Biomass burning drives atmospheric nutrient redistribution within forested peatlands in Borneo

Biomass burning plays a critical role not only in atmospheric emissions, but also in the deposition and redistribution of biologically important nutrients within tropical landscapes. We quantified the influence of fire on biogeochemical fluxes of nitrogen (N), phosphorus (P), and sulfur (S) in a 12 ha forested peatland in West Kalimantan, Indonesia. Total (inorganic + organic) N, NO 3 − –N, NH 4 + –N, total P, PO 4 3 − –P, and SO 4 2 − –S fluxes were measured in throughfall and bulk rainfall weekly from July 2013 to September 2014. To identify fire events, we used concentrations of particulate matter (PM10) and MODIS Active Fire Product counts within 20 and 100 km radius buffers surrounding the site. Dominant sources of throughfall nutrient deposition were explored using cluster and back-trajectory analysis. Our findings show that this Bornean peatland receives some of the highest P (7.9 kg PO 4 3 − –P ha−1yr−1) and S (42 kg SO 4 2 − –S ha−1yr−1) deposition reported globally, and that N deposition (8.7 kg inorganic N ha−1yr−1) exceeds critical load limits suggested for tropical forests. Six major dry periods and associated fire events occurred during the study. Seventy-eight percent of fires within 20 km and 40% within 100 km of the site were detected within oil palm plantation leases (industrial agriculture) on peatlands. These fires had a disproportionate impact on below-canopy nutrient fluxes. Post-fire throughfall events contributed >30% of the total inorganic N ( NO 3 − –N + NH 4 + –N) and PO 4 3 − –P flux to peatland soils during the study period. Our results indicate that biomass burning associated with agricultural peat fires is a major source of N, P, and S in throughfall and could rival industrial pollution as an input to these systems during major fire years. Given the sheer magnitude of fluxes reported here, fire-related redistribution of nutrients may have significant fertilizing or acidifying effects on a diversity of nutrient-limited ecosystems.


Introduction
Coupled changes in climate and land use are altering fire regimes in tropical forest regions (Siegert et al 2001, Cochrane 2003, Bowman et al 2009, Hansen et al 2009, Margono et al 2014, with consequences for global, regional, and local biogeochemical cycles (Crutzen and Andreae 1990, Andreae and Merlet 2001, van der Werf et al 2010. For example, over the past two decades, Indonesia has transitioned from a historically long fire-return interval regime (Goldammer 2007) to become one of the most important sources of biomass burning emissions worldwide (van der Werf et al 2010). In 2015, carbon dioxide (CO 2 ), methane (CH 4 ), and nitrous oxide (N 2 O) CO 2 -equivalent emissions from Indonesian fires were estimated to exceed Japan's 2013 fossil fuel CO 2 emissions (Global Fire Emissions Database 2015).
Although fires in Indonesia are anthropogenic-ignited during drier conditions primarily to clear vegetation for agriculture--these land-clearing fires and associated wildfires have increased dramatically in Sumatra and Kalimantan since the early 1990s as a direct result of governmental land-use policies, particularly logging and forest degradation, economic incentives for industrial agriculture, and governmentsponsored transmigration, among other factors (Siegert et al 2001, Curran et al 2004, Dennis et al 2005, Langner et al 2007, Murdiyarso and Adiningsih 2007, Page et al 2009. El Niño Southern Oscillation (ENSO) conditions exacerbate dry season droughts and result in nonlinear increases in fire activity (van der Werf et al 2008). Mounting evidence suggests that fire regimes in Sumatra and Kalimantan are again changing; large severe fires can occur following brief dry spells and during non-ENSO years , Gaveau et al 2014. Moreover, an increasing proportion of these fires occur within peatlands (Miettinen et al 2011a, Marlier et al 2015.
Indonesian peatlands (∼200 000 km 2 ) comprise ∼5% of global peatland area and store an estimated 57 Gt carbon (C), ∼9.4-12% of the global peat C pool (Page et al 2011). These ecosystems are highly sensitive to reductions in precipitation--especially extended ENSO-associated droughts (Field et al 2009)--that lower the water table, thereby leading to peat desiccation and enhancing peat flammability (Usup et al 2004, Wooster et al 2012. Due to high rates of forest conversion and peatland drainage associated with industrial-scale agricultural expansion (e.g., oil palm), peatlands have become increasingly vulnerable to fire (Hooijer et al 2006, Miettinen et al 2011a, Margono et al 2014, Turetsky et al 2015. Peatland fires not only turn peatlands from C sink to source, but they also release large amounts of other gases and particles to the atmosphere; peat emissions are often orders of magnitude greater than those from other land-cover types (Christian et al 2003, Iinuma et al 2007, van der Werf et al 2010, Akagi et al 2011. For example, peat combustion is a significant source of biologically important nutrients, including nitrogen (N), phosphorus (P), sulfur (S), and potassium (K) (Andriesse 1988) that are often stored in peat for millennia (Weiss et al 2002). Emitted gases, particles, and chemicals are eventually delivered to downwind ecosystems, either dissolved in precipitation (wet or fog deposition) or directly in dry form (dry deposition; Weathers and Ponette-González 2011). After hitting the forest canopy, exchange via uptake and/or leaching can occur before nutrients are deposited to soils in throughfall (Ponette-González et al 2014), resulting in the redistribution of nutrients across the landscape (Ponette-González et al 2010a).
Ombrogenous peatlands are particularly sensitive to changes in atmospheric nutrient loading because they derive nutrients exclusively from the atmosphere and are also often nutrient limited (Tipping et al 2014). Research in boreal and temperate peatlands indicates that increases in atmospheric deposition can alter ecosystem C fluxes through effects on species composition, productivity, decomposition, and CH 4 flux (Gauci et al 2004, Bragazza et al 2006, Limpens et al 2008, Frolking et al 2011. Given the magnitude of peatland fire emissions and the potential consequences of resultant deposition for biogeochemical cycling, here we address the following question: what is the influence of fires on biogeochemical fluxes of three biologically important plant nutrients (N, P, S) to adjacent, nutrient-limited peatland ecosystems? Specifically, what is the magnitude of fire-related N, P, and S fluxes to peat soils and what are the most important sources of N, P, and S in throughfall?
Climatic data compiled from Supadio International Airport in Pontianak (8 km from site) indicate that mean annual rainfall  is 3197±481 mm (±SD), with considerable intra-and inter-annual precipitation variability due to local land-sea breezes, seasonal monsoons, and ENSO events (Qian et al 2013). Total annual rainfall ranges nearly two-fold, 2329-4912 mm, from strong dry El Niño to strong wet La Niña years. Monthly rainfall is highest during the onset of the northeast monsoon (Oct-Dec: 314-359 mm mo −1 ) and lowest when the southwest monsoon prevails (Jun-Sep: 186-250 mm mo −1 ). Fire events mirror these rainfall patterns with few fires recorded during wet months and peak burning during drier periods (Vadrevu et al 2015).

Fire, land use, and air quality observations
The MODIS Active Fire Product (Giglio et al 2003) was used to determine fire locations within 20 and 100 km radius buffers surrounding the forested peatland. Fire counts are likely underestimated with these MODIS data because clouds and smoke as well as forests with high LAI obscure fire detection (Roy et al 2008). Moreover, smoldering fires typical of peatlands are impossible to discern from flaming fires based on MODIS data, and are likely to be undetectable if they occur belowground (Elvidge et al 2015). Within 100 km surrounding this site, we digitized provincial and district agricultural oil palm lease records. Using peatland distribution information from RePPProT (1990) and Wetlands International (Wahyunto and Subagjo 2004), we then compiled fire hotspots by: (1) oil palm lease versus non oil palm lease; and, (2) soil type (i.e., mineral or peatlands; figure 1(c)). Daily observations of particulate matter concentration <10 μm diameter (PM 10 ) were obtained from the government agency Badan Lingkungan Hidup Kota Pontianak (23 km from site).

Throughfall chemical fluxes
Across the 12 ha forested peatland, we measured dissolved total N, -NO 3 -N, + NH 4 -N, total P, -PO 4 3 -P, -SO 4 2 -S, and Cl − in throughfall, water that flows through plant canopies and carries nutrients and pollutants from atmospheric deposition and canopy processing to soils. Using a spherical densiometer, the study site was stratified by canopy cover (0%-35% open, 35%-65% intermediate, 65%-100% closed). Six throughfall collectors were established randomly within each canopy cover class, for 18 total collectors (figure S1). Three bulk rainfall deposition collectors (i.e., collectors that remain open between sampling events) were established in an adjacent clearing without canopy cover. Rainfall volume was recorded every five minutes with a 20 cm diameter tipping bucket rain gauge (Rainwise, Rainew 111) connected to a HOBO event logger (Onset, UA-003-64).
Throughfall and bulk deposition collectors were constructed with a 15 cm diameter funnel set on a PVC tube 1 m aboveground (after Weathers et al 2006, Ponette-González et al 2010b. Funnels were rinsed with deionized water and connected to plastic tubing, which drained into a 5 l plastic jug. A polywool filter was placed inside the funnel to prevent sample contamination. Water samples were collected weekly from July 2013 to September 2014. After each sample week (i.e., event), water volume was measured and a 300 ml aliquot was collected in a plastic bottle, wrapped in aluminum foil, and immediately frozen. Frozen samples were shipped to the Indonesian Institute of Sciences Limnology Laboratory in Bogor, Indonesia, where they were analyzed for total N (inorganic+organic N), -SO 4 2 -S, and Cl − following standard protocols (APHA 1975(APHA , 2005(APHA , 2012, text S1).
Organic N was estimated as the difference between total N and dissolved inorganic N (DIN, -NO 3 -N+ + NH 4 -N) and organic P as the difference between total P and -PO 4 3 -P. Chloride was used to determine the seasalt- Volume-weighted mean (VWM) bulk rainfall and throughfall concentrations were computed for each sample event (i.e., week) or overall: where i is the collector, conc is the solute concentration (mg l −1 ) and precip is the rainfall or throughfall amount (mm). Bulk rainfall deposition and throughfall chemical fluxes for each event were then calculated by multiplying VWM concentrations by water volume (de Souza et al 2015). To compute the weighted mean and standard deviation of constituent concentrations across all weeks, as well as post-fire and normal weeks, we used the SDMTools package in R (VanDerWal et al 2014). From weekly VWMs, we calculated overall mean and standard deviation, weighting by the total volume across all collectors each week. We applied weighted least squares regression to determine whether concentrations between post-fire and normal weeks were significantly different.

Chemical signatures and sources
Hierarchical cluster analysis was used to identify groups of throughfall events with similar chemical composition. Prior to analysis, non-normally distributed data were log-transformed and all data were standardized by subtracting the mean and dividing by the standard deviation of each variable (Templ et al 2008). Cluster analysis was performed on -SO 4 2 -S, and Cl − concentrations using Ward's clustering method. Organic P was correlated with -PO 4 3 -P and thus excluded to minimize redundancy. After clustering, sources of elements in throughfall were explored using various methods. We examined between-cluster differences in molar ratios (N/P, using Kruskal-Wallis tests with post-hoc Steel Dwaas comparisons and within-cluster Pearson correlation coefficients among variables. Significance was set at p<0.05. Analyses were performed using JMP v12 (SAS Institute, Cary, NC, USA). We also applied NOAA's Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT; Draxler and Rolph 2015) model to compute 168-h backward air mass trajectories for each throughfall event. Trajectories were computed at the surface (100 m), within the boundary layer (500 m), and above the boundary layer (1500 m).

Dry periods and fire
During the 15 month (65 week) study (July 2013-September 2014), total rainfall was 2819 mm (188 mm mo −1 ), well below the long-term monthly average for this region (266 mm mo −1 ). Six dry periods occurred that varied in duration and severity (figure 2(a), table S1). These dry periods were associated with fire pulses--defined here as two or more weeks with 10 fire hotspots detected within 20 km of the site--that also differed in magnitude, proximity to the focal site, and land-cover source (figure 2(b), table S1).
Total hotspots ranged 20-fold (10-196 hotspots) within 20 km of the site, and seven-fold (138-968 hotspots) within 100 km of the site. Approximately 80% of fires (20 km) and 53% of fires (100 km) were detected within oil palm leases, which covered 64% (20 km) and 49% (100 km) of the land area. A disproportionate number of fires was also detected within peatlands: 95% of fires versus 65% of land area within 20 km and 76% of fires versus 42% of land area within 100 km of the site. Seventy-eight percent (20 km) and 40% (100 km) of fires were located within oil palm leases on peatlands. PM 10 levels closely tracked the fire pulses and exceeded 150 μg m −3 , the Indonesian daily ambient air quality standard, during all fire events 3.2. Throughfall N, P, and S fluxes From July 2013 to September 2014, a total of 44 throughfall events (i.e., weeks) were sampled. Average VWM concentrations of all solutes measured in throughfall over the study period were significantly greater than in rainwater (p<0.005, Wilcoxon signed-rank test). Orthophosphate-P showed the most dramatic enrichment in throughfall concentration. Concentrations of -PO 4 3 -P were 35-fold higher, while total P and organic P were 13-fold and four-fold higher, respectively, in throughfall than in bulk rainfall (table S2). For total N, -NO 3 -N, + NH 4 -N, DIN, and -SO 4 2 -S, throughfall to bulk rainfall enrichment ratios ranged 1.4-2.3, indicating increasing concentrations with passage of water through the forest canopy.
Throughfall water flux was 95% of incoming rainfall (2688 mm). Notably, total P input, which averaged 0.024 kg P −1 ha −1 wk −1 in bulk rainfall (table S2), was 12-fold higher in throughfall, 0.29 kg P −1 ha −1 wk −1 (table 1). Dry deposition and/or canopy leaching (throughfall-bulk deposition) comprised 92% of the total P, 97% of the -PO 4 3 -P and 76% of the organic P flux to the soil. The form of P delivered in rainwater differed markedly from that in below-canopy throughfall. Orthophosphate-P increased from~33% of total P in rainwater to >78% of total P in throughfall, indicating that -PO 4 3 -P is deposited primarily in dry form and/or that organic P is converted to -PO 4 3 -P after deposition (Tipping et al 2014).
Total throughfall -NO 3 -N and + NH 4 -N fluxes were 117% and 110% higher (table 1), respectively, than bulk rainfall deposition of these ions (table S2). On average, mean DIN flux to soils was 0.25 kg ha −1 wk −1 . Differences between throughfall flux and bulk deposition were greater for DIN (5.8 kg ha −1 /15 mos) than for organic N (4.7 kg ha −1 / 15 mos), indicating higher dry deposition and canopy exchange of DIN than of organic N. However, organic N represented a greater fraction of total N than DIN in both bulk rainfall (69%, table S2) and throughfall flux (60% , table 1). Total -SO 4 2 -S flux was two-fold higher than total N flux. Bulk rain S deposition was 31 kg ha −1 /15 mos, of which 45% was nss-   Table 1. Volume-weighted mean (VWM) throughfall concentrations (mg l −1 ) and throughfall fluxes to soils for post-fire throughfall events (n=6, preceding weeks 10 fires within 20 km buffer followed by rain), normal events (n=38), and all events sampled from July 2013 to September 2014 in a rainfed forested peatland in West Kalimantan, Indonesia. Contribution of post-fire and normal events to the total throughfall flux is shown over the entire 15 month sampling period. Means±SD are shown for VWM. Asterisk ( * ) indicates significant differences in VWM between post-fire throughfall and normal events (p<0.05). Throughfall fluxes may not add due to rounding.
Throughfall flux (kg ha −1 ) Nineteen to 34% of the total throughfall chemical flux over the 15 month period was accounted for by six post-fire events (table 1). During these events, over a third of 'biologically-available' nutrient deposition occurred: We also compared total N, total P, and S fluxes sampled in throughfall during two water years (October 2012-September 2013, October 2013-September 2014; text S2) with similar rainfall but a three-fold difference in total annual fire counts 20 km from the site (figure 3). Total P and -SO 4 2 -S in throughfall were two-fold higher, while total N was only slightly greater, during the high-fire compared to the low-fire year (figure 3). The slight increase in total N likely reflects the prevalence of organic N in throughfall, which was little affected by fire (table 1).

Chemical signatures and sources
Cluster analysis discerned five unique throughfall event clusters (table 2)  contained three of six post-fire throughfall events, with most samples collected during peak biomass burning months in Indonesia and northern Australia (July-September; figure 4). Backward trajectories confirmed prevailing southeasterly winds that originated in northern Australia and then passed over southern Kalimantan before arriving at the focal site.
For Cluster 2, the mean    (table 2) and, as with Cluster 1, throughfall -SO 4 2 concentrations were strongly positively correlated with organic P (r=0.72, p=0.0086). Sulfate concentrations were also correlated with -PO 4 3 (r=0.75, p=0.0053). Cluster 4 throughfall events were primarily influenced by southeasterly winds and otherwise variable winds in January, April, and May (figure 4). All events were temporally subsequent to two major fire pulses, suggesting a mixture of burning (Cluster 1), pollution, and agricultural sources (Cluster 2).
Cluster 5  ratio and distances traveled by the air masses were notably less than the other clusters, suggesting that throughfall chemistry was influenced mainly by local biomass burning rather than by long-range sources. and throughfall flux (7.9 kg ha −1 yr −1 ) indicates that the vast majority (97%) of inorganic P deposited to peatland soils was the result of dry deposition, a finding similar to observations from mature Neotropical dry forests (Das et al 2011).
Although our measured throughfall flux of 8.7 kg DIN ha −1 yr −1 agrees well with modeled estimates of total (wet+fog+dry) DIN deposition for regions of Southeast Asia (4-10 kg N ha −1 yr −1 ; Vet et al 2014a, 2014b), the proposed critical N load limit for tropical forests is 5-10 kg N ha −1 yr −1 . Negative effects on species diversity, soil N cycling, and streamwater are expected beyond this limit (Pardo et al 2011), especially on highly acidic peatland soils.
Total annual S flux to this peatland was also extraordinarily high, totaling 42 kg S ha −1 yr −1 , of which we estimated 19 kg ha −1 yr −1 was nss--SO 4 2 -S. Assuming -SO 4 2 in throughfall is a robust index of total (wet+fog+dry) atmospheric deposition in this peatland site (Weathers et al 2006, Ponette-González et al 2010a, our sample site ranks among the regions with the highest predicted total S deposition globally (i.e., East Asia 20-50 kg S ha −1 yr −1 ; Vet et al 2014aVet et al , 2014b. The S flux reported here is comparable to estimates for forests located in Southern China's 'urban acid islands', areas <67 km from large cities where acid deposition exceeds critical loads (Du et al 2015).
Major global shifts in geographic source areas and ecosystem sinks for biologically important nutrients, including N, P, and S have occurred over the past three decades (Fowler et al 2013, Vet et al 2014a, 2014b. Emissions inventories and long-term network data show reductions in SO 2 and NO x in North America and Europe with concomitant declines in the deposition of S and oxidized N (JRC/PBL 2011, Lehmann and Gay 2011, Tørseth et al 2012, US EPA 2014. In contrast, emissions and deposition of these compounds are rising in many Asian regions due to rapid urbanization, industrialization, as well as agricultural intensification and expansion (Dentener et al 2006, Vet et al 2014a, Du et al 2015. Although empirical data on P deposition are patchy, Asian ecosystems may become the new P deposition 'hotspots' (Wang et al 2015). Our measurements highlight this geographic redistribution of nutrients and pollutants over Asian ecosystems (Dentener et al 2006, Vet et al 2014a, Wang et al 2015.

Fire drives the redistribution of N, P, and S
Here, we show that fires have a disproportionate influence on below-canopy loading of N, P, and S to forested peatland soils. More than 30% of the inorganic N and P and nearly 20% of the nss--SO 4 2 -S in throughfall was deposited following major local and regional fire pulses (table 1, figure 2), although these throughfall events comprised only 14% of total sampled events. Dissolved inorganic nitrogen, -NO , 3 and -PO 4 3 were the solutes whose concentrations and fluxes were most enhanced in post-fire throughfall.
It is well established that tropical biomass burning influences local and regional atmospheric chemistry and deposition, particularly N and P (Lobert et al 1990, Mahowald et al 2005. For example, experimental and observational studies of Indonesian peat fires find that -NO 3 is a major component of peat smoke aerosol (See et al 2007), while a dominant product of smoldering peat is ammonia gas (NH 3 ; Christian et al 2003). Biomass burning is also an important source of rainwater and aerosol P (Maenhaut et al 2002, Mahowald et al 2008 and -PO 4 3 compared with non-biomass burning periods (Balasubramanian et al 1999, Sundarambal et al 2010. Hsu et al (2014) collected aerosols downwind of the June 2013 peat fires in Riau Province; concentrations of total and soluble P in these samples were four-fold higher than in aerosol samples influenced by East Asian dust and pollution outflows.
While we cannot evaluate the relative contribution of biomass burning versus dust to our measured P fluxes, we note that soil dust may contribute to P loading during fire events. Three potential sources include: (1) dust emissions from biomass burning (Andreae et al 1998); (2) long-range dust transport from arid and semi-arid regions in East Asia (Lin et al 2007); and, (3) long-range dust transport from Australia (McGowan and Clark 2008). We are unaware of any study that has quantified the contribution of fires to atmospheric mineral dust emissions in Indonesia. Dust outflows from East Asia are most prevalent during the northeasterly monsoon (December-March), when only one of the fire events in this study occurred, and their influence on the southern South China Sea is thought to be minimal (Lin et al 2007). However, dust transport from Lake Eyre, Australia, to Borneo (June-August) is coincident with the peak biomass burning period in Southeast Asia (McGowan and Clark 2008). Therefore, it is possible that some of the P we measured in post-fire throughfall was derived from dust.
Because several studies report elevated -SO 4 2 in rainwater and aerosol during peat fires (Ikegami et al 2001, See et al 2007, we were surprised to find that S concentrations were not enhanced in post-fire throughfall events. Two potential explanations include high local or regional S emissions from nonpeat sources, and variability in fire emissions chemistry. First, fossil fuel and volcanic S emissions may have masked the influence of peat burning on atmospheric S fluxes. In Pontianak, the provincial capital of West Kalimantan (23 km from site), oil combustion was found to be the most important source influencing aerosol chemistry (Maenhaut et al 2002). Second, atmospheric emissions from flaming and smoldering fires differ in their chemical composition (Lobert et al 1990). In this study, we used MODIS Active Fire data that cannot distinguish between flaming and smoldering fires (Elvidge et al 2015). Smoldering fires are the fire type associated with high S emissions. Therefore, our correlations between fire detections and throughfall S concentrations possibly reflect the chemistry of flaming rather than smoldering fires.
Around our study site, land use--particularly industrial-scale agriculture--was a dominant driver of peatland burning. Fires were concentrated within oil palm leases on peatlands within 20 km (78% of fires versus 49% of land area) and 100 km (40% of fires versus 19% of land area) of the focal site. Yet, our data suggest that long-range transport of fire emissions from southern Kalimantan and northern Australia also influenced deposition at this site (figure 4). Land clearing and burning for oil palm plantation development are the dominant fire source not only in our study region, but also across Kalimantan and Sumatra (Curran et al 2004, Miettinen et al 2011b, Carlson et al 2012, Gaveau et al 2014, Marlier et al 2015. Atmospheric emissions from these activities are projected to increase over the next decade (Carlson et al 2012, Miettinen et al 2012. The enhanced N, P, and S fluxes we documented in association with agricultural peat fires during a non-ENSO year indicate that emissions from larger and more severe fires typical of strong ENSO years (figure S2) could result in dramatically increased deposition to adjacent ecosystems. Our results showed that a three-fold increase in detected fires resulted in a twofold increase in annual total P and S flux to peatland soils (figure 3). We also captured the influence of a severe fire pulse from January to March 2014 (196 fires 20 km from the site) on throughfall fluxes (table S1). Following this single fire event, 13% of DIN, 16% of -PO 4 3 -P, and 8% of -SO 4 2 -S measured over the study period was delivered to peatland soils as throughfall flux (figure 2).

Conclusions
In this study, we show that biomass burning in tropical peatlands plays a largely unrecognized yet critical role in the redistribution of major limiting nutrients within tropical landscapes. Our estimates indicate that as much as 30% or more of the annual load of inorganic N and P to Bornean peatlands occurred following fire events and that fire-emission-deposition of S could rival that of industrial pollution during high-fire strong ENSO years. As such, biomass fires may have significant fertilizing or polluting effects on recipient ecosystems, thereby altering the contribution of peatlands to climate change (Frolking et al 2011). While the ecosystem effects of biomass burning deposition remain uncertain in tropical regions, the sheer magnitude of the fluxes measured here indicate that firedriven nutrient redistribution may have unforeseen consequences across a diversity of ecosystems, including forested peatlands as well as agroecosystems, wetlands and streams, oligotrophic ocean waters, and coral reefs. gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.ready.noaa.gov) used in this publication. We appreciate D Astiani and her students in the Department of Forestry & Agricultural Sciences, University of Tanjungpura, Pontianak, West Kalimantan, for previous fieldwork and soil and vegetation analyses. N Lisnawati, Y Purwanto, Ruspita, Supardi, WI Suci, A Rohman, R Hartono, U Abdurrahman, and D Firnata provided invaluable support. LMC and APG gratefully acknowledge financial support from the NASA Land Cover/Land-Use Change Program (# NNX11AF08G). Previous NASA support to LMC and teams (NNG05GB51G, NNX07AK37H) provided resources for the establishment and monitoring of this peatland site with surrounding land use. LMC received critical support from the John D and Catherine T MacArthur Foundation, Santa Fe Institute, Stanford University, and Yale University.