Forests, peat, and past fires: understanding the drivers of past fire in Amazonian forests

Abstract

The Amazon Basin is a biome of global importance due to its crucial role in hydrological cycling, biodiversity, and climate stabilisation over long periods. However, over the last decades, it has been threatened by unprecedented disturbances from climate change and anthropogenic activities. Fire has been one of the most significant threats to the area, and this is especially true for carbon-rich peatlands that have been degraded or drained. A better understanding of past fire regimes in Amazonian peatlands can inform current fire regulation and management. This thesis uses palaeoecological proxies to reconstruct fire history over thousands of years to unravel the response of peatland ecosystems to possible fire drivers, including climatic variables and human activities. Pan-tropical peatland records were synthesised to reflect the regional variability in peatland burning. An overall decline in peatland burning was observed in the Neotropics over the past two millennia, probably associated with cooler and wetter climates. In contrast, drier and warmer climates with more peatland burning can be observed during the period, such as in years with intensified El Niño-Southern Oscillation (ENSO) activity. On a continental scale, the increased fire trend during the 20th century in the Neotropical ecoregion is not as pronounced as those located in the Indomalayan and Australasian ecoregions. The comparison between peatland and wider landscape burning across the tropics showed lower burning levels in peatland ecosystems, suggesting that peatlands are less susceptible to fires, mostly likely due to naturally permanently waterlogged conditions. The reconstructions of fire history from seven peatlands in western Amazonia confirm the impact of ENSO activity. These peatland records allow the reconstruction of fires over the last 4,300 years, using the combination of charcoal and pyrogenic carbon (PyC) proxies. Two main phases of fire history were identified: one with high-intensity fires between 4,300 and 3,000 cal yr BP and another with low-intensity fires for the last 2,000 years. Among these seven peatlands, three palm swamp forests were selected for a detailed exploration of vegetation and fire dynamics over the last 2,600 years. The absence of fire during possible dry periods reflects the fire-rare nature of peatland ecosystems. Additionally, subtle changes in vegetation even during periods of more fires suggest a certain resilience of these water-logged ecosystems to fire disturbances. Although there was no direct evidence of human activities in our records, the possible human impact inferred from the presence of macro-charcoal in a wet climate should not be ignored. PyC, a legacy of past fires, was estimated in peatlands across the Amazon Basin based on compiled datasets. While uncertainties and limitations remain, basin-wide predictions could contribute to a better understanding of regional and global carbon cycles and model predictions of future carbon changes. This thesis demonstrates the influence of climate, humans and vegetation on fires recorded in lowland Amazonian peatlands, suggesting that increased burning could result from warmer, drier climates and/or climate anomalies (intense drought pressure), available fuels (aboveground biomass and dried peat), and more importantly, human activities. Given that the Amazon Basin experiences more drought, land use changes and fires in modern times, peatland conservation and fire management are needed to preserve these diverse and carbon-rich ecosystems, particularly in climatically dry years and to avoid similar trajectories to those of Indonesian peatlands, which experience high-frequency fires that are complex to manage and control.