by Alaric Fisher, Parks & Wildlife Commission of
the NT and Adam Liedloff, CSIRO Sustainable Ecosystems
From Managing for healthy country in the VRD, Tropical
Savannas CRC 2000.
Modelling the trade-off between fire and grazing |
model | Methods | Results | Discussion | Modelling vegetation change
Satellite information is being used to assist in the monitoring
of a number of rangeland sites across the VRD. The attraction of
using satellite information is that it enables the extrapolation of
land condition and trend information across large areas and reduces
the need for intensive ground based sampling.
Click here for a large version of
Regional trend product for the Ord-Victoria
region from 1992-97 using the average of Landsat MSS band 1 + band
2. The area analysed is 66,550km2 of the 126,000 km2 within four
mosaiced Landsat scenes.
Light green represents areas where vegetative
cover has increased; areas in red a decrease in cover and dark
green represents stable cover. In this example, fire scars have not
been removed and much of the red is attributed to burnt
Fire history vectors overlain on this image
would aid in identifying areas affected by fire. Clearly these data
show that over most of the Ord-Victoria region cover has increased.
This trend can be directly attributed to an exceptional run of good
seasons from 1993-97.
Work was undertaken to determine whether there is a correlation
between land condition assessed with the aid of satellite
information, and vertebrate biodiversity. The idea was to determine
what species are more likely to be found in sites classified under
land condition criteria ranging from "poor" to "good". This would
enable much broader scale assessments of vertebrate impacts to be
made from satellite information, without the need for labour
intensive field based sampling.
While some differences were found in the type of vertebrates
found at "good" and "poor" sites, the relationships were quite
weak. Unfortunately, the effects of canopy cover and landscape
variations complicated assessments of vertebrate impact. Much more
work will need to be done to use this technology for assessment of
Fire and grazing are known to influence the tree-grass character
of tropical savannas. Preliminary results from computer modelling
has confirmed that savannas in the VRD are reliant on intense fire
to maintain their open nature and suppress woody vegetation.
Moderate grazing, however, reduces forage and therefore provides
insufficient fuel loads to maintain intense fire. A balance
therefore needs to be found which takes account of the relationship
between fire and grazing intensities.
Computer modelling enables a range of scenarios to be studied
that would either be far too costly or not feasible with field
experiments alone (See completed CRC project, Modelling and
landscape change link below). In this project, the SAVANNA
model was used to simulate the structure and function of savannas
in the VRD over a 40-year period. The model incorporates factors
such as plant population and primary production (the amount of
plant biomass such as roots, leaves and stems produced over time)
and the role of light, water, weather, fire, soils and grazing
pressure. Weather, soils and vegetation information from Kidman
Springs was fed into the model and this allowed comparison with
results from previous field-based studies.
Simulations were performed on two 500 metre by 500 metre plots.
One was on red calcareous loam soils supporting eucalypt woodlands
and the other on black clay soils with a ribbon/blue grass
grassland. A range of fire regimes and grazing pressures were
simulated to determine the response of the two vegetation types to
disturbance. These included mild and intense fires to represent
controlled early dry season fires and uncontrolled late dry season
fires respectively. These fires also varied in frequency. Grazing
was simulated using moderate and heavy grazing pressures.
In the absence of fire, the model revealed a gradual increase in
trees and shrubs, particularly on the red soil sites. Tree cover
was not influenced by mild fire, but was reduced by intense fire.
Shrubs, however, were more susceptible and cover was found to
decline with any fire regime. Fire not only reduced woody
vegetation cover, but also changed the structure of woody plant
populations. The proportion of the smallest trees and shrubs was
found to increase as those of intermediate size were killed and
replaced by seedlings and re-growth. Fire also led to a decline in
perennial grass, which was most noticeable in red soil
Like fire, the two levels of grazing simulated also caused a
decline in perennial grasses at the red soil site. The more
resilient black clay grassland was able to withstand these levels
of grazing with only slight declines in yield.
These results suggest that intense fire may be an economic means
of controlling woody plant populations. Fires will, however, reduce
pasture biomass and this must be considered if pasture condition is
important for grazing. Moderate to high grazing pressure not only
reduces the chance of intense fires, but may also mean that even
mild fires used to stimulate grass shoots and improve pasture may
not be possible, allowing woody vegetation to increase.
It may not be sufficient to select a simple fire regime in
management and expect all facets of the savanna to be maintained. A
trade-off between the desirable outcomes of fires, such as reduced
woody growth and improved areas of pasture, and the undesirable
effects on pasture condition, must be balanced.
Computer modelling is being used in this project to explore how
rangeland vegetation might change over long periods of time under
different management and climate change scenarios. The model being
developed is designed to overcome some of the limitations of
existing models so that more specific and reliable management
conclusions can be drawn. One of the key management issues to be
studied with the model is the trade-off between grazing and fire in
different savanna environments. A wide range of environmental
information will be fed into the model and an assessment will be
made regarding what scale and level of detail is necessary to
provide the best results.
The model will simulate long term vegetation dynamics in terms
of the amount and type of dry matter produced. It will explore the
role of plant traits such as root: shoot ratios, rooting depth, and
associated growth potential, in combination with the distribution
of resources in different rangeland environments over space and
Feedback mechanisms between vegetation and soils will be
factored into the model as well as changes in the composition and
physiological characteristics of vegetation. The latter will
include changes in the structure of the canopy and root system of
plants and the composition of the actual plant tissue.
Changes in vegetation composition will focus on the proportions
of annual or perennial grasses and forbs, shrubs and trees, growth
and death rates, conditions for the recruitment, establishment,
maturing and reproduction of vegetation, and other factors which
facilitate the growth potential or competitive ability of pasture
plants. This study was part of the now complete CRC research
project Modelling Landscape Change .