Assessment on land use changes and livelihood transition under the hydropower dam construction in Paunglaung Township, Southern Shan Highlands, Myanmar

We examined agricultural land use changes and their impacts on the local livelihoods of upstream farmers after the Paunglaung hydropower dam construction in Myanmar. Landsat and GeoEye images, acquired in 2011 and 2018, were used to detect agricultural land use changes. A marked decrease in agricultural land (5.77% of farmland and 9.64% of swidden) was observed, whereas orchards and plantations were introduced as new farming practices. Using the household survey, the income strategies were categorized by different wealth groups, and the socioeconomic condition was compared among the groups. We also compared the annual household income before and after dam construction. A significant change in income strategies after the dam construction were observed in low‐income households, but not in medium and high‐income households. Local people engaged in diversified livelihood strategies after the dam construction, mainly in relocated villages. Besides, farmland lost due to dam construction caused the significant insufficiency of rice self‐consumption and consequently resulted in greater dependency on the market of rice and their products. This study highlighted land use changes due to dam construction and its impacts on the livelihood of local farmers, contributing to planning the land use and livelihood strategies for affected farmers to ensure their livelihood security.


| INTRODUCTION
Upland farming is getting attention since the early 2010s, due to raising issues of livelihood security and resilience among people currently depending on swidden (alternatively shifting cultivation) (Heinimann et al., 2017;Hurni, 2013;Kohler & Romeo, 2013).Swidden cultivation is still being widely practiced in the mountains of Southeast Asia (Heinimann et al., 2017;Mertz et al., 2009;Schmidt-Vogt et al., 2009;Van Vliet et al., 2012).Their trends, drivers and impacts of changes in swidden cultivation at global level were assessed by (Van Vliet et al., 2012;Van Vliet et al., 2013).However, some details of the agricultural land use change at local level were still missing.Therefore, there is a need to well understand the local level transition in swidden agriculture, in relation to the changes in its land use and socioeconomic impacts of these changes on mountain dwellers, for sustainable land management at local level.
Globally, different scenarios of agricultural land use changes, the rapid evolution of agricultural systems, and farmland abandonment have been reported in several previous studies (Arvor et al., 2012;Kanazawa et al., 2006;Lobao & Meyer, 2001;Vanwambeke et al., 2007).Drastic changes in agricultural land use and agricultural land use change have been increasingly growing in mountainous areas in mainland Southeast Asia (Cramb et al., 2009;Castella et al., 2013;Swe & Nawata, 2020).Changes in agricultural practices such as the crop expansion with commercial crops and the ecological intensification with non-commercial crops in Mato Grosso, Brazil, were also reported by Arvor et al. (2012).Mostly, changes in agricultural practices in rural sectors led to cash cropping in accordance with the market demand (Vanwambeke et al., 2007).The Boserup (1965) theorized the agricultural land use change in terms of demographic trend (in-out migration) in upland areas.
In case of Myanmar, one of the world's most ethnically diverse countries, the ethnic groups commonly practice the traditional swidden cultivation (Win, 2005), with the zero-tillage upland farming on a rotational basis.There, agricultural land use change in the uplands and changes in the rural livelihoods of swidden farmers are mainly concerned with the local government's policy, forest conservation activities, and complex land tenure systems.This study will highlight the agricultural land use change in both upland and lowland farms, and livelihood changes of the farmers after the village relocation due to hydropower dam construction.
In general, hydropower dams have been considered as a key to a country's development by contributing to water supply, irrigation, and energy generation since the 1950s (Galipeau et al., 2013).Presently, about 16% of the global energy requirements are met by hydropower dams (Morimoto, 2013).Of these, Asia is a major region of hydropower development (Morimoto, 2013).However, several studies reported that dams induce significant environmental and socioeconomic impacts on the local communities (Beck et al., 2012;Galipeau et al., 2013;Kirchherr & Charles, 2016;Rousseau et al., 2017;Yankson et al., 2017).Manyari and de Carvalho (2007) and Chen et al. (2015) stated that the environmental impacts of large-scale hydroelectric projects in the Amazon region included the interruption of the natural streamflow, deforestation, and forest degradation.Besides, land use and land cover changes caused by dam construction, especially in primarily agricultural landscapes, have impacted farmers living near construction sites.Similar case was documented in Manwan hydropower dam in China; the dam caused a significant reduction in forest cover and farmlands being submerged under reservoirs (Zhao et al., 2010).Likewise, forced displacement and farmland loss were reported as major problems induced by hydropower dam constructions in Sri Lanka and west Java in Indonesia (Manatunge et al., 2009).
On the other hand, the World Commission also recognized dam construction as an important issue in sustainable development and developed a strategic assessment to mitigate socio-environmental impacts caused by dams (Finer & Jenkins, 2012).Also in Myanmar, the hydropower dam construction projects severely impacted local communities.These impacts include human right violations and adverse effects on livelihoods, food security, and health (Brown, 2015).Rampant economic growth has led to the establishment of hydropower dams to meet the increasing demand for electricity by industries and factories.In Myanmar, most dams were built in areas inhabited by ethnic communities primarily in Shan, Kachin and Karen States that were dependent on natural resources, which resulted in land conflict between ethnic groups and the government (Aung et al., 2021).Paunglaung hydropower dam was built in 2013 to supply electricity to the capital of Myanmar with the aid of Chinese, Swiss and British firms (Brown, 2015).Villages adjacent to the dam site were relocated to the surrounding mountainous areas, resulting in loss of land possession and displacement of local communities.Therefore, it is urgently needed to reconcile and adjust the power demand and livelihoods sustainability.However, assessing the spatial changes and socioeconomic conditions of relocated villages after dam construction is not well-documented.Also, there is no proper strategic assessment of the spatial and socioeconomic changes in circumstances for the Paunglaung dam after its construction.In this context, examining the considerable impacts, changes in livelihood and faming system, and investigating the linked factors for new adoption strategies are essential for future rural development and sustainable land management.
Therefore, we hypothesized that village relocation under the dam construction will induce some changes in agricultural land use, farming practices and their livelihood income strategies.In this study, we examined the agricultural land use changes, highlighting loss of permanent farmland and swidden transformation after the Paunglaung dam construction.We also investigated the socioeconomic conditions before and after dam construction and clarified the livelihood strategies of local farmers, by wealth status, subsequent to dam construction so that implementing sustainable agricultural land use planning and ensuring the food security of the relocated local farmers could be called for.

| Study area
This study was conducted in Pinlaung Township of Taunggyi District, Southern Shan State, Myanmar.Specifically, the area is in the Paunglaung reserved forest, occupying 163,551 ha (Figure 1).The area is undulating and hilly with an elevation of 180-1800 m above mean sea level; it is covered by evergreen forest and mixed deciduous forests (Mon et al., 2009).Paunglaung reserved forest is crucial timber supply area for both domestic and commercial purpose (Mon et al., 2009).The average monthly rainfall and mean air temperature of the study site collected from the nearest stations was shown in

| Mapping agricultural land use change after hydropower dam construction
We collected ground truth data to map swidden, fruit tree orchards, plantations, and lowland paddy fields using GPS (GARMIN Darkota 20).
Field data collection was carried out four times (November-December in 2018; January-February; May-June; September-October in 2019).
We performed manual delineation of lowland paddy fields and settlements area with the help of Google Earth image archives and ground-truthing.Subsequently, these delineated areas were masked out to avoid confusion with the young orchards, young plantations, and swidden fields.
We also classified land cover and detected their changes in the entire reserved forest, using freely available Landsat images (30 m resolution) taken in January 2011 (before dam construction) and February 2018 (after dam construction) by maximum likelihood method.We while the newly introduced orchards and plantations were taken as an additional category for 2018.We subsequently conducted postclassification analysis, that is, refinement with topographic factors.The final classified images were smoothed with a 3 Â 3 majority filter tools such as "boundary clean," and "nibble."The workflow of the analysis using images was detailed in Figure 3.All these land cover and land use mappings were done using ENVI 5 and ArcGIS 10.6 (ESRI) software.

| Accuracy assessment for image analysis
Accuracy assessment was conducted by following the good practices for area estimation and accuracy assessment by Olofsson et al. (2014).
Stratified random sampling was applied to determine the number of ground truth for each land use categories by using the sampling technique by Cochran (1977).According to sampling technique by Cochran (1977), total of 401 ground truth points was needed to acquire the user's accuracy of 0.8 and standard error of the estimated overall accuracy of 0.02.The necessary ground truth points for each land use categories were 6 points for permanent farmland, 92 points for water bodies, 30 points for swidden fields, 199 points for forest areas, 18 points for settlement areas, and 56 points for orchards and plantation.We collected the ground truth data from the ground survey, topographic map (scale 1:50,000) obtained from the Forest Department, the Ministry of Natural Resource and Environmental Conservation (MONREC), Google Earth imageries, and GeoEye imageries.Accuracy assessment of land cover classification using the confusion matrix, and comparison between the reference data and the corresponding results of the classification were carried out.We computed overall, producer and user accuracies and Kappa Coefficient (Tables 1 and 2).

| Socioeconomic surveys and data analysis
Out of the 23 relocated villages, we selected three villages, anonymously recorded as TLM, LL, and TPK, as samples for socioeconomic assessment (Figure 1).Villages were selected based on the accessibility.We conducted in-depth household interviews to 52, 38, and 32 respondent households which were randomly selected and corresponded to over 30% of the total households in LL, TPK, and TLM, respectively.LL and TLM are located on the main road connecting Naypyidaw and Pinlaung town, while TPK is about 5 km away from the main road (Figure 1).These villages were formerly located inside the reserved forest and villagers were listed as illegal encroachers Collected data on 18 socioeconomic variables (Table 3) were used to calculate the descriptive statistics, based on sampled respondent households (N = 122).Also, we assessed the relationship between contributed income sources and annual household income by Spearman's rank-order correlation.
In Moreover, to clarify the causal impact of dam construction on the socioeconomic condition of villagers, average total household income of three relocated villages were compared with that of five nonrelocated villages derived from the previous study (Cho, 2018) by using one sample t-test.Those five villages in the previous study were assumed to be existed before dam construction as control, because those five villages were located nearby our study site, but those villages were not relocated due to dam construction.
Also, to find out the rice sufficiency, the quantity of rice purchased annually by each household from the market was converted into monetary value and weighed that value with the total annual income.We then estimated the affordability of each household to buy rice (hereafter termed as "rice sufficiency") by comparing household-wise rice production and consumption.
Besides, collinearity is tested among the socioeconomic variables, and principal component analysis (PCA) was conducted on 10 selected variables to explore their inter-relationships and summarize the multiple variables to understand the underlying pattern of socioeconomic conditions in the three studied villages.PCA is considered useful for summarizing multiple variables based on particular characteristics (Maung & Yamamoto, 2008).We extracted the prominent factors for each component by using the Varimax rotation option in IBM SPSS Statistics.PC scores derived from principal components were analyzed to calculate Fisher's Least Significant Differences (LSD) which will be used to determine the differences in farmers' adapted income strategies among three villages.One-way ANOVA, one sample t-test and PCA were performed in IBM SPSS Statistics for Windows, Version 20.0 released by IBM Corp.

| Land use change detection in dam area
In order to detect the land use changes in the ROI of the dam area, land cover change analysis was conducted using the Landsat image.
According to the results of Landsat image analysis of the dam area, the area of farmland notably showed a sharp reduction within 7 years because farmlands were submerged under the expanded water bod- Overall accuracy = 92% Kappa coefficient = 0.88

| Agricultural land use changes
Similarly, detection of agricultural land use change using GeoEye imageries (25 km 2 focal area) also revealed that the area of farmland abruptly decreased from 278 ha to 35 ha within 7 years due to the expanded water bodies (Figure 5, Table 4).The settlement area showed a remarkable increase from 48 ha to 136 ha even though the 23 villages were inundated in the water body.This is because the targeted ROI  4).Notably, transition analysis also showed that the settlement areas and water body increased from 2011 to 2018 (Figure 5, Table 4).

| Socioeconomic conditions after dam construction
According to interview responses, the average family size per household was 4.35, while the average number of laborers and children was 4.34 and 1.13 per household, respectively.The average age of household's head is about 46 and the average education level is primary education.Regarding land possession, the average areas of paddy field, swidden, rubber plantation and fruit tree orchards were 0.  5).
Spearman's correlation analysis gave the result that there was a significant positive correlation between the total household annual income and non-farm income (r s = 0.73, p < 0.01; Figure 6a).A similar relationship was observed between total household annual income and animal husbandry income (r s = 0.74, p < 0.01; Figure 6b).
T A B L E 3 Characteristics of socioeconomic variables.Conversely, a weak but significant correlation was seen between total household annual income and the rubber plantation income (r s = 0.20, p ≤ 0.05; Figure 6c).NTFPs income was not significantly correlated with the total household annual income (Figure 6d), while income from swidden was significantly correlated with the total household annual income (r s = 0.32, p < 0.01; Figure 6e).

| Distribution of income share among the three wealth status groups
Comparing the income share among the three wealth status groups, swidden was the main contributor for LIG and MIG (Table 6).Income from NTFPs also contributed more income to the LIG when compared to other groups.HIG was mainly contributed from non-farm job income (Table 6).

| Comparison of socioeconomic conditions before and after dam construction
The comparative analysis using one sample t-test showed that the average total income of the villages was significantly different before and after dam construction (Table 7) resulting in increased income after dam construction.When we analyzed the differences among the household wealth status before and after dam   sample villages from the other nearby villages were also found in this study.Moreover, some of the households from TPK village moved to the LL and TLM villages because of better accessibility and job opportunities.

| Rice sufficiency
According to the interview survey, the main occupation of the local people was lowland paddy and upland swidden farming prior to the dam construction, and most of the farmers could produce the rice for their self-sufficiency.This result concurs with the findings of land use change analysis, and it approved that farmland loss happened under dam construction and consequently induced the food insecurity of local farmers.In this study, we also assessed whether households produced adequate rice for their self-consumption, and whether they can earn sufficient money to purchase the rice needed for their sustenance after the dam construction.Therefore, the amount of cash needed to buy the rice was deducted from annual absolute income.
Some farmers were unable to earn enough money to buy rice in LL and TLM villages (Table 8).Very few households, 10% of HHs from LL and 26% of HHs from TLM villages, could produce enough rice for their self-consumption, but most had to purchase rice from the market.The farmers from TPK village could afford to buy rice for consumption even though they lost their farmland loss under the dam construction.

| Farmers' adaptation to dam-induced land use changes and household strategies
According to PCA analysis scores, the first component accounted for the highest variance, explained 25%, while the second, third, and fourth component explained 22%, 18%, and 14%, respectively (Table 9).PC1 highlighted the dependency of non-farm job activities, PC2 was related to engagement in newly introduced agricultural practices, PC3 depended on swiddening while PC4 was related to the household profile and consumption.These scores were also used to determine the differences in income strategies among the three villages (Table 10).
ANOVA results showed that engagement in non-farm strategies was significantly different among the three villages (Table 10).High dependency on the non-farm strategies was observed in TPK village, followed by TLM and LL villages.This is possibly because farmers from the TPK village have to find enough money to buy for their household rice consumption as we discuss in Table 8.LSD post hoc test showed that dependency on the non-farm income strategies between LL and TPK was significantly different at the 95% confidence level, but there was no significant difference between LL and TLM, or between TLM and TPK.Besides, villagers from TLM engaged mainly in practicing the newly introduced agriculture, followed by TPK and LL villages.Villagers from LL highly relied on the swidden, followed by TLM and TPK villages.In the case of swidden dependency, there was no significant difference among the three villages ( Values derived from previous study (Cho, 2018) were assumed to be the average total household income before dam construction.
T A B L E 8 Rice sufficiency assessments for three villages.

| Agricultural land use changes and socioeconomic conditions before and after dam construction
The most significant impact of dam construction on the local community was the loss of permanent farmlands.Notably, within the focal 25 km 2 area, 242 ha permanent farmland was submerged under the dam (Figure 5, Table 4).In this study, sample-based accuracy assessment was applied following the good practices for area estimation by Olofsson et al. (2014) since the pixel-based accuracy assessments are leading to bias Cochran (1977).Most of the farmers lost their permanent farmlands that were along the river when the areas were submerged in 2018.Jackson and Sleigh (2000) also reported a similar finding wherein 34,000 ha of agricultural land were inundated due to the construction of the Three Gorges dam in China.In Vietnam, too, 15,738 people lost their land and houses due to the construction of the Ban Chat dam in 2014 (Rousseau et al., 2017).
Affected community members became to rely on the market more for their daily consumption after dam construction.These results were in agreement with the findings of Rubin et al. (1999).
Farmers could not produce sufficient amounts of rice for their selfconsumption, and consequently needed to earn cash to offset this deficiency.It highlighted that some households could not even afford to buy rice after dam construction.All the farmers from TPK could earn the cash income needed to buy the rice from the market because they could engage in diverse household income strategies, and their economic condition was getting better compared to other two villages.Therefore, it can be said that dam construction resulted in improved food security in TPK village.
On the other hand, swidden extent conspicuously decreased probably due to limited land availability and difficult accessibility after the dam construction.Besides, a shift from the fallow land to the Community Forest (CF) under the management of the Forest Department.The swidden transformation into a more permanent form of agriculture was observed in this study.A similar phenomenon of swidden transformation was also found out in the Bago Mountains, Myanmar (Swe & Nawata, 2020).According to the interview survey, prior to the dam construction, the main occupation of the local people are lowland farming and upland swidden framing.Since before the dam construction, the swiddeners in this area have been growing the cash crops, that is, turmeric.This means that the swiddening in Moreover, the swiddeners stated that the price of turmeric had declined, and therefore swiddening alone could not secure their subsistence.
Prior to the dam construction, the main occupation was lowland paddy and upland swidden farming.After they lost their farmland, most of them need to do odd jobs.Due to dam construction, access to farmlands and natural resources was cut off for the swiddeners.
Consequently, the motorboat became an essential vehicle for swiddeners to commute between their swidden fields along the reservoir and their homes.Those who could not afford motorboats inevitably abandoned their swidden farms and had to find alternative employment as hired laborers for existing swidden farmers or take up odd jobs.In case of Vietnam, the Handai farmers started to engage in swiddening in the sloping land adjacent to the floodplain after losing their farmland (Rousseau et al., 2017).
The agricultural land use change map of 2018 showed an increase in orchards and plantations where market-oriented crops such as rubber, betel nut, coffee, pineapple, and banana were recently introduced in these villages after dam construction.The major driving forces for the introduction of new agricultural practice were the farmland loss after dam construction, marginalization of swidden by the Forest Department, changes in market demand, and contract farming introduced by some non-government organizations (NGOs).Coffee plantations were mainly introduced by some NGOs in this study site.
A similar phenomenon was observed in Laos, and the Laos government supported land use segregation with agricultural intensification through the regeneration of forest cover and promoted the swiddeners into farm enterprise by allocating the land zoning, agribusiness, contract farming, and plantation concessions (Castella et al., 2013).
The introduction of market-oriented crops and commercial plantation could possibly relate with the political transition in Myanmar followed by the opening of its economy because the time of construction of the dam coincided with the political transition of Myanmar.This study did not consider controlling the effect of open market economy on the agricultural land use changes and socioeconomic changes.
In this study, TLM and LL villages were located beside the main road, and they have better chances to access the new farming technologies disseminated from outside.Even though some farmers adopted newly introduced agricultural practices such as rubber plantation and fruit tree orchards, the majority could not make the initial investment to do so.However, income contributions from these activities were still limited because those plantations or orchards were still young.Conversely, the correlation between the income from rubber plantation and total annual income showed that the maximum income from rubber could contribute approximately 10% to the total income; this could be assumed to be a feasible economic adaptation to daminduced impacts (Figure 6c).
Considering livelihood, other income sources such as non-farm income and animal husbandry were highly correlated with the total household income (Figure 6a,b); these livelihood activities became their main economically viable income strategies after dam construction.Possibly because displaced and landless farmers did not have technical and financial capacities for such newly introduced agricultural practices, the number of households engaged in such activities was still limited.On the other hand, swidden income still continued to contribute significantly to the total income, suggesting that it functions as a safe livelihood option for swiddeners (Figure 6e).When we compared the income share among the three wealth status groups after dam construction, LIG and MIG were mainly contributed by swidden income (Table 6).It confirmed that swidden still remains as a safety net for the local people in the study sites.Income from NTFPs also contributed more income to the LIG when compared to other groups (Figure 6d).Moe and Liu (2016) and Aye et al. ( 2017) also reported a similar result with this finding.Shackleton and Sheona (2006) also added that the forest resource income significantly contributed to the LIG.Villagers from this study site collected the orchids, bamboo, elephants-yams, cane, as well as broom grass, and sold them to local merchants.In contrast, HIG obtained more earnings from non-farm job activities.The turmeric harvested from the swidden field and NTFPs were sold to the local merchants belonging to HIG or MIG (Table 6).
The increased water bodies after dam construction provided alternative income sources such as seasonal boating to the tourism site for the LIG during the summer season.However, that new alternative can secure the livelihood sustainability of LIG groups.Besides, higher alternative income sources such as restaurants, motels, and home shops were mainly engaged by HIG and MIG.Therefore, annual household income significantly increased especially in the MIG and HIG after the dam construction highlighting that dam construction impacted the marginal households in LIG (Table 7).After the dam construction, the government provided official land title of the relocated village areas, public hospitals, and schools to the relocated local communities.However, the official approval of allocated land ownership seemed to be unsecured according to interview survey.The government also compromised to access the electricity and an electrical meter box to the land lost households.Even after they had access to electricity, the relocated villagers continued collecting fuelwood from surrounding forests.The average consumption of fuelwood was relatively high in this study (10.25 m 3 per household in 2018).In Katha District of Myanmar, the average fuelwood consumption was 3.4 m 3 in the results of Hlaing et al. (2017).This implies that most of the villagers could not afford electricity and therefore still relied on the adja-

| Farmers' adaptation strategies to agricultural land use change
Our interpretation of PCA results showed that farmers typically adopted three types of income strategies of the local people after the dam construction.These were: dependency on non-farm job activities, engagement in newly introduced agricultural practices and dependency on swiddening (Table 9).
The adaptation strategies of the local people were different depending on the village.Non-farm strategies were significantly different across the three villages (Table 10).As TPK village was far from the main road, farmers from TPK village had motorboats and engaged in season-based tourism by renting motorboats to transport tourists to the waterfall located in the northern part of the dam during summer.Since most of the farmers from TPK lost their lowland farm, they relied more on non-farm job activities consequently.TLM villagers engaged mainly in newly introduced agriculture.As TLM was located on the main road, farmers from that village had more opportunities to introduce orchards and plantations.Villagers from LL village highly relied on swidden, which highlights that swidden still remained as a safety net (Table 10).

Figure 2 (
Figure 2 (Swe & Funakawa, 2023).The Paunglaung multi-purpose dam was constructed on the Paunglaung river, approximately midway between Naypyitaw, which is the capital of Myanmar, and Pinlaung town.Its construction was commenced in 2003 and completed in 2013.It resulted in submergence of 9708 acres (39.3 km 2 ) including 23 villages that were relocated to rocky mountainous area nearby.The relocated villages were collectively renamed as Paunglaung Myo Thit, and later notified as official villages by the Ministry of Social Welfare, Relief and Resettlement.Majority are the Burmese, and overlaid the classified images with the masked polygons of farmland and categorized into six land cover classes viz.open forest, closed forest, farmland, roads and settlement, water body, and bare land.Later, to detect the changes in upland swidden fields and lowland permanent paddy fields in detail, we utilized high-resolution Geo-eye images (1.6 m resolution) acquired in January 2011 (noted 2011 in the F I G U R E 1 Location of study sites in the Southern Shan State.[Colour figure can be viewed at wileyonlinelibrary.com]Monthly averages of rainfall (Avg RF) and air temperature (Avg Air Temp) in the Paunglaung, southern Shan Highlands over 1998-2018 (data sources: https://climatecharts.net/for the Paunglaung).later analysis) and February 2019 (noted 2018 later in the analysis).For agricultural land use change map, we particularly focused on the 25 km 2 area covering the agricultural lands of three selected villages sampled in this study.Prior to determining the 25 km 2 region of interest (ROI) for land use analysis, we confirmed the proximity of the agricultural land use, where the villagers from the three sample villages were practicing before and after the dam construction, through the focus group discussion (FGD), key informant interview (KII) and indepth interview survey.Image acquisition dates were selected considering the time of dam construction, image availability, cloud cover, and harvest season to avoid possible confusion in the spectral reflectance between swidden fields and permanent farmland with the surrounding orchards and forests.For the agricultural land use change map, the GeoEye images were classified into five land cover classes for 2011,

(
The Pyihtaungsu Hluttaw, 2018; The State Law and Order Restoration Council, 1992) before dam construction.The villagers mainly engaged in swidden cultivation for cash income and lowland farming for their home consumption.The village territory and its agricultural lands belonged to the reserved forest area which was managed by the Forest Department (FD).Myanmar FD regarded the swidden as one of the factors of deforestation because of its slashing and burning the vegetations and shifting nature.Besides, the Ministry of Forestry (1996) regarded the swidden as the major constraints of the forest policy imperatives, as described in the Forest Policy 1995, "discourage the shifting cultivation practices, which are causing extensive damages to the forests …."(Swe & Nawata, 2020).The FD has tried to mitigate the swidden cultivation areas practiced inside the reserved forest and transformed them into community forests and monoculture teak plantations, which are managed by the local community and Forest Department under the guidance of the Community Forest Instruction 1995(Swe & Nawata, 2020).In the selected sample villages, interviews were also carried out four times using semi-structured questionnaire.The questionnaire covered five main topics: general information of the household, agricultural land use history, agricultural land use change after dam construction, household income, and rice yield from each land category, including costs and benefits of each agricultural land use.Focused group discussions and key informant interviews were also conducted to validate the information.
addition, to check out the average income share by the livelihood activities among wealth status, we stratified the total sampled respondent households (N = 122) into three groups of almost equal sample size by their total household income and compared among the three groups.Based on the total household income, we stratified 40 sampled respondent households into low-income group (LIG), another 40 into medium-income group (MIG), and the remaining 42 into high-income group (HIG).The comparison on income share by livelihood activities among three wealth status was done by one-way ANOVA.
belonged to the proximity area centering the current locations of three sample villages.The changes in settlement area resulted from GeoEye imageries analysis were opposite to Landsat imageries analysis because government relocated all the dam affected villages to sample area of this study and some other villagers also shifted to this area because of accessibility.Area of swidden remarkably decreased from 330 ha to 185 ha during the same period, and in 2018, most of the swidden fields could be found only near settlement areas, along the reservoir and near the newly established rubber plantations and fruit tree orchards.Most of the local people in this area mainly grow turmeric for cash income, and some farmers grow upland rice in swidden fields for their own consumption.They earned the cash income by selling turmeric to merchants who directly came to buy at the villages.According to the interview survey to the merchants who are trading the turmeric, the turmeric was exported to China market.The abrupt introduction of orchards and plantation were observed in 2018, that is, after the dam construction.In accordance with the interview survey, the farmers started to introduce the rubber plantation, beetle nut plantation, coffee plantation and tea orchards after the dam construction based on the market demand and some are introduced by government and NGO planning.Conversely, newly introduced orchards and plantations were largely detected in 2018 images, and these were mainly established by swiddeners near their swidden fields (Figure5, Table

f
Income selling from pig, piglet, and chicken.F I G U R E 4 Land cover changes to water body from 2011 to 2018 using Landsat images.[Colour figure can be viewed at wileyonlinelibrary.com]F I G U R E 5 Map showing the agricultural land use changes from 2011 to 2018 using GeoEye images.[Colour figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 6 (a) Correlation between non-farm income and total income.(b) Correlation between animal husbandry income and total income.(c) Correlation between income from rubber plantation and total income.(d) Correlation between income from NTFPs and total income.(e) Correlation between income from swidden and total income.[Colour figure can be viewed at wileyonlinelibrary.com]T A B L E 6 Result of income share by different livelihood activities in percentage (%) among three different income level groups (N = 122).
this area have already intensified compared to other mountainous areas in Myanmar.The local farmers used to earn profitable income by selling the turmeric before the dam construction.According to the interview survey with the villagers and merchants, the quality and size of the turmeric remarkably reduced because of limited land availability, which led to the shorter fallow age of swidden after the dam construction.The longest fallow period reduced to 3 years; consequently, most of the respondent swiddeners mentioned the lower quality of turmeric due to repeated growing.The fallow period was relatively shorter compared to the other studies in traditional swidden cultivations areas in Myanmar(Swe & Nawata, 2020).Another possible factor for the shorter fallow period could be population growth and internal movement from other areas.This finding concurs withBoserup Agrarian Change Theory (1965).The available land for swidden and swidden income became limited because of indirect impact from dam construction, increasing population, shorter fallow age, and forest conservation activities such as the introduction of monoculture teak plantation, community forests, newly introduced orchards, and teak plantation.However, the land use right and land ownership for those newly introduced rubber plantations, and fruit tree orchards were still in uncertain in this study area according to interview survey and KII.
cent forests.Technically, those villages are located inside the strictly managed reserved forest, and they have limited access to the forest, especially illegal timber extraction is strictly prohibited and NTFPs collection is granted under several conditions by laws and regulations by the Ministry of Natural Resource and Environmental Conservation (MONREC).However, increased dependencies on casual jobs, especially displaced farmers and landless farmers, might lead to increased dependence on the surrounding forests for collecting more NTFPs and engaging in illegal timber extraction.
After the dam construction, local communities were engaged in diverse livelihood strategies.Besides agriculture, local communities also engaged in non-farm activities such as animal husbandry, casual jobs, and NTFP collection, and these livelihood options significantly contributed to household earnings.Some households engaged in newly introduced agricultural practices, which markedly contributed to their household income.Currently, most of the households engaged in non-farm jobs.However, all three villages lacked stable household income strategies after the dam construction.Even swidden still remain as one of their agricultural activities, the farmers changed their farming systems by introducing the fruit tree orchards and plantations.Introduction for new farming also related to the limited availability of fallow land, shortening the fallow period, and consequent factors of low productivity and food insecurity.Additionally, lowland agricultural land use change was linked to the drastic land use changes because of dam construction.Swidden and upland farm changes, such as swidden transformation into a more permanent form of agriculture and intensification, were related to the forest conservation activities, contract farming, changes in demographic trends, and market access.After the dam construction, adaption strategies and the livelihood system of low-income household groups became more diversified, they subsequently had to take up casual jobs.Our study also showed that non-farm jobs were the prominent income strategies that contributed to household earnings.Moreover, engagement in newly introduced agricultural practices can also be one of the feasible alternatives to supplement the household income.The latter could be challenging because of limitations in land availability, initial investment, and road access to farmland areas.Therefore, infrastructure development for accessibility to their farmlands, resolution of land use conflicts between the Forest Department and local communities, as well as incentive policy for integrated agro-based business, are required in these areas.
Accuracy assessment for the land cover change map of Paunglaung reserved forest and agricultural land use change of 25 km 2 area from 2011 and 2018.Accuracy assessment of agricultural land use change map of 25 km 2 area in 2018 (GeoEye image).
ies.Farmland extent of 788 ha, settlements of 1361 ha, and bare land of 304 ha was changed into water bodies.Analyzing the increase in water bodies vis-a-vis other land use categories after dam construction showed that closed forest area of 1602 ha and open forest area of 641 ha were converted into water bodies (Figure4).T A B L E 1

Ks) Income from Swidden (Ks)
Descriptive statistics of socioeconomic variables in three relocated villages in 2018 (N = 122).
1 Income from selling rubber.2Incomefromselling the fruit harvested from fruit tree orchards.3Incomefromsellingcropscultivated in swidden fields (turmeric).4IncomefromsellingNTFPs(elephant-footyam, Bamboo shoots, orchids, broom grass).5Incomefromnon-farmjob (hired labor, governmental staff, restaurant, carpenter, bus car conductor, merchant, shopkeeper, timber trading, and hiring motorboat, and hotel).6Incomeselling from pig, piglet, and chicken.construction, the significant increase in income was observed in the MIG and HIG, whereas the LIG showed the decrease after the dam construction.The former two villages were located beside the road, and they have better chances for non-farm job opportunities and technological dissemination for introducing new agricultural practices.According to the interview survey, local in-migration to the Table 10).Comparison of the average total household income before and after dam construction (one sample t test, N = 122).
T A B L E 9 Results from principal component analysis based on the correlations of 10 social variables.Comparison of income strategies among the three villages.