Biodiversity publications archive

Country in flames

Proceedings of the 1994 symposium on biodiversity and fire in North Australia - Biodiversity series, Paper no. 3
Deborah Bird Rose (editor)
Biodiversity Unit
Department of the Environment, Sport and Territories and the North Australia Research Unit, The Australian National University, 1995

A healthy savanna, endangered mammals and Aboriginal burning

RW Braithwaite
CSIRO, Division of Wildlife and Ecology, Darwin


Tropical savanna is the first child of the wet-dry tropics. The extreme alternation of wet and dry seasons produces the savanna vegetation (eg Frost et al. 1986, Mott et al. 1985). Other vegetation types are largely the result of extra water accumulation at particular places in the landscape, either as surface water at low points in the landscape or as groundwater where the geology and geomorphology allow water to return to the surface. In the wet tropics where rainfall is relatively evenly distributed over the year, rainforest is the dominant vegetation type.

In many parts of the tropical world, rainforest has been cleared and/or been changed by frequent burning. This may result in a derived savanna vegetation, one which will gradually return to rainforest if kept free of fire. It is from such situations that data have been collected from Africa and South America (eg Gillon 1983, Medina & Huber 1992). In contrast in Australia, the area of wet tropics is very small in comparison with the vast area of wet-dry tropics. Consequently the vast Australian savanna area is not a rainforest waiting to develop if only people would stop burning (eg Bowman et al. 1988). For example, the present area of monsoon forest (the local rainforest/closed forest type) in Kakadu National Park is only 0.5% of the northern Stages I and II. Studies of Russell-Smith (1985) suggest that in the wetter times 5000 years ago it might have been four times as widespread as at present still only 2% of the area.

A wet-dry tropical climate not only makes the survival of rainforest trees difficult but predisposes the vegetation to fire. An annual period of productivity during the wet season inevitably produces growth which dries out and becomes flammable during the dry season. The wet-dry tropics have abundant lightning the highest annual rates of anywhere in Australia (Prentice 1978). Naturally it co-occurs with the wet season, but is most likely to cause fires early in the wet season when the vegetative fuel is still dry and rain generally light and localised.

As I have pointed out in an earlier study (Braithwaite & Estbergs 1985), the pre-Aboriginal fire regime would have been different in timing from the Aboriginal one which moved the main time of burning to the early dry season rather than the early wet season. Although lightning strikes still occurred after Aboriginal people took control of the landscape, much less fuel is available if considerable burning has occurred earlier in the year through human action. Although the biological impacts of burning at different times are quite different (see below), the point here is that fire has dominated this landscape for millions of years and is not a post-Aboriginal phenomenon of only tens of thousands of years. The fires have, as a result of the wet-dry tropical climate, been annual and must have had a huge impact on the evolution of the North Australian fauna and flora. In other words, fire has been an important evolutionary force for millions of years.

Furthermore, through the consistent high frequency, it has been a rather different evolutionary force from fire in arid or temperate Australia. All other components of fire regime, particularly severity, stratum type, area of fire, and patchiness within the fire area, relate in a general sense to the fire frequency (see Table 1). It is this different evolutionary force which over many millennia has moulded very different life histories and behaviour in relation to fire in tropical plants and animals. Thus the relationship between fire regime and biodiversity in the tropics, at least for the dominant savanna vegetation type, is likely to be fundamentally different from that of temperate and arid climatic regions.

Table 1: A comparison of components of fire regimes in the arid/temperate and tropical climates
Attribute Temperate/arid Tropical
Frequency 4 400 yrs 1 2 yrs
Time of year more restricted less restricted
Severity to 100,000 kW/m to 20,000 kW/m
Stratum ground & canopy ground
Area larger smaller
Patchiness less more

What is savanna?

This vegetation type is the dominant type for the seasonally dry or wet-dry tropics. It covers about 10% of the earth's land surface. It varies greatly in tree and grass cover. Its structural and functional characteristics are primarily determined by plant-available moisture and plant-available nutrients. Fire and herbivory are seen as secondary determinants or modifying factors (Frost et al. 1986).

In Northern Australia, herbivory by native species is low, and in excess of 99% of it is by invertebrates (Braithwaite, unpublished). Cattle raising is only possible in the low fertility native pastures of north-western Australia with nutrient supplementation either directly by feeding supplements or pasture modification (eg Winter 1990). Herbivory contributes substantially to generating heterogeneity in Africa where it is substantial (eg Belsky 1986). In north-western Australia, it is fire which is the main generator of heterogeneity. The principal exception is along the seasonal creeklines where differential moisture availability creates greater heterogeneity than further upslope.

Other vegetation types are primarily the result of accumulation of water in low-lying parts of the landscape (eg wetlands), access by roots to groundwater (monsoon forest), or substrate composition (eg rocky escarpment plant communities). They are small in area and generally fragmented. Percentages of country burned each year are highest in the savanna with a range of 48 71% being measured in Kakadu in different years. The equivalent figures for wetland and rockland (escarpment) are 38 48% and 8 68% respectively (AN Andersen, pers comm)

Savanna in Kakadu is about 65% of the area. In most other areas, there are less rocky and wetland communities and an even greater percentage of savanna. These other vegetation types are islands in a sea of savanna. In landscape ecology terms, savanna is the matrix. It follows from this that management of this landscape is fundamentally management of the savanna.

As a result of this prevailing frequent fire regime and other derived or direct aspects of the environmental context, a range of special features have evolved in the biota of the savanna. Specifically, these include:

  1. Low flammability of canopy due to low volatility of essential oils;
  2. Non-accumulation of fuel beyond 2-3 years due to termites and high decomposition rates (GD Cook, pers comm);
  3. High vegetative regenerative capacity and little regeneration by seed germination after fire;
  4. Substantial deciduousness of trees, particularly with trees of pan tropical genera of the middle layer;
  5. Great variability between species in timing of flowering and fruiting (RJ Williams, pers comm);
  6. Substantial availability of fleshy fruit as well as woody fruit;
  7. High synchrony of seasonal leaf flushing between species within major strata;
  8. A mixture of fauna of both arid and wet tropical origins with a great seasonal range of breeding times between species of birds, lizards and mammals.

Major gradients in the savanna

The savanna vegetation is far from uniform. There are three major gradients in the savanna. Major attributes of these gradients are listed in Table 2.

Table 2. Correlates of the major gradients in the savanna
Gradient Scale (km) Increasing Decreasing
Coast-arid 103 temperature seasonality total rainfall
    inter-year rainfall variation rainfall seasonal
    reptile richness mean tree height
    mammal species loss projective foliage cover
    immense mammal richness plant richness
    latitude bird richness
      mammal richness
      frog richness
      termite richness
Lowland-rockland 102 elevation soil fertility
    endemicity productivity
      fire frequency
Seep-ridge 101 plant richness patchiness
    elevation mammal richness
      mammal abundance
      water availability

From the coast inland

This is the gradient from the wet end (c.1600 mm rainfall pa) to the dry end (450 mm pa) of the savanna landscape. In the Northern Territory, it occurs over a thousand kilometres distance and is the subject of special study under the name of North Australian Tropical Transect (NATT).

From the lowlands to the rocklands

This gradient extends from the wetlands associated with the river floodplains to the monsoon forests, paperbark and other 'margin' communities to the savanna and then to the sandstone escarpment or rockland communities. This gradient extends over about 10 50 km and is complicated and may be broken into wetland-margin, margin-savanna and savanna-rockland gradients, each with its own distinct set of correlates (Braithwaite 1990, Braithwaite et al. 1990). Major soil differences are seen between the fertile black cracking clays of the lowlands, the humic gleys (A type of grey soil prone to waterlogging) of the margin communities, the infertile lateritic soils of the savanna and the sandy skeletal soils of the rocklands.

Within the savanna, from seasonal creeklines to low

This is an altitudinal gradient of about 30 50 m over a distance of 0.5 1 km. The water of the seasonal stream is used by animals for drinking and enhances the period of productive growth during the dry season (Braithwaite 1990).

Finer-scale heterogeneity

Tributaries of the seasonal drainage lines of 5-10 km join in the usual dendritic fashion. These tributaries are <0.5 km and joined by other smaller tributaries and so on. Thus very minor topographic differences contribute to heterogeneity in the savanna.

Savanna plants and animals

At first sight the savannas look as though they are composed of relatively few plant species. The savannas, however, are rich in grasses, particularly annuals. There are many ephemeral herbs, which appear above ground for a brief time, usually in the wet season or early dry season. Further, many of the grasses and woody plants look very similar to others in their group. On a half square km piece of savanna (including a section of creekline), we obtained 70 species of woody plants, 55 species of grasses and sedges, and over 150 species of herbs (Braithwaite, unpublished). Relative to other vegetation types in the area, the savannas are rich in mammals, lizards and insects (Braithwaite 1990, Braithwaite et al. 1990).

Fire and habitat selection

In many parts of the world, typically, infrequent but intense fires kill many plants thereby creating opportunities for new species to colonise from seed already present or blown or carried onto site (Table 3). Thus fire may remove a mature forest and replace it with a newly planted young forest. The process of replacement of dominant species is debated (Gill et al. 1981), but it is clear that both the structure and composition continue to change through time over decades and sometimes over hundreds of years. The resources available to animal species changes considerably over time. Thus the habitat changes over the years and the abundance of different animal species changes also. In many cases the animal composition may be totally different at one end of the successional series versus the other. Therefore fire has a clear influence on habitat diversity and the range of species which can be carried in an area of bush. This is the usual model of fire-driven habitat selection seen in arid, mediterranean and temperate climatic regions and incidentally is the subject of most of the literature on animals and fire. Examples include mammals, birds and lizards (eg Gill et al. 1981).

Such a model is just not possible with frequent fires of low intensity. The highest intensity fire we have recorded in Kakadu is 18 000 kWm-1, less than a fifth of that of the fires of highest intensity in southern Australia. Most fires in Kakadu are between 2000-5000 kWm-1 (RJ Williams, pers comm).

Typically the changes in vegetation generated by fire in the tropical savannas are not dramatic and recovery is rapid. Within two weeks to a month, perennial grasses have usually resprouted, new leaves have appeared on the trees, new and prolific suckers have developed on shrubs and the underground lignotubers. This new vegetation is grown from reserves that exist in the plants and can be five times as rich in nutrients as the foliage it replaced. The removal of the old rank foliage also improves the accessibility of many resources to animals, including seeds and animal prey for predators and scavengers. Thus the fire itself and the short-term recovery phases following over the next two months are exploited by many animals (eg Gillon 1983, Braithwaite & Estbergs 1987). Many predators hunt around the fire itself, scavengers clean up any carcasses, granivores find seeds easily, herbivores feed on the new nutritious foliage, insectivores feed on herbivorous insects attracted to the new foliage and so on. Some species find the conditions created by the fires so favourable that they breed during the dry season (eg James & Shine 1985).

Recovery after fire in savannas is typically rapid, however, the nature of the impact and recovery varies with the co-related time of year and intensity of the fire. Thus for the rest of the year, the structure and resource availability of the habitat created by the fire and the short-term recovery of the vegetation remains until the next wet season, and this suits different species to differing degrees. For example, different lizard species survive and prosper to different degrees following fires at different times of year/intensities (Braithwaite 1987).

Similar results are being obtained with mammals in the Kapalga Fire and Water Experiment in Kakadu National Park (Andersen & Braithwaite 1992, Table 4). While there is preference for all experimental fire types (early dry season, late dry season, progressive fires through the dry season [simulation of Aboriginal burning] and no fires set [called natural]) by different species, the preference with some species changes from year to year. This is thought to relate to the strong effect different yearly rainfall patterns have on food resources. In effect, we are examining the interaction between years and fire type or regime.

The other important outcome is the large number of species and indeed total numbers of mammal individuals selecting the natural (no fire) treatment sites. Interestingly, there was no significant influence of any of the four fire regimes on mammal species richness after two years (Braithwaite, in press b).

In addition to the tropical models of fire and short-term changes, and the habitat changes due to time of year/fire intensity (Table 3), resource access opportunities may result from the removal of dominant species of animal. For example, the plunge in population numbers of Rattus tunneyi after a very severe fire in 1986 allowed a dramatic period of high survival for Pseudomys delicatulus, a species which appears to breed continuously as if waiting to quickly exploit such an opportunity (Braithwaite & Brady 1993). It is also important to recognise that this phenomenon has not emerged during the poorer times of the Kapalga Fire and Water Experiment, again highlighting the importance of the rainfall patterns of the years concerned.

In order to maintain the current intact biota of Northern Australia (eg Woinarski & Braithwaite 1990), it is necessary to maintain the heterogeneity of the savanna. The mammals are the group which appears to be most vulnerable to extinction.

The richness of mammals is positively related to habitat diversity at both canopy and ground levels (Braithwaite, in press a). It is clear that good habitat diversity is essential to their persistence and probably the persistence of most groups and thus biodiversity.

Table 3: Models of habitat partitioning involving fire
Animal selection of habitat created by: Temporal scale (days) Spatial scale (m)
Successional seres 103 104 103 104
Fire and short-term recovery 10-2 101 103 105
Fire intensity and time of year of burn 102 101 102
Disturbance of animal populations 102 102
Table 4: Responses of mammals to four regimes in the Kapalga Fire Experiment
Fire treatment Year 1 Year 2
Natural Dasyurus hallucatus Dasyurus hallucatus
  Rattus tunneyi Isoodon macrourus
    Antechinus bellus
    Melomys burtoni
    Trichosurus vulpecula
Early Trichosurus vulpecula Rattus tunneyi
Late Antechinus bellus  
Progressive Melomys burtoni  
No differentiation Isoodon macrourus Rattus colletti
  Rattus colletti  

Aboriginal burning

Burning by traditional Aboriginal people was and is done for a wide variety of reasons (Braithwaite 1991b, see below). Almost all Aboriginal burning was incidental to other activities. It permeated all other activities, it was part of every day. It is integral to traditional life.

Purposes of Aboriginal burning

Future management

Contemporary managers have a wealth of issues to consider (see below) and satisfaction of some may conflict with the satisfaction of others. In other words, compromises will be necessary. The exact emphasis given to the various issues will change with region and types of economic activity conducted there, and it is right that it does so. For example, the rural sector near a city will rate protection of people and property and smoke abatement as a more pressing issue than would a national park where nature conservation issues will be given greater prominence.

Nonetheless, in both cases a strategic approach is needed. The aim should never be to burn all of a region. This would be unnecessary waste. Control
of fire in the landscape should always be the aim. With Aboriginal management, control was the outcome, whether it was thought about in those terms or not. Further, that some areas remain unburnt is an important conservation goal (see above). Maximising habitat diversity is a reasonable overall goal for fire management for conservation. It is also in effect the outcome of traditional Aboriginal management. It is in fact to be expected that a long-standing management regime is likely to be what the fauna and flora is adapted to.

Fire management issues


Heterogeneity is a defining characteristic of savanna and whether viewed from the perspective of fire control or maintenance of biodiversity, its maintenance must remain the keystone of fire management in any landscape dominated by savanna.


I thank the numerous people who have contributed to my understanding of the issues and also Alan Andersen, Garry Cook and Dick Williams for allowing me to quote their unpublished work. This is CSIRO Tropical Ecosystems Publication Number 830.


Andersen AN & Braithwaite RW, 1992. Burning for the conservation of the Top End's savannas, in I Moffatt & A Webb (eds), Conservation and Development Issues in Northern Australia, North Australia Research Unit, ANU, Darwin, 117 122.

Belsky JA, 1986. Population and community processes in a mosaic grassland in the Serengetti, Tanzania, Journal of Ecology 74, 841 56.

Bowman DMJS, Wilson BA & Hooper RJ, 1988. Response of Eucalyptus forest and woodland to four fire regimes at Munmarlary, Northern Territory, Australia, Journal of Ecology 76, 215 32.

Braithwaite RW, 1987. Effects of fire regimes on lizards in the wet-dry tropics of Australia, Journal of Tropical Ecology 3, 265 75.

Braithwaite RW, 1990. The implications of Australia's unique biota for ecological processes, Journal of Biogeography 17, 347 54.

Braithwaite RW, 1991a. Aboriginal fire regimes of monsoonal Australia in the nineteenth century, Search 22, 247 249.

Braithwaite RW, 1991b. The human face of fire, unpublished paper.

Braithwaite RW, in press a. On the importance of habitat heterogeneity in tropical savanna and its maintenance by anthropogenic fires, in L McCaw, W Burrows & GR Friend (eds), Landscape Fires 93, Res. Bull, CALM, Perth.

Braithwaite RW, in press b. Biodiversity and fire in the savanna landscape, in OT Solbrig, E Medina & JF Silva (eds), Biodiversity and Savanna Processes: A Global Perspective, Springer-Verlag, Berlin.

Braithwaite RW & Brady P, 1993. The delicate mouse: a continuous breeder waiting for the good times, Australian Mammalogy 16, 93 6.

Braithwaite RW & Estbergs JA, 1985. Fire patterns and woody vegetation trends in the Alligator Rivers region of northern Australia, in JC Tothill & JJ Mott, Ecology and Management of the World's Savannas, Australian Academy of Science, Canberra, 359 364.

Braithwaite RW & Estbergs JA, 1987. Firebirds of the Top End, Australian Natural History 22, 298 302.

Braithwaite RW, Friend GR & Wombey JC, 1990. Reptiles and amphibians, in MG Ridpath, C Haynes, MAJ Williams (eds), Monsoonal Australia: Landscape, Ecology and Man, AA Balkema, Rotterdam, 109 124.

Frost P, Medina E, Menaut JC, Solbrig O, Swift M & Walker B (eds), 1986. Responses of Savannas to Stress and Disturbance: A Proposal for a Collaborative Programme of Research, Biology International, Special Issue 10.

Gill AM, Groves RH & Noble IR, 1981. Fire and the Australian Biota, Australian Academy of Science, Canberra.

Gillon C, 1983. The fire problem in tropical savannas, in F Bourliere (ed), Tropical Savannas, Elsevier, Amsterdam, 617 41.

James C & Shine R, 1985. The seasonal timing of reproduction: a tropical-temperate comparison in Australian lizards, Oecologia (Berlin) 6, 464 74.

Medina E & Huber O, 1992. The role of biodiversity in the functioning of savanna ecosystems, in OT Solbrig, HM van Emden & PGWJ van Oordt (eds), Biodiversity and Global Change, Monograph. No. 8, International Union of Biological Sciences, Paris, 139 58.

Mott JJ, Williams J, Andrew MH & Gillison AN, 1985. Australia savanna ecosystems, in JC Tothill & JJ Mott (eds), Ecology and Management of the World's Savanna, Australian Academy of Science, Canberra, 56 82.

Prentice SA, 1978. Lightning risk, Search 9, 222 9.

Russell-Smith J, 1985. A record of change: studies of Holocene vegetation history in the South Alligator region, Northern Territory, in MG Ridpath & LK Corbett (eds), Ecology of the Wet-Dry Tropics, Proceedings of the Ecological Society of Australia, Vol. 13, Ecological Society of Australia, Canberra, 131 202.

Winter W, 1990. Australia's northern savannas: a time for change in management philosophy, Journal of Biogeography 17, 525 9.

Woinarski JCZ & Braithwaite RW, 1990. Conservation foci for Australian birds and mammals, Search 21, 65 8.