2016 Theses Doctoral
Biophysical and anthropogenic contributions to fire disturbance dynamics on the peat-swamp landscape, Indonesia
Fires have been increasing in size and frequency across the tropics in recent decades, particularly in tropical peatland areas. Indonesia has the largest amount of tropical peat carbon globally. Fires in fuel-rich tropical peatlands are a major source of carbon emissions, have serious consequences for human health, destroy or degrade habitat, and result in high economic costs. There have been many calls for a better understanding of the relative contributions of the biophysical and anthropogenic factors that drive fire, as this understanding would contribute to the success of efforts to reduce these fires. This dissertation uses remote sensing, fieldwork, and modeling to explore the dynamics of fire disturbance in Indonesia and investigates this disturbance from the framework of coupled human and natural systems, where complex interactions between the social and the biophysical are explicitly considered. Chapters One and Two assess both the influence of various human and biophysical factors to fire probability (Chapter One) and ignitions (Chapter Two) on a peat-swamp forest area in Central Kalimantan, Indonesia, equivalent to a third of Kalimantan's peatland area. A Bayesian modeling approach is used in Chapter One to estimate the effects of atmospheric dryness, human access, vegetation, and hydrology on the probability of fire occurrence. The potential for peatland restoration to offset the impacts of climate on fire occurrence is also explored. I find that climate is the most important factor driving fire occurrence, which is consistent with the findings in many other parts of the tropics. However, two human-driven factors are almost as significant as the influence of climate: drainage canals, which were put in place as part of a failed agricultural project and have lowered the water table; and woody vegetation, which has decreased over time. Chapter Two inspects the oft-asserted claim that escaped fires from oil palm concessions and smallholder farms near settlements are the primary sources of fire ignitions. We evaluate fire origin and spread, and find that most fires originate in non-forest, compared to oil palm concessions, and relatively few originate close to settlements. Moreover, most fires started within oil palm concessions and in close proximity to settlements stay within those boundaries. However, fire ignition density in oil palm concessions and close to settlements is high. Furthermore, increased anthropogenic activity in close proximity to oil palm concessions and settlements produces a detectable pattern of fire activity. These results refute the claim that most fires originate in oil palm concessions, and that fires escaping from oil palm concessions and settlements constitute a major proportion of fires in this study region. However, there is a potential for these land use types to contribute more substantially to the fire landscape if their area expands. Chapter Three examines the potential for the financial incentive mechanism of Roundtable on Sustainable Palm Oil (RSPO) certification, which prohibits the use of fire on certified concessions, to reduce fire activity on oil palm concessions. We examine if RSPO-certified concessions have reduced fire activity in Sumatra and Kalimantan, the leading producers of oil palm both within Indonesia and globally. We also evaluate if this pattern changes with increasing likelihood of fires. These questions are particularly critical in fuel-rich peatland areas, of which approximately 46% was designated as oil palm concession as of 2010. We find that fire activity is significantly lower on RSPO certified concessions than non-RSPO certified concessions when the likelihood of fire is low (i.e., on non-peatlands in wetter years), but not when the likelihood of fire is high (i.e., on non-peatlands in dry years or on peatlands). These chapters advance our understanding of how anthropogenic factors influence the controls of fire in Kalimantan and Sumatra, both directly (i.e., human-caused ignitions) and indirectly (i.e., changing the susceptibility of the landscape to ignitions and to burning). The findings presented in this dissertation indicate that oil palm concessions are associated with high fire probability (Chapter One) and a substantial amount of ignitions and relatively high ignition density (Chapter Two). One of the more pointed ways to target fire on oil palm concessions is through RSPO certification; however, we find that certification is only effective when fire likelihood is already low, suggesting that, in order for this mechanism to reduce fire, more assistance may be needed to control fires in dry years and on peatlands (Chapter Three). Non-forested, degraded areas contribute much more to fire activity than oil palm on this landscape; these areas experience the greatest number of ignitions, have highest ignition density, and are the primary source of forest fires (Chapter Two). Furthermore, the declines in vegetation and the hydrological alteration in these degraded areas contribute substantially to fire occurrence (Chapter One). Effective fire management in this area, including fire prevention and suppression efforts, should therefore target not just oil palm concessions and smallholdings around settlements, but should also focus strongly on non-forested, degraded areas – and in particular those near oil palm concession boundaries and outside the immediate vicinity of settlements – where fire probability is high and where ignitions and fires escaping into forest are most likely to occur. Rehabilitation of the degraded landscape through restoring hydrology and replanting will be key to fire reduction, and can offset the effects of climate on fire in this landscape. The methodological approaches in this dissertation demonstrate ways in which remote sensing and analytical technologies can be used to answer complex questions about coupled human and natural systems that fuse social and environmental data, for both theoretical and management applications. Chapter One uses biophysical information from remotely sensed products and fieldwork with information about human access on the landscape and integrates them together with Moderate Resolution Imaging Spectroradiometer (MODIS) Active Fire detections under a Bayesian framework. Chapters Two and Three use a novel technique to cluster remotely sensed data on fire occurrence (MODIS Active Fire detections) into fire events so that ignitions can be isolated. This technique allows us to answer questions related to fire origin, spread, and impact that cannot be investigated by evaluating fire detections alone. This dissertation addresses a gap in knowledge regarding the anthropogenic contributions to increased fire probability and to ignitions in peat swamp, and the approaches could be applied to other degraded peatland areas in Indonesia that are candidate sites for restoration (e.g., under the newly established Peatland Restoration Agency), and to degraded peatlands that experience a novel fire regime in other parts of the tropics. Furthermore, this dissertation evaluates the capacity for RSPO certification to reduce fire activity on oil palm concessions across Sumatra and Kalimantan, Indonesia, and the analyses conducted could be applied to landscapes in other parts of the tropics experiencing oil palm development. In conclusion, the research findings presented in this dissertation are a product of combining social and environmental data and evaluating this data with a suite of classic and novel modeling approaches. This dissertation is presented in the hope that it contributes to our understanding of fire dynamics in the globally important peat-swamp forest, Indonesia, and thus our capacity to manage these disturbances.
- Cattau_columbia_0054D_13543.pdf binary/octet-stream 5.06 MB Download File
- Academic Units
- Ecology, Evolution, and Environmental Biology
- Thesis Advisors
- DeFries, Ruth S.
- Ph.D., Columbia University