|
V. Research priorities
Fire
An understanding of the relationship between fire frequency and
rainforest dynamics, and the spatial and temporal distribution
patterns of fire, are central to the issue of rainforest management.
They are also central to the management of timber resources and
protection. The project should be developed in three parts, relating
to the compilation of data, a retrospective study of fire and
disturbance, and a statistical analysis of fire probabilities.
Data requirements
A study should be developed that addresses the ignition probabilities
of different forest types, changes in ignition probabilities with
age and silvicultural treatment, the spatial variability of fire
intensity within a fire perimeter, and the relationship of fire
patterns to topography. The ability to estimate the lengths of
rainforest stands in the GIS database adjacent to other forest
types would need to be developed. The acquisition of data on
the distribution of fire impacts within the fire perimeter may
be limited to the 1983 fires, and to particular regions in Victoria
where the aerial photography is of sufficient quality.
Recommendation
A retrospective study of disturbance impacts
A retrospective study of fire and harvesting would provide critical
information on the behaviour of fire in relation to rainforest,
and in relation to forest management practices. Data may be available
from East Gippsland or the Central Highlands, using GIS data and
aerial photography from the 1983 fires. The study would require
details of patterns of fire within the perimeter. The results
could be used to evaluate historical patterns of fire events in
Victorian forests at different spatial scales, and to identify
the factors that predispose a rainforest stand to fire. Most
estimates of recovery times following fire are speculative, and
disturbance response dynamics of rainforest and mixed forest could
be included in this study.
Recommendations
Statistical analysis of fire probabilities
Effective fire management will be enhanced by a knowledge of the
statistical properties of wildfires. This information could be
used to evaluate the risk of loss of rainforests that results
from different management alternatives in different forest types.
The study would require most of the data outlined under 'data
requirements' above. It should use available data to make a first
approximation of impact on rainforest on the basis of statistical
models, and then develop a spatially explicit model of fire risks.
The results of this study are likely to have useful consequences
for timber resource management and fire suppression, as well as
rainforest protection.
Recommendation
Windthrow
Windthrow of eucalypts and rainforest trees in the rainforest
buffer on the edge of clear fall areas has important consequences
for physical and ecological processes within the buffer and in
core rainforest. Not least, windthrow is implicated as a causal
mechanism in the infection of cool temperate rainforests by Myrtle
wilt, and as a mechanism increasing risk of fire in rainforest.
The extent and severity of windthrow on Victorian clear fall
boundaries is undocumented. The ability to manage for windthrow
will depend on an understanding of such factors as soil type,
slope, aspect, orientation to the dominant wind direction, stand
density, stand size and stand shape, with the aim of producing
a predictive function for the probability of windthrow of a tree
as a function of landscape and site factors. This information
could be used to ameliorate the effects of windthrow during planning
of harvesting operations, and it may have an impact of future
prescriptions for rainforest buffers.
Recommendation
Disease
There are numerous questions to be answered in relation to the
question of the effect of management activities on the distribution
and prevalence of the disease. Monitoring and a bioclimatic study
of the potential distribution of Phytophthora cinnamomi
are probably sufficient protection measures for diseases caused
by that pathogen. The general themes developed by Cameron and
Turner (1994) in relation to myrtle wilt require urgent clarification
and testing. They include studies to address the following points.
The current status of myrtle wilt in Victoria
A study should be implemented to survey a significant sample of
cool temperate rainforest stands and mixed forest stands (emergent,
mature or senescent eucalypts over a rainforest canopy) in the
State that include N. cunninghamii with the aim of establishing
baseline disease status data. Within each stand, multi-phase
sampling involving low-level aerial photo interpretation and ground
truthing of stratified random samples should be used to estimate
the frequency of the disease within size classes of N. cunninghamii.
The sample should be sufficiently intense to provide an estimate
of the rate of disease within a stand that is 95% certain to be
within 10% of the true value. The method should be cost-effective,
and the process should be repeated at least biennially for the
next 20 years. The x,y co-ordinates of all samples should be
recorded so that the information could also be used for other
studies, detailed below. The data could be used to answer such
allied questions as, is the rate of the disease in the Otways
forests higher than the rate in floristically similar callidendrous
forests of Tasmania, and, are small trees relatively resistant
to the disease?
Recommendation
Associations between human activities and the current distribution
of the disease in Victoria
A retrospective analysis of anthropogenic disturbance of cool
temperate rainforest stands and mixed forests should be undertaken.
The study should incorporate estimates of the frequency of the
disease within stands, as well as recruitment and survival rates
of plants, and measurements of other contributing factors such
as the site disturbance history, topography, exposure, aspect,
co-dominant species, stand density, disease history, distance
to other stands, and disease status of nearby stands. A subset
of these stands should be monitored at fixed intervals for changes
in the infection rate, and to provide information for spatial
analyses of the patterns of new infections. One of the most important
processes in the development of a monitoring program is to implement
power tests of proposed monitoring programs, preferably with information
from pilot studies (Stoline and Cook 1986, Millard 1987, Tversky
and Kahneman 1982b). The data from the monitoring program and
the survey of the distribution of the disease could be used to
address allied questions such as, which factors predispose N.
cunninghamii forest stands to the disease?
Recommendation
The moisture status of wilt infected communities
The question of the moisture status of wilt affected forests impinges
directly on the suggestion that affected rainforest communities
have elevated fire risks. The project should measure a number
of microclimatic variables including light, temperature, litter
moisture, vapour pressure deficit, humidity and understorey cover
of vascular plants and non-vascular cryptograms. These measurements
should be made in both wilt affected and uninfected sites. The
sample sites should be chosen randomly and independently within
the constraint of infection status, and any variables that may
contribute to moisture status should be measured concurrently.
Recommendation
The sensitivity of disease processes to environmental factors
As noted above, the conclusions reached by CNR on the hazard represented
by disease spread in Victoria's rainforest depend on the accuracy
of prognoses concerning site and individual susceptibilities,
and on the rate of movement of the disease among stands leading
to new point infections. Simulation of disease processes is in
general well developed. The value of such an exercise in the
context of disease in Victorian rainforests is that it will identify
the processes and parameters that contribute most to the establishment
and spread of disease. The results would enhance the ability
to predict and if necessary, contain impacts on rainforests.
Computer based simulation models are likely to be most valuable
because of problems of tractability in geographically explicit,
stochastic models (Bailey 1967), particularly in highlighting
the short-term (non-asymptotic) behaviour of disease spread (Mollison
1977). Such models should be formed using mechanisms of growth
(local spread, infection) and contact distribution leading to
the identification of movement patterns, quarantine restrictions
and monitoring activities that result in different thresholds
for endemic, epidemic and pandemic disease (Mollison and Kuulasmaa
1985). The likelihood of achieving such states could be evaluated
using different assumptions about species and site susceptibilities.
Stochastic, structured, spatially explicit matrix models for
disease spread in human populations (eg., Rvachev and Longini
1985) may make suitable starting points. Recruitment and death
rates collected in the above studies could be used to parameterise
these models. They would provide answers to questions related
to the sustainable rate of disease infection, and the sensitivity
of the disease to different management practices such as protection
of wounds, timing of operations, and removal of dead trees.
Recommendation
Edge effects
The impact of harvesting on environmental variables and on specific
species should be investigated with appropriately designed sampling
and monitoring studies (see, for example, the studies above).
The data should be used to develop a statistical model of the
effects of clear-felling operations on physical variables (light,
wind, temperature, moisture) and biotic variables, as a function
of stand attributes including logging history, structural characteristics,
topography, stand size and shape and so on (eg., Vanclay 1989).
The data from this study could be used to establish the parameters
of the regeneration niche of rainforest species. The species
included in the sample should include plants, soil and canopy
invertebrate fauna (eg., Basset and Arthington 1992, Kitching
et al. 1993). Both standard statistical analyses, spatial
analyses, and gradient analyses (eg., Gosz 1992) should be employed
to measure environmental change at a range of spatial scales.
Such analyses will provide information on the physical environmental
variables constraining ecosystem dynamics within the ecotone,
the biota that operate within these constraints, and the interactions
among these biota. For example, the distribution and abundance
of weed species, and changes in their abundance through time,
will be established. The direction of successional change may
be investigated by comparing the recruitment patterns evident
in gaps caused by the natural deaths of adults (eg., Lertzman
1992). The results of this study will allow edge effects to be
quantified (Bradshaw 1992) and will have consequences for the
future prescription of rainforest buffers.
Recommendations
|
