Table 1.
Fire-use purpose categories. The table indicates whether a purpose is the primary reason for burning, or a co-benefit. It also indicates the number of examples of each fire-use purpose (N) in the LIFE database. Smith et al. ( [8], Table 1) provide examples from the literature for each lower-tier fire-use purpose category. See S1 Appendix for a more detailed description of each purpose, as well as a description of the geographical distribution of case studies in LIFE.
Fig 1.
Maps showing case study locations of fire use practices with known purpose(s) for which spatiotemporal data was available, separated by the type of spatiotemporal data available at each case study location: a) case study locations with burned area data (165 fire use practices); b) case study locations with return interval data (298 fire use practices); c) case study locations with data on months when burning takes place (572 fire use practices). Background shading indicates seasonality type (precipitation-driven, combination-driven, temperature driven), as described in Section 2.4. The colour of points indicates higher-tier fire use purpose (A = agriculture, P = pastoralism, HF = hunting and fishing, G = gathering, C = charcoal and fuelwood production, M = movement, HW = health and wellbeing, S = social signals). Where a case study described multiple fire use purposes at the same location, fire use purposes were plotted in the order listed above and so only the last fire use purpose is visible. S1 Fig and S2 Fig in S2 Appendix show maps enlarged over Africa and south-east Asia, to better display these regions where the density of case studies is very high.
Fig 2.
Process-flow of methods used for seasonality analysis.
Potential evapotranspiration (PET) and precipitation data were sourced from ERA5 reanalysis products [35].
Table 2.
Overview of data available for area burned, fire return interval, and seasonality (i.e., months of burning) for fires set for different purposes. ‘n’ indicates the number of fire use practices with data contributing to the analysis. Mean, median, and standard deviation (SD) are only provided where n ≥ 6. For fire use practices with information on months of burning, we separately consider practices located within precipitation-driven, temperature-driven, or combination-driven climate seasonality zones. ‘Not applicable’ refers to those purposes that are typically co-benefits rather than the primary reason for fire use and would therefore not be expected to directly structure when and where fire is applied to the landscape, or for fire return interval where fires are typically one-off occurrences. See Table 1 for fire-use purpose codes.
Fig 3.
Boxplots of best available estimates of central tendency for fire return interval by fire purpose category.
The figure only includes boxplots for those fire purpose categories for which there were six or more fire use practices with data on fire return interval. Y-axis limited to exclude outliers.
Fig 4.
Boxplots of best available estimates of central tendency for burned area by fire purpose category.
The figure only includes boxplots for those fire purpose categories for which there were six or more fire use practices with data on burned area. Y-axis limited to exclude outliers.
Fig 5.
Seasonality of burning for fire practices recorded in areas with precipitation-driven seasonality.
For each month, the thickness of the shaded bar is proportional to the fraction of records with that fire use purpose that reported burning taking place in that month. See S3 Appendix for data on number of records of burning per month for each fire use purpose. The figure only includes those fire purpose categories for which there were six or more fire use practices with data on the months in which burning takes place. Light shading indicates that at least 6 but fewer than 20 fire use practices contributed data to the analysis for that fire purpose. Dark shading indicates that there were 20 or more fire use practices contributing data. Months are defined relative to the driest month in the annual seasonal cycle (= month 0). The mean precipitation seasonal cycle (mean across case study locations over the time period 1990-2020) is shown in the background.
Fig 6.
Seasonality of burning for fire practices recorded in areas with combination-driven seasonality.
For each month, the thickness of the shaded bar is proportional to the fraction of records with that fire use purpose that reported burning taking place in that month. See S3 Appendix for data on number of records of burning per month for each fire use purpose. The figure only includes those fire purpose categories for which there were six or more fire use practices with data on the months in which burning takes place. Light shading indicates that at least 6 but fewer than 20 fire use practices contributed data to the analysis for that fire purpose. Dark shading indicates that there were 20 or more fire use practices contributing data. Months are defined relative to the month with the lowest sum of precipitation and PET in the annual seasonal cycle (= month 0). The mean precipitation and PET seasonal cycles (mean across case study locations over the time period 1990-2020) are shown in the background.
Fig 7.
Seasonality of burning for fire practices recorded in areas with temperature-driven seasonality.
For each month, the thickness of the shaded bar is proportional to the fraction of records with that fire use purpose that reported burning taking place in that month. See S3 Appendix for data on number of records of burning per month for each fire use purpose. The figure only includes those fire purpose categories for which there were six or more fire use practices with data on the months in which burning takes place. Light shading indicates that at least 6 but fewer than 20 fire use practices contributed data to the analysis for that fire purpose. Dark shading indicates that there were 20 or more fire use practices contributing data. Months are defined relative to the month with the lowest PET in the annual seasonal cycle (= month 0). The mean PET seasonal cycle (mean across case study locations over the time period 1990-2020) is shown in the background.