Open Access Theses

Permanent URI for this collectionhttps://hdl.handle.net/1885/3

To view all theses in this collection, select one of the 'Browse by' options (Issue Date, Author, Title, Subject, Title or Type (of thesis). You can also enter your keyword/s into the text box above and click on Search.

ANU theses are harvested by the National Library of Australia's Trove service and other search engines, making them fully discoverable online.

Find Australian theses.
Full instructions available here

Submit your thesis (Approved ANU research theses only)

Please note: The Abstracts displayed in item metadata are in many cases truncated. For the full Abstracts, see the thesis document files.

Browse

Now showing 1 - 20 of 16289

Access status: Open Access ,
Advancing Graph Neural Networks: Expressivity Enhancement, Information Flow Optimization, and Dynamic Process Modelling
(2026) Hevapathige, Asela
Graph Neural Networks have emerged as powerful tools for learning representations from graph structured data, yet fundamental limitations constrain their effectiveness across diverse applications. This thesis addresses three critical challenges in graph representation learning: improving structural expressivity, optimizing information propagation, and modeling dynamic processes. First, we address the expressivity limitations of standard message passing neural networks, which are bounded by the 1 Weisfeiler Lehman test. Using permutation invariant graph partitioning, we develop Graph Partitioning Neural Networks that capture complex structural interactions by explicitly modeling how different graph components interact. Our theoretical analysis demonstrates that this approach efficiently approaches 3 WL expressivity while remaining computationally tractable. Next, we develop two complementary frameworks for adaptive depth allocation in GNNs. The first uses learnable Bakry Emery curvature to capture both structural properties and diffusion dynamics, showing that nodes with higher curvature require fewer message passing iterations for effective representation learning. The second provides theoretical foundations linking neighborhood characteristics to optimal aggregation strategies under different homophily conditions, revealing that aggregation requirements depend on the balance between same label neighbors, opposite label neighbors, and local degree distribution. Both frameworks remove the need for separate architectures for homophilic and heterophilic graphs. We then address limitations of existing diffusion based GNNs through two new frameworks. The Generalized Opinion Dynamics Neural Framework unifies multiple opinion dynamics models, incorporating node specific stubbornness, dynamic neighborhood influence, and structural regularization to enable diverse convergence behaviors including single consensus, multi consensus, and individualized consensus. For influence maximization, we develop Deep Sheaf Networks that model propagation as a sheaf diffusion reaction process, introducing pointwise and coupled dynamics operators to capture both intrinsic node updates and neighborhood interactions across progressive and non progressive diffusion models. A subgraph based optimization method reduces the combinatorial search space while accounting for overlapping influence among nodes. In summary, the theoretical contributions of this thesis provide rigorous foundations for understanding structural interactions, information flow optimization, and dynamic process modeling in graph neural networks. Extensive experiments across node classification, graph classification, graph regression, influence estimation, and influence maximization tasks show that our approaches consistently outperform existing methods.
Access status: Open Access ,
SHIELD: A SHapley and Information-theory based framework for Equitable Learning via Dissimilar feature grouping
(2025) Yun, Hyeonggeun
Machine learning models are increasingly applied in clinical and biomedical settings, due to their ability to capture an intricate underlying structure, yet their complexity can obscure critical insights and risk propagating biases that disproportionately affect vulnerable populations. Hence, this thesis introduces SHIELD: A SHapley and Information-theory based framework for Equitable Learning via Dissimilar feature grouping, which combines dissimilarity-driven feature grouping with interpretable latent representations to mitigate proxy bias and enhance equitable learning of the resulting model. By constructing a dissimilarity matrix based on conditional mutual information (CMI), features are grouped to weaken correlations that might encode sensitive attributes, reducing redundant signals that may contribute to unfairness. This automated approach is more efficient than clustering similar features and fixing problematic groups post-hoc. Group-specific autoencoders learn latent representations that summarise each group’s unique information while preserving a decoder-weight mapping back to the original features. This enables precise SHapley Additive exPlanations (SHAP) value decomposition, leading to interpretable feature-level attributions despite dimensionality reduction. Experiments on four benchmark clinical datasets demonstrated that the proposed grouping approaches, greedy, Bicriterion, and K-plus anticlustering, achieved notable improvements in fairness metrics and produced more evenly distributed feature importance compared to raw features and traditional baselines. This was evident as grouping on average led to 9.47% improvement in the distance from origin of bias quadrant, which accounts for both explanation and prediction bias. In addition, the fairness overview score, which considers other typical fairness metrics as well, was improved by 2.42% when grouped by dissimilarity. While a modest reduction of 3.43% in accuracy and 5.16% in F1-score was observed, it remained within acceptable limits for clinical applications, demonstrating the feasibility of this fairness-performance trade-off. Overall, SHIELD provides a principled framework that integrates dissimilarity-based grouping, latent representation learning, and explanation-level auditing to promote equitable and explainable machine learning for health informatics.
Access status: Open Access ,
Frogs in farmland: Ecology in changing agricultural landscapes
(2026) Littlefair, Michelle
Amphibians are globally threatened, with habitat loss and fragmentation due to agricultural activities, posing substantial threats to many species. This thesis investigates the role of farm dams as habitats for amphibians within agricultural landscapes. I focused on the habitat value of farm dams, adult frog distribution, reproductive dynamics, and landscape features. Despite growing interest in human-made habitats in agricultural landscapes, there has been no comprehensive review of the literature on farm dams as habitat. This leaves gaps in our understanding of their broader biodiversity implications, and their value as habitat - particularly for frogs. While farm dams have been known to provide habitat for frog species, management interventions to improve dams as habitat have not been explored. In particular, enhancing dams by excluding livestock and encouraging native vegetation growth, to support adult frog populations and improve reproductive output. Additionally, little is known about the temporal and spatial dynamics of frog populations across agricultural landscapes. By addressing these gaps, this thesis examines the habitat value of farm dams for frogs, specifically how enhancing farm dams may support frog populations and reproductive success when compared to control dams.
Access status: Open Access ,
Lifetime measurements of the 2^{+}_{1} states in even-even platinum isotopes with A = 180-186
(2026) Kono, Rikako
The neutron-deficient even-mass platinum isotopes have long been of interest for their transitional shapes and the emergence of shape coexistence. The more neutron-rich ^{190-198}Pt isotopes have weakly deformed, oblate structures in their ground states and show some evidence of triaxiality. In contrast, the lighter mid-shell isotopes ^{178-186}Pt exhibit prolate ground states and clear evidence of shape coexistence, with ^{188}Pt as a transition point between two regions. These platinum isotopes present a notably complex evolution of shapes and shape coexistence that would benefit from further investigation. The lifetimes of 2^+_1 states and the B(E2; 2^+_1\rightarrow0^+_1) reduced transition strengths are valuable to give insight into the magnitude of the nuclear deformations and their shape. However, there are significant discrepancies in 2^+_1 lifetimes for the shape-coexistence region, ^{180-186}Pt, from the Evaluated Nuclear Structure Data File (ENSDF) database and more recent measurements using methods such as the recoil distance method (RDM) and the Generalised Centroid Difference (GCD) method. To address this puzzle, we performed direct-timing lifetime measurements of the 2^+_1 states in ^{180-186}Pt using exponential-convoluted Gaussian fitting and GCD methods applied to \gamma-fast timing data taken with LaBr_3(Ce) detectors at the Australian Heavy Ion Accelerator Facility (HIAF) in August and September 2023. The 2^+_1 lifetimes obtained from both the convolution and GCD methods are in agreement, yielding results of 533(26)~ps, 561(14)~ps, 479(6)~ps and 350(4)~ps for ^{180-186}Pt, respectively. The measurements for ^{180}Pt and ^{186}Pt in the present study are consistent with ENSDF, and those for ^{182}Pt align with a recent result. However, a discrepancy is observed between our measurement and the ENSDF data for ^{184}Pt, suggesting to revisit the lifetime measurement for this isotope. Our experimental values were then compared to the results obtained from an empirical multi-band-mixing model to clarify the underlying shape assignments. The calculation considering shape coexistence reproduced the experimental B(E2; 2^+_1\rightarrow0^+_1) values well, thereby further emphasising the presence of shape coexistence in these regions. A new passive-shielding array for LaBr_3 detectors for future lifetime measurements has been designed, built and incorporated into the CAESAR gamma-ray detector array. The first test experiments with the new shields were performed in October 2025, and the results are currently being analysed.
Access status: Open Access ,
Mastering old age : Buddhist practice and techniques of the self among elderly laywomen in Ho Chi Minh City
(2016-09) Le, Hoang Anh Thu
This thesis is an anthropological study of old age and Buddhism in Vietnam. It explores how Vietnamese women utilise Buddhist practice to navigate the late stage of life. Embedded in the context of a fast growing aging population and the present lack of studies that focus on aging in Vietnam, this research looks to contribute to the understanding of aging and the life course, and the role of religious practice in informing and reconstructing the everyday experience of being old in Vietnam. This research shows that Buddhism is a regime of practice that elderly women follow in their everyday life. As a daily practice, Buddhism shapes women’s experience of being old, by transforming the temporal and spatial structure of their everyday lives, and extending their social networks and engagements far beyond their families and feminine roles at home. Buddhist imaginaries and conceptualisations of life redefine the way women conduct themselves in their everyday lives and perceive and live out their roles at home and in wider society. In the process, elderly women also transform Buddhist practice to make it respond to their lived realities. Buddhism is lived out by elderly women not as a doctrine but as a practical way of life. The way of life that Buddhism carves out for its elderly adepts is embedded in the socioeconomic landscape of the contemporary socialist market economy in Vietnam and is imprinted with its practitioners’ gender- and age-specific habitus. This research finds out that Buddhism is also a set of techniques that work on the self, through which elderly women redefine their relational personhood in their interactions with their families and the wider society. Buddhism provides its elderly adepts with self-mastery resources which enable them face the dreadful thresholds of the final stage of life. As a technique of the self, Buddhism enables elderly practitioners to adjust their identities and personhood, which in turn give them a sense of purposefulness, agency and control over the various realities that emerge in the course of old age, such as physical decline, ambiguities in identity and loss of agency at the end stage of life. Old age is therefore mastered with Buddhist practice, and is transformed from a life stage of marginality and infirmity, to an experience of continuing personal development and empowerment.
Access status: Open Access ,
Learning Realistic Asset-Oriented Human Interaction and Animation
(2026) Wang, Rong
Modelling diverse and complex human activities is a fundamental research topic in computer vision and graphics, as it enables a wide range of applications such as extended reality, digital humans, video game and movie creation. In many realworld scenarios, human activities often involve additional object assets, for example, interactions with handheld items such as tools or sports equipment, or animations of avatars wearing complex apparel like coats or dresses. Meanwhile, generating such interaction and animation scenes conditioned on the given assets in a realistic manner can be challenging, as it requires in-depth understanding of the spatial and temporal relationships between the humans and assets. In this thesis, we tackle this challenge of generating asset-oriented human interaction and animation at a higher level of realism, offering an immersive user experience for downstream applications. In the first part of the thesis, we focus on the task of interacting hand-object pose estimation. To ensure that the reconstructed hand and object poses conform to contact dynamics, we develop novel physics-inspired network architectures by explicitly incorporating contact heuristics as an inductive bias in the network design. In the first work, we reformulate hand-object interaction modelling from previous approaches that encode sparse keypoint-level dependencies to a new representation based on actual surface-level contact points. This allows us to capture proximity relationships between hand and object vertices using dense graph attention, enabling the network to automatically infer plausible contact regions. To further enforce real-world physical constraints, in our second work, we extend to incorporate more complex contact heuristics, i.e. stable grasping against gravity. Since contact dynamics are well implemented in modern physics simulators, we propose to distill their knowledge from simulation and transfer it to the base pose estimation model, which is the key for reconstructing simulation-aligned stable configurations. In the second part of the thesis, we investigate effective approaches for generating realistic animations of clothed human avatars. Since human bodies and apparels exhibit distinct dynamical properties, our key insight is to decompose the animation of a holistic clothed human avatar into independent, interpretable components that can be modelled and supervised separately. In this first work, we begin with stylised characters wearing complex accessories, where we animate their rigid body motion and non-rigid apparel motion via skinned deformation and auto-regressive vertex displacement respectively, allowing us to produce visually appealing dynamical effects for the apparels. Next, we extend this approach to real humans reconstructed from casually captured phone photos. Specifically, we develop a universal clothed human model that jointly decomposes personalised avatar shapes, skinning weights, and pose-dependent cloth deformations, achieving superior model robustness and generalisability. Finally, we apply the proposed decomposition principle to improve cloth animation generation with fine-grained dynamics such as wrinkles and folds. In this last work, we represent cloth motion via decomposed low-frequency posed shapes and high-frequency wrinkle details, which mitigates spectral bias in neural networks and facilitates the learning of high-fidelity cloth movements. In summary, this thesis presents five works aimed at improving the reconstruction realism of asset-oriented human interaction and animation. We validate our methods across multiple applications, including interacting hand-object pose estimation, 3D motion transfer, and animatable clothed human reconstruction. Extensive experiments demonstrate that our approaches achieve superior fidelity and plausibility compared to existing methods, advancing the state of the art in understanding human behaviours involving diverse real-world assets.
Access status: Open Access ,
Epidemiological investigation of diseases of public health importance in Victoria, Australia
(2017-07) Arnott, Alicia
In this thesis, I present the projects and activities I have undertaken as a Master of Philosophy in Applied Epidemiology (MAE) Scholar in Victoria between February 2015 and September 2016. I was placed with the Communicable Disease Epidemiology and Surveillance (CDES) Unit at the Victorian Department of Health and Human Services, and the Victorian Infectious Diseases Reference Laboratory (VIDRL) Epidemiology Unit. Through these placements I experienced the day-to-day activities of a state public health unit as well as an applied public health research environment. I conducted a cohort study to identify the source of a large, highly publicised outbreak of Salmonella Typhimurium at a five-star hotel in Melbourne, which was identified as raw-egg mayonnaise used in sandwich fillings served at a High Tea. For my data analysis project, I investigated the epidemiology of legionellosis in Victoria between 2000 and 2015 to determine whether the ubiquity of the Legionella pneumophila specific urinary antigen test is creating an 'epidemiological blind spot' for non - Legionella pneumophila infections. I found that whilst this method was ubiquitous prior to and during the study period, the rate of infection with non-Legionella pneumophila species did not decline in parallel as expected if urinary antigen testing routinely precluded their detection. I evaluated the complex Victorian influenza surveillance system, which involved analysis of data captured by the system from 2005 - 2014. As a whole, the system was found to perform well and the data collected were used to inform public health activities. Geographic representativeness of syndromic ILI data was high, demonstrating its utility as a surveillance tool, and the widespread uptake of molecular diagnostic testing enhanced overall system sensitivity, timeliness and flexibility. In addition, the data collected enabled robust estimates of seasonal vaccine effectiveness to be determined which informs local public health action and global vaccine development. However, important deficiencies that prevent the system from achieving a number of its objectives were identified. Most importantly, the ability of the current system to guide planning and implementation of policy and to detect and control outbreaks is limited, and laboratory testing denominator data are not available to facilitate interpretation of seasonal trends and true influenza incidence. My epidemiological project was an analysis of whether registering with Spleen Australia [formerly the Victorian Spleen Registry (VSR)], which provides education, clinical guidance and health promotion reminders, reduces the incidence of overwhelming post splenectomy infection amongst Victorian registrants without a spleen. By conducting a survival analysis, I found that VSR registration was indeed associated with a highly significant (p<0.001) 88% reduction in the incidence of severe infection amongst splenectomised registrants. Finally, I present two teaching activities conducted during my MAE: a session on how to conduct outbreak investigation interviews and a lesson from the field highlighting the utility and pitfalls of culture-independent diagnostic testing when investigating an outbreak of Shigella. This thesis provides an account of my MAE experience, fulfills the requirements of the program and outlines the contribution my work has made to public health in Victoria.
Access status: Open Access ,
The ecology and conservation of southern squirrel glider populations in agricultural landscapes
(2016-04) Crane, Mason James
Much of the world's terrestrial environment is transformed by human activities, particularly agriculture. Within these anthropogenic landscapes some wildlife persists but, with varying degrees of success. Increasingly, there is a realisation of the importance of natural and seminatural features within agricultural landscapes for biodiversity conservation. For species whose geographical range or habitat preference largely coincides with landscapes transformed by other land uses, understanding key habitats, how they are used, and the threats posed to these habitats is essential. This work focuses on the southern population of the squirrel glider Petaurus norfolcensis, a genetically distinct subpopulation with a geographic range that largely coincides with the sheep/wheat belt of eastern Australia. This thesis explores the use of relictual and seminatural habitats of the species in highly modified agricultural landscapes, via radio-tracking collared individuals. The research aims to examine key habitats and their use, and to close some of the knowledge gaps relating to their ecology and conservation. I describe crucial denning and feeding habitats of the species. Detailed field measurements and logistic regression models were used to identify key physical characteristics of these habitats and their biophysical context. Large, healthy Eucalyptus trees were significant in all models. Gliders used numerous den sites, often sharing with other individuals. Den tree selection was often associated with the location of nocturnal activity, indicating the role of den trees in facilitating efficient foraging. The majority of nocturnal activity took place in the canopy of Eucalyptus trees, with a strong preference for those trees in flower. In the absence of flowering, gliders focused their feeding activity in Eucalyptus trees close to drainage lines. These results highlight the importance of habitats in different parts of the landscape and of maintaining connectivity across them, but particularly, the importance of the mesic parts of the landscape in sustaining populations between flowering events. Gliders used four distinct countryside elements: linear roadside remnants, native vegetation patches (typically on travelling stock reserves), native tree plantings, and scattered trees. Gliders were found to survive entirely within these four elements, with some individuals able to persist in just one element (with the exception of native tree plantings). I demonstrate that the relative habitat value of each element varied, evident from the preferential selection of elements and the impact of their availability on home range size. Scattered trees had a disproportionately high value when compared to other countryside elements. My thesis presents new knowledge of key habitats of the squirrel glider, their use and threats (including a detailed examination of the threat of wildfire on scattered trees). This information has direct conservation implications for the species and adds to the broader discussion on conservation in agricultural landscapes. The thesis is concluded with a synthesis of the new and existing knowledge of southern squirrel gliders and an assessment of the strengths, weaknesses, opportunities and threats, apparent in the conservation efforts for the species. I recommend ten points of action to improve the conservation outcomes of the southern squirrel glider population.
Access status: Open Access ,
Journey to excellence : characteristics of sustainable elite athlete organisations
(2015-07) Stephens, Clive
In the past, management of elite athletes has focused on a coach-athlete (CA) relationship model. More recently, there has been a shift in thinking, to accommodate inputs from the multidisciplines of science, medicine and technology. To better understand the effect of the expanding resources, relationships and networks in elite athlete management, this thesis develops and defines the concept of an Elite Athlete Organisation (EAO), and from an organisational perspective, seeks to answer the question, “What are the organisational factors, practices, and resources that together enable an EAO to be effective and to consistently build a sustainable competitive advantage "? The research was conducted over a three and a half year period, utilising a concurrent mixed method approach (longitudinal case studies and a specially developed survey instrument). Data was collected from a cohort of twelve elite athlete organisations (six key members from each organisation (n = 72 in total). The thesis is an original contribution that validates the emergence and importance of an EAO by comparing the Top 10 and Top 50 EAOs in elite sport. The thesis empirically validates the organisational factors that influence or impeded organisational effectiveness. The primary theoretical contribution is to theory development of an EAO, which sees it as a tightly coupled emergent system made up from the totality of organisational factors, artefacts networks and individuals. Essentially, an EAO is a high functioning knowledge based multidisciplinary entity, delivering a range of specialised integrated services to elite athletes who are capable of podium positions at world-class events. Secondly, the thesis contributes to the discussion of effective management practices in multidisciplinary environments.
Access status: Open Access ,
Embodied Remedies: Plant Medicine in the Art of Fiona Hall, Janet Laurence and Lauren Berkowitz
(2026) Blake, Rebecca
This dissertation examines the work of three contemporary Australian women artists - Fiona Hall, Janet Laurence and Lauren Berkowitz - who all explore the themes of health, healing and botanical medicine in their works of art. Focusing primarily on their works from the 1990s and 2000s - a period marked by a growing belief in the infallibility of medicine, counter-culture scepticism of this paradigm, and a growing interest in alternative approaches to wellness - this dissertation examines how botanical medicine functions as a recurring, yet critically under-explored motif in their artworks. The central claim of this dissertation is that Hall, Laurence and Berkowitz use medicinal plant materials, alongside references to botanical medicine, not merely for aesthetic or symbolic purposes, but as a means to examine the critical cultural place of plants in therapeutic healing and to question the ways in which these plants have been appropriated by scientific medicine. Hall, Laurence and Berkowitz all draw on the long history of plant medicine to provide feminist perspectives on the medicalisation of contemporary society. Through installations, sculptures, photographs, paintings and mixed-media works that foreground the sensory, ecological and historical dimensions of medicinal plants, these artists explore the entanglements between nature, colonialism, capitalism and care. They use the themes of medicine, science, alchemy and magic to forge connections between nature and culture. This study situates art as a vital tool for reimagining medical care and agency beyond the confines of neo-liberal value systems. By tracing the intersections between art, nature and healthcare in their practices, this study contributes to a deeper understanding of how contemporary art can serve as a site for reimagining relationships between health, illness and the natural world.
Access status: Open Access ,
Understanding Australian Orchids and their Mycorrhizal Fungi: Macro to Microevolutionary Perspectives
(2026) O'Donnell, Ryan
The Orchidaceae is one of the most speciose flowering plant families on Earth, and questions surrounding the family's extraordinary diversification through time continue to captivate biologists. The family is renowned for its diversity of floral forms and intricate pollination systems, and for their obligate dependence on symbiotic associations with mycorrhizal fungi for seed germination. Australia plays host to some of the world's most unusual terrestrial orchids, many of which exhibit a remarkable degree of both above and below ground specificity with respect to their pollinator and fungal associates. The majority of studies into the Australian orchid flora and its associated funga have thus far been narrow in scope with a focus on either a macro- or microevolutionary scale. In this thesis, I studied the Australian terrestrial orchid flora and its associated funga as a unified system spanning from the macro to microevolutionary scale. Beginning at the macroevolutionary scale, I synthesised phylogenomic data with over 70 years of orchid mycorrhizal fungal research incorporating fungal sequences, morphology, and germination data to detangle recalcitrant evolutionary relationships within the terrestrial orchid tribe Diurideae. I uncovered an unprecedented degree of fungal niche specificity and demonstrated that fungal symbiont preferences in the Diurideae are phylogenetically structured and can be used to support certain topological hypotheses despite confounding evolutionary histories. At population scale, I focused on a species complex of sexually deceptive greenhood orchids (Pterostylis; Pterostylidinae) where cryptic species are likely, and performed a population genomic study to determine levels of genomic variation between putative species and whether there is support for existing species hypotheses. Genotyping-by-sequencing revealed the presence of discrete genetic clusters, providing molecular support for morphologically delimited taxa. In this chapter I also provide the first preliminary evidence in the literature for hybridisation between species of Pterostylis within a clade where pollination by sexual deception has been experimentally confirmed and is predicted to be the dominant mode of pollination. Moving to the fungal side of the orchid-mycorrhizal interaction, I present two studies with a close focus on one of the core orchid mycorrhizal fungal families: Ceratobasidiaceae (Agaricomycetes; Cantharellales). Firstly, I performed a taxonomic review of the various anamorph-teleomorph typified generic names within the Ceratobasidiaceae. Using publicly-available ex-type sequence data, I inferred a phylogeny which reiterated the paraphyly of several genera within the family and formalised the synonymisation of several genera under a unified Rhizoctonia. Finally, to investigate fungal species boundaries at the genome-scale, I sequenced several putative operational taxonomic units of Rhizoctonia known to associate with species of Pterostylis using long-read sequencing. I present 38 new high quality whole-genome assemblies of Rhizoctonia. Using these data, I found clear genomic divergence thresholds across the Ceratobasidiaceae, allowing us to describe at least six new species of Rhizoctonia. I show that divergence thresholds currently in use do not adequately represent the level of diversity within the family, and highlight the discovery of several lineages that are entirely new to science.
Access status: Open Access ,
How Yolngu talk about art: Aspects of narratives, metaphors and colour terms from the Ganalbingu (Western Yolngu) perspective
(2026) Li, Haoyi
This thesis examines intersections between the language and art practices of the Ganalbingu clan (Western Yolngu) based on eight months of fieldwork in Ramingining and Gapuwiyak, North Arnhem Land, in the Northern Territory of Australia. By studying the discourse around traditional paintings and fibre art of Western Yolngu communities, or how Yolngu talk about art, this thesis investigates aspects of the form and content of narratives about paintings; linguistic and conceptual metaphors; and the semantics of colour terms and aesthetics. This analysis in turn sheds light on Yolngu worldviews, cultural values and pedagogical processes, triangulating between language and art. Based on prolonged engagement with Western Yolngu communities whose perspectives are under-documented in the literature, this thesis contributes to a more complete understanding of the diversity in the Yolngu region, where current understanding is skewed towards Miwatj (Eastern) Yolngu. The main theoretical contributions of this thesis are two-fold: it adopts a contemporary Western Yolngu conceptualisation of 'art', known as gamununggu, which challenges European understandings of art and enriches the cross-cultural metacategory of art (Chapter 4). Through a study of the visual language of Ganalbingu paintings alongside spoken language and other modalities of expression such as bunggul (ceremonial dance) and manikay (song), a multimodal model of communication is put forward which encapsulates this complexity (Chapter 8). The interdisciplinary framing of this thesis is necessary in order to achieve these research aims, as the visual language underlying the interpretation of paintings and spoken language in storytelling are inherently interconnected. The thesis structure reflects the natural pedagogical progression of a non-Indigenous learner of Yolngu knowledge. It begins with a general introduction to Yolngu language and art, referencing relevant literature, and the interdisciplinary approach necessitated by this reflection of Yolngu categories of knowledge (Chapter 1). It then outlines the methods of data collection and general theoretical underpinnings (Chapter 2). A synchronic snapshot of the geography, society, language and art of Ramingining and Gapuwiyak is outlined next, grounded in historical documentation from the literature (Chapter 3). Key terms and theoretical frameworks about art are then set up (Chapter 4), followed by an overview of the discourse (Chapter 5) outlining the most important themes and structures, which is the foundation on which other analyses are built. Subsequently, two aspects of the discourse are analysed in depth: conceptual and linguistic metaphors, mainly based on body-part terms (Chapter 6) and the way colours are named and referred to in discussions about art, as one component of the visual language within multimodal expressions of aesthetics (Chapter 7). Finally, these findings are synthesised in a model of communication, drawing together findings from the entire thesis and contextualising them within the literature on ceremony (Chapter 8).
Access status: Open Access ,
The Social Structure of Geoeconomics: Institutional Transformations Underpinning Weaponisation of Economic Interdependence in East Asia after the 2008 Financial Crisis
(2025) Jaknanihan, Arrizal
Why have geoeconomic tensions continued to rise in East Asia despite the deepening level of economic interdependence in the region? This thesis analyses the global and regional transformation of institutions that contributed to the rise of the weaponisation of economic interdependence in East Asia after the 2008 global financial crisis. It uses the concept of primary institution (i.e. normative principles ordering international society) from the English School of International Relations to conceptualise institutional transformations underpinning weaponised interdependence. The thesis finds that the transformation of the market (i.e. liberal market ideology) as a primary institution after the 2008 crisis has changed the way states view the security value of economic interdependence. The declining influence of liberal market ideology after the crisis has increasingly allowed and legitimised subordination of economic relations to the state’s security and nationalist imperatives, allowing the weaponisation of economic interdependence to rise during conflict. Through the case of China’s economic coercion in East Asia, the thesis demonstrates that the deepening and weaponisation of economic interdependence came together as the downstream impact of declining influence of the market in ordering international society. This thesis contributes to the study of geoeconomics by analysing weaponised interdependence as part of broader institutional changes unfolding in the global and regional levels.
Access status: Open Access ,
Peptide hormone regulation and synthesis
(2015-09) Caron, Karine
The work presented in this thesis focuses on the study of peptide hormones, their post-translational biosynthesis and regulation, and the synthesis of unnatural peptide hormone analogues using cell-fee protein synthesis. In the first instance, the enzymatic activity of peptidylglycine a-amidating monooxygenase (PAM) was studied, an enzyme that catalyses the final post-translational step for many hormones. A novel whole-cell assay is described in Chapter 2, utilising HPLC, which allows simultaneous detection of multiple species of calcitonin (CT). Regulation of PAM activity was examined through measurement of variation in CT and pro-CT levels, in a small cell lung carcinoma cell line (DMS53), to evaluate the effect of PAM inactivators on CT amidation, and in extension, on cancer cell survival. It was found that CT levels relative to CT precursor levels could not be decreased below a certain level in a dose-dependent manner, which signifies the presence of a mechanism for maintenance of homeostasis. Furthermore, through the identification of CT degradation products generated in DMS53 cells medium, it was possible to further probe the degradative pathways for CT in this cell line. To better understand the interactions of PAM with natural substrates, the binding affinities of a series of peptide prohormones were determined, in Chapter 3, using PAM from DMS53 cell medium. While the role of the penultimate amino acid in substrate recognition has been previously reported for short synthetic substrates, the work presented herein highlights the role of the entire peptidic sequence in substrate recognition. It was found that though trends are similar, the extent of the effect of the penultimate amino acid on substrate binding is exaggerated in synthetic substrates, when compared to natural substrates. Furthermore, this variation in peptide hormone binding does not appear to govern the relative levels of those hormones in vivo. Due to our interest in the study of peptide hormones, a new methodology was developed as an alternative to laborious solid phase synthesis for the production of peptide hormones and peptide analogues. As presented in Chapter 4, cell-free protein synthesis, which is commonly used for the production of natural proteins and proteins containing non-canonical amino acids, was applied as a simple and cheap means for generating peptides and unnatural peptide analogues. To do this a new strategy was employed, making use of fusion partner methodology. An expression plasmid was designed and produced, which expresses the peptide of choice as a fusion with a soluble protein, through an enterokinase cleavable linker. This allowed production of CTG and CTG analogues containing the non-canonical amino acids chloro-Val, chloro-Ile, chloro-Tyr, fluoro-Leu and fluoro-Phe. Finally, with the aim of making this new technique more efficient and cost-effective for the production of peptides, in Chapter 5, the use of acetyl phosphate to drive nucleoside triphosphate-dependent protein synthesis together with the generation of the expensive and unstable nucleoside triphosphates was undertaken. This process required characterisation of the endogenous phosphorylating enzymes of the commonly used E. coli BL21(DE3) star strain used for cell-free protein synthesis.
Access status: Open Access ,
Characterisation of flagellar mastigoneme components of phytophthora nicotianae
(2015-04) Hee, Wei Yih
Plant diseases caused by Phytophthora species pose significant threats to agriculture and natural biodiversity throughout the world. Motile, flagellate zoospores of Phytophthora and most Oomycete species play a key role in pathogen dissemination and initiation of infection of host plants. Tripartite tubular hairs called mastigonemes on the zoospore anterior flagellum are responsible for cell motility by reversing the thrust of flagellar beat. The objective of the work presented in this thesis was to identify and characterise components of Phytophthora mastigonemes. In previous studies, peptide sequencing and electron microscopy using an antibody directed towards mastigonemes of Phytophthora nicotianae zoospores had identified a gene, PnMas2, which encodes a protein in the tubular shaft of the mastigoneme. Bioinformatic analysis revealed that PnMas2 belongs to a protein family consisting of three or four members. In the current study, degenerate primers were used to clone PnMasl and PnMas3, the two other members of the Mas family in the P. nicotianae genome. Homologues of the three PnMas genes were identified in the genomes of 30 species of Stramenopiles, the major protist assemblage to which the Oomycetes belong. Phylogenetic analyses indicated that the Mas family can be divided into three sub-families, namely Masl, Mas2 and Mas3. Extensive BLAST analyses of Stramenopile and other eukaryotic genomes demonstrated that the Mas protein family occurs only within the Stramenopile taxon and is not found in those few Stramenopile species that do not produce flagellate cells. Quantitative real-time PCR showed that all three Mas genes are expressed during asexual sporulation but that relative levels of expression at other stages of the asexual life cycle differ for the three genes. This suggests that the Mas proteins may have different functions apart from their role in flagellar mastigonemes. Monoclonal antibodies (mAbs) specific for PnMasl and PnMas3 were generated to study the localisations and protein-protein interactions of the two putative mastigoneme proteins. On immunoblots, anti-PnMasl and anti-PnMas3 each reacted with a single polypeptide with a relative molecular weight close to that predicted from the amino acid sequence of the protein - 64 kDa for PnMasl and 24 kDa for PnMas3. Immunoblotting of zoospore protein extracts in native gels indicated that PnMasl and PnMas2 both PnMas proteins has provided clues as to their function. The new information resulting from my study will contribute to increase the understanding of Phytophthora zoospore motility and subsequently the mechanism of Oomycete disease dissemination.
Access status: Open Access ,
Characterising Algorithm Debt in Machine and Deep Learning Systems
(2026) Simon, Iko-Ojo
The integration of Machine and Deep Learning (ML/DL) into modern software systems has transformed application domains such as finance, healthcare, and autonomous technologies. However, the complexity of ML/DL algorithms, the unique development pipeline, and the stochastic training process introduce challenges beyond those found in traditional software development, compromising the long-term reliability of ML/DL systems. These challenges, often due to design choices, lead to Technical Debt (TD). Algorithm Debt (AD) in ML/DL systems is a TD type caused by inefficient algorithmic decisions, leading to poor scalability and model degradation. Despite its impact, AD remains underexplored in ML/DL systems due to its novelty, with limited and fragmented studies, creating a knowledge gap. To address this gap, this thesis aims to characterise AD in ML/DL systems. It defines AD, investigates its causes, effects, and mitigation strategies, using a sequential mixed-methods approach. The first study, a Systematic Literature Review (SLR), synthesised evidence from 44 included studies published across IEEE, Springer, ACM, and ScienceDirect. This review produced an extended definition for AD in ML/DL systems and revealed that AD manifests in three ways: poor model scalability, inefficiency, and degradation. The SLR also identified nine ``AD smells'' (i.e., indicators) that lay the foundation for subsequent studies. Next, an experimental study was conducted to explore automated AD detection in code comments using ML/DL approaches. A labelled dataset of self-admitted TD from seven open-source repositories, identified through the SLR, was used to train and evaluate ML/DL classifiers. However, the results were constrained by the limited nature of the available dataset, which was labelled based on pre-existing, narrower definitions of AD. This underscored the inadequacy of current data for understanding AD and justified the subsequent need for comprehensive studies. Following the ML/DL experiments, two studies were conducted to gain a deeper understanding of AD through the real-world experiences of ML/DL practitioners and researchers. The first study involved semi-structured interviews with 21 participants. The second study used a questionnaire completed by 65 respondents. These studies revealed the causes, effects, and mitigation strategies of AD in ML/DL systems. Key findings showed that awareness of AD among practitioners is limited. Also, poor model scalability was identified as the major effect of AD, with continuous practitioner education emerging as a key mitigation strategy. In summary, this thesis provides an enhanced definition of AD and characterises its causes, effects, and mitigation strategies in ML/DL systems. It reveals that AD is a socio-technical phenomenon shaped by both intrinsic ML/DL system properties and engineering decisions. It also introduces a novel, empirically grounded framework for AD through methodological triangulation. The findings underscore the need for continuous practitioner education to address knowledge gaps within the software engineering community, alongside proactive strategies for managing AD. Ultimately, this work lays the foundation for building scalable and reliable AI systems, bridging the gap between research and practice in software engineering. Keywords: Technical Debt, Algorithm Debt, Machine Learning, Deep Learning, Software Engineering, Automated Detection, AD Smells
Access status: Open Access ,
The cultural context and poetic tradition of some of the vernacular lyrics from MS. Harley 2253
(1973) Willetts, Marjorie R.
"King John", wrote A. A.. Milne with unhistorical simplicity, "was not a good man»" Probably not for he was one of the first royal bureaucrats*
Access status: Open Access ,
Some aspects of the biology of Velacumantus australis (Quoy and Gaimard) (Gastropoda: Potamididae)
(1967) Ewers, William
Velacumantus australis (Q,uoy and Gaimard) (Gastropoda : Potamididae) is common in estuaries, coastal lakes and sheltered bays along the eastern coast of Australia, south of the Tropic of Capricorn, along the eastern part of the Victorian coast and the southwest part of the coast of Western Australia.
Access status: Open Access ,
Double Burden of Food Insecurity and Poor WASH Access: Health Impacts on Mothers and Children in Coastal Bangladesh.
(2026) Mondal, Shuvagato
Bangladesh, among the world's most climate-vulnerable nations, faces increasing challenges to ensure food security and universal access to water, sanitation and hygiene (WASH) facilities. This doctoral research investigates the dual burden of household food insecurity and inadequate WASH access and their combined impacts on child and maternal nutritional outcomes in climate-sensitive coastal regions of Bangladesh. Using a multidimensional framework, this study aimed to identify socioeconomic and demographic determinants of food insecurity and WASH access, measure prevalence and examine their combined effects on nutritional health among disadvantaged coastal populations. The study adopted a mixed-methods design, integrating secondary and primary data analyses. Secondary data from Bangladesh Demographic and Health Survey 2017- 18 were analysed to explore the links between socioeconomic inequalities, sanitation and childhood malnutrition. A cross-sectional household survey using a three-stage cluster sampling strategy recruited 471 mother-child pairs for primary data. Quantitative analyses, including multivariable logistic regression and structural equation modelling (SEM), identified individual, household, environmental-level factors, and assessed the dual burden of food insecurity and inadequate WASH access on undernutrition. In addition, qualitative insights from focus group discussions and key informant interviews explored community perspectives, underlying drivers, consequences, self-efficacy and policy implications regarding food and WASH insecurity. A thematic content analysis approach revealed deeper contextual insights into lived experiences and sociocultural factors shaping vulnerability, informing evidence-based, community-focused and context-specific interventions. In coastal Bangladesh, prevalence of child stunting (31.4%) and wasting (8.5%) exceeded national estimates, influenced by age, maternal education, household wealth, sanitation, child morbidity and maternal undernutrition. Food insecurity affected nearly 29% of households, with a majority lacking at least one basic WASH service. Lower chance of experiencing food insecurity was in households with younger head (40 years or less) [Adjusted Odds Ratio (AOR): 0.42] compared to older group, having educated mothers (AOR: 0.22) compared to uneducated and higher monthly income (AOR: 0.09) compared to lower income. WASH access was linked to household size, wealth, location, head's occupation and education. Poorer households had significantly lower odds (AOR: 0.24) of safe water access than the wealthiest, while access was higher in southeastern coastal part (AOR: 1.67) than those in southwest. Larger households (AOR: 0.61) and fishermen headed (AOR: 0.30) were less likely to have improved sanitation access than smaller and headed by service-holders. Lower adults' educational attainment was associated with reduced access to basic hygiene and combined WASH services. Among coastal communities, 54.4% of children were stunted, 25.2% underweight and 9.4% wasted, with 61.6% exhibiting at least one anthropometric failure. Among mothers, one-third were obese and 34.5% anaemic. Child malnutrition was strongly associated with younger age, lower household wealth, food insecurity, poor water and sanitation access, maternal underweight and recent occurrence of diarrhea. SEM demonstrated that combined effect of food security and WASH access had a larger positive effect on both child's and maternal better nutritional outcomes, with standardised coefficient values of 0.50 and 0.68, respectively. This thesis highlights the urgent need for integrated, multisectoral, and context specific policy interventions to combat malnutrition in coastal Bangladesh. It provides a robust foundation of evidence to support equity-focused health and nutrition strategies that tackle the interconnected issues of food insecurity and inadequate WASH access among geographically vulnerable communities.
Access status: Open Access ,
Characterisation of the SIX6 effector protein from Fusarium oxysporum f. sp. lycopersici
(2026) Khambalkar, Pravin
The phytopathogenic fungus Fusarium oxysporum f. sp. lycopersici (Fol) causes vascular wilt disease in tomato. During host colonisation, Fol secretes multiple effector proteins into the xylem sap, including fourteen SIX (Secreted In Xylem) proteins. Three of these SIX proteins function as the avirulence determinants Avr1 (SIX4), Avr2 (SIX3) and Avr3 (SIX1) recognised by tomato resistance proteins I, I-2 and I-3, respectively. Despite identifying the I-7 resistance gene, which encodes a leucine-rich repeat receptor protein, its corresponding avirulence gene remains undetermined. This work initially aimed to identify Avr7 by evaluating untested SIX genes and novel effector candidates. Effector candidates from transcriptomic analysis of infected tomato roots were assessed via Agrobacterium-mediated co-expression with I-7 in Nicotiana benthamiana. I-7-dependent leaf cell death would indicate recognition. None of the examined candidates exhibited I-7-dependent cell death, but SIX6 induced cell death independently of I-7 (Chapter 2). This observation prompted experiments to characterise SIX6 function in planta. In collaboration with Daniel Yu from the Williams Laboratory at ANU, Fol SIX6 was expressed in E. coli and recombinant protein purified for functional analyses. Purified recombinant Fol SIX6 induced cell death in N. benthamiana and tomato (Chapter 3). Subsequently, SIX6 homologues from F. oxysporum ff. spp. cubense TR4, melonis and vasinfectum (Foc, Fom and Fov SIX6) were expressed and purified, and found to induce cell death responses in N. benthamiana and tomato (Chapter 3). Testing purified Fol, Foc and Fom SIX6 proteins in foliar tissues of representatives from multiple taxonomic families (Solanaceae, Cucurbitaceae, Brassicaceae and Leguminosae) revealed broad yet differential plant sensitivity (Chapter 3). Notably, Fol and Fov SIX6 induced stronger cell death in the corresponding tomato and cotton hosts than vice versa. VIGS of known plant defence-signalling components in N. benthamiana was used to test whether Fol SIX6-induced cell death might constitute a plant immune response. VIGS targets included pattern recognition co-receptors BAK1 and SOBIR1, and mitogen-activated protein kinases WIPK and SIPK. VIGS-mediated depletion of these signalling components failed to attenuate Agrobacterium-mediated Fol SIX6-induced cell death (Chapter 4). These experiments were complemented by challenging mutant tomato lines compromised in defence signalling with Fol SIX6 protein. The ethylene-insensitive Never Ripe mutant and sun1-1 mutant deficient in EDS1-mediated resistance exhibited cell-death responses comparable to wild-type controls following SIX6 infiltration (Chapter 4). These findings suggested SIX6 induced cell death through a mechanism distinct from these plant defence pathways. A microscopic investigation of cellular responses leading to SIX6-induced cell death revealed that Fol SIX6 protein causes stomatal opening in tomato and N. benthamiana before visible cell death (Chapter 5). Physiological measurements demonstrated that Fol SIX6 infiltration significantly increased stomatal conductance and transpiration rates in N. benthamiana leaves, regardless of light conditions (Chapter 5). Confocal fluorescence microscopy following Agrobacterium-mediated expression of a vacuole-targeted fluorescent protein revealed that SIX6 treatment caused vacuolar shrinkage and leakage in N. benthamiana leaf epidermal cells (Chapter 5). An examination of the effect of Fol SIX6 on N. benthamiana and tomato roots showed that Fol SIX6 protein caused electrolyte leakage (Chapter 6). In vitro growth assays on medium supplemented with purified Fol SIX6 protein resulted in significant inhibition of primary root elongation in tomato seedlings (Chapter 6). These findings suggest SIX6 exerts physiological effects on root tissues that may contribute to Fol pathogenicity during infection.

Browse

Recent Submissions