AbstractFor naturalised invasive weeds there is a recognised need both to identify priority areas for research and weed control or mitigation efforts, and to improve our understanding of their impacts. In this study I focused on an invasive African pasture grass Urochloa mutica para grass in the Mary River wetlands, located in the wet-dry tropics of northern Australia. I addressed the need to clarify research and weed management priorities by creating a habitat suitability model (HSM) for para grass and used the output from the HSM in a simple model of spread. Predictions from the spread model, based on habitat suitability and proximity to existing populations of para grass, were used to estimate the relative risk of invasion for different wetland habitats. Two habitat suitability modelling approaches (generalised linear modelling and Bayesian inference) were compared to provide a more robust insight into the effect of analytical approach on assessment of risk of invasion. Spread modelling was also used to explore the effect of alternative management responses to estimates of the relative susceptibility to para grass invasion.
In adopting this approach I recognised that the models developed represented a trade-off between reality, generality and precision. In this study the trade-off is between ecological precision (e.g. patch scale effects of local variation in environmental variables) and the utility of landscape-scale models of habitat suitability. I contend that the benefits of being able to assess invasion risk and hence weed research and management priorities at the landscape scale, offsets the limitations associated with the precision of the predictions. Both the habitat suitability and spread modelling were implemented within a GIS to facilitate spatially explicit modelling at the landscape scale.
Following the assessment of the relative susceptibility of wetlands habitats to para grass invasion, I quantified the current impact of para grass on wetland habitats using plot-based flora survey to assess changes in floristic composition in response to varying para grass density. Survey plots (n=374) were placed along 55 kilometres of transects were sampled in mid-wet season between 2001 and 2003. The results were used to measure current impacts on floristic composition and infer landscape-scale effects of para grass invasion in the absence of management intervention.
The study showed (1) para grass has invaded large areas of the wetlands and severely reduced habitat diversity in many of these areas (2) there is potential for much larger, currently intact, areas to be invaded in the future (3) the areas at greatest risk of invasion play an important role in the conservation of several key species in Northern Australia (4) current land and weed management practices favour cattle production and will compromise habitat conservation goals unless this imbalance is redressed (5) geographic information systems have an important role to play in the development of decision support tools to support adaptive management of natural resources. However, their potential is constrained by their limited capacity to readily perform statistically robust and ecologically meaningful modelling.
|Date of Award||Mar 2007|
|Supervisor||Peter Whitehead (Supervisor)|