This project builds on a previous study that quantified forest expansion and grassland contraction in Litchfield National Park, a subcoastal region of the Australian monsoon tropics, based on digitized aerial photographs taken in 1941 and 1994. Changes in fire regimes, groundwater and atmospheric CO2
were hypothesized to be possible drivers. Lack of ground-truthing meant that the nature and causes of the observed vegetation dynamics remained uncertain. To tackle this problem, the present study involved a field survey of 248 transects within the same study site. One-way ANOVAs of floristic and abiotic variables revealed significant differences between the 1994 vegetation types as well as between landscape setting classes as defined by the previous study. This provided field evidence that an expansion of Closed Forest and contraction of Grassland had occurred. To evaluate the hypotheses for the driver of the observed change, a TWINSPAN analysis facilitated the separation of rainforest and non-rainforest species groups. Subsequent ANOVAs found a lower density of rainforest species in areas where the vegetation had declined and an increase in areas where the vegetation had thickened. The percentage of sites of each vegetation type in 1994 occurring on dry substrates was found to be inconsistent with the groundwater hypothesis. GIS analyses revealed a relationship between the distance of each sample point from the other vegetation types in 1941 which was consistent with the fire hypothesis. The average fire frequency of each trajectory for the period 1990 to 2000, calculated from a GIS database, was correlated to the direction of vegetation change. The ethnographic and contemporary management of the study site from interviews with a local resident and a park ranger were found to be consistent with the fire hypothesis. A proxy data source of historical groundwater fluctuations from the Katherine River was found to be consistent with changes in groundwater driving the vegetation dynamics. However, the evidence suggested that even if groundwater or CO2
were significant influences, the landscape has been modified substantially by the incidence of fire and probably changes in fire pattern. It is acknowledged that interactions between fire, soil moisture and CO2
are highly complex and therefore difficult to evaluate with the currently available data. Nonetheless, an interaction between fire and soil moisture in the vegetation dynamics is hypothesized. These results are discussed in relation to significant problems in landscape ecology and the current management regime of the study site. It is concluded that that fire management will be critical in determining future shifts in vegetation boundaries, and that future biodiversity conservation will depend on value judgments as to which vegetation types should be protected.
|Date of Award
|David Bowman (Supervisor)