Impacts of fire and prospects for recovery in a tropical peat forest ecosystem.

Uncontrolled fires place considerable burdens on forest ecosystems, compromising our ability to meet conservation and restoration goals. A poor understanding of the impacts of fire on ecosystems and their biodiversity exacerbates this challenge, particularly in tropical regions where few studies have applied consistent analytical techniques to examine a broad range of ecological impacts over multiyear time frames. We compiled 16 y of data on ecosystem properties (17 variables) and biodiversity (21 variables) from a tropical peatland in Indonesia to assess fire impacts and infer the potential for recovery. Burned forest experienced altered structural and microclimatic conditions, resulting in a proliferation of nonforest vegetation and erosion of forest ecosystem properties and biodiversity. Compared to unburned forest, habitat structure, tree density, and canopy cover deteriorated by 58 to 98%, while declines in species diversity and abundance were most pronounced for trees, damselflies, and butterflies, particularly for forest specialist species. Tracking ecosystem property and biodiversity datasets over time revealed most to be sensitive to recurrent high-intensity fires within the wider landscape. These megafires immediately compromised water quality and tree reproductive phenology, crashing commercially valuable fish populations within 3 mo and driving a gradual decline in threatened vertebrates over 9 mo. Burned forest remained structurally compromised long after a burn event, but vegetation showed some signs of recovery over a 12-y period. Our findings demonstrate that, if left uncontrolled, fire may be a pervasive threat to the ecological functioning of tropical forests, underscoring the importance of fire prevention and long-term restoration efforts, as exemplified in Indonesia.

Ecosystem-scale methane flux in tropical peat swamp forest in Indonesia.

Data on ecosystem-scale methane (CH4 ) fluxes in tropical peatlands are currently lacking in the global CH4 budget. Although the waterlogged Indonesian peatlands contain the largest share of peat carbon in South-East Asia, ecosystem-scale CH4 budgets have not yet been reported, although these peatlands have the potential to emit CH4 . We observed 1-year variations in the ecosystem-scale CH4 flux in an undrained secondary peat swamp forest in central Kalimantan, Indonesia, using the eddy covariance method. We found that the peat swamp forest switched from being a CH4 sink during the dry season (as low as -8.9 mg C m-2  day-1 ) to a source of CH4 during the wet season (up to 10.7 mg C m-2  day-1 ), and this was dependent on changes in the groundwater level (GWL). The high GWL during the wet season enhanced the anaerobic CH4 production in the surface layer that had more labile organic matter. However, the CH4 emission also increased when the GWL dropped during dry spells in the wet season. The annual CH4 budget in the studied tropical peat swamp forest (0.09-0.17 g C m-2  year-1 ) was much lower than that in northern, temperate, and subtropical wetlands. We found that CH4 fluxes had almost no effect on the global warming gas budget of the peat swamp forest, and values were only a few percent less than the CO2 fluxes at the same site. In addition, we conducted anaerobic soil incubation experiments to examine the effect of land-use change on CH4 production. The results indicated much higher CH4 production potential in undrained forest soil than in drained or drained and burned ex-forest soils. However, although CH4 production decreased in drained soils relative to undrained soils, conserving pristine peat swamp forests with high GWLs is important to suppress global warming because CO2 emissions increase in drained peatlands.

Factors affecting oxidative peat decomposition due to land use in tropical peat swamp forests in Indonesia.

The increasing frequency of fire due to drainage of tropical peatland has become a major environmental problem in Southeast Asia. To clarify the effects of changes in land use on carbon dioxide emissions, we measured oxidative peat decomposition (PD) at different stages of disturbance at three sites in Central Kalimantan, Indonesia: an undrained peat swamp forest (UF), a heavily drained peat swamp forest (DF), and a drained and burned ex-forest (DB). PD exhibited seasonality, being less in the wet season and greater in the dry season. From February 2014 to December 2015, mean PD (±SE) were 1.90±0.19, 2.30±0.33, and 1.97±0.25µmolm-2s-1 at UF, DF, and DB, respectively. The groundwater level (GWL) was a major controlling factor of PD at all sites. At UF and DF, PD and GWL showed significant quadratic relationships. At DB, PD and GWL showed significant positive and negative relationships during the dry and wet seasons, respectively. Using these relationships, we estimated annual PD from GWL data for 2014 and 2015 as 698 and 745gCm-2yr-1 at UF (mean GWL: -0.23 and -0.39m), 775 and 825gCm-2yr-1 at DF (-0.55 and -0.59m), and 646 and 748gCm-2yr-1 at DB (-0.22 and -0.62m), respectively. The annual PD was significantly higher in DF than in UF or DB, in both years. Despite the very dry conditions, the annual PD values at these sites were much lower than those reported for tropical peat at plantations (e.g., oil palm, rubber, and acacia). The differences in the relationship between PD and GWL indicate that separate estimations are required for each type of land. Moreover, our results suggest that PD can be enhanced by drainage both in forests and at burned sites.

Evapotranspiration of tropical peat swamp forests.

In Southeast Asia, peatland is widely distributed and has accumulated a massive amount of soil carbon, coexisting with peat swamp forest (PSF). The peatland, however, has been rapidly degraded by deforestation, fires, and drainage for the last two decades. Such disturbances change hydrological conditions, typically groundwater level (GWL), and accelerate oxidative peat decomposition. Evapotranspiration (ET) is a major determinant of GWL, whereas information on the ET of PSF is limited. Therefore, we measured ET using the eddy covariance technique for 4-6 years between 2002 and 2009, including El Niño and La Niña events, at three sites in Central Kalimantan, Indonesia. The sites were different in disturbance degree: a PSF with little drainage (UF), a heavily drained PSF (DF), and a drained burnt ex-PSF (DB); GWL was significantly lowered at DF, especially in the dry season. The ET showed a clear seasonal variation with a peak in the mid-dry season and a large decrease in the late dry season, mainly following seasonal variation in net radiation (Rn ). The Rn drastically decreased with dense smoke from peat fires in the late dry season. Annual ET forced to close energy balance for 4 years was 1636 ± 53, 1553 ± 117, and 1374 ± 75 mm yr(-1) (mean ± 1 standard deviation), respectively, at UF, DF, and DB. The undrained PSF (UF) had high and rather stable annual ET, independently of El Niño and La Niña events, in comparison with other tropical rainforests. The minimum monthly-mean GWL explained 80% of interannual variation in ET for the forest sites (UF and DF); the positive relationship between ET and GWL indicates that drainage by a canal decreased ET at DF through lowering GWL. In addition, ET was decreased by 16% at DB in comparison with UF chiefly because of vegetation loss through fires.

Carbon dioxide emissions through oxidative peat decomposition on a burnt tropical peatland.

In Southeast Asia, a huge amount of peat has accumulated under swamp forests over millennia. Fires have been widely used for land clearing after timber extraction, thus land conversion and land management with logging and drainage are strongly associated with fire activity. During recent El Niño years, tropical peatlands have been severely fire-affected and peatland fires enlarged. To investigate the impact of peat fires on the regional and global carbon balances, it is crucial to assess not only direct carbon emissions through peat combustion but also oxidative peat decomposition after fires. However, there is little information on the carbon dynamics of tropical peat damaged by fires. Therefore, we continuously measured soil CO2 efflux [peat respiration (RP)] through oxidative peat decomposition using six automated chambers on a burnt peat area, from which about 0.7 m of the upper peat had been lost during two fires, in Central Kalimantan, Indonesia. The RP showed a clear seasonal variation with higher values in the dry season. The RP increased logarithmically as groundwater level (GWL) lowered. Temperature sensitivity or Q10 of RP decreased as GWL lowered, mainly because the vertical distribution of RP would shift downward with the expansion of an unsaturated soil zone. Although soil temperature at the burnt open area was higher than that in a near peat swamp forest, model simulation suggests that the effect of temperature rise on RP is small. Annual gap-filled RP was 382 ± 82 (the mean ± 1 SD of six chambers) and 362 ± 74 gC m(-2)  yr(-1) during 2004-2005 and during 2005-2006 years, respectively. Simulated RP showed a significant negative relationship with GWL on an annual basis, which suggests that every GWL lowering by 0.1 m causes additional RP of 89 gC m(-2)  yr(-1) . The RP accounted for 21-24% of ecosystem respiration on an annual basis.