Assessing the threat posed by an invasive African grass Urochloa mutica (Forssk) Nguyen to biodiversity conservation in the Mary River wetlands, Northern Territory

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