Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2022 Mar 31;17(3):e0263313. doi: 10.1371/journal.pone.0263313

The burning island: Spatiotemporal patterns of fire occurrence in Madagascar

Tristan Frappier-Brinton 1, Shawn M Lehman 2,*
Editor: Ji-Zhong Wan3
PMCID: PMC8970516  PMID: 35358197

Abstract

Anthropogenic fire use is widespread across Madagascar and threatens the island’s unprecedented endemic biodiversity. The vast majority (96%) of lemur species are already threatened with extinction, and Madagascar has already lost more than 44% of its forests. Previous conservation assessments have noted the role of fire in the rampant deforestation and habitat degradation across Madagascar, but published, quantified data on fire use across the island are incredibly limited. Here, we present the first quantification of spatiotemporal patterns in fire occurrence across Madagascar using VIIRS satellite fire detection data. We assess which regions of Madagascar have the most prevalent fire use, how fire use is changing over time, and what this means for Madagascar’s remaining forest ecosystems. An average of 356,189 fires were detected every year in Madagascar from 2012–2019, averaging 0.604 fires/km2. Fire use was near-ubiquitous across the island, but was most prevalent in the western dry deciduous forests and succulent woodlands ecoregions. Fire frequency in the eastern lowlands was highest around the remaining humid rainforest, and fire frequency was increasing over time around much of the remaining humid and dry forest. We found that 18.6% of all remaining forest was within 500 m of a fire within a single year, and 39.3% was within 1 km. More than half of remaining forest was within 1 km of a fire in a single year in the dry deciduous forests, succulent woodlands, and mangroves ecoregions. However, fire frequency within national park protected areas was, on average, 65% lower than their surroundings. Only 7.1% of national park forest was within 500 m of a fire within one year, and 17.1% was within 1 km, suggesting that national parks are effective at reducing fire frequency in Madagascar’s tropical forests.

1. Introduction

1.1 The issue of fire in the tropics

Fires are contributing to the destruction and degradation of tropical forests worldwide [1–3] and are only expected to become more severe with climate change [4, 5]. Even lush rainforests that used to experience natural fires once in hundreds to thousands of years are now burning regularly [1, 6–9]. Recent tropical forest fires in the Amazon Basin [9] and Southeast Asia [10] have illustrated fire’s devastating effects on tropical forests ecosystems. Most native tree species in tropical forest are not adapted to withstand fires and suffer high immediate and delayed mortality [11, 12]. However, there are few data on the effects of fires on forest cover in Africa.

Africa is the most fire-threatened continent on Earth. Fires in Africa comprise 70% of the total annual burned area worldwide [13] and are responsible for 50% of global carbon emissions due to fire [8]. The impacts of forest fires are of particular concern in mega-diverse regions like Madagascar, which has some of the most concentrated endemic biodiversity across taxa of any nation [14, 15]: 80–90% of species in Madagascar are found nowhere else on Earth [16]. Most (96%) of Madagascar’s 107 endemic lemur species are already threatened with extinction due to anthropogenic pressures and almost a third (31%) are considered critically endangered, the final classification before “extinct in the wild” [17]. It is estimated that 88% of species in Madagascar, including lemurs, cannot survive outside of forests [18], which makes the destructive impacts of fire on tropical forest a concern for both floral and faunal conservation in Madagascar.

Madagascar has lost more than 44% of its forests since 1953 [19]. While there is still debate on the extent of original forest before the 1950s, Madagascar may have already lost more than two thirds of its original forest cover [20]. Previous research has noted the widespread use of anthropogenic fires across Madagascar, primarily for agriculture land management [21, 22], and the destructive impacts fires have on Madagascar’s already threatened forests [23–25]. However, there are few published studies that quantify patterns of fire occurrence across the island. In this paper, we present the first analysis to quantify the distribution and seasonal timing of fires across the entire island of Madagascar based on satellite VIIRS data.

1.2 Madagascar’s fire regime

Fires in Madagascar are overwhelmingly started by humans [22, 26]. Fire exposure is now a major source of mortality for many Malagasy forest species [26], including species that evolved alongside natural fire occurrence [27]. The frequency of fires is now considerably higher than the frequency that these ecosystems evolved to cope with naturally, and humans have spread fire into habitats that were previously too wet or too dry to support burning [28]. Consequently, the fire regime across most of Madagascar is considered to be “Very Degraded”, the worst possible classification [29].

The Malagasy people have typically used fire to clear forests for agricultural crops and to rejuvenate pastures for cattle [22]. Grassland fires are most common in the central plateau and western lowlands in the dry season (~May-November) [21]. Most grass fires are lit in the second half of the dry season to replace the standing lignified grasses with new growth shoots that are more nutritious for cattle [22]. However, many grassland fires spread into nearby forests, either intentionally or unintentionally, where they contribute to habitat degradation and the conversion of primary forest into secondary forest and, eventually, grassland [26]. Most fires occur in the late dry season, when they have higher intensity, making their spread into forests more likely and more destructive [26].

Intentional burning of forest to clear land for agriculture is particularly common in the eastern lowland rainforests, where more than 50% of original forests are believed to have been lost to slash-and-burn agriculture [19]. Forests are cut in the mid-late dry season, then residual biomass is burned near the end of the dry season (October-November) [22]. Slash-and-burn agriculture also threatens western dry forests for cropland conversion, particularly for the growing of maize [30, 31]. Most soils in Madagascar can only support crop growth for a few years before they must be abandoned and left to fallow so the soil nutrients can recover [32]. This fallowing cycle, combined with Madagascar’s rapid population growth of 2.7% per year [33], drives the continuing expansion of slash-and-burn agriculture into new lands, creating constant pressure on existing forests [34]. The time period between fallow cycles has also decreased from 8–15 years to a largely unsustainable 3–5 years, which is increasing fire frequency and accelerating the conversion of natural forests into secondary grassland [32]. Across the island, anthropogenic fires are also used for the herding of zebu cattle [35], charcoal production, particularly in the southwest of the island [31], and lime clay production, particularly in coastal mangrove regions [36].

1.3 Madagascar’s ecoregions

Madagascar’s topography has given rise to many unique ecoregions with different climates, species compositions, and susceptibilities to fire (Fig 1A). Here, we consider seven ecoregions based on the World Wildlife Fund ecoregion assessment [37]: Dry deciduous forests, Subhumid forests, Lowland forests, Succulent woodlands, Spiny thickets, Ericoid thickets, and Mangroves. The western dry deciduous forests are composed of forest fragments surrounded by grassland savannas [30]. These forests experience a long, pronounced dry season, which typically runs from May through November. The subhumid forest ecoregion is found along the central plateau of the island (>1000 m altitude). There is little contemporary forest across this ecoregion [19], and it is currently dominated by grasses and shrubs. The eastern lowland forests ecoregion contains humid rainforest, with a much shorter dry season from approximately September to December. Across the island, dry season length tends to increase towards the west coast. The southwestern succulent woodlands and spiny thickets both have longer dry seasons lasting from approximately May to November. Towards the southwest, the climate is more arid, understory growth is less continuous, and vegetation is increasingly dominated by aphyllous shrubs, particularly in the southernmost spiny forests. Ericoid thickets are montane, shrub-dominated habitats found only in small areas above ~2000 m altitude [34]. Here, dry seasons are less pronounced, and habitats are relatively isolated from anthropogenic disturbance, meaning this may be the only ecoregion where fires are predominantly caused by lightning strikes rather than anthropogenic origins [34]. Mangroves are found in patches of coastal, estuarine forest along the west coast of Madagascar, surrounded by dry deciduous forests and succulent woodlands ecoregions.

Fig 1. Madagascar ecoregions and annual fire frequency.

Fig 1

Open in a new tab

(A) The ecoregions of Madagascar, based on WWF ecoregion classifications [37] Republished from [37] under a CC BY license, with permission from WWF. (B) Average annual fire frequency (number of fires per km2 per year) from 2012–2019 VIIRS fire detection data with 1 km resolution.

Madagascar is home to stunning levels of microendemism across taxa [38], meaning that all ecoregions have unique species compositions and conservation outlooks. Mangroves, Tropical and Subtropical Dry Broadleaf Forests, and Tropical and Subtropical Moist Broadleaf Forests are all found in Madagascar, and have been identified as the three most fire-sensitive habitat types globally [29]. Here, we present the first quantification of fire occurrence across Madagascar using VIIRS fire detection. The objectives of our research were to 1) quantify fire frequency across Madagascar and how it is changing over time, 2) determine which ecoregions are most impacted by fire use and the fire seasonality within each ecoregion, and 3) assess the proximity to fire of remaining forests in each ecoregion and whether conservation efforts in national parks are reducing the fire pressure on protected forests.

2. Methods

2.1 VIIRS data

Fire data were obtained from the VIIRS (Visible Infrared Imaging Radiometer Suite) 375 m-resolution dataset (VNP14IMGT), which are managed by NASA and NOAA, and available to the public at https://firms.modaps.eosdis.nasa.gov/download/. The VIIRS instrument, launched in October 2011, orbits the Earth onboard the Suomi-National Polar-orbiting Partnership (S-NPP) satellite, detecting thermal anomalies/fire hotspots within 375 m pixels. There is currently a second VIIRS instrument orbiting Earth onboard the NOAA-20 satellite, but data collection did not begin until Jan 2020, which was outside of our study period. The VIIRS 375 m satellite fire-detection algorithm uses five single-gain channels, primarily using the brightness temperature of the infrared I4 (3.55–3.93 μm) and I5 (10.5–12.4 μm) channels for fire detection/identification, and channels I1 (0.6–0.68 μm), I2 (0.846–0.885 μm), and I3 (1.58–1.64 μm) for discrimination of clouds, water, and sun glint. Data were downloaded for Madagascar for 2012–2019. During this time frame, the VIIRS S-NPP satellite scanned the island for fires approximately every 12 hours, at ~10:30am and ~10:30pm, allowing for daytime and nighttime fire detection. For the purposes of this paper, the term “fire” will be used to refer to each detected 375 m fire hotspot. A single burning front can result in multiple fire hotspots, but to do so the front must extend multiple 375 m pixels at the same time of satellite overpass or burn for longer than ~12 hours (the return time of the satellite).

VIIRS data undergo onboard preprocessing and post-processing by NASA’s LANCE (Land, Atmosphere Near-real-time Capability for EOS) to maximize the accuracy of fire detections and locations. The refined methods and the main I4 band’s low saturation temperature of 367 K make this the most accurate and comprehensive fire detection database currently available. The VIIRS system is able to detect as many as 5-10x more fires than the MODIS system [39–41], and global estimates of false positive rates for nominal confidence pixels are only 0.03% of detected fires [42]. Global false positive estimates for low confidence fires are ~6%, and these false positives are clustered in industrial areas with highly reflective buildings, particularly in Eastern China [42]. Analysis of low confidence fire pixels in southern Africa showed they represented real fires of lower size/intensity, often bordering higher intensity fires [42]. False positives from industrial areas are not expected to have a notable impact on our study for two reasons: 1) there are few industrial areas in Madagascar, and 2) these areas tend to be close to cities and thus not near the remaining forests analyzed here. Low confidence pixels represent ~11% of all fires detected globally, and ~15% of fires within Madagascar. To obtain the most complete analysis possible of fire locations within Madagascar, all VIIRS fire detection in Madagascar were downloaded and analyzed. Locations of low confidence pixels were also examined visually to confirm that they corresponded to the locations of fire use indicated by nominal and high confidence pixels.

2.2 VIIRS processing/analysis

VIIRS data were downloaded in csv format and imported as points into ArcGIS ArcMap. A kernel density map at a resolution of 1 km was generated using all VIIRS points from 2012–2019. Change over time in fire frequency was calculated by inputting 5 km resolution kernel density maps for each year 2012–2019 into the Curve Fit tool [43] which fit a linear regression to the number of fires by year in each 5 x 5 km pixel. The output raster contained the average rate of change (slope) for the number of fires per year for every pixel on the map. To standardize these values by the background fire frequency, rate of change values were divided by the average fire density across the study period (2012–2019) and multiplied by 100 to convert the rates of change into a percentage. As such, the shown rates of change in percent represent a constant rate of change (as a % of the 2012–2019 average fires/km2), and not an exponential/compounded rate of change. The ecoregion map was obtained from the World Wildlife Fund [37], with minor shapefile modifications made following Steffens et al. [44]. All fire points were spatially joined to the ecoregion they occurred within. Due to slight discrepancies in the detailed mapping of coastline between the ecoregions map and other datasets, 0.02% of total fires and 0.5% of total forest was not included within any ecoregion. All analyses were performed using the WGS 1984 UTM Zone 38S projection, except for area calculations, which used the Laborde projection.

2.3 Tree cover and conservation

Tree cover data were obtained at a scale of 30 m pixels from Vieilledent et al. [19]. The most recent dataset available shows estimated tree cover in 2017. These data were compared to the VIIRS detected fires in 2017 and the ecoregion boundaries to determine how threatened the remaining forest is by fire in each ecoregion. Buffers were created for 500 m, 750 m, 1 km, 1.5 km, 2 km, 2.5 km, and 3 km distances around all VIIRS fire points, and the proportion of forest in each ecoregion falling within these buffer distances was calculated in ArcMap. These values were also calculated separately for forest within the borders of a national park. National park boundaries were obtained from the Protected Planet database [45] in July 2019.

3. Results

3.1 Distribution of fires

VIIRS detected an average of 356,189 fires per year from 2012–2019, averaging 0.604 fires/km2 per year. We found that 90.9% of fires were detected during the day, with only 9.1% detected at night. Daytime fire detection time ranged from 9:15 AM to 11:50 AM and averaged 10:32 AM. Nighttime fire detection time ranged from 9:00 PM to 11:15 PM. Average time of detection for night fires was 9:29 PM, as later satellite overpasses tended to detect fewer fires.

Average density of fires was highest in western Madagascar, particularly in the dry deciduous forest ecoregion, and in the northern part of the succulent woodlands ecoregion (Fig 1). During 2012–2019, average fire density in some locations reached as high as 4.1 fires/km2 per year, with the highest values recorded northwest of the spiny thickets ecoregion, corresponding to the area immediately northeast of the city of Toliara. The largest area of fire frequency above 1.5 fires/km2 was around the city of Morondava, encompassing both the Menabe Antimena Protected Area and Kirindy Mitea National Park. Regions of high fire frequency in the east were predominantly located around the last remaining strip of humid forest along the central escarpment (Fig 2B).

Fig 2. Rates of change in fire frequency and remaining forest cover.

Fig 2

Open in a new tab

(A) Rate of change in VIIRS fire frequency for each 5 km x 5 km pixel from 2012–2019, expressed as a percentage of the average fire frequency (fires per km2) for that pixel in the 2012–2019 period. (B) Madagascar’s remaining forest cover (in black) from 2017 from Vieilledent et al. [19] superimposed onto the data shown in (A) of change in fire frequency over time.

There were very few areas that were not affected by fire. In central Madagascar, the largest fire-free area was in the urban region surrounding the capital city of Antananarivo, extending down to another large city of Antsirabe. Other large areas of limited fire use were in the southwest of the island, in the spiny thickets and southern succulent woodlands, and in the northeast, near Masoala and Makira protected areas. Fire use was relatively low on the Masoala peninsula, but fire frequencies on the unprotected east coast of the peninsula were more than double than those inside the adjacent Masoala National Park. Areas of low fire use can also be seen along the eastern coast, a region with high population density [46], particularly around the city of Toamasina.

3.2 Changes in fire frequency over time

Across the entire island, there was no clear change in the total number of VIIRS detected fires over the study period (2012–2019), and MODIS fire detection data also suggested that the total number of fires island-wide has remained fairly constant since 2003 (S1 Fig). However, there were distinct regional changes in fire frequency (Fig 2). From 2012–2019, fire frequencies increased in the dry deciduous forests ecoregion and the southern portion of the central plateau/subhumid forests ecoregion. We found that 65% of land in the dry deciduous ecoregion saw a trend of increasing fire use from 2012–2019, compared to 49% of land in mangroves, 40% in subhumid forests, 38% in succulent woodlands, 23% in ericoid thickets, 16% in lowland forests, and 13% in spiny thickets ecoregions. Most of the lowland forests ecoregion saw decreases in fire frequency, but areas of distinct increases corresponded to the location of the few remaining lowland forests (Fig 2B). There were also increases in fire frequency around most of the remaining dry forest and some areas of northern humid forest.

3.3 Fire seasonality and frequency by ecoregion

During 2012–2019, the dry deciduous forests ecoregion recorded the highest average fire frequency with 0.80 fires/km2 per year, followed by succulent woodlands (0.67), subhumid forests (0.59), mangroves (0.57), lowland forests (0.41), spiny thickets (0.40), and ericoid thickets (0.13) (Table 1). Most fires occurred in the late dry season, with 69.3% of all fires occurring between August and November. Fire seasons in the western ecoregions started earlier than those further east, with all ecoregions except for lowland forests and ericoid thickets seeing increases in fire use immediately after the end of the western wet season in April (Fig 3). Fire frequencies in dry deciduous forests, subhumid forests, and succulent woodlands increased steadily throughout the dry season (April—October) until the beginning of rain onset in late November, with very low fire frequency by December. Lowland forests and ericoid thickets did not experience increases in fire frequency until the end of the eastern wet season in August. However, fires in lowland forests rapidly increased from September-November to reach similar frequencies as the eastern ecoregions. The fire season in lowland forests lasted later into the year, with fire use not fully decreasing until February, relating to the later onset of rains in the eastern lowlands [47]. 80% of fires in the lowland forest ecoregion occurred from October to December. The spiny thickets ecoregion had slight increases in fire use starting in April, but fire use increased precipitously in October, with 56% of annual fires occurring in just October and November. Surprisingly, the highest average fires per km2 for any month in any ecoregion was November in Mangroves, with 0.150 fires/km2 in one month.

Table 1. Summary of fire occurrence and proximity to remaining forest within each ecoregion.

Ecoregion Total Area (km2) Remaining Forest (km2) Total Fires per Year Annual Fires per km2 % Forest within 500m of fire % Forest within 1km of fire % Forest within 2km of fire
Dry deciduous forests 150,824 14,696 120,158 0.80 32.0 60.5 84.5
Subhumid forests 198,745 15,195 116,690 0.59 13.2 30.8 57.5
Lowland forests 111,396 26,191 45,801 0.41 12.9 32.2 59.7
Succulent woodlands 79,402 9,989 53,199 0.67 32.0 57.9 81.8
Spiny thickets 43,097 16,369 17,101 0.40 12.4 28.7 53.8
Ericoid thickets 1,273 392 162 0.13 4.3 10.2 24.7
Mangroves 5,113 1,177 2,920 0.57 29.6 51.9 69.6
TOTAL 589, 850 84, 460 356, 189 0.60 18.6 39.3 64.9
Open in a new tab

Fig 3. Seasonality of fire frequency by ecoregion.

Fig 3

Open in a new tab

The average number of VIIRS fires detected per month in each ecoregion (expressed in the number of fires per km2 per month) over the 2012–2019 time period.

Values for total fires and fires/km2 calculated using the annual averages from 2012–2019. Ecoregion boundaries based on the WWF ecoregion map [37]. Percent values for distance to fire calculated using the distance of each forest pixel to the nearest fire using 2017 VIIRS fire data and 2017 forest cover from Vieilledent et al. [19].

3.4 Fires around remaining forest

In 2017, of the remaining 84, 460 km2 of forest on the island, 18.6% (15,681 km2) was within 500 m of a fire, 39.3% (33,169 km2) was within 1 km, and 64.9% (54, 774 km2) was within 2 km. Forests in certain ecoregions were notably more threatened than others (Table 1). More than one quarter of remaining forest in the most threatened ecoregions, dry deciduous forests, succulent woodlands, and mangroves, was within 500 m of a fire in 2017 (32.0%, 32.0%, and 29.6%, respectively), and more than half was within 1 km (60.5%, 57.9%, and 51.9%).

Of the 8, 293 km2 forest within national parks, only 7.1% (585 km2) was within 500 m of a fire, 17.1% (1417 km2) was within 1 km, and 38.6% (3200 km2) was within 2 km. The proportion of national park forest within any distance to fire was lower than the island average for all ecoregions and all distance classes (Fig 4 and Table 2). However, the sample size of forest within a national park for mangroves and ericoid thickets were both small (< 15 km2 of forest). Total fire frequency was 65% lower within national parks than the fire frequency within a 2 km buffer area surrounding each park, and 58% lower than their respective ecoregion average. Land within 2 km of the border of a national park had fire frequencies that were an average of 24% higher than their respective ecoregion averages.

Fig 4. Distance of remaining forest to the nearest fire.

Fig 4

Open in a new tab

The percent of all remaining forest across the island not within a national park (left) and exclusively within a national park (right) that was within each buffer distance of the closest fire in 2017. Values calculated using VIIRS 2017 fire detections and 2017 forest cover from Vieilledent et al. [19].

Table 2. Differences in distance from forest to nearest fire by national park status.

Ecoregion % of Total Forest within a National Park Forest not within a National Park Forest within a National Park
% Forest within 500m of fire % Forest within 1km of fire % Forest within 2km of fire % Forest within 500m of fire % Forest within 1km of fire % Forest within 2km of fire
Dry deciduous forests 8.30 33.3 62.7 86.1 17.3 35.6 67.2
Subhumid forests 11.16 13.5 31.5 58.0 11.1 25.8 52.8
Lowland forests 15.58 14.9 36.7 66.3 2.2 7.9 24.2
Succulent woodlands 7.91 33.8 60.8 84.4 11.1 25.0 51.5
Spiny thickets 2.89 12.7 29.4 54.9 1.2 5.1 18.2
Ericoid thickets 3.44 4.5 10.6 25.6 0.0 0.0 0.0
Mangroves 0.90 29.9 52.2 70.0 1.1 10.9 22.7
TOTAL 9.82 19.8 41.7 67.7 7.1 17.1 38.6
Open in a new tab

The proportion of remaining forest within each distance of the closest fire within 2017, split by whether or not the forest was protected by a national park. Values calculated using VIIRS 2017 fire detections and 2017 forest cover from Vieilledent et al. [19].

4. Discussion

4.1 Patterns of fire occurrence

Average fire frequency was consistently high throughout the dry deciduous forests ecoregion, reflecting the frequent use of fire as a tool for pasture maintenance in western Madagascar [22, 26]. The high fire frequencies surrounding remaining forests in the east are likely indicative of slash-and-burn agriculture that has already destroyed more than 50% of Madagascar’s eastern rainforests [19]. The quantification of fire use presented here confirms previous identifications of the most fire-threatened ecosystems. The forests surrounding Toliara are believed to be under heavy fire pressure from charcoal production [31], and agricultural burning is a major threat to conservation efforts in Kirindy-Mitea National Park, where 6% of all remaining forest burned within a single dry season [48].

The northeastern humid forests around Masoala and Makira protected areas represent the largest remaining sections of intact forest on the island (Fig 2B), so it is reassuring for conservation efforts to see infrequent fire use across this region. However, the visible increases in fire frequency outside the borders of Masoala National Park suggest that these forests still may become threatened by fire use, and that national parks may help mitigate this damage. The lack of fire use in the large area surrounding Antananarivo, however, is likely the result of minimal vegetation remaining to burn rather than a natural ecosystem free from fire use. In the southwest, the large area of minimal burning may be due to a lack of continuous flammable fuels to enable fire spread in this arid vegetation type.

4.2 Changes in fire use over time

Analysis of the rate of change in fire frequency between 2012 and 2019 revealed that fire frequencies are increasing throughout the dry deciduous forests ecoregion, and disproportionately around the last remaining strip of humid lowland forests that runs north-south along the edge of the central escarpment/subhumid forests ecoregion. The forests in these areas are already subject to fire frequencies that are likely unsustainable for tropical forest. Research from mainland Africa suggests increasing population size decreases fire use in areas of high population density, but increases fire use in areas of low population density [49]. Population density in Madagascar is believed to be highest around Antananarivo, and along the eastern coast [46], which largely saw decreases in fire frequency. Population density is believed to be lowest across the western dry deciduous forests ecoregion and the southern central plateau/subhumid forests ecoregion [46], which saw increases in fire frequency. While high population densities have reduced total burned area worldwide [50], this is usually the result of the conversion of natural, burnable land into agricultural or urban land, and does not necessarily indicate reduced fire pressures on remaining habitat. As populations in Madagascar continue to expand [33], the heavy reliance of local people on the land for agriculture [22] and extractive resources (primarily wood fuel) [51] will continue to increase the fire pressures on remaining forests, which can be seen around the remaining eastern humid forest.

4.3 Ecoregion fire regimes

Considerable fire use pervaded all of Madagascar’s ecoregions. Total annual fire use was highest in the western dry deciduous forests ecoregion, and second highest in succulent woodlands, despite a large area of minimal fire use in the south of the ecoregion. Ericoid thickets were the only ecoregion with low average fire frequency (3 to 6 times lower than any other ecoregion) and a fire season that did not increase continuously throughout the dry season, supporting the idea that fires in ericoid thickets are predominantly started by lightning, as opposed to anthropogenic fires [34]. Fire seasonality in all other ecoregions peaked in the late dry season of each ecoregion. Fire season was most condensed in the lowland forests ecoregion, but here most fires represent deforestation that may have actually occurred months prior to biomass burning [22].

Fire frequency was surprisingly high in coastal mangroves, with the highest average single-month (November) fire frequency of any ecoregion. Mangrove forests mitigate erosion [52] and provide resources and aquaculture opportunities to local people [53], but in Madagascar they are being harvested at unsustainable rates [54–56]. The noteworthy fire use in this semi-aquatic ecosystem reflects the high anthropogenic pressures that Madagascar’s mangroves face. However, this analysis may not encapsulate all of Madagascar’s mangroves, as fragments of mangrove forest also extend along the western coast of Madagascar, outside of the WWF’s “Mangroves” ecoregion [19].

4.4 Implications of fire use for Madagascar’s forests

Based on VIIRS fire detections, 40% of remaining forest in Madagascar was within 1 km of an active fire within a single year. The overuse of fire has already contributed to the extensive deforestation and habitat degradation that Madagascar has experienced over the past few decades [19, 20], and our analysis suggests that forests in Madagascar are still heavily threatened by anthropogenic fires. Slash-and-burn agriculture is widespread across the east [32] and west [30, 31] of Madagascar, and the cycles of fire use and crop growth deplete soil nutrients and drastically reduce the number of forest species [26, 57]. In grasslands, fires eliminate patches of scrub or forest cover that would otherwise reduce erosion [58] and spread into forest edges, leading to their conversion into secondary grassland [26]. Grassland fires can spread into forests as crown fires that result in the immediate mortality of canopy and understory vegetation, or more commonly as surface fires that only burn understory biomass [2]. However, the damage caused by surface fires can result in delayed mortality of more than one third of tree species within 3 years [59] and reduce canopy cover by 10–40% in humid forests [3, 11]. Reduced canopy cover increases fuel drying, and tree mortality increases fuel availability, which both create a positive feedback loop where subsequent fires are more intense and more likely to occur [7]. Once fires facilitate the conversion of forests into grasslands, the loss of soil nutrients and burning of newly established shoots make it near-impossible for the natural re-establishment of forests [19].

4.5 An underestimation of fire use

While these data show an alarming number of fires across all of Madagascar’s ecoregions, VIIRS data likely underestimate the total number of fires across the island. The S-NPP satellite collected VIIRS data around 10:30am and 10:30pm each day, but most fires in Madagascar burn in the afternoon [60, 61], so the VIIRS data likely missed a considerable proportion of total fires. Afternoon fires would only be detected by these VIIRS data if they were started before ~10:30 am, or continued to burn into the night. Globally, only a small fraction of daily fires continues to burn overnight [61], and in these Madagascar data, nighttime detections represented only 9.1% of annual fires. This also means that few fire pixels would be counted twice by VIIRS, as it would require fire to burn into the night and for a minimum of 12 hours.

Surface fires may also be under-detected in these VIIRS data. Satellite fire detection datasets have struggled to detect small understory fires that are obscured by forest cover and/or smoke emissions [62, 63]; however, the low detection threshold of VIIRS should reduce these errors [64]. Reduced detection abilities may result in an underestimation of fires in forested regions compared to open grasslands, which could result in an underestimation of fires within national parks, which have disproportionately higher forest cover. Thus, if fires within forest are under-detected, the proportion of remaining forest impacted by fire is even greater than what is estimated here by VIIRS detections.

Cloud cover reduces the ability of VIIRS satellites to detect active fires [65] and would have contributed to the underestimation of the number of fires across Madagascar. This underestimation may have disproportionately affected regions or time periods with greater cloud cover, particularly during the rainy season(s) or in the eastern lowland forests ecoregion where the rainy season lasts longer into the year. Thus, while the increased fuel moisture during the rainy season would reduce fuel flammability, estimates of fire use during periods of heavy cloud cover may be further deflated due to under-detection.

4.6 Conservation implications and future directions

Given the negative effects of fire occurrence on forest cover, it is reassuring for conservation outlooks to see here that fire frequencies within Madagascar’s national parks were 58% lower than the averages for their ecoregions, and were an average of 65% lower than areas directly surrounding each park. The proportion of forest within 500 m of a fire was notably lower within national parks (7.1% inside vs 19.8% outside, 2.8 times lower), as was the proportion of forest within 1 km of a fire (17.1% vs 41.7%, 2.4 times lower). These results are in line with previous studies that suggest conservation efforts can reduce the number of fires across the tropics [66], and in individual regions of Madagascar [67]. However, most of Madagascar’s protected areas are found outside of national parks in special reserves and community-managed protected areas [68]. These additional protected area types vary considerably in their ability to establish and enforce boundaries [68], and thus were not included in this analysis. Understanding the ability of all protected area types to protect their ecosystems from fires will require future analyses.

Conservation methods used to protect forests from fires vary considerably by region. In some protected areas surrounded by savanna (e.g., Ankarafantsika National Park), park managers use prescribed burns in the early dry season at the park’s forest-grassland boundary to create fire breaks and reduce fuel loads for the late dry season (Steffens TS, personal communication) when fires pose the greatest risk to remaining forests [26]. Restricting fire use is often necessary for conservation, but complete bans on fire use that do not consider the larger factors leading to fire use, like poverty and the need to expand or rejuvenate agricultural lands, have been criticized for their limited effectiveness [22, 26]. However, conservation areas that prioritize the livelihoods of local communities have also been criticized for their limited ability to preserve ecosystems [35, 68–70]. Ultimately, research suggests that funding amount [67] and number of guards per km2 of land [66] are the most important factors to protect remaining ecosystems from fires.

To fully understand the threat that rampant fire use poses to Madagascar’s conservation efforts, we need more data on how different tree and faunal species respond to fire in their habitats. Studies of fire-related mortality of Madagascar’s tree species are few [71], but research suggests that even tree species that evolved alongside fire are now threatened by too frequent burning [27]. Studies of faunal responses to fire in the tropics are incredibly limited [72, 73], particularly for primates which are one of Madagascar’s top conservation priorities in lemurs. Fires decrease the production of fruits and flowers [12, 27, 74] which many lemurs rely upon, and can exacerbate hunting pressures by clustering species at surviving food trees [12]. Many lemur populations already face hunting at unsustainable levels [75, 76]. Fire can also create new forest edges [2, 77], contributing to Madagascar’s issue of rampant forest fragmentation [19], which can isolate small, unsustainable populations [78] and negatively impact edge-averse species [79]. Fire has long been discussed as a contributing factor to the rapid decline of Madagascar’s biodiversity [23–25], and we hope that this first large-scale quantification of the frequency and distributions of fires across Madagascar will help inform discussions on the current levels of fire use.

Supporting information

S1 Fig. Total number of fires detected per year in Madagascar by VIIRS and MODIS datasets.

Total number of fires detected by MODIS (left axis) from 2003–2019 and VIIRS (right axis) from 2012–2019. VIIRS detected an average of 5.6 times more fires than MODIS.

(TIF)

Click here for additional data file. (184.9KB, tif)
S1 Data

(ZIP)

Click here for additional data file. (40.2MB, zip)
S2 Data

(ZIP)

Click here for additional data file. (40.3MB, zip)
S3 Data

(ZIP)

Click here for additional data file. (42.5MB, zip)
S4 Data

(ZIP)

Click here for additional data file. (42.6MB, zip)
S5 Data

(ZIP)

Click here for additional data file. (43.3MB, zip)
S6 Data

(ZIP)

Click here for additional data file. (45.6MB, zip)
S7 Data

(ZIP)

Click here for additional data file. (51.1MB, zip)
S8 Data

(ZIP)

Click here for additional data file. (53MB, zip)
S1 File

(PDF)

Click here for additional data file. (74.9KB, pdf)

Acknowledgments

We are grateful for the NASA FIRMS system for making the VIIRS fire data publicly available, and for the WWF for allowing us to use their ecoregion data. We would also like to thank University of Toronto colleague Fernando Mercado Malabet for his assistance and expertise with spatial analysis in ArcGIS.

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

SML, No grant number, Discovery Grant from the Natural Sciences and Engineering Research Council of Canada, https://www.nserc-crsng.gc.ca/index_eng.asp. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Goldammer JG, editor. Fire in the tropical biota: Ecosystem processes and global challenges. Springer Science & Business Media; 2012. [Google Scholar]
  • 2.Cochrane MA. Fire science for rainforests. Nature, 2003; 421: 913–919. doi: 10.1038/nature01437 [DOI] [PubMed] [Google Scholar]
  • 3.Laurance WF. Slow burn: the insidious effects of surface fires on tropical forests. Trends Ecol Evol. 2003; 18: 209–212. [Google Scholar]
  • 4.Silvestrini RA, Soares-Filho BS, Nepstad D, Coe M, Rodrigues H, Assunção R. Simulating fire regimes in the Amazon in response to climate change and deforestation. Ecol Appl. 2011; 21: 1573–1590. doi: 10.1890/10-0827.1 [DOI] [PubMed] [Google Scholar]
  • 5.Jolly WM, Cochrane MA, Freeborn PH, Holden ZA, Brown TJ, Williamson GJ, et al. Climate-induced variations in global wildfire danger from 1979 to 2013. Nat Commun. 2015; 6: 1–11. doi: 10.1038/ncomms8537 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Saldarriaga JG, West DC. Holocene fires in the northern Amazon basin. Quat Res. 1986; 26: 358–366. [Google Scholar]
  • 7.Cochrane MA, Alencar A, Schulze MD, Souza CM, Nepstad DC, Lefebvre P, et al. Positive feedbacks in the fire dynamic of closed canopy tropical forests. Science. 1999; 284: 1832–1835. doi: 10.1126/science.284.5421.1832 [DOI] [PubMed] [Google Scholar]
  • 8.Van der Werf GR, Randerson JT, Giglio L, Collatz GJ, Mu M, Kasibhatla PS, et al. Global fire emissions and the contribution of deforestation, savanna, forest, agricultural, and peat fires (1997–2009). Atmos Chem Phys. 2010; 10: 11707–11735. [Google Scholar]
  • 9.Davidson EA, de Araújo AC, Artaxo P, Balch JK, Brown IF, Bustamante MM, et al. The Amazon basin in transition. Nature. 2012; 481: 321–328. doi: 10.1038/nature10717 [DOI] [PubMed] [Google Scholar]
  • 10.Nellemann C, editor. The last stand of the orangutan: state of emergency: illegal logging, fire and palm oil in Indonesia’s national parks. UNEP/Earthprint; 2007.
  • 11.Cochrane MA, Schulze MD. Fire as a recurrent event in tropical forests of the eastern Amazon: effects on forest structure, biomass, and species composition. Biotropica. 1999; 31: 2–16. [Google Scholar]
  • 12.Barlow J, Peres CA. Effects of single and recurrent wildfires on fruit production and large vertebrate abundance in a central Amazonian forest. Biodivers Conserv. 2006; 15: 985–1012. [Google Scholar]
  • 13.Giglio L, Randerson JT, Van Der Werf GR. Analysis of daily, monthly, and annual burned area using the fourth-generation global fire emissions database (GFED4). J Geophys Res Biogeosci. 2013; 118: 317–328. [Google Scholar]
  • 14.Myers N, Mittermeier RA, Mittermeier CG, Da Fonseca GA, Kent J. Biodiversity hotspots for conservation priorities. Nature. 2000; 403: 853–858. doi: 10.1038/35002501 [DOI] [PubMed] [Google Scholar]
  • 15.Brooks TM, Mittermeier RA, da Fonseca GA, Gerlach J, Hoffmann M, Lamoreux JF, et al. Global biodiversity conservation priorities. Science. 2006; 313: 58–61. doi: 10.1126/science.1127609 [DOI] [PubMed] [Google Scholar]
  • 16.Goodman SM., Benstead JP. Updated estimates of biotic diversity and endemism for Madagascar. Oryx. 2005; 39: 73–77. [Google Scholar]
  • 17.IUCN 2021. The IUCN Red List of Threatened Species. Version 2021–1. [cited 2021 Jun 1] https://www.iucnredlist.org.
  • 18.Allnutt TF, Ferrier S, Manion G, Powell GV, Ricketts TH, Fisher BL, et al. A method for quantifying biodiversity loss and its application to a 50-year record of deforestation across Madagascar. Conserv Lett. 2008; 1: 173–181. [Google Scholar]
  • 19.Vieilledent G, Grinand C, Rakotomalala FA, Ranaivosoa R, Rakotoarijaona JR, Allnutt TF, et al. Combining global tree cover loss data with historical national forest cover maps to look at six decades of deforestation and forest fragmentation in Madagascar. Biol Conserv. 2018; 222: 189–197. [Google Scholar]
  • 20.Harper GJ, Steininger MK, Tucker CJ, Juhn D, Hawkins F. Fifty years of deforestation and forest fragmentation in Madagascar. Environ Conserv. 2007; 34: 325–333. [Google Scholar]
  • 21.Kull CA. Isle of fire: the political ecology of landscape burning in Madagascar. University of Chicago Press; 2004. [Google Scholar]
  • 22.Kull CA., Laris P. Fire ecology and fire politics in Mali and Madagascar. In: Cochrane MA, editor. Tropical Fire Ecology. Springer-Verlag; Berlin Heidelberg; 2009. pp 171–226. [Google Scholar]
  • 23.Ganzhorn JU, Lowry PP, Schatz GE, Sommer S. The biodiversity of Madagascar: one of the world’s hottest hotspots on its way out. Oryx. 2001; 35: 346–348. [Google Scholar]
  • 24.Irwin MT, Wright PC, Birkinshaw C, Fisher BL, Gardner CJ, Glos J, et al. Patterns of species change in anthropogenically disturbed forests of Madagascar. Biol Conserv. 2010; 143: 2351–2362. [Google Scholar]
  • 25.Morelli TL, Smith AB, Mancini AN, Balko EA, Borgerson C, Dolch R, et al. The fate of Madagascar’s rainforest habitat. Nat Clim Change. 2020; 10: 89–96. [Google Scholar]
  • 26.Bloesch U. Fire as a tool in the management of a savanna/dry forest reserve in Madagascar. Appl Veg Sci. 1999; 2: 117–124. [Google Scholar]
  • 27.Alvarado ST, Buisson E, Rabarison H, Rajeriarison C, Birkinshaw C, Lowry PP II, et al. Fire and the reproductive phenology of endangered Madagascar sclerophyllous tapia woodlands. S Afr J Bot. 2014; 94: 79–87. [Google Scholar]
  • 28.Burney DA. Climate change and fire ecology as factors in the Quaternary biogeography of Madagascar. In: Lourenço WR, editor. Biogéographie de Madagascar. Paris: Editions de l’ORSTOM; 1996. pp 49–58. [Google Scholar]
  • 29.Shlisky A, Alencar AA, Nolasco MM, Curran LM. Overview: Global fire regime conditions, threats, and opportunities for fire management in the tropics. In: Cochrane MA, editor. Tropical Fire Ecology. Springer-Verlag; Berlin Heidelberg; 2009. pp 65–83. [Google Scholar]
  • 30.Scales IR. Farming at the forest frontier: Land use and landscape change in Western Madagascar, 1896–2005. Environ Hist Camb. 2011; 17: 499–524. [Google Scholar]
  • 31.Waeber PO, Wilmé L, Ramamonjisoa B, Garcia C, Rakotomalala D, Rabemananjara ZH, et al. Dry forests in Madagascar: neglected and under pressure. Int For Rev. 2015; 17: 127–148. [Google Scholar]
  • 32.Styger E, Rakotondramasy HM, Pfeffer MJ, Fernandes EC, Bates DM. Influence of slash-and-burn farming practices on fallow succession and land degradation in the rainforest region of Madagascar. Agric Ecosyst Environ. 2007; 119: 257–269. [Google Scholar]
  • 33.United Nations. World Population Prospects 2019; 2019 [cited 2021 Jun 1] https://population.un.org/wpp/DataQuery/
  • 34.Gautier L, Tahinarivony JA, Ranirison P, Wohlhauser S. Vegetation. In: Goodman SM, Raherilalao MJ, Wohlhauser S, editors. The terrestrial protected areas of Madagascar: Their history, description, and biota. Antananarivo: Association Vahatra; pp. 207–242. [Google Scholar]
  • 35.Klein JR, Reau B, Kalland I, Edwards M. Conservation, development, and a heterogeneous community: The case of Ambohitantely Special Reserve, Madagascar. Soc Nat Resour. 2007; 20: 451–467. [Google Scholar]
  • 36.Scales, H. Madagascar’s Mangroves: The Ultimate Giving Trees. Hakai Magazine. [Cited 2021 Jun 1]. https://www.smithsonianmag.com/science-nature/the-giving-trees-madagascar-180960865/
  • 37.Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GV, Underwood EC, et al. Terrestrial Ecoregions of the World: A New Map of Life on Earth: A new global map of terrestrial ecoregions provides an innovative tool for conserving biodiversity. BioScience. 2001; 51: 933–938. [Google Scholar]
  • 38.Wilmé L, Goodman SM, Ganzhorn JU. Biogeographic evolution of Madagascar’s microendemic biota. Science. 2006; 312: 1063–1065. doi: 10.1126/science.1122806 [DOI] [PubMed] [Google Scholar]
  • 39.Sharma A, Wang J, Lennartson EM. Intercomparison of MODIS and VIIRS fire products in Khanty-Mansiysk Russia: Implications for characterizing gas flaring from space. Atmosphere. 2017; 8: 95. [Google Scholar]
  • 40.Vadrevu K, Lasko K. Intercomparison of MODIS AQUA and VIIRS I-Band fires and emissions in an agricultural landscape—Implications for air pollution research. Remote Sens. 2018; 10: 978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Vadrevu KP, Lasko K, Giglio L, Schroeder W, Biswas S, Justice C. Trends in vegetation fires in south and southeast Asian countries. Sci Rep. 2019; 9: 1–13. doi: 10.1038/s41598-018-37186-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Schroeder W, Oliva P, Giglio L, Csiszar IA. The New VIIRS 375 m active fire detection data product: Algorithm description and initial assessment. Remote Sens Environ. 2014; 143: 85–96. [Google Scholar]
  • 43.De Jager NR, Fox TJ. Curve Fit: a pixel-level raster regression tool for mapping spatial patterns. Methods in Ecol Evol. 2013; 4: 789–792. [Google Scholar]
  • 44.Steffens TS, Ramsay MS, Mercado Malabet FM, Lehman SM. The Effects of Forest Loss and Fragmentation on non-Volant Mammals in Madagascar. In: Goodman SM, editor. The New Natural History of Madagascar. Princeton: Princeton University Press; Forthcoming. [Google Scholar]
  • 45.Juffe-Bignoli D, Burgess ND, Bingham H, Belle EMS, De Lima MG, Deguignet M, et al. Protected planet report 2014. UNEP-WCMC: Cambridge, UK. 2014. [Google Scholar]
  • 46.CIESIN—Center for International Earth Science Information Network—Columbia University. Gridded Population of the World, Version 4 (GPWv4): Population Density, Revision 11. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC); 2018 [cited 2021 Jun 1].
  • 47.Dunham AE, Erhart EM, Wright PC. Global climate cycles and cyclones: consequences for rainfall patterns and lemur reproduction in southeastern Madagascar. Glob Change Biol. 2011; 17: 219–227. [Google Scholar]
  • 48.Axel AC. Burned area mapping of an escaped fire into tropical dry forest in western Madagascar using multi-season Landsat OLI data. Remote Sens. 2018; 10: 371. [Google Scholar]
  • 49.Archibald S, Roy DP, van Wilgen BW, Scholes RJ. What limits fire? An examination of drivers of burnt area in Southern Africa. Glob Change Biol. 2009; 15: 613–630. [Google Scholar]
  • 50.Andela N, Morton DC, Giglio L, Chen Y, van der Werf GR, Kasibhatla PS, et al. A human-driven decline in global burned area. Science. 2017; 356: 1356–1362. doi: 10.1126/science.aal4108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Minten B, Sander K, Stifel D. Forest management and economic rents: Evidence from the charcoal trade in Madagascar. Energy Sustain Dev. 2013; 17: 106–115. [Google Scholar]
  • 52.Thampanya U, Vermaat JE, Sinsakul S, Panapitukkul N. Coastal erosion and mangrove progradation of Southern Thailand. Estuar Coast Shelf Sci. 2006; 68: 75–85. [Google Scholar]
  • 53.Friess DA, Rogers K, Lovelock CE, Krauss KW, Hamilton SE, Lee SY, et al. The state of the world’s mangrove forests: past, present, and future. Annu Rev Environ Resour. 2019; 44: 89–115. [Google Scholar]
  • 54.Rakotomavo A, Fromard F. Dynamics of mangrove forests in the Mangoky River delta, Madagascar, under the influence of natural and human factors. Forest Ecol Manag. 2010; 259: 1161–1169. [Google Scholar]
  • 55.Jones TG, Ratsimba HR, Ravaoarinorotsihoarana L, Glass L, Benson L, Teoh M, et al. The dynamics, ecological variability and estimated carbon stocks of mangroves in Mahajamba Bay, Madagascar. J Mar Sci Eng. 2015; 3: 793–820. [Google Scholar]
  • 56.Scales IR, Friess DA. Patterns of mangrove forest disturbance and biomass removal due to small-scale harvesting in southwestern Madagascar. Wetl Ecol Manag. 2019; 27: 609–625. [Google Scholar]
  • 57.Brand J, Pfund JL. Site-and watershed-level assessment of nutrient dynamics under shifting cultivation in eastern Madagascar. Agric Ecosyst Environ. 1998; 71: 169–183. [Google Scholar]
  • 58.Sankaran M, Hanan NP, Scholes RJ, Ratnam J, Augustine DJ, Cade BS, et al. Determinants of woody cover in African savannas. Nature. 2005; 438: 846–849. doi: 10.1038/nature04070 [DOI] [PubMed] [Google Scholar]
  • 59.Barlow J, Peres CA, Lagan BO, Haugaasen T. Large tree mortality and the decline of forest biomass following Amazonian wildfires. Ecol Lett. 2003; 6: 6–8. [Google Scholar]
  • 60.Goldammer JG, Pfund JL, Helfert MR, Lulla KP, the STS-61 Mission Crew. Use of the earth observation system in the space shuttle program for research and documentation of global vegetation fires: A case study from Madagascar. In: Levine JS, editor. Biomass Burning and Global Change. Cambridge, MA: MIT Press; 1996. pp 236–240. [Google Scholar]
  • 61.Giglio L. Characterization of the tropical diurnal fire cycle using VIRS and MODIS observations, Remote Sens Environ. 2007; 108: 407–421. [Google Scholar]
  • 62.Roy DP, Boschetti L, Justice CO, Ju J. The collection 5 MODIS burned area product—Global evaluation by comparison with the MODIS active fire product. Remote Sens Environ. 2008; 112: 3690–3707. [Google Scholar]
  • 63.Giglio L., Schroeder W, Justice CO. The collection 6 MODIS active fire detection algorithm and fire products. Remote Sens Environ. 2016; 178: 31–41. doi: 10.1016/j.rse.2016.02.054 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Csiszar I, Schroeder W, Giglio L, Ellicott E, Vadrevu KP, Justice CO, et al. Active fires from the Suomi NPP Visible Infrared Imaging Radiometer Suite: Product status and first evaluation results. J Geophys Res Atmos. 2014; 119: 803–816. [Google Scholar]
  • 65.Polivka TN, Wang J, Ellison LT, Hyer EJ, Ichoku CM. Improving nocturnal fire detection with the VIIRS day–night band. IEEE Trans Geosci Remote Sens. 2016; 54: 5503–5519. [Google Scholar]
  • 66.Bruner AG, Gullison RE, Rice RE, Da Fonseca GA. Effectiveness of parks in protecting tropical biodiversity. Science. 2001; 291: 125–128. doi: 10.1126/science.291.5501.125 [DOI] [PubMed] [Google Scholar]
  • 67.Tabor K, Jones KW, Hewson J, Rasolohery A, Rambeloson A, Andrianjohaninarivo T, et al. Evaluating the effectiveness of conservation and development investments in reducing deforestation and fires in Ankeniheny-Zahemena Corridor, Madagascar. PLOS ONE. 2017; 12: e0190119. doi: 10.1371/journal.pone.0190119 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Gardner CJ, Nicoll ME, Birkinshaw C, Harris A, Lewis RE, Rakotomalala D, et al. The rapid expansion of Madagascar’s protected area system. Biol Conserv. 2018; 220: 29–36. [Google Scholar]
  • 69.Rasolofoson RA, Ferraro PJ, Jenkins CN, Jones JP. Effectiveness of community forest management at reducing deforestation in Madagascar. Biol Conserv. 2015; 184: 271–277. [Google Scholar]
  • 70.Gardner CJ, Jasper LD, Eonintsoa C, Duchene JJ, Davies ZG. The impact of natural resource use on bird and reptile communities within multiple-use protected areas: evidence from sub-arid Southern Madagascar. Biodivers Conserv. 2016; 25: 1773–1793. [Google Scholar]
  • 71.Hill DM. Forest restoration in Eastern Madagascar: Post-fire survival of select Malagasy tree species M.Sc. Thesis, University of Minnesota. 2018. https://www.proquest.com/docview/2130166076
  • 72.Parr CL, Chown SL. Burning issues for conservation: a critique of faunal fire research in Southern Africa. Austral Ecol. 2003; 28: 384–395. [Google Scholar]
  • 73.Barlow J, Silveira JM. The consequences of fire for the fauna of humid tropical forests. In: Cochrane MA, editor. Tropical Fire Ecology. Springer-Verlag; Berlin Heidelberg; 2009. pp. 543–556. [Google Scholar]
  • 74.Peres CA. Ground fires as agents of mortality in a Central Amazonian forest. J Trop Ecol. 1999; 15: 535–541. [Google Scholar]
  • 75.Dunham AE, Erhart EM, Overdorff DJ, Wright PC. Evaluating effects of deforestation, hunting, and El Niño events on a threatened lemur. Biol Conserv. 2008; 141: 287–297. [Google Scholar]
  • 76.Golden CD. Bushmeat hunting and use in the Makira Forest, north-eastern Madagascar: a conservation and livelihoods issue. Oryx. 2009; 43: 386–392. [Google Scholar]
  • 77.Cochrane MA, Laurance WF. Fire as a large-scale edge effect in Amazonian forests. J Trop Ecol. 2002; 18: 311–325. [Google Scholar]
  • 78.Saunders DA, Hobbs RJ, Margules CR. Biological consequences of ecosystem fragmentation: a review. Conserv Biol. 1991; 5: 18–32. [Google Scholar]
  • 79.Lehman SM, Rajaonson A, Day S. Edge effects and their influence on lemur density and distribution in southeast Madagascar. Am J Phys Anthropol. 2006; 129: 232–241. doi: 10.1002/ajpa.20241 [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Ji-Zhong Wan

21 Oct 2021

PONE-D-21-27734The burning island: Spatiotemporal patterns of fire occurrence in MadagascarPLOS ONE

Dear Dr. Lehman,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Dec 05 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Ji-Zhong Wan

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at 

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and 

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please provide the specific VIIRS data you extracted to use in the study, either by adding a URL to a data repository in the Methods section or by uploading a supplemental file

3.We note that Figure 1 in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

1. You may seek permission from the original copyright holder of Figure 1 to publish the content specifically under the CC BY 4.0 license.  

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

Additional Editor Comments (if provided):

Please revise the manuscript according to the reviewers' comments fully.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper analyzes VIIRS satellite fire detection data to assess which regions of Madagascar have the most prevalent fire use, how fire use is changing over time, and what this means for Madagascar’s remaining forest ecosystems.

In my opinion, the topic is very interesting and worth investigating. I Have just a few minor comments:

- The abstract should be more assertive in describing the novelty of the paper;

- The abstract and Introduction should mention the methodology adopted in the research;

- There should be a small section for conclusions.

Reviewer #2: This study quantifies VIIRS hotspot detection, a proxy for fire use, on Madagascar, summarized by different ecoregions. It is well written and interesting There is no statistical modelling of drivers of fire use in different locations, which would provide further insight into fire use. However, the study provides an important summary/quantification of the issue, which sets the stage for greater future research. The authors are also clear about their method, their aim with the study and the discussion flows well from the results. The authors have summarized and presented the information well, and I believe this will be a valuable paper to researchers interested in human fire use, slash and burn agriculture and deforestation in the tropics.

I have several minor comments – mainly some clarification around certain text and some further explanation that is required to improve the manuscript - see attached document.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Attachment

Submitted filename: comments_to_authors.pdf

Click here for additional data file. (217.4KB, pdf)
PLoS One. 2022 Mar 31;17(3):e0263313. doi: 10.1371/journal.pone.0263313.r002

Author response to Decision Letter 0

7 Dec 2021

We would like to thank our reviewer for their knowledgeable and insightful feedback that greatly improved the clarity of our manuscript. Please find an itemized list below of how we integrated specific comments into our revised manuscript.

Figure 1 copyright concerns: There were concerns expressed in the feedback of our manuscript that the ecoregion map used in Figure 1 may be subject to previous copyright that would impede the publication of these data in PLOS ONE under the PLOS ONE’s CC BY license. However, the data used here were obtained from the World Wildlife Fund’s ecoregions map, which is available to the public to be used and edited under an active CC BY license. Slight edits to the borders of some ecoregion shape files were made following the regions used in the following article which SM Lehman also authored:

Steffens TS, Ramsay MS, Mercado Malabet FM, Lehman SM. The Effects of Forest Loss and Fragmentation on non-Volant Mammals in Madagascar. In: Goodman SM, editor. The New Natural History of Madagascar. Princeton: Princeton University Press; Forthcoming.

The editing of these creative commons data obtained from the WWF does not change the copyright, and this Steffens et al. article was simply cited so that any edits of the WWF CC BY data can be tracked accordingly. We apologize that this was not made clear in the text of our manuscript, and we have made revisions to help clarify this in our Figure 1 caption and the section of the Methods describing the ecoregions map.

The WWF Ecoregion map/data can be obtained online at: https://databasin.org/datasets/68635d7c77f1475f9b6c1d1dbe0a4c4c/

General comments

These are my main comments to the authors.

• In my opinion, you need to be clear how you are defining a “fire” in this paper. It looks like you are saying one hotspot is one fire, but one fire is usually defined (in fire behaviour/fire history studies) as the single continuous burnt area. That is, one wildfire can have multiple hotspots detected at once, e.g. in the picture below (VIIRS image overlayed with hotspots) I would say the green/yellow polygon (the eventual fire boundary) is one fire, and (at this particular time) there are multiple hotspots. So, I’d say you need to clearly define how you are using the word “fire”, or change your use of fire to hotspot or another term in the text and in table 1 for example.

Response: We would like to thank the reviewer for pointing out that we did not adequately clarify this key definition to our paper. We have added a clearer description of how we will be using the term “fire”, and the details of VIIRS fire hotspots to Methods section 2.1 (Line 140)

• Adding some explanation about how clouds may have affected the results is needed. Are there any areas/ecoregions of Madagascar that a consistently cloudier than other areas? If so, you could expect that VIIRS detection would be lower here because clouds may shielded fires from detection by VIIRS. This may be important for the results, e.g. if the lowlands are cloudier, than VIIRS may under-detect fires in this region. I think some acknowledgement of this issue in the methods or in section 4.5 is warranted.

Response: This is an excellent point, and we have added a brief paragraph to Section 4.5 (line 379) describing how cloud cover may disproportionately affect VIIRS detections at certain locations or in certain times of year.

Specific comments

Line 42: It would be useful to add some specific examples of how or why fires have negative impacts on tropical forests (e.g. trees easily killed by fire because they are not adapted), rather than only the broad statement about “devastating effects” etc.

Response: Example added in text.

Line 43: “the most fire-threatened continent on Earth”. I’d add this to the next paragraph, as the previous paragraph doesn’t provide support for this statement.

Response: Changed

Line 100 and 102: What does “ca.” represent?

Response: Changed “ca.” to “approximately”

Line 102: The sentence starting “In the southwest, the climate becomes” has unclear wording. It sounds like the aphyllous shrubs continue south. I suggest rewording this sentence.

Response: Changed

Line 125: (Section 2.1) Need to add a sentence or few words saying that there are two VIIRS instruments on separate satellites: Suomi NPP and NOAA-20. It is stated that you only used S-NPP, but it is worth saying there is another satellite now, but it is newer and had not collected data for your study period.

Response: Clarification added in text.

Line 134: Similar to the previous point, better to clarify which satellite instrument. “VIIRS satellite” could refer to either, better to say VIIRS S-NPP or something similar.

Response: Changed

Line 144: Were there any industrial/urban areas false detection that needed to be excluded for your study area? Perhaps this was accounted for anyway when doing the intersection with the ecoregion polygons

Response: Addressed in line 153

Line 236: Just a question, could this later peak be because mangroves take a long time to dry out, but when they do they are very flammable?

Response: This is an interesting question, and one that we hope to future studies to determine the reasons behind the observed patterns in fire occurrence across ecoregions.

Line 239: I believe the fig 3 caption should say “expressed in the number of fires per km2 per month)

Response: Changed

Line 265: I found this chart a little confusing. This may be better as a bar chart i.e each buffer distance along the x axis and % along the y axis, and then a small bar for each ecoregion (repeated at each buffer distance). E.g. the “grouped bar chart” here: https://www.r-graph-gallery.com/48-grouped-barplotwith-ggplot2.html. Otherwise, add the points that make up each line so the buffer categories are clear.

Response: We have added points to the figure that denote each buffer distance, as well as horizontal lines at 100%, 50%, and 25% of remaining forests to make the differences between plots more clear and easier to interpret. We believe that the number of buffer distances included in the analysis would make a bar chart too expansive to include for all seven ecoregions, and a line graph allows for more data to be conveyed. We believe this is much clearer with the suggested addition of points along each line.

Line 267: Saying “that was within each buffer distance of the closest fire” would make this clearer.

Response: Changed

Line 271: Saying “effect” in the Table 1 caption implies a statistical model, where this just seems to be showing differences in summary statistics. I think you should say “Differences in distance to nearest fire by National park status” or something similar.

Response: Changed

Line 288: Reassuring in terms of what?

Response: Clarification added “reassuring for conservation efforts”

Line 298: “Analysis of the rate of change in fire frequency between 2012 and 2019 revealed that fire frequencies are increasing throughout the dry deciduous forests ecoregion“ I suggest adding a table (or adding to an existing table) the mean change for each ecoregion type. I think this statement is just based of looking at the maps. It would be good to see some numbers.

Response: We agree that seeing the numbers for this statement help to make it more compelling. We have added the percentage of land within each ecoregion that saw an increasing trend in fire use across the study period to the Results section 3.2 (line 223)

Line 360-366: As in the general comment, how does cloud influence this? Are the lowlands cloudier, so hotspots undetected there.

Response: Clarified in an added paragraph in Discussion Section 4.5

Also, you should not use the word “alarming” without explaining why. I would reword this final sentence (Line 366) to something like: If fires within forest are underestimated, the impact of fire on the forests is greater than can be determined/estimated by VIIRS detections..”, or something more similar. Response: Changed

Line 369: Avoid “reassuring” without further explanation. Or just explain that it is a positive in terms of conservation outcomes

Response: Clarification added

Line 372: Is the 19.8% for non-national parks? Also probably better to say 2.8 times lower, rather than 2.8 fold decrease.

Response: Clarified, and changed

Also, could it be national parks are more commonly in less fire-prone areas, e.g. steep/hilly areas more inaccessible to farmers, therefore less fires ignited there due to the terrain rather than national park status. Add a note if you think that could be influencing the results.

Response: This is a very interesting topic, and one that we hope to explore more fully in future studies. Determining the reasons behind lower fire frequency within national parks will ideally require a pixel matching approach, but these methods are currently hindered in Madagascar by the lack of an accurate, high-resolution population density map that would allow habitats with comparable anthropogenic pressures, climate, and topography to be compared across protected statuses. Without such an analysis we do not know if topography/inaccessibility may be contributing to these results, and while some national parks along the eastern escarpment do have steep topography, national parks across the north and west are largely located on flatter land below the Central Plateau and close to the coast.

Line 389: What is guard density? Could you please clarify.

Response: Clarification added

Attachment

Submitted filename: Response to Reviewers_2.docx

Click here for additional data file. (20KB, docx)

Decision Letter 1

Ji-Zhong Wan

18 Jan 2022

The burning island: Spatiotemporal patterns of fire occurrence in Madagascar

PONE-D-21-27734R1

Dear Dr. Lehman,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Ji-Zhong Wan

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors improved their paper inline with previous comments from the reviewrs. I have no further comments.

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Acceptance letter

Ji-Zhong Wan

23 Mar 2022

PONE-D-21-27734R1

The burning island: Spatiotemporal patterns of fire occurrence in Madagascar

Dear Dr. Lehman:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Ji-Zhong Wan

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Fig. Total number of fires detected per year in Madagascar by VIIRS and MODIS datasets.

    Total number of fires detected by MODIS (left axis) from 2003–2019 and VIIRS (right axis) from 2012–2019. VIIRS detected an average of 5.6 times more fires than MODIS.

    (TIF)

    Click here for additional data file. (184.9KB, tif)
    S1 Data

    (ZIP)

    Click here for additional data file. (40.2MB, zip)
    S2 Data

    (ZIP)

    Click here for additional data file. (40.3MB, zip)
    S3 Data

    (ZIP)

    Click here for additional data file. (42.5MB, zip)
    S4 Data

    (ZIP)

    Click here for additional data file. (42.6MB, zip)
    S5 Data

    (ZIP)

    Click here for additional data file. (43.3MB, zip)
    S6 Data

    (ZIP)

    Click here for additional data file. (45.6MB, zip)
    S7 Data

    (ZIP)

    Click here for additional data file. (51.1MB, zip)
    S8 Data

    (ZIP)

    Click here for additional data file. (53MB, zip)
    S1 File

    (PDF)

    Click here for additional data file. (74.9KB, pdf)
    Attachment

    Submitted filename: comments_to_authors.pdf

    Click here for additional data file. (217.4KB, pdf)
    Attachment

    Submitted filename: Response to Reviewers_2.docx

    Click here for additional data file. (20KB, docx)

    Data Availability Statement

    All relevant data are within the manuscript and its Supporting Information files.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES