Understanding forest cover dynamics is important for a nation’s environmental, social and political commitments. In the past decade, Myanmar had the highest deforestation rate, in mainland South East Asia (Hansen et al., 2013). Further, in 2009, Myanmar embarked on a landmark political change from military regime to democratic transition which significantly impacted its forest cover. Myanmar also ranks first with respect to forest fires in South/Southeast Asia. In Myanmar, forest cover loss and fire are intrinsically linked through the traditional taungya system of slash and burn. Thus, quantifying factors controlling forest fires in Myanmar is an important topic that needs attention. Although the Myanmar government established protected areas throughout the country to conserve forests, their effectiveness remains unevaluated.

This dissertation aims to understand the current status of forest cover dynamics in Myanmar. The five chapters in this dissertation address the impact of the political transition on forest cover loss and fragmentation, fire disturbance in tropical evergreen and deciduous forests including the factors controlling vegetation fires in the protected and non-protected forests. The dissertation contributes to the existing knowledge in land cover and land use change science (LCLUC),

ii

especially the impact of institutional changes on forest cover in the tropics. The analysis of the relationship between forest loss, fire and effectiveness of the protected areas addressed in the study, contributes to regional knowledge on fire and conservation science respectively.

The findings of this dissertation depict that in Myanmar, the political transition to democracy significantly influenced its forest cover. Our analysis showed that during 2001-2014, a total loss of 2,030,101 ha of forest occurred at the rate of 145,007.21 ha/year with a linear increase of 15,359 (±1793) ha/year. The observed increase in variance in between 2008-2011 coincides with political transition period which started with the formation of the new Constitution in 2008 and ended with the military government handing over power to the democratic government in 2011. Analysis of trend and variance patterns of two landscape fragmentation metrics (Number of Patches and Mean Patch Area) at the provincial level show the influence of the political transition on landscape fragmentation. The impact of political transition was more pronounced in provinces associated with plantations and urban areas. Among the rubber producing States, the border States, Shan, Kayah, and Kayin were more impacted compared to inland Mon. Tanintharyi and Bago Regions showed higher variance in residuals of both metrics before the transition occurred due to the military government supported oil palm and teak plantations. Fragmentation and the variance in fragmentation metrics in Kachin increased post 2008. Apart from plantation areas, urban areas like Yangon and Mandalay showed high fragmentation post 2009 period after the new government was formed. We attribute the forest loss and fragmentation to the economic and structural reforms of the democratic government, specifically to the increased granting of agricultural concessions and logging for plantations.

iii

A study of the fire regime from 2003 to 2012 using MODIS satellite data suggested March as the peak of the fire season with 12900 km2 of Burned Area (BA) and 95000 fire counts. Forests accounted for majority (41.3%) of the total BA and most fires (89.7%) resulted in medium or high vegetation disturbance. A higher negative correlation between BA and Gross Primary Productivity (GPP) was reported for deciduous forests than for evergreen forests (r=0.49 vs r = 0.36, p ~ 0). A maximum decrease in 29% of original GPP (2007-2012) was observed in the evergreen forest patches. The scale-dependent correlation analysis suggested significant BA-GPP correlation at 1 × 1 degree, as compared to finer resolutions. These results highlight the significance of fires impacting carbon cycle.

An in-depth analysis of fire causative factors in Myanmar was studied. The mean fire density in non-protected areas was found to be two times more than in protected areas. Fire-land cover partition analysis suggested dominant fire occurrences in the savannas (protected areas) and woody savannas (non-protected areas). The five major fire causative factors in protected areas in descending order were found to be population density, land cover, tree cover percent, travel time from nearest city and temperature. The causative factors in non-protected areas were population density, tree cover percent, travel time from nearest city, temperature and elevation. The fire susceptibility analysis showed distinct spatial patterns with central Myanmar as a hot spot region of vegetation fires. Results from propensity score matching suggested that forests within protected areas have 11% less fires than non-protected areas. These findings provide information to policy makers about the current forest loss, forest fragmentation and forest fire hotspots, status of forest conservation and can be used to inform, update or evaluate policies. These findings are timely and can guide policy makers to arrive at best management strategies as the new government is formulating policies and laws and amending old ones to aid forest conservation.