The ecological and physiological drivers and the development of improved landscape detection of the invasive perennial, Eragrostis curvula, across the southern tablelands of NSW, Australia

Australia's prosperity hinges on its grasslands, the backbone of its agricultural endeavours, with $29.6 billion in livestock and livestock product value between 2017-2018. Yet, the invasion of undesirable grassland species, or weeds, poses a formidable threat to these ecosystems. One such invader is Eragrostis curvula (Schrad.) Nees, an invasive perennial tussock grass that has encroached upon approximately 15% of the Snowy Monaro region in NSW, Australia. This thesis addresses the pressing ecological and practical challenges posed by E. curvula while focusing on the specific research conducted during the PhD journey. Here, we critically review the literature on E. curvula, emphasising its ecological impact, forage quality, and taxonomic classification, with the goal of providing clarity in these areas. Fieldwork in 2019, conducted in the Snowy Monaro Region, examines the relationship between E. curvula's biomass and cover, soil attributes, and above-ground species richness. The research uncovers correlations, particularly in terms of E. curvula's persistence during drought conditions. Further investigations explore the impact of E. curvula on soil seed banks and above-ground plant diversity, shedding light on the challenges it poses to native species. In controlled glasshouse conditions, the study compares herbicide-resistant and susceptible populations of E. curvula, highlighting differences in early development morphological and physiological traits. This research underscores the importance of integrated management strategies to balance agricultural productivity and biodiversity in the face of herbicide-resistant species. Machine learning and remote sensing technology are utilised as predictive models are employed to map the distribution of E. curvula in the Snowy Monaro Region. The insights gained provide valuable guidance for targeted resource allocation in invasive species management. Finally, the research explores the utilization of convolutional neural networks (CNN) to detect E. curvula using drone imagery, achieving a detection accuracy of 65.4%. This thesis presents a focused and comprehensive exploration of the ecological and physiological drivers of E. curvula. It offers valuable insights for land managers and researchers grappling with this invasive species, contributing to the ongoing efforts to manage and mitigate its impact effectively.