-IPCC was calculated with the following formula: where CF d is the IPCC-defined contributing factor (exposure, sensitivity, or adaptive capacity) for community d, M d i is the major component for community d indexed by i, W m i is the weight of each major component, and n is the number of major components in each contributing factor. After calculating the exposure, sensitivity, and adaptive capacity, the following equation was used to calculate LVI-IPCC: where LVI-IPCC d is the LVI for community d expressed using the IPCC vulnerability framework, e d is the calculated exposure score for community d (equivalent to the natural disasters and climate variability major component), a d is the calculated adaptive capacity score for district d (weighted average of the sociodemographic, livelihood strategies, and social network major components), and S d is the calculated sensitivity score for district d (weighted average of the heath, food, water, and housing and land tenure major component).

Climate change adaptation strategies
The collected respondent data on the daily livelihood activities, farming system, and indigenous knowledge that are related to climate change adaptation were cross-checked during the field visit and documented in the present study.

RESULTS AND DISCUSSION
People living in different riverine areas of Bangladesh have been suffering from recurrent catastrophes which cause large damages every year and ultimately affect the overall livelihood pattern of the local communities. So, understanding local climate, perceptions, and adaptive behavior of household levels bring enhanced insights and information relevant to a policy that helps to address in the face of climate change.
3.1. Historical climate change scenario over 39 years Rainfall and temperature are two climatic parameters that profile the structure of any socioecological system. In this study, historical changes in temperature and rainfall were evaluated for the Sylhet district, Bangladesh. Meteorological records from this region indicate that temperature has increased significantly (p ¼ 0.0001) over the period 1980-2019 (see Figure 2), although there has been no significant trend in mean annual rainfall. A more accelerated temperature change was also observed in the last 30 years (Raihan et al. 2015;Rahman & Lateh 2017). Another study reported by Basak et al. (2013) also demonstrated the indication of increasing temperature in the northeast region of Bangladesh. The trend of the temperature indices represents that the mean annual maximum temperature is increasing in recent decades with the highest rise in 2014 ( 1980-1992 and 1993-2005). Khan et al. (2019) reported in their study that the average monthly maximum temperature and minimum temperature have increased significantly by 0.35 and 0.16°C/decade, respectively. Using the trend line equation, Rahman & Lateh (2016) also found that the annual maximum and minimum temperatures have increased from 1971 to 2010. From the last 39 years, the mean annual minimum temperature was found the lowest in 1992 and the highest in 2016. Apart from that, a statistically significant positive trend of mean annual minimum temperature has also been detected equally as the mean annual maximum temperature (see in Figure 3). The mean annual minimum temperatures of 17.04, 17.67, and 18°C have been detected between the years 1980-1992, 1993-2005, and 2006-2018, respectively, which indicates the gradual rise of mean annual minimum temperature. Hence, it might be claimed that in this region, distinct warming trends were observed, and mostly the maximum temperature has a larger influence on the annual warming.
In contrast to the significant changes in temperature, there is a negative trend in the mean annual rainfall which is statistically insignificant (p ¼ 0.5, see in Figure   Water flow in Tanguar haor comprises mainly of water from the river systems situated south of the Sunamganj district. Water is also received from streams flowing from upstream northern hills in India. Water levels within Tanguar haor regularly rise during wet seasons as water flows from the Surma rivers into the haor (Bagchi et al. 2020). The water level of this haor area increased due to the increase of rainfall and water level of Surma river which ultimately experiences flash flood events (Muzaffar & Ahmed 2007). In Tanguar haor, flash flood event occurs frequently when the river water level exceeds the danger level (DL). The Surma at Sumanganj exhibited a fast rise and drop in different periods of the monsoon (see Table 3). In the year 2019, the water level of the river Surma at Sunamganj poured beyond its DL (8.25 m) for 7 days during the monsoon. Recorded results show its highest peak of 8.64 m PWD on 11 July, which was 39 cm above its DL. The maximum water level of Surma at Sunamganj was 8.47 m PWD on 5 July 2018 which was 22 cm above its DL (8.25 m). During the whole monsoon in 2017, the DL of the river Surma at Sunamganj continues for 15 days and recorded its highest peak of 9.19 m PWD on 14 August which was 94 cm above its DL (8.25 m). In 2016, Sumangnaj showed the rapid rise in water levels from mid-April with the abrupt jump in sometime in May and June with its maximum peak of 9.15 m PWD on 21s July, which was 90 cm above the DL. On 20 August 2015, the water level of Sunamganj flows 83 cm above DL (8.25 m), and the highest peak was 9.08 m PWD. Throughout this year, the water level crossed the DL two times and remained in a total of 31 days. Besides, the Sunamganj district experiences the highest annual average water level anomaly   Table 3 (Bagchi et al. 2020). Exposure of this area may be affected by flash floods that devastate the key production sector (i.e., agriculture) and hence threaten the livelihood of the haor households (BHWDB 2012). According to the department of agriculture extension, Tanguar haor has been affected most extensively due to the flash flood of 2017. The total affected area was 30,100 ha, and the total damage of standing crops in the boro field was 18,200 ha. Due to enormous damages of rice production, 33% of the population need to displace. Property loss of population was 65%. Research demonstrated that due to flash floods, 48% of people suffered from a shortage of drinking water. Different water-borne diseases have broken out and trigger the vulnerability. Around 10% of people were affected by diseases because of poor health facilities during the post-disaster period. This perhaps implies that the needy people suffer the hardest time during floods due to inadequate means of protection measures and also the loss of property and income (Parvin et al. 2016). Another study by Islam et al. (2020) evaluates the risk and resilience level of flash floods and reported that 55% of people of Tahirpur in Tanguar haor have low resilience that renders them more vulnerable. So, the higher the number of floods, the higher the vulnerability and impacts caused by the hazards will be higher also. Accordingly, in future studies, a flood hazard map (intensity and probability of occurrence) along with a flood damage risk map will deliver additional insights. In this study, the exposure to natural disasters to climate change was evaluated by detecting the impacts due to climate variability and change using the respondent perceptions of hazards and coping methods during the last 5 years.

Perceptions of local communities on changing pattern of climatic impacts
All of the 200 respondents comprising of each village were asked if they understand the changes in temperature, rainfall, and duration of the summer season, rainy season, winter season, and the resultant impacts or the changing pattern of climatic impacts they had experienced within the last 5 years or not. Figure 5 presents the respondents' perception regarding climate changes in the study area. During field investigations, it is found that more than 87% of the respondents were not concerned about climate change and its impacts on the human and nonhuman environment. Most of them believed that the natural events are coming from God and, hence, it cannot be dodged; mostly God will determine who would suffer and who would be safe. The local respondent opinion is consistent with the findings of Kamal et al. (2018) and Anik & Khan (2012). Though the results revealed that none of the respondents had any prior idea about climate change, but they were able to respond to the questions when the terminology was made understood to them. Respondents perceived that both the intensity of hazards and vulnerability of the people have increased considerably over the last few decades. They reported that the heat intensity has increased immensely during summer and winter, which does not occur timely over the last 5 years ( Figure 6). Since most of the people of the haor region live on low-lying land, the income source of both the farmers and fishermen are disturbed and often destroyed. The study areas highly face flood and riverbank erosion hazards. Besides, due to the floods and drought in the dry season, most of the farmers can cultivate rice just once a year. Kamal et al. (2018) also reported in their findings that they primarily rely on dry season boro rice crops as their primary source of income.

Model 01: LVI
The actual and standardized values along with maximum and minimum values of subcomponents of five villages are presented in Table 4. The aggregate score (Model 01) suggests that Tahirpur is more vulnerable than Jamalganj due to the higher LVI values. Results reveal that Tahirpur is more vulnerable to adaptive capacity in terms of sociodemographic profile, Uncorrected Proof livelihood strategies, and social network components compared to Jamalgonj (see Table 5). The sociodemographic major component was composed of three subcomponents (percentage of female-headed households, the average age of household head, and percentage of households have no education). It represents that the dependency ratio is higher in Tahirpur (0.280).
In the case of livelihood components, most households (87%) stated that they have been engaged in agriculture as a source of income in Tahirpur, whereas the case is a little bit different in Jamalgonj. Around 92% of people relying on fishing. The social network indicators reflect that households do not get training on adaptation or local government assistance. According to the respondent, they did not receive any early warning about the flood.
In the case of sensitivity, both villages show an almost similar LVI score. Findings of overall food vulnerability scores demonstrate nearly similar outcomes such as in Tahirpur 0.396 and Jamalgonj 0.398. Households of Tahirpur reported a  Journal of Water and Climate Change Vol 00 No 0, 10 Uncorrected Proof longer average time to get health facilities and show a higher LVI score (0.567) than Jamalgonj (0.347). Households of Tahirpur stated that they highly face malnutrition and diseases during the disaster period and suffering from chronic illness. It is found that Jamalgonj (LVI ¼ 0.250) is more vulnerable than Tahirpur (LVI ¼ 0.188) in terms of the water resources component. Households of Jamalgonj reported using unsafe water for their daily needs and suffering from recurrent waterborne diseases. This could be due to their low level of education and awareness and longer traveling time to reach safe water sources. The vulnerability of housing and land tenure (1.00) is the same for both Upazilas. In terms of land ownership, many households lacked a basic form of land tenure. Therefore, due to limited property rights and housing conditions (fundamentally raw house made of bamboo, mud, and straw), living in Tanguar haor is also very strenuous (Haque & Kazal 2008). Roy (2010) stated that in the Jamalgonj area, 83% of fishers' house structures are Kacha as they use bamboo for roofing materials and only 17% are semipaca (floor and wall made by bricks and roof made by woods cover). Based on the average reported number of floods that have been occurred in the last 5 years, Tahirpur (0.46) has the highest score compared to Jamalgonj (0.41).

Uncorrected Proof
Floods induced exposure are among the greatest devastating natural disaster in Tanguar haor and cost many lives every year. The findings of the study revealed almost parallel vulnerability scores for the natural disaster component, based upon the average number of reported floods or flash floods at Tahirpur (3.4) and Jamalgonj (3.2) (see Table 4). Understanding how these components and indicators influence the vulnerability of livelihoods, we have set out some questions. In response to the questions related to exposure (natural disaster and climate variability), the households of these two Upazilas conveyed that they do not receive a warning before any disaster which largely affects the livelihood pattern of the people residing in these areas. Regarding the constraints in mitigating flood risks, household reported that 5% of people in Tahirpur and 2.5% in Jamalgonj did not receive any early warning systems due to lack of proper infrastructure and primitive broadcasting system. They reported that due to poor warning systems they could not get adequate time for safety measures to protect themselves and their livelihoods from flood risks. Kamal et al. (2018) stated similar findings that only 7.5% of respondents had access to prevailing weather forecasting and attempt to make required arrangements for the impending flood. Furthermore, nearly 38% of the total respondent reported acute injury or death because of the most severe natural disaster. The effect of the devastating scenario of the climatic event (e.g., flash flood, heavy rainfall, river erosion, cyclone, and drought) on every aspect of socioeconomic life at Tanguar haor is also observed by Rahaman et al. 2019. Regarding health facilities, households from Tahirpur (28.3%) and Jamalgonj (26.6%) reported longer average time to get health facilities and experienced the greatest incidence of chronic illnesses due to the shortage of advanced medication facilities during the flooding period. All villages in this study area shared almost a similar degree of vulnerability to natural disasters and climate variability, since these villages belong to the same geographical area. The results of these two Upazilas for major components are presented in the spider diagram (see Figure 7).

Model 02: LVI-IPCC
The LVI-IPCC analysis produced a similar result with LVI analysis within the range from À1 to þ1. According to the LVI-IPCC analysis, people of Tahirpur Upazila (0.137) are most vulnerable than Jamalganj Upazila (0.121). A vulnerability spider triangle diagram shows the scores for adaptive capacity, exposures, and sensitivity for Tahirpur and Jamalganj Upazilas (see Figure 8). Results from Model 02 suggested that the two communities shared a very similar degree of vulnerability. The model however suggested that Tahirpur is more vulnerable than Jamalganj in terms of exposure, adaptive capacity, and sensitivity.  Journal of Water and Climate Change Vol 00 No 0, 13

Uncorrected Proof
Tahirpur was more exposed to natural disasters and climate variability (0.46); on the other hand, Jamalgonj was less exposed to natural disasters and climate variability (0.41). The adaptive capacity of Tahirpur (0.21) is greater than Jamalganj (0.17), and the climate sensitivity of households of Tahirpur is 0.54, which is also greater than Jamalganj (see Table 6).

Identified adaptation strategies in and along with Tanguor haor
Tanguar haor is the distant and backbreaking places of Bangladesh. Traditionally, people of Tanguar haor are communally and structurally downgraded inside and outside of their homes. Communication to major cities is narrow due to the economic condition of the people and behindhand infrastructure. Therefore, the people of Tanguar haor commence various coping methods using their capabilities, resources, familiarity, and their surrounding environment. As the people living in haor areas faced many challenges in their daily life, they preferred to make their household areas slightly above the water level. So, the household areas could hardly affect by flashflood during the rainy season. Kamal et al. (2018) reported in their study that flashflood causes huge crop damage (∼800,000 t Boro rice; worth the US $450 million) and brings immense danger to the haor people particularly who work as sharecroppers or landless laborers. The result of the study showed that some respondents were highly cautious about the flood and did not want to lose their crops and preferred to homestead gardening. People of Tahirpur and Jamalganj Upazilas are constructed with various infrastructures to protect themselves from natural hazards. After any hazards, they built a bamboo bridge (Sako) or brick bridge on the road, a water dam to reserve water for the dry season, and two or more storeyed buildings for education in the time of flood. Anik & Khan (2012) also reported in their studies that the people of the northeastern region tried to adapt themselves to climate change by introducing various adaptation strategies such as crop diversification, duck rearing, floating garden, and construction of embankments by their knowledge.
Rarely flashflood arises early in the haor areas so that people could not get time to harvest their crops which results in food scarcity and high price and threats to local livelihood. To ensure food security, weather forecasting assisted them in controlling and timing of plants. Also, haor people allow a single crop throughout a year; therefore, the recurrent unemployment problem is very critical here (Haque & Kazal 2008;Kamal et al. 2018;Rahaman et al. 2019). But the situation was a bit different in Tahirpur and Jamalganj. As the haor allows a single crop throughout a year, most farmers were interested in other crops along with cereal crops. Since climate change is affecting the sowing and harvesting time of crops, they prefer to cultivate diversified crops in their agricultural fields. Nowadays, people changing their crop production. They are producing varieties of crops like mung, bean, lentil, linseed, oilseed, potato, spices, pumpkin, corns, and onion, ginger, and garlic with the help of the PROTIC project. Pest outbreak is a great problem in these areas. To get rid of this problem, farmers used various pesticides, though it has a huge environmental impact. Most of the respondents said that the changes in rainfall, raising the temperature, and shifts in time of seasonal onset were the main causes of frequent outbreaks of pests and diseases.
In our study, about 87.5 percent of Tahirpur and 92.5 percent of Jamalganj villagers' secondary occupations are fishing. But due to climate change or uneven climate variability, they are changing their fishing location and fishing technique. The common fishing practice in the study areas during the waterlogging period was cage aquaculture. According to them, the earlier species of fish are no longer available. In our study, they are rearing various types of domestic animals like cows, hens, and ducks. The average number of livestock they have own is 2.2 for Tahirpur and 4.88 for Jamalganj Upazila, and the average number of livestock items is 0.975 and 0.925, respectively, for Tahirpur and Jamalganj. But the maximum item of livestock for both Upazila is 3. In Tanguar haor, people cannot switch their formal livelihood activities into alternative sources. About 30% of the total respondents are getting 1,500-3,000 Takas per month and nearly 39% are getting 3,000-5,000 Takas as a day labor per month (Haque & Kazal 2008), but in this study, Tahirpur was found economically vulnerable as the most workers get more or less 5,000 Takas per month for working in agricultural land and fisheries, whereas people at Jamalganj found financially healthy because the rate of workers who were getting below 5,000 Takas is 2.5% per month. But the area is floodprone and often struck by flash floods which makes them more vulnerable. Therefore, improving socioeconomic conditions may reduce vulnerability and enhance their adaptive capacity (Sebesvari et al. 2016). Skill development training for poor people could be an effective way during seasonal unemployment. Beside developing communication networks, constructing infrastructure and access to organizations for participation can diversify their livelihood and create income opportunities which in turn enhance their resilience. Though Vittal et al. (2020) identified the consequences of combined information from the social vulnerability (SoV) and hydro climatic hazards where they stated that hazards are potential to substantial contribution to the growing risk and not SoV. As a result, focusing the magnitude of hazard and vulnerability in the increasing risk region, policymaker should emphasize individual resilience with an economic capable risk reduction plan.

CONCLUSION
In this paper, participants' view and perception of the changing climate were also confirmed by the evidence from temperature and rainfall data analysis. Results indicate that the area has been experiencing seasonal fluctuations over the past 39 years. We found that the causal factors (exposure, sensitivity, and adaptive capacity) independently present variations in their indices for Tahirpur and Jamagonj Upazilas, and no major variation is observed for the total livelihood index. Local livelihoods of the communities were vulnerable with regard to sociodemographic profile, access to food, health, and climate variability. Various factors, such as geographical isolation, lack of education, underdeveloped infrastructures, lack of advanced health facilities, lack of awareness, lack of support from local government, and lack of alternative livelihood during extreme events, had made the community more vulnerable to climate change. Households with low levels of human, physical, natural, financial, and social capitals were found to have limited adaptive capacity to meet the challenges during the disaster periods. Strengthening and improving the indicators identified in this paper could help to reduce the livelihood vulnerability of these two Upazilas. Government and nongovernment organizations should take several initiatives like training for income-generating activities, credit or loan facilities should be provided with ease or at a lower rate of interest that can help them to establish various agro-based industries, fisheries, and other sectors for the diversification of their livelihood options. They should provide an educational program to increase awareness regarding the natural disaster, early warning signs to the community members in making farming decisions, and give empowerment to the women to take a prominent role in climate adaptation strategies. However, due to lack of proper data on individual component, it was not possible for us to compare the ratio of each indicator's response on vulnerability. Future research should be undertaken based on this.