Humans dominated biomass burning variations in Equatorial Asia over the past 200 years: Evidence from a lake sediment charcoal record

Highlights

• There are very few paleofire records from Equatorial Asia.

• This makes estimating burning and benchmarking fire models for this region difficult.

• We present a new, charcoal-based paleofire record from East Java.

• Our record challenges assumptions about fire’s controls.

• Fire in Equatorial Asia did not respond to climate and population growth uniformly.

Abstract

Wildfire impacts ecosystems, climate, carbon cycling, societies, and human health. Quantification of these impacts relies upon climate and fire models, which are constrained by historical observations that are limited to the past 30 years. But in regions where records are sparse, like Equatorial Asia (EQAS), fire activities are assumed to be insignificant before the 1960s. We present a 200-year charcoal record from Lake Lading, Indonesia, which shows substantial fire variability since the 19th century. We identify a significant role of humans in controlling fire activity in Java, which could potentially extend to other parts of EQAS. These results contradict assumptions made in current fire emissions estimates and suggest an oversimplification of the spatiotemporal complexity of fire in EQAS before the 1960s. Our study highlights the need for more high-resolution charcoal records in the tropics to improve fire models and emissions estimates.

Introduction

Wildfire and biomass burning around the world have significant effects on terrestrial ecosystems (Wilcove et al., 2013; Margono et al., 2014), the carbon cycle (Van Der Werf et al., 2004), atmospheric aerosols (Jiang et al., 2016; Hamilton et al., 2018), human health (Marlier et al., 2013, 2015; Knorr et al., 2017), and the economy (Glauber and Gunawan, 2016). Though the impacts of fire in recent decades are relatively well-documented by satellite measurements of burned area, longer term dynamics between fire, vegetation, climate, and human activities remain elusive (Bowman et al., 2009). Understanding the impacts of fire and its relationship with these factors in the past and future relies on fire models (Li et al., 2019) and historical estimates of biomass burning emissions (van Marle et al., 2017a). Emissions estimates are particularly important as they provide boundary conditions for state-of-the-art climate models (e.g., Golaz et al., 2019; Danabasoglu et al., 2020; Sellar et al., 2020) and are the best means of evaluating fire models.

Historical estimates of biomass burning emissions rely upon multiple data sources. Satellites measure burned area in recent decades (Van Der Werf et al., 2017), and the Global Fire Emissions Database (GFED) uses satellite measurements of burned area combined with the Carnegie-Ames-Stanford Approach model to obtain emissions estimates from 1997 to present (Van Der Werf et al., 2017). For approximations prior to 1997, multiple approaches are used. These include (1) scaling GFED estimates with global population estimates (Dentener et al., 2000), (2) combining satellite products and historical records (field measurements, aerial photography, fire-fighter reports, and tree ring reconstructions (Mouillot and Field, 2005; Mieville et al., 2010), and (3) combining satellite products, data from the literature, historical records and numerical models (Schultz et al., 2008). As part of phase 5 of the Coupled Model Intercomparison Project (CMIP), Lamarque et al. (2010) combined results from Schultz et al. (2008), Mieville et al. (2010) and an earlier version of GFED to produce what were then state-of-the-art biomass burning emissions estimates spanning from 1850 to 2000.

More recent syntheses and modeling efforts have enabled the use of charcoal records and fire models to estimate historical biomass burning emissions (van Marle et al., 2017a). The latest estimates span from 1750 to 2015 and are currently used as part of phase 6 of CMIP (CMIP6). This estimate combined GFED post-1997 (Van Der Werf et al., 2017), 6 fire models, visibility observations (Field et al., 2009; van Marle et al., 2017b), and charcoal records from the Global Charcoal Database (Marlon et al., 2016). This estimate divided the globe into 17 regions. Prior to 1997, emissions in North America and Europe were estimated using charcoal records that were scaled to fire models, whereas other regions were estimated solely using fire models. The only exceptions were Equatorial Asia (EQAS) and the arc of deforestation (ARCD; i.e. tropical South America), wherein estimates from the mid-20th century to 1997 were based on visibility records and were assumed to be insignificant prior to this period (thus held constant at the lowest decadal average). The constant biomass burning estimation in EQAS and ARCD assumes that fire in these regions scales with population density.

Although increases in fire activity are often assumed to scale with an increase in human population density, this positive relationship has been questioned (Doerr and Santín, 2016). Indeed, charcoal records (Marlon et al., 2008, 2016), satellite products (Andela et al., 2017) and model simulations (Arora and Melton, 2018; Ward et al., 2018) have all shown a decreasing trend in global fire in the recent past despite an increase in population density (early to mid 20th century for model simulations and charcoal records; 21st century for satellite products). This decrease is proposed to be related to agricultural expansion and intensification, and active fire suppression (Marlon et al., 2008; Andela et al., 2017; Arora and Melton, 2018). In contrast to the globally-consistent inverse trend amongst different datasets, observational records suggest a positive relationship between human populations and fire in the tropics (Field et al., 2009; van Marle et al., 2017b), lending support to the assumption made by van Marle et al. (2017a) regarding biomass burning emissions in EQAS and ARCD prior to the mid-20th century. This contrast highlights an important knowledge gap in our understanding of fire in the tropics. Indeed, there is no significant trend in burned area in Southeast Asia (i.e. EQAS) and South America (including ARCD) during the satellite era (Andela et al., 2017), though there are very few charcoal records from these areas to provide a longer-term perspective of fire dynamics (Marlon et al., 2016). Thus, it remains unclear whether biomass burning emissions from EQAS and ARCD were insignificant prior to the mid-20th century and whether the assumption that population density and fire activities are positively related in these regions holds true.

In the case of EQAS, the assumption of a positive relationship between human population and fire activity is based on visibility observations from Kalimantan and Sumatra (Field et al., 2009). Indeed, notable fires in EQAS in recent decades occurred mostly on these two islands undergoing rapid development by humans (van der Werf et al., 2008; Huijnen et al., 2016). Yet, population density and growth in Indonesia were focused in Java prior to the mid-20th century (Fearnside, 1997). The population densities of Sumatra and Kalimantan were also very low prior to the transmigration that occurred in the mid-20th century. This raises the question of whether visibility observations from Kalimantan and Sumatra reliably represent broader EQAS fire conditions prior to the mid-20th century. In fact, past studies have documented large-scale settlement and cultivation in Java since the late 18th century, which suggests the possibility of large-scale fire activity for landscape management purposes (Smiet 1990 and references therein). Furthermore, a charcoal record collected near Java that spanned from 350 to 3279 years before present suggests potential significant fire activity in the region (Poliakova et al., 2017). It is also unclear whether trends in population density of Kalimantan and Sumatra accurately represent that of the broader EQAS region, as the population of other islands in EQAS, such as Java, was denser and grew earlier than on these two other islands. Each of these uncertainties can bias our understanding of the relationship between population density and fire. Therefore, it is imperative to investigate whether there are non-negligible fire emissions in EQAS prior to the mid-20th century. In this study, we present analyses of sedimentary charcoal preserved in East Javanese lacustrine sediments during the last 200 years. We aim to assess whether assumptions regarding fire activity in Java prior to the mid-20th century hold true and to identify the implications of our findings for a better understanding of fire emissions in EQAS.