Fire Scar Maps Research Articles

Assessing the suitability of a surveillance fauna-monitoring program for detecting future changes in reptile occupancy

Context Ecological surveillance monitoring typically targets multiple taxonomic groups by using standardised sampling across large spatial scales. Although surveillance monitoring confers advantages over hypothesis-driven monitoring in its broad taxonomic and spatial scope, the approach has been criticised for its disconnect from ecological management and failure to provide insights on the drivers of ecological change Aims To assess the adequacy of a plot-based general fauna-monitoring program for sampling reptiles as indicators of ecosystem health in a semi-arid upland region of the Northern Territory, Australia. Methods We surveyed reptiles at 90 sites, stratified between major landform and vegetation types, and using standard fauna-sampling methods, across the 2568 km2 Tjoritja National Park in the MacDonnell Ranges. We compiled a full inventory of the reptile fauna of the study area and identified species with potential utility as ecological indicators. We then used single-season occupancy models and power analyses to evaluate the adequacy of sampling for detecting potential future changes in occupancy. Key results We detected 57 of the 68 reptile species known from the protected area, 17 of which are potentially useful indicators of ecological health, mostly related to fire management. There was insufficient power to detect moderate (50%) future changes in reptile occupancy for all but the single most detected species. For the two ecological indicator species with sufficient detections for occupancy modelling, a positive association with a keystone structure (dense spinifex grass) was confirmed. However, increasing detection probability or the number of surveys would result in only minor improvements in power to detect occupancy change in these species. Conclusions Although reptiles are potentially useful indicators of ecological health, particularly in relation to fire regimes, the number of sites required to detect future changes in reptile occupancy by using standardised plot-based monitoring in this protected area is prohibitively high. Implications Our results suggest that once ecological associations are understood, monitoring ecological health remotely by using techniques such as fire-scar mapping to track proportions of long-unburnt vegetation should be considered over labour-intensive surveillance monitoring for reptiles. Targeted monitoring of threatened and other reptile species of conservation or cultural concern may also be warranted.

Fuel dynamics and rarity of fire weather reinforce coexistence of rainforest and wet sclerophyll forest

Rainforest, wet sclerophyll forest and eucalypt forest exist as Alternative Stable States moderated by fire. Eucalypt forest is flammable and subject to regular fire. Insights into the flammability of rainforest and wet sclerophyll forest are difficult to obtain because of the rarity of fire, particularly in sub-tropical and tropical environments. This study aimed to determine the relative flammability of rainforest and wet sclerophyll forest with a canopy of Lophostemon confertus in subtropical Australia. Fires in 2019, coinciding with extreme fire weather and following a long period of drought, burnt rainforest and wet sclerophyll forest in south-east Queensland. The weather record indicates these conditions recur about three to four times a century. Remapping of the vegetation overlaid with the fire-scar mapping, indicated that the 2019 fires burnt 23 % of rainforest, much of which was disturbed, but 60 % of wet sclerophyll forest suggesting greater flammability of the latter. This study revealed that above-ground biomass and litter fuel loads are more than twice as high in wet sclerophyll forest than rainforest. The litter of Lophostemon confertus, a Myrtaceae tree dominating the upper canopy, on average decomposes at less than half the rate of rainforest litter, contributing to the higher fuel loads and thus greater flammability of wet sclerophyll forest compared to rainforest. The Alternative Stable States of wet sclerophyll and rainforest are partly stabilised by the fire retardance of rainforest, the rarity of weather conditions suitable for fire in the subtropical environment and the higher fuel loads in wet sclerophyll forest that allow occasional fire. Wet sclerophyll forest requires rare fire events and the legacy of these fires over hundreds of years is evident in the multiple cohorts of ‘giant’ canopy trees.

Characterization of forest fire frequency using fire scar mapping of temporal satellite data for forest fire management

One of the most complex problems facing in tropical forests, particularly in deciduous forests, is the recurrent incidence of fire. It is well known that fire caused extensive damage in the forest ecosystem by quantitatively and qualitatively. To reduce occurrences of forest fire, proper management of fire is highly important which entails mapping of forest fire frequency and identification of suitable area for watchtowers. In the present study, fire frequency analysis of Melghat Tiger Reserve, Maharashtra was done for the last seven years (2014-2020) based on the fire scar on the temporal Landsat data during fire season (January-June). Fire frequency analysis shows that an area of 1053.64 ha (0.52%) of the reserve was burned all seven years followed by 3050.53 ha (1.49%) for six times, 3849.52 ha (1.88%) for five times, 5520.04 ha (2.70%) for four times, 11845.63 ha (5.80%) for three times, 36863.52 ha (18.03%) for two times, 70126.33 ha (34.31%) for once and 72093.02 ha (35.27%) remains unburned all these seven years. The fire frequency map generated was used as an input for prioritizing the locations of watch towers as well as prioritizing grazing closure areas and fires lines. Identification of suitable sites for locating new watchtowers has been done by integrating and modeling of forest fire frequency map, existing watchtowers and viewshed analysis in GIS. Based on these results only thirteen watchtowers were categorized under retainable among the existing watchtowers and 27 new watchtowers are proposed to cover the entire area.

Remote spatial analysis lacking ethnographic grounding mischaracterizes sustainability of Indigenous burning regime

Abstract: Scientific research that purports to evaluate Indigenous fire regimes in the absence of ethnographically contextualized ecological data runs the risk of exacerbating the fire blame game and providing evidence to support distorted narratives advanced by anti-Indigenous advocates. Spatial analysis of fire scars in Indigenous territories can be an effective tool for characterizing cultural fire regimes in terms of distribution and frequency, especially when qualified by linkages to different local ecosystems. A recently published article drew on fire scar mapping from satellite imagery to assess anthropogenic fire distribution and frequency in the Pimentel Barbosa Indigenous Land, Central Brazil. The authors use their findings to characterize A'uwẽ (Xavante) use of fire as unmanaged and a model of unsustainable use of cerrado resources. In this article, we discuss Aguiar & Martins's recent paper in light of our long-term research on A'uwẽ hunting with fire in the Pimentel Barbosa Indigenous Land, arguing that A'uwẽ hunters do burn according to established cultural protocols, manage their use of fire for conservationist purposes, and do not cause environmental degradation by burning.

A multi-scale assessment of fire scar mapping in the Great Victoria Desert of Western Australia

Fire management is increasingly acknowledged as a necessary tool to maintain diversity in desert environments such as the Great Victoria Desert of Australia, but it needs to be informed by accurate fire history data. We compared and assessed the utility of Landsat-derived and Moderate Resolution Imaging Spectroradiometer (MODIS)-derived burnt area mapping (30 m and 250 m resolution, respectively) for sub-regional, landscape and habitat scale management. We did so by using Sentinel-2-derived, 10 m resolution, burnt area mapping as a reference, to determine the most appropriate product to support land management planning. At the landscape scale, Landsat had significantly lower average omission and commission errors (3.4% and 8.0%, respectively) compared with that of MODIS (42.2% and 19.9%, respectively). At the habitat scale, Landsat burnt area percentage was more accurate, in plots of 500 m × 500 m (root mean square error (RMSE) 0.6% to 8.6%), but offered lower accuracy when estimating partially burnt habitat plots of 120 m × 120 m (RMSE 14.1% to 23.9%). Only Landsat-derived fire scar mapping provided enough detail to produce reliable fire history maps to inform fire management and biodiversity conservation operations at a sub-regional scale, landscape scale and a habitat scale of 500 m by 500 m.

A spatial vulnerability assessment of monsoonal wetland habitats to para grass invasion in Kakadu National Park, northern Australia

Kakadu National Park and its wetlands are World Heritage and Ramsar listed and are at risk from invasive grasses. However, it appears that not all habitats and native vegetation are equally at risk. We conduct a spatial risk assessment for para grass (Urochloa mutica) invasion across seasonally inundated habitats of the 258 km2 Magela Creek floodplain within Kakadu National Park using Landsat 5 TM time-series imagery. Two maps, representing water depth and fire history, were derived from the imagery using object-based image analyses. Depth was modelled using a linear regression relationship established between 254 known water depth locations and the multi-date spectral index values of segmented image objects at corresponding locations (R2 = 0.67; p < 0.0001). Binary fire-scar maps were then produced for each year of a 10-year period using visual interpretation and nearest neighbour classification. A map of the incidence of annual fire over this period was then calculated from the sum of the maps, overlaid. The maps were integrated in a GIS with an existing Landsat vegetation map to measure spatial inter-relationships between para grass, native vegetation, depth and fire. With a highly clustered distribution pattern, para grass occupied 1388 ha or 6% of the total floodplain area. However, its optimal depth habitat, estimated to be from 1.1 to 1.4 m, occurred over a much larger area (7180 ha) or 30% of the floodplain. Only 2% of this optimal area was actually occupied by para grass. Together the low occupancy of ‘optimal-depth’ habitat and a highly clustered distribution of para grass strongly suggested that, if left uncontrolled, it has capacity to spread further and eventually occupy much larger areas of this floodplain at high density. Landsat provided spatial information of suitable scale and accuracy to understand the landscape ecology of para grass; and from which to design and conduct further research, or trial management interventions to protect wetland vegetation at risk to weed invasion.

Landscape fragmentation, severe drought, and the new Amazon forest fire regime.

Changes in weather and land use are transforming the spatial and temporal characteristics of fire regimes in Amazonia, with important effects on the functioning of dense (i.e., closed-canopy), open-canopy, and transitional forests across the Basin. To quantify, document, and describe the characteristics and recent changes in forest fire regimes, we sampled 6 million ha of these three representative forests of the eastern and southern edges of the Amazon using 24 years (1983-2007) of satellite-derived annual forest fire scar maps and 16 years of monthly hot pixel information (1992-2007). Our results reveal that changes in forest fire regime properties differentially affected these three forest types in terms of area burned and fire scar size, frequency, and seasonality. During the study period, forest fires burned 15% (0.3 million ha), 44% (1 million ha), and 46% (0.6 million ha) of dense, open, and transitional forests, respectively. Total forest area burned and fire scar size tended to increase over time (even in years of average rainfall in open canopy and transitional forests). In dense forests, most of the temporal variability in fire regime properties was linked to El Nino Southern Oscillation (ENSO)-related droughts. Compared with dense forests, transitional and open forests experienced fires twice as frequently, with at least 20% of these forests' areas burning two or more times during the 24-year study period. Open and transitional forests also experienced higher deforestation rates than dense forests. During drier years, the end of the dry season was delayed by about a month, which resulted in larger burn scars and increases in overall area burned later in the season. These observations suggest that climate-mediated forest flammability is enhanced by landscape fragmentation caused by deforestation, as observed for open and transitional forests in the Eastern portion of the Amazon Basin.

Fire scar mapping for disaster response in KwaZulu-Natal South Africa using Landsat 8 imagery

This study assessed the potential of the new Landsat 8 multispectral imagery in rapidly mapping fire scars to aid disaster management response teams in emergency efforts. Maximum likelihood and iso cluster algorithms where used to classify burnt and unburnt areas in KwaZulu-Natal, South Africa. The Landsat 8 sensor successfully classified burnt and unburnt areas with overall accuracies ranging from 80% to 93.33% on independent test datasets. Farms and communities affected by the wildfires were overlaid with the classified maps in order to determine the extent of each farm burnt. Maps were created for disaster management response teams in order to identify critical farms and communities in need of assistance. The study indicates the operational use of the new Landsat 8 data in fire scar mapping for disaster response. The result is critical for fire scar mapping in South Africa using freely available Landsat 8 multispectral data.

Mapping spatial and temporal patterns of Mediterranean wildfires from MODIS

Wildfires are part of the Mediterranean ecosystem, however, in Israel all wildfires are human caused, either intentionally or un-intentionally. In this study we aimed to develop and test a new method for mapping fire scars from MODIS imagery, to examine the temporal and spatial patterns of wildfires in Israel in the 2000s and to examine the factors controlling Israel's wildfire regime. To map the fires we used two ‘off-the-shelf’ MODIS fire products as our basis—the 1km MODIS Collection 5 fire hotspots, the 500m MCD45A1 burnt areas—and we created a new set of fire scar maps from the 250m MOD13Q1 product. We carried out a cross comparison of the three MODIS based wildfire scar maps and evaluated them independently against the wild fire scars mapped from 30m Landsat TM imagery. To examine the factors controlling wildfires we used GIS layers of rainfall, land use, and a Landsat-based national vegetation map. Wildfires occurred in areas where annual rainfall was above 250mm, mostly in areas with herbaceous vegetation. Wildfire frequency was especially high in the Golan Heights and in the foothills of the Judean mountains, and a high correspondence was found between military training zones and the spatial distribution of fire scars. The use of MODIS satellite images enabled us to map wildfires at a national scale due to the high temporal resolution of the sensor. Our MOD13Q1 based mapping of fire scars adequately mapped large (>1km2) fires with accuracies above 80%. Such large fires account for a large proportion of all fires, and pose the greatest threats. This database can aid managers in determining wildfire risks in space and in time.

Comparison of burnt area estimates derived from satellite products and national statistics in Europe

Burnt area maps based on satellite observations are frequently used in calculations related to fire regime, such as those of carbon dioxide emissions. Nevertheless, burnt area estimates between products vary widely, and validation against independent data is scarce, especially for Europe. Here we compare two active fire maps (the ATSR World Fire Atlas and the Moderate Resolution Imaging Spectroradiometer (MODIS) Active Fire Product) and two fire scars maps (the L3JRC and the MODIS Burned Area Product) to independent national statistics taken from 22 European countries between 1997 and 2008. We also tested the coincidence between satellite products derived by calculation of the fraction of active fires that were confirmed by a subsequent drop in reflectance. As a large proportion of fire pixels (between 40% and 66%, depending on the product) is located on urban land or crop fields, filtering out fires located on these land uses greatly improves the agreement between satellite-based burnt area estimates and national statistics and it also improves the coincidence between satellite products. The MODIS Active Fire Product appears to be most suitable for use as a proxy for burnt area patterns, showing a high correlation to national statistics (R2 = 0.9), relatively low spatial and temporal heterogeneity and only a slight underestimation of the total burnt area (19 000 ha year–1). Unfiltered products show cases of substantial wildfire overestimation in all products, mainly attributable to anthropogenic activity, in the case of active fire products, and drought-induced vegetation dieback, in that of fire scar maps. Thus, filtering out fires on anthropogenic land uses seems to be essential when analysing patterns of forest fires from satellite observations. However, if agricultural fires are to be included, a combination of MODIS Active Fire and MODIS Burned Area products is recommended. We obtained that such combination shows low temporal and spatial heterogeneity and the highest coincidence between satellite products (25%), although the correlation to national statistics is not very high (R2 = 0.67) and clearly underestimates the total burnt area (187 000 ha year–1).

Delineating a managed fire regime and exploring its relationship to the natural fire regime in East Central Florida, USA: A remote sensing and GIS approach

A managed fire regime on John F. Kennedy Space Center, Florida and surrounding federal properties was mapped using time series satellite imagery and GIS techniques. Our goals were to: (1) determine if an image processing technique designed for individual fire scar mapping could be applied to an image time series for mapping a managed fire regime in a rapid re-growth pyrogenic system; (2) develop a method for labeling mapped fire scar confidence knowing a formal accuracy analysis was not possible; and (3) compare results of the managed fire regime with regional information on natural fire regimes to look for similarities/differences that might help optimize management for persistence of native fire-dependent species. We found that the area burned by managed fire peaked when the drought index was low and was reduced when the drought index was high. This contrasts with the expectations regarding the natural fire regime of this region. With altered natural fire regimes and fire-dependent species declining in many pyrogenic ecosystems, it is important to manage fire for the survival of fire-adapted native species. The remote sensing and GIS techniques presented are effective for delineating and monitoring managed fire regimes in shrub systems that grow rapidly and may be appropriate for other fire-dependent systems world wide.

Effects of band combinations and GIS masking on fire-scar mapping at local scales in east-central Florida, USA

The fire-adapted vegetation in east-central Florida provides habitat for many threatened and endangered species, such as the Florida scrub-jay (Aphelocoma coerulescens). Accurate fire occurrence records are critically important for better understanding the relationship between fire and vegetation structure. The rapid growth rates of fire-adapted vegetation in east-central Florida make it difficult to capture detailed fire scars with remote sensing data acquired weeks after the fires. The objective of this study is to develop a reliable remote sensing approach for accurately mapping burned areas in Florida scrub vegetation at the National Aeronautics and Space Administration (NASA) John F. Kennedy Space Center (KSC) and Merritt Island National Wildlife Refuge (MINWR). Landsat thematic mapper (TM) data acquired on 21 April 1987 were used for classification experiments. Geographic information system (GIS) data layers of fire management units (FMUs) with known fire occurrence (presence or absence) were used to mask the original remote sensing data or thematic maps following classification. A separation index (SI) was used to evaluate each individual band for its power to discriminate unburned and burned areas. Twelve classifications with selected band groups derived from Landsat TM data with different geographic extents were compared using an error matrix method. The classification of the four most suitable bands derived for the entire KSC-MINWR area resulted in the highest accuracy. The final map product was generated by overlaying the classified map with the FMU data layer and masking out FMUs that did not burn. This paper addresses a number of issues relevant to the classification of burned areas and includes the effect of geographic extent (GE effect) of remote sensing data on classification, determining the best bands for classification, and cleaning classification results using GIS masking. It also serves as the first published effort to map fire scars in the Florida scrub and flatwoods vegetative communities of the southeastern US using image processing techniques.

Spatial patterns of fire occurrence in Catalonia, NE, Spain

In this paper, we analyse spatial patterns of fire occurrence in Catalonia (NE Spain) during 1975–98. Fire scar maps, discriminated by means of 30–60 m resolution remote sensing imagery, have been used as a source of fire occurrence. We employ several visual or analytical approaches to interpret fire occurrence in this region, such as those of Minnich and Chou (1997), Ricotta et al. (2001) or Krummel et al. (1987). Crucial spatial patterns such as fire size distribution, fire frequency distribution, spots and residual vegetation islands are documented. In addition, several geographical layers were overlaid with burned area maps in order to determine interactions between fire occurrence and environmental parameters such as altitude, slope, solar radiation, and burned land cover. Assuming that fire occurrence is well determined by such a posteriori empirical factors we detect areas most prone to fire in this region and aim to enhance the local forest management and conservation plans.

Mapping fire scars in a southern African savannah using Landsat imagery

The spectral, spatial and temporal characteristics of the Landsat data record make it appropriate for mapping fire scars. Twenty-two annual fire scar maps from 1972–2002 were produced from historical Landsat imagery for a semi-arid savannah landscape on the South Africa–Botswana border, centred over Madikwe Game Reserve (MGR) in South Africa. A principal components transformation (PCT) helped differentiate the spectral signal of fire scars in each image. A simple, nonparametric, supervised classification (parallelepiped) of the PCT data differentiated burned and unburned areas. During most years, fire occurrences and the percentage of area burned annually were lowest in Botswana, highest in MGR, and intermediate in South Africa outside MGR. These fire scar maps are aiding MGR managers, who are endeavouring to restore a more active fire regime following decades of fire exclusion.

Statistical analysis of fire frequency models for Catalonia (NE Spain), 1975 - 1998) based on fire scar maps from Landsat MSS data

This paper estimates fire frequency in Catalonia (NE Spain) for the last quarter of the 20th Century (1975–1998) from historical burned area maps. Remote sensing images provided perimeters of fires ≥ 30 ha, which were used to characterize the temporal patterns of fire occurrence in Catalonia. Several fire frequency models were used to reproduce the observed pattern of wildfires occurrence in the study period. Natural fire rotation period was estimated to be 133 years. Poisson tests were carried out to check random fire occurrence either along the time period or across the analysed region. Observed fires were not randomly generated either in space or in time, despite being sampled using two different plot sizes. This sampling design was also used for Mean Fire Interval (MFI) analysis, which allowed us to significantly fit a Weibull distribution to the observed proportion of fire intervals (for both sample sizes), enabling us to estimate the hazard of burning, mortality, and survivorship functions. Finally, MFI was also applied to forest regions of Catalonia, which are defined according to forest management plans based on their homogeneous climatic conditions. Such an analysis revealed relevant differences in forest management and their consequences on fire occurrence.

A dynamic algorithm for wildfire mapping with NOAA/AVHRR data

A wildfire-mapping algorithm is proposed based on fire dynamics, called the dynamic algorithm. It is applied to daily NOAA/AVHRR/HRPT data for wildland areas (scrub, chaparral, grassland, marsh, riparian forest, woodland, rangeland and forests) in California for September and October 1999. Daily AVHRR images acquired from two successive days are compared for active fire detection and burn scar mapping. The algorithm consists of four stages: data preparation; hotspot detection; burn scar mapping; and final confirmation of potential burn scar pixels. Preliminary comparisons between the result mapped by the dynamic algorithm and the fire polygons collected by the California Department of Forestry and Fire Protection through ground survey indicate that the algorithm can track burn scars at different developmental stages at a daily level. The comparisons between wildfire mapping results produced by a modified version of an existing algorithm and the dynamic algorithm also indicate this point. This is the major contribution of this algorithm to wildfire detection methods. The dynamic algorithm requires highly precise registration between consecutive images.

Reliability of biomass burning estimates from savanna fires: Biomass burning in northern Australia during the 1999 Biomass Burning and Lightning Experiment B field campaign

This paper estimates the two‐daily extent of savanna burning and consumption of fine (grass and litter) fuels from an extensive 230,000 km2region of northern Australia during August‐September 1999 encompassing the Australian continental component of the Biomass Burning and Lightning Experiment B (BIBLE B) campaign [Kondo et al., 2002]. The extent of burning for the study region was derived from fire scar mapping of imagery from the advanced very high resolution radiometer (AVHRR) on board the National Oceanic and Atmospheric Administration (NOAA) satellite. The mapping was calibrated and verified with reference to one Landsat scene and associated aerial transect validation data. Fine fuel loads were estimated using published fuel accumulation relationships for major regional fuel types. It is estimated that more than 43,000 km2was burnt during the 25 day study period, with about 19 Mt of fine (grass and litter) fuels. This paper examines assumptions and errors associated with these estimates. It is estimated from uncalibrated fire mapping derived from AVHRR imagery that 417,500 km2of the northern Australian savanna was burnt in 1999, of which 136,405 km2, or 30%, occurred in the Northern Territory study region. Using generalized fuel accumulation equations, such biomass burning consumed an estimated 212.3 Mt of fine fuels, but no data are available for consumption of coarse fuels. This figure exceeds a recent estimate, based on fine fuels only, for the combined Australian savanna and temperate grassland biomass burning over the period 1990–1999 but is lower than past estimates derived from classification approaches. We conclude that (1) fire maps derived from coarse‐resolution optical imagery can be applied relatively reliably to estimate the extent of savanna fires, generally with 70–80% confidence using the approach adopted here, over the major burning period in northern Australia and (2) substantial further field assessment and associated modeling of fuel accumulation, especially of coarse fuels, is required.

Spatial distribution of carbon sources and sinks in Canada’s forests

Annual spatial distributions of carbon sources and sinks in Canada’s forests at 1 km resolution are computed for the period from 1901 to 1998 using ecosystem models that integrate remote sensing images, gridded climate, soils and forest inventory data. GIS-based fire scar maps for most regions of Canada are used to develop a remote sensing algorithm for mapping and dating forest burned areas in the 25 yr prior to 1998. These mapped and dated burned areas are used in combination with inventory data to produce a complete image of forest stand age in 1998. Empirical NPP age relationships were used to simulate the annual variations of forest growth and carbon balance in 1 km pixels, each treated as a homogeneous forest stand. Annual CO2 flux data from four sites were used for model validation. Averaged over the period 1990–1998, the carbon source and sink map for Canada’s forests show the following features: (i) large spatial variations corresponding to the patchiness of recent fire scars and productive forests and (ii) a general south-to-north gradient of decreasing carbon sink strength and increasing source strength. This gradient results mostly from differential effects of temperature increase on growing season length, nutrient mineralization and heterotrophic respiration at different latitudes as well as from uneven nitrogen deposition. The results from the present study are compared with those of two previous studies. The comparison suggests that the overall positive effects of non-disturbance factors (climate, CO2 and nitrogen) outweighed the effects of increased disturbances in the last two decades, making Canada’s forests a carbon sink in the 1980s and 1990s. Comparisons of the modeled results with tower-based eddy covariance measurements of net ecosystem exchange at four forest stands indicate that the sink values from the present study may be underestimated.

Patch-mosaic burning: a new paradigm for savanna fire management in protected areas?

The shift in ecological thinking, from equilibrium to non-equilibrium processes has been accompanied by a move to encourage heterogeneity rather than homogeneity in landscapes. Spatial and temporal heterogeneity is thought to be a major source of biotic diversity, and disturbances such as fire, producing heterogeneity are now recognised as being important. A patch-mosaic system of burning is based on the premise that fire pattern is a surrogate for diversity, and produces a range of patches in the landscape with unique patch characteristics and fire histories. A patch-mosaic system of burning is supported historically and empirically through field studies. However, there is a need for more research into the effects of various aspects of patch and fire variables on biotic diversity, especially in savannas where our understanding is particularly poor. Landscape-scale experiments, like those to be established in the Kruger National Park, South Africa are necessary to test different burning regimes. Challenges to patch- mosaic burning include determining the 'natural' range of variation for fire parameters, implementing random ignitions, and cost-effective fire scar mapping at the appropriate resolution. An adaptive management approach should be adopted to deal with the ignorance and uncertainties that characterise the management of savanna ecosystems. This should be applied with both modelling and monitoring as key elements in this process.