Extensive fires in Australia’s northern spinifex deserts – investigating the 2023 ‘Black Spring’ and the influence of indigenous fire management

Introduction

Fire is a natural ecological disturbance factor in many ecosystems worldwide (He et al. 2019). Nevertheless, marked disruptions to fire regimes can be problematic even in fire-adapted ecosystems, if the new fire regime has characteristics that exceed the bounds within which species and ecosystems evolved (Whelan 1995). Disruptions in fire frequency, severity and season caused by habitat loss, fragmentation and degradation, changes to anthropogenic ignition patterns, and now changes in climate, are contributing to species declines (Kearney et al. 2019; Bowman et al. 2020b; Legge et al. 2023) and ecosystem collapse (Bergstrom et al. 2021). Fire regime changes include an increasing incidence of extremely extensive, severe fires (Linley et al. 2022).

Despite the fact that fire is especially prevalent in many Australian ecosystems (Murphy et al. 2019), large areas of habitat are being deleteriously affected by extensive, severe fires. A textbook case of such fires in Australia occurred during the ‘Black Summer’ of 2019–2020, when over 100,000 km², or almost 20% of the eastern and southern temperate and subtropical forests burnt (Legge et al. 2023). Over the previous 18 years, the average annual fire extent in these forests had been just 2% (Bowman et al. 2020a). These fires caused substantial conservation detriment to many species (Gallagher et al. 2021; Marsh et al. 2021; Legge et al. 2022) and elicited an outpouring of concern from the general public (van Eeden et al. 2020; Rumpff et al. 2023). The proximity of the Black Summer fires to the most heavily populated parts of Australia raised the profile of their impacts on people as well as natural environments, leading to international media coverage. The Black Summer fires herald an uncharted future of increasing frequency of intense and extensive fires in Australian forests (Nolan et al. 2020; Canadell et al. 2021). However, fire regimes are also changing in other biomes that are further from the public eye and therefore receive less public and scientific attention.

Most fire in Australia occurs in the tropical savannas and desert biomes (Murphy et al. 2013). This is true even for the 2019–2020 season, when 200,000 km² (mostly in the tropical savannas, but also in woodlands and desert) burnt in addition to the 100,000 km² of the temperate and subtropical forests (Bowman et al. 2020a). The potential for extensive fire in savannas and deserts depends largely on ground fuel availability, which in turn depends largely on accumulated rainfall since the last fire (Murphy et al. 2013; Ruscalleda-Alvarez et al. 2023). In the tropical savannas of northern Australia, the annual monsoon reliably causes large rainfall totals during each year’s wet season (600 to >1000 mm), resulting in grass growth and thus fuel loads that can carry fires every 1–3 years during the intervening dry seasons. Moving south through the deserts, the influence of the monsoon weakens and the average annual rainfall decreases (from 600 to <200 mm pa), becomes less seasonal, and as a result the frequency of fire diminishes (Murphy et al. 2013; van Etten and Burrows 2018). While rainfall in most Australian deserts is low on average, they do experience irregular episodes of widespread, prolonged rainfall, linked to global climate cycles (Verhoeven et al. 2020). These rainfall events drive a pulse of growth in the grass layer that readily carries fire at the end of the next dry season, when dry lightning strikes are common. Desert fire regimes therefore includes periods where these prolonged rainfall events provide the pre-requisite conditions for extensive fires (Edwards et al. 2008; Turner et al. 2011).

The fire regimes associated with rainfall patterns in the tropical savannas and desert were modified for many thousands of years by the traditional burning practices of Indigenous people, who used fire intensively for purposes such as signalling, ceremony, clearing access, hunting and promoting food and medicinal plants (Yibarbuk et al. 2001; Pike 2008; Vaarzon-Morel and Gabrys 2009). In the tropical savannas, progressive burning throughout the year tended to stop the spread of fire later in the dry season. In the deserts, burning practices created fine-scale patchworks of vegetation in different stages of regrowth post-fire. There was probably geographic and temporal variation in where traditional fire practices occurred (Kimber and Friedel 2015), and their effect on the fire regime (Wright et al. 2021). Nevertheless, the outcomes of traditional fire practice on desert vegetation can clearly be seen in aerial photos of the western and northwestern deserts from the 1940s and 1950s, indicating that Aboriginal inhabitants still living a traditional lifestyle actively burned large areas (Burrows et al. 2006; Bliege Bird et al. 2020; Blackwood et al. 2022). Traditional fire practice continues around some small remote desert communities, and fire patterns at these sites resemble those evident from the historical photos (Bliege Bird et al. 2012, 2018; Paltridge et al. 2025).

As Indigenous people across northern and central Australia moved off their traditional lands into missions, towns, settlements and pastoral stations between ~1870 and 1980, traditional fire management practices waned. Fires became more extensive, frequent and severe (Russell-Smith et al. 2009; Legge et al. 2011; Paltridge et al. 2025), contributing to biodiversity loss (Woinarski et al. 2011; Legge et al. 2023). The deterioration of fire regimes has been arrested and reversed in many parts of northern Australia during the last 15 years, as large-scale fire management programs have emerged, particularly on Indigenous lands (Edwards et al. 2021). These programs seek to achieve interlinked social, cultural and conservation objectives, and are facilitated by greenhouse gas reduction offsets (Ansell and Evans 2019; Evans and Russell-Smith 2019). They have caused a marked reduction in the extent of uncontrolled fire and encouraged more pyro-diverse landscapes (Perry et al. 2016; Wysong et al. 2022; Vigilante et al. 2024).

Large-scale fire management programs have developed more slowly in the desert, partly because there is currently no access to emissions reduction payments (Wysong et al. 2022). Despite this disadvantage, large-scale fire management has been expanding from some nuclei sites (Ruscalleda-Alvarez et al. 2023). The management is driven by biocultural priorities but is carried out by Indigenous rangers, blending traditional understandings of landscape/fire interactions with contemporary methods including aerial incendiaries dropped from aircraft and satellite technology. In the northern deserts, rainfall is higher and more strongly seasonal because of the influence of the monsoon, and prescribed burning typically occurs in March/April, soon after the summer wet season. Rainfall is lower (on average) and less seasonal in more southern deserts and prescribed burning typically occurs during the cool winter months (June–August). This program of fire regime change includes fine-scale ground burning around sensitive areas, including culturally significant sites and habitats of threatened species such as bilbies Macrotis lagotis, night parrots Pezoporus occidentalis and great desert skinks Liopholis kintorei (Ngururrpa Rangers et al. 2024; Paltridge et al. 2025; Southwell et al. 2025). However, the essential feature of the program is to burn Country (a term signifying all aspects of the land and its living things, and the knowledge, cultural practices and responsibilities related to them) in a controlled way, to break up large areas of vegetation that are at the same stage of post-fire recovery. This ensures that wildfires cannot spread evenly across these areas when temperatures and winds rise later in the dry, hot season (September–November).

The outcomes of this large-scale fire management in Australian deserts that applies traditional knowledge but uses contemporary methods are still poorly reported, because most fire programs are recent relative to average fire return intervals (6–30 years). Evaluations of some of the first programs to start (around 2010) do suggest a shift in fire patterns (Ruscalleda-Alvarez et al. 2023). However, many programs started more recently; for example, in the Great Sandy Desert, ranger groups began large-scale burning around 2019 (e.g. Legge et al. 2024; Ngururrpa Rangers et al. 2024).

Widespread and prolonged rain occurred in Australia’s northern deserts and some parts of the low rainfall tropical savannas during 2022–2023 (BoM 2024). The rain caused biomass accumulation that resulted in very extensive desert fires in the following dry season (June–November 2023). Despite their size, and their potential impacts on cultural and biodiversity values, they were not acknowledged as a significant national event. In this paper, we describe the extent of the 2023 northern desert fires, and their relationship to lightning strikes. We also examine fire spread patterns to explore whether the emerging, large-scale fire management programs being delivered by ranger groups in one part of these northern deserts influenced the fires.

Methods

To explore whether the 2023 fires in the northern desert were unusual, we used satellite imagery to examine the fire patterns in three large desert bioregions through a 23-year time series, and to compare the 2023 fires in the northern deserts to the fire patterns in the adjacent topical savannas in the same year. We sourced information on lightning ground strikes to explore the likelihood that extensive fires will occur, when fuel conditions allow. Finally, using a case study approach, we examine the fire occurrence and spread of the two most extensive fires, which occurred in regions with contrasting levels of prescribed burning by Indigenous rangers.

Study area

This study focuses on the Great Sandy Desert, Tanami, and Davenport Murchison Ranges bioregions, which together cover 712,000 km², and encompass most of northern spinifex deserts (Fig. 1a). Their ecosystems are dominated by spinifex Triodia spp. hummock grasslands with areas of shrubland (Acacia spp., Grevillea spp., Hakea spp.) and scattered trees (Corymbia spp., Eucalyptus spp., Owenia reticulata, Allocasuarina decaisneana), occasionally intersected by paleodrainage lines and dry salt lakes (Thackway and Cresswell 1997). The climate of these bioregions is hot and arid. The average annual rainfall grades from 600 mm in the north, to 200 mm in the south. Most rain (>90%) falls in the wet season (December–March). In 2023, these bioregions experienced ‘very much above average’ rainfall deciles in the period mid-2022 to mid-2023 (BoM 2024) (Fig. 1b).

Fig. 1.

(a) The study area showing the burnt area extent in 2023 in grey; the tropical savanna bioregions (green) (and in which the high and low rainfall zones are either side of the >1000 m isohyet (blue)); and the three northern spinifex desert bioregions (red), Great Sandy Desert (1), Tanami (2) and Davenport Murchison Ranges (3). (b) The blue shading shows the ‘very much above average’ rainfall deciles in the period mid-2022 to mid-2023 (BoM 2024). The grey shading underlaid shows areas burnt in 2023. (c) The two case study fires (in black) within the three northern spinifex desert bioregions (in grey outline). Regional population centres are also shown.

To provide context, we also assess the 2023 fire extent of the tropical savanna bioregions to the north, considering both the high rainfall (>1000 mm) and lower rainfall zones (>600, <1000 mm) (Fig. 1a). The tropical savannas are mostly comprised of tall open forests to open woodlands, with smaller areas of grassland. Within this matrix, wetland, riparian, rainforest pocket and heathland ecosystems occur in smaller areas.

Results

Fire extent

Fires burnt 294,393 km² (41.2%) of the Great Sandy Desert, Tanami and Davenport Murchison Ranges bioregions in 2023, or roughly three times the fire extent of the Black Summer fires in eastern and southern Australia (Legge et al. 2023). Over the 2000–2023 time series, the average annual fire extent was 15%. The 2023 fire extent was the second largest area burnt through the 24-year mapped period, surpassed only by the 2011 fire extent (Fig. 2). The three desert bioregions show pulses of high fire extents in 2000–2002, 2011–2012, and 2023, each following widespread, above-average rainfall caused by La Niña weather conditions (Fig. 2).

Fig. 2.

The proportion burnt (orange bars) and the rainfall (blue) for each year, between 2000 and 2023, for the three northern spinifex desert bioregions (Great Sandy Desert (GSD), Tanami (TAN) and Davenport Murchison Ranges (DMR) separately and together, as well as the tropical savannas (high and low rainfall zones).

In the low-rainfall (600–1000 mm) tropical savannas, the average annual fire extent over the 2000–2023 time series was 24%. Similarly to the desert bioregions, fire extent in the tropical savannas also pulsed upwards in 2000–2001, 2011–2012, but less dramatically in 2023 (31.3% burned), despite high preceding rainfall in some areas. In the high-rainfall savannas the fire extents have been more even across years; the 2023 fire extent (33.8%) was the seconded lowest in the time series after 2017 (Fig. 2).

Fire event timing

Across all three northern spinifex desert bioregions, fires occurred sporadically between March–August, covering small areas (Fig. 3). Fire activity began to increase from early September, noticeable first in the Davenport Murchison Ranges. From mid-late September through to November, fire extent across the desert bioregions increased markedly. Over 75% (218,967 km²) of the year’s fire extent in the three desert bioregions burnt in 60 days before 18 November 2023. Periods of higher winds and temperatures led to the most rapid fire spread. For example, in early November, 27,732 km² (23% of the total) of the Great Sandy Desert burnt in just 5 days. In the Tanami, 29,928 km² (20% of the total) burnt in 5 days from the 22 October (Fig. 3).

Fig. 3.

Burnt area extent (blue lines) and ground lightning strikes (orange bars) by day of year (Julian date) for the Great Sandy Desert (GSD), Tanami (TAN) and Davenport Murchison Ranges (DMR) bioregions separately, then together.

Ground strike lightning records

The Vaisala lightning network recorded over 1,138,000 ground strikes over the three northern spinifex desert bioregions between September–November. Of these, 25,810 strikes occurred across the Great Sandy Desert and Tanami bioregions in October, and over 1 million in November (Figs 3, 4). This overlapped with the period of the greatest fire extent in both bioregions. In contrast, Davenport Murchison Ranges experienced very little lightening until November, after most of the burning had already occurred, when 30,943 ground strikes were recorded.

Fig. 4.

The density of lightning ground strikes during October and November across the three northern spinifex desert bioregions. September data have been omitted because there were few ground strikes in this month.

Case studies

Two of the largest single fire events in the three northern spinifex desert bioregions in 2023 are investigated as case studies (Fig. 1c). The area east of Tennant Creek in the Davenport Murchison Ranges has experienced little prescribed burning in previous years (R. Fisher, pers. comm.). In contrast, much of the Great Sandy Desert bioregion has experienced large-scale prescribed burning since 2019. For example, in 2022–2023, Indigenous rangers dropped over 115,000 incendiaries over this area (H. Cliff, pers. comm.) (Fig. 5).

Fig. 5.

Outcomes of prescribed burning (in orange and red) by the Nyangumarta, Karajarri, Nyikina Mangala and Ngurrara Rangers in the Great Sandy Desert (GSD), over 2021–2023. Grey lines are the bioregion boundaries. The 2023 fires are shown in grey, with the case study fire in darker grey.

The fire in the Great Sandy Desert (one ignition) burnt over 41,800 km² (10.5% of the bioregion, and 35% of the overall 2023 fire extent in the bioregion) (Fig. 6a). Starting in the northeast, near the Canning Stock Route, this fire burnt for >50 days from early September to late October, travelling over 600 km (in a straight line) before meeting other fires near the coast, south of Broome. Fire spread mostly occurred over areas that were last burnt in 2017 or before; conversely, fire spread was usually prevented in areas burnt between 2021 and earlier in 2023 (Fig. 6b).

Fig. 6.

Final extent and spread of fires in the (a) Great Sandy Desert and (c) Davenport Murchison Ranges case studies. The year of the previous fire (0 (burned earlier in 2023) to 11 years prior) in (b) Great Sandy Desert and (d) Davenport Murchison Ranges. The outline of the fires is shown by a black line; a 10 km buffer has been added around the perimeters so that the year of previous burns where the 2023 fires stopped is visible.

The fire (two ignitions) in the Davenport Murchison Ranges bioregion burnt 25,000 km² (43% of the bioregion, and 82% of the overall 2023 fire extent in the bioregion) (Fig. 6c). This fire burnt for >60 days from early August to mid October. Whilst starting in, and mostly affecting, the Davenport Murchison Ranges bioregion, the fire also burnt >200 km east from its starting point into the eastern Tanami bioregion. Again, fire spread occurred in older time-since-fire areas (2017 and older) but appeared to be limited in the few areas burnt in 2021 and early 2023 (Fig. 6d).

Overall, most of the area burnt in both fires had been burnt more than 4 years before (Fig. 7). In each fire, the buffer around the fire boundary contained vegetation that burnt relatively more recently (Fig. 7). However, the fire spread in these two cases studies differed markedly, in that the Great Sandy Desert fire was continually blocked and re-routed by patches of vegetation that had been burnt in earlier fires between 2017 and early 2023 (Fig. 6b). This ‘stuttering’ spread pattern was not evident in the Davenport Murchison Ranges bioregion, where patches burned between 2017 and early 2023 were rare (Fig. 6d).

Fig. 7.

For each of the (a) Great Sandy Desert (GSD) and (b) Davenport Murchison Ranges (DMR) bioregions, the graphs show the proportions of the 2023 fire extent that covered areas that had last burnt 0–1, 2–4, and >4 years before the 2023 fire. The orange bars show the values from within the fire extent boundary; the grey bars show the values including the 10 km buffer around the fire extent boundary.

Discussion

The collation of annual fire extents across the three bioregions that cover the northern spinifex deserts show that extensive fire in the late dry season (September–November) made 2023 one of the biggest fire years in the northern deserts since 2000 (Fig. 2). Over 40% of the three bioregions burned, well above the annual average of 15%. The total area burned exceeded 294,000 km², which is triple the area burned (100,000 km²) in the eastern and southern forests during the Black Summer fires (Legge et al. 2022). The desert fires moved extremely quickly at times, for example covering over 25,000 km² in 5 days, on two occasions. Despite their cumulative size, the 2023 desert fires barely registered in the national media. The limited media that did occur focussed on the threat of fire to the town of Tennant Creek in the Northern Territory (e.g. Dick 2023; Spina-Matthews 2023) but most Australians are likely unaware of the existence or scale of these fires.

Fire risk in spinifex deserts increases substantially after periods of widespread, prolonged rainfall (Edwards et al. 2008; Turner et al. 2011; Ruscalleda-Alvarez et al. 2023). These rainfall conditions happen episodically, usually every 5–10 years (van Etten and Burrows 2018). In the year leading up to the 2023 fires, the northern spinifex deserts experienced ‘very much above average’ rainfall (Fig. 1b) which caused a rapid increase in spinifex biomass and greater fuel connectivity. Local land managers and regional fire authorities were aware of the impact of this rain on spinifex biomass and thus fire risk, with a senior fire agency spokesperson warning of impending, extensive fires in the Northern Territory (Hislop 2023).

Our analysis of lightning data shows that, given the sheer volume and density of ground strikes, ignitions are inevitable; therefore, when sufficient spinifex biomass and fuel connectivity exists, extensive fires are to be expected. The 2023 desert fires stimulated some discussion in the local media of the role of arson in starting the fires (Ellis 2023). Reducing human ignitions in the late dry season is important and should be approached through community engagement and education. However, even without arson there are multiple and widespread ignitions from lightning strikes, and once started, continuous high fuel loads mean that fires will spread extensively.

Despite the repeated and predictable confluence of rainfall, high spinifex biomass and lightning, we suggest that extensive desert fires are not inevitable. The case study comparison of fire occurrence and spread in the Great Sandy Desert versus the Davenport Murchison Ranges bioregions demonstrates the influence that prescribed burning can have on fire outcomes. Since 2019, Indigenous ranger teams in the Great Sandy Desert have conducted large-scale prescribed burning annually, at the end of the wet season (March–May). These intentional burns affected the progression of wildfire in late 2023, when compared to the wildfire in the Davenport Murchison Ranges (Fig. 6), where prescribed burning in previous years had been much more limited (Fig. 3). These differences are not caused by differences in topography or vegetation mosaics – one of the characteristics of the Great Sandy desert bioregion is the homogeneity of its spinifex sand dunes and plains (Thackway and Cresswell 1997).

In the northern savannas, annual, large-scale prescribed burning programs over the past 10–20 years have transformed fire patterns by reducing fire size, frequency and severity of fire, and smoothing out fluctuations in fire extents between years, including in 2023 (Fig. 2, Ansell and Evans 2019; Evans and Russell-Smith 2019; Wysong et al. 2022; Vigilante et al. 2024). Achieving this result in deserts, which has lower and more variable rainfall, is challenging partly because the level of prescribed burning effort required will vary substantially between years, making work planning difficult. In particular, the prescribed burning effort needed after prolonged, widespread rainfall across the huge areas involved, is daunting. Whereas fire managers might burn about 5% of their management area in a year with average rainfall, after high rainfall, they may need to increase their effort 5–10 fold over a short period with attendant logistic hurdles and costs. Finally, road access for aviation fuel drops and crew exchanges is extremely constrained in the desert. Nevertheless, after just a few years of work, the rangers of the Great Sandy Desert have made a measurable difference to wildfire spread. With ongoing refinement of their prescribed burning delivery, they are likely to increasingly modify overall fire patterns, including after above-average rainfall years. For example, the prevailing winds in the late dry season are southeasterly, as shown by the direction of fastest fire spread in Fig. 6a. Strategic fire breaks, aligned north-south, may help limit the spread of future wildfires.

For millenia, Indigenous traditional fire practices markedly influenced the desert fire regime (Burrows et al. 2006; Bliege Bird et al. 2020; Blackwood et al. 2022). Changes to fire regimes occurred during the 20th century, as people moved off Country and Indigenous burning practices were interrupted. The loss of these fire practices may have contributed to biodiversity losses in the desert (Legge et al. 2023). For Indigenous custodians, restoring intentional burning is fundamental for improving spiritual and cultural, as well as environmental, health (Vaarzon-Morel and Gabrys 2009; Robinson et al. 2021, 2023). Large-scale fire management is likely to become increasingly challenging: climate change projections for the northern deserts suggest higher and more variable rainfall (CSIRO and Bureau of Meteorology 2022), which may increase the frequency of conditions that trigger extensive fire. However, the case study we present here suggests that supporting Indigenous rangers to restore fire practices using contemporary methods could have a substantial effect on desert fire patterns and may limit the extent and damage caused by fires in the northern spinifex desert.