Results from CSIRO’S Kapalga fire
experiments raise some intriguing questions, writes Greg
Miles, a ranger from Kakadu Park with more than 20 years’
experience. He says apparent contradictions can be explained when
seen from a perspective of a landscape under invasion by
Role of spear grass in the
system | Spear grass on the march |
Wet-season burning | Kentia
palms protected | Conclusion
| Kapalga Fire
Wet-season burning in Kakadu
Photo: Greg Miles
In March 2000, a workshop was held summarising the vast amount
of work on fire research undertaken at CSIRO’s Kapalga
Research Station (see box on this page). Some valuable detail was
revealed that placed more pieces into an otherwise sparsely filled
jigsaw. However, many of the speakers detailed ongoing ecological
processes in the Kapalga experiment that are apparently
inexplicable: the eucalypt E. miniata should be either
extinct or vary rare—instead it is neither.
The results showed that young and old eucalypts are most at risk
of being killed by hot fires; E. miniata reproduces only by
seed and most seedlings are killed by fire be it early or late dry
season. Frequent fire has a negative affect on seed production.
There was also approximately a 5 per cent loss of nitrogen each
year from the biota under an annual fire regime used at
When examined together, and knowing that Kapalga has a history
of hot and frequent fires, this information tell us that E.
miniata (at least) should be either extinct or rare in the
study area. This coupled with the finding of nutrient loss tells us
that the Kapalga woodlands should rapidly be converting to
grasslands or at least host only crippled and stunted woody
vegetation. Why is this not so? To my mind, this is a great
To resolve this and other mysteries posed by the experiment I
propose to look at the results from another perspective, in which
Kapalga is seen as part of a landscape being transformed by spear
From an ecological point of view, local native spear grasses (
Sorghum spp.) could be considered indigenous weeds in the
One only need observe their distribution and density in relation
to other grass species to get an inkling of this. The grass
occupies an extraordinarily wide range of habitats from high,
lateritic woodlands, seasonally inundated flats through to
sandstone sand-flats and scree slopes. Its hyper-abundance gives
the appearance of a plant in disharmony with local floristics.
Fire in the woodlands of Kakadu is largely about spear grass. It
constitutes around 40 to 50 per cent of the annual fuel load and
the majority of the free-standing fuel, so without spear grass,
fires would be much less intense. It is not uncommon to see lowland
forest standing in spear grass suffer 100 per cent scorch as early
as late April.
Leaf litter is the second major contributor to the annual fuel
budget. Leaf litter fires (without spear grass) tend to have a low
scorch height and intensity, although in extreme weather
conditions, a heavy accumulation of litter can produce very hot
fires 1 . Despite its major influence on annual fire,
little is known about spear grass in terms of changes in its
distribution and density over time.
The role of native sorghum in Top End fire regimes needs more
Some authors have alluded to an expansion in the distribution of
spear grass over the past 70 to 90 years. They attribute this to
disruption of traditional burning caused by the depopulation of the
region by the 1890s and that without a resident and stable
Aboriginal population, traditional burning largely came to an end.
In its place were many decades of hot, late-season fires which were
unchecked by traditional early burning 2 . There is
evidence to suggest that this same scenario is now occurring in
central Arnhem Land and the Kimberley 3 .
Anecdotal evidence from resident old timers also suggests that
spear grass was less common 40 years or so ago than it is today
This apparent invasion of spear grass could be having profound
and disturbing consequences. It may be promoting more ferocious and
frequent fires that have been implicated in the thinning of the
woodlands canopy (a finding reinforced by the Kapalga study). At
the same time, this new fire regime may be promoting yet more spear
In many parts of Australia, trees will invade grasslands if
fire, especially traditional fire, is not maintained 5 .
For most of the 20th century, however, the opposite was happening
in the Kakadu region. That is, because of hot and frequent woodland
fires, the tree-scape thinned and monsoon forests shrunk,
apparently in favour of habitats dominated by spear grass.
I believe that in many parts of the Top End not subject to
robust fire-management regimes, this negative scenario has
significantly worsened since the mid-1980s. At this time the
buffalo population suddenly and spectacularly crashed as a result
of the National Brucellosis and Tuberculosis Eradication Campaign.
Spear grass, which was probably thinned due to trampling by buffalo
in the wet season, was apparently released from the constraints of
bovine activity and increased in density throughout the region.
Thus, it is my view that from the mid 1980s, dry-season fires
became hotter and possibly more frequent with a resulting impact on
fire-sensitive ecological communities (such as lowland monsoon
forests). However, this increase in fire frequency and, more
particularly, temperature, has been subtle and not
recognised by the casual observer.
The problem is clear enough and most people agree that fire in
Top End spear grass landscapes occurs too frequently. A fire regime
that burns the same areas one year out of two or three, curtails
recruitment of many plant species and, on balance, has harmful
ecological consequences. While a rise in spear grass density and
distribution seems to be a factor here, it is generally agreed that
this situation is also due to the high frequency of fires lit by
land managers, and by careless tourists or arsonists.
What is difficult is finding agreement upon a management regime
that will reduce this problem. It is clear to most Top End land
managers that fire is an annual inevitability and they must plan
for it. Unfortunately this backs managers into the ‘I will
burn it before someone else does’ mentality. Right now this
is, sadly, probably the best policy, even though it forces managers
to burn a lot of country every year. As a result, some places get
burnt every year. To turn this negative cycle around only one
method seems to hold any promise: wet-season burning (WSB).
For the uninitiated, wet-season burning is used during December
or January (in Kakadu) to clear a given area of spear grass. A fire
at this time will kill newly germinated grass seedlings growing
among the stalks of last season’s parent grass. As sorghum is
an annual, almost all seed in the soil will germinate in the early
wet season. Following the fire, there will be little or no seed
remaining and the affected area will be largely free of spear grass
(and therefore fire) until the grass population re-asserts itself.
This may take three to seven years. In the interim, any fire will
effectively be a leaf litter fire.
WSB involves risk. To be effective, spear grass must be retained
over a full dry season in order to supply enough fuel to carry fire
in December/January. The risk occurs during the late dry season
when areas, which have been saved for WSB, are ignited prematurely
by lightning or (more commonly) arsonists.
Young Kentia Palms at the West Alligator River
in Kakadu are protected from fire by a wet-season burn
Photo: Greg Miles
Kentia palms protected
An excellent case study of WSB in Kakadu was carried out near
the West Alligator River crossing on the Arnhem Highway adjacent to
Kapalga. Here, Top End endemic Gronophyllum ramsayi palms
were roasted every year by hot spear grass fires. In 1993 Kakadu
rangers subjected the area to a wet-season burn—a well-timed
fire that rendered annual spear grass almost extinct locally.
Without any further fire suppression the area remained untouched by
wildfire until July 2000. Precisely what is called for—and
all from one strategic wet-season burn.
One study on the effect of WSB on woodland floristics concluded
that any benefit from WSB would be relatively short lived
6 . But it has taken many years for rangers to refine
WSB methods to achieve the best result. It is probable that the
one-hectare area used by the floristics study was not large enough
to achieve results such as those described for the
WSB is part of an intensively applied and sophisticated
prescribed burning regime in Kakadu that is starting to reverse the
impacts of spear grass. Evidence drawn from aerial photography and
fire-monitoring plots, maintained by Kakadu rangers under the
guidance of Dr Jeremy Russell-Smith, point to a cooling of fires
and subsequent improvement in the general structure of the
woodlands in many areas of the Park.
To return to the mystery of the surviving eucalypts, I would
suggest it is because the various fire regimes documented at
Kapalga (this would probably apply to much of the centre of the Top
End) were relatively recent and atypical. In other words, the
Kapalga fire scenario, which may be exterminating the eucalypts, is
simply not old enough for us to witness its inevitable and ultimate
consequence. The decline in eucalypt recruitment and loss of
nitrogen found at Kapalga was a product of relatively recent fires
fuelled by spear grass. Arguably, this situation did not exist 200
It is proposed here that the careful and sustained use of WSB in
the woodlands across the Top End will provide the best means
available to reduce both the intensity and frequency of fires. I am
unaware of any other practical method available which can be used
on a landscape scale.
However, if wet-season burning is to be widely adopted it must
be better understood. To date we know little about its broader
ecological impacts. It is paradoxical that the main—if not
only—ecological tool available to significantly manipulate
fire on a landscape scale has attracted so little scientific
attention. Such attention is needed to answer important, and as yet
poorly understood, questions about wet-season burning.
CSIRO’s Kapalga Fire Experiment in Kakadu National Park
covered more than 250 square kilometres, and tested the four major
fire regimes common in the Top End. The fire treatments, which took
place between 1990 and 1994, were an early dry season burn; three
‘progressive’ burns through the dry season; a late
dry-season burn; and no fire.
Two experiments showed that many plants and animals showed
little or no response at all to the varying fire treatments,
indicating much savanna biota is highly resilient to fire. In many
cases the greatest contrast was between burnt (whether early or
late) and unburnt, indicating that fire frequency is a particularly
Much of the country is burnt every one or two years, but the
Kapalga results indicate that many plants and animals require a
fire frequency of at most once every three to five years.
1. Dick Williams pers. comm.
2. Russell-Smith, J., Whitehead, P., Cook, G., Hoare, J. (in press) 'Response of Eucalyptus-dominated savanna to frequent fires: lessons from Munmarlary, 1973-1996', Ecological Monographs.
3. Ian Morris pers. comm.
4. The late Fred Pocock and Ian Morris pers comm.
5. Peter Stanton, 1997, 'The role of fire: a conservation manager's perspective', in Fire in the Management of Northern Australian Pastoral Lands, Occasional Publication No. 8, Tropical Grassland Society of Australia, pp. 48-49.
6. Brennan, 1997, 'The impact of wet season burning on herbaceous plants in savanna woodlands in the Jabiru area', in Proceedings of Bushfire 97,CSIRO, p 24.