Managing fire risk during drought: the influence of certification and El Niño on fire-driven forest conversion for oil palm in Southeast Asia

Abstract

Abstract. Indonesia and Malaysia have emerged as leading producers of palm oil in the past several decades, expanding production through the conversion of tropical forests to industrial plantations. Efforts to produce "sustainable" palm oil, including certification by the Roundtable on Sustainable Palm Oil (RSPO), include guidelines designed to reduce the environmental impact of palm oil production. Fire-driven deforestation is prohibited by law in both countries and a stipulation of RSPO certification, yet the degree of environmental compliance is unclear, especially during El Niño events when drought conditions increase fire risk. Here, we used time series of satellite data to estimate the spatial and temporal patterns of fire-driven deforestation on and around oil palm plantations. In Indonesia, fire-driven deforestation accounted for one-quarter of total forest losses on both certified and noncertified plantations. After the first plantations in Indonesia received RSPO certification in 2009, forest loss and fire-driven deforestation declined on certified plantations but did not stop altogether. Oil palm expansion in Malaysia rarely involved fire; only 5 % of forest loss on certified plantations had coincident active fire detections. Interannual variability in fire detections was strongly influenced by El Niño and the timing of certification. Fire activity during the 2002, 2004, and 2006 El Niño events was similar among oil palm plantations in Indonesia that would later become certified, noncertified plantations, and surrounding areas. However, total fire activity was 75 % and 66 % lower on certified plantations than noncertified plantations during the 2009 and 2015 El Niño events, respectively. The decline in fire activity on certified plantations, including during drought periods, highlights the potential for RSPO certification to safeguard carbon stocks in peatlands and remaining forests in accordance with legislation banning fires. However, aligning certification standards with satellite

Forest and Land Fires Are Mainly Associated with Deforestation in Riau Province, Indonesia

“…Analysis of Sumatran fires in 2013 found that 58% of fires occurred on land that had been forest 5 years previously [9]. Across Indonesia, 25% of forest loss in oil palm concessions experienced coincident fire the same year or one year before forest loss [30]. In Riau, active fires were found to occur on average 58 ± 10 days before loss of natural forest [31], further confirming the very tight association between fire and forest loss.…”

Characterizing Tropical Forest Cover Loss Using Dense Sentinel-1 Data and Active Fire Alerts

“…We identified fire-related forest cover loss by co-locating clumped forest cover loss pixels with active fire alerts. This approach is consistent with approaches used in previous studies combining active fire alerts with remote-sensing-based forest-change products [5,10,16,19]. We used the 2015 forest land cover (Figure 1; [35]) of Indonesia's Ministry of Environment and Forestry to separate fire-related changes for different natural forest classes.…”

“…Riau has the highest forest-cover-loss rates in Indonesia [1] mainly driven by expansion and conversion to oil palm, acacia, coconut, and rubber plantations [34]. Although forbidden by law, fire use for forest removal is still wide-spread in Riau [5,10]. Figure 1 depicts the 2015 forest land cover map of Indonesia's Ministry of Environment and Forestry [35] that we used as forest benchmark map for this study.…”

“…fires in Indonesia are human-induced, and only very few are natural [5,8]. While prohibited by law, fire use is widespread in both industrial and small holder plantations (e.g., palm oil) and are often used to expand cultivatable land [5,9,10]. Fire is used for a diversity of land management practices and its temporal relationship with forest-cover loss can vary largely [8,11,12].…”

“…When compared to the 1 km MODIS alerts, the higher resolution VIIRS alerts show improved response to small-scale fires [18]. Active fire alerts combined with mono-temporal Sentinel-1 based forest cover loss information [19] or annual Landsat based forest-cover-loss products [5,10,16,20] have been used to identify fire-related forest-cover loss. Relying on annual forest cover loss information, however, does not allow a detailed assessment of the temporal relationship between fires and forest-cover loss.…”