School of BioSciences - Theses
Permanent URI for this collection
Search Results
1 - 10 of 34
-
ItemEffects of temporally heterogeneous stress on individual marine invertebrates( 2021)Some environmental conditions may be stressful, adversely affecting the growth and reproduction of organisms. The impact of stress, however, may vary with its (1) timing, relative to an individual’s stage of life, (2) duration, or the total length of stress, and (3) frequency, or the number of times an individual is exposed. I used the colonial bryozoan Watersipora subatra and the solitary ascidian Styela plicata to study the response of individuals to stress that varies in timing, duration and frequency. I exposed these invertebrates to stress using elevated concentrations of trace metals, W. subatra to copper in the field, and S. plicata to cadmium and zinc via seawater or food in the laboratory. To expose W. subatra to variable stress, and assess their growth and reproduction, I applied stress (1) at different times within a single life-history stage (early reproductive maturity), (2) that varied in duration at different life-history stages (juvenile, young adult or mature adult), and (3) that differed in duration and frequency for adults at two field locations. To assess the uptake and loss of metals with repeated exposure, I exposed S. plicata twice to cadmium and/or zinc in seawater. Individuals performed differently within a life-history stage. When stressed at the onset of embryo production, fewer adult colonies reproduced, but when stressed as adults in the later stages of brooding or producing subsequent bouts of offspring, colonies decreased in size (fragmented) and released fewer larvae compared with unstressed colonies. Generally, juveniles, young adults and mature adults responded according to life-history stage, not the nature of the stress. Young adults produced more embryos than mature adults, whilst mature adults produced smaller larvae with greater settlement success than young adults. Once reproductive, juveniles produced fewer embryos and were less likely to release larvae than both young and mature adults. The duration of stress affected individuals’ reproduction and, sometimes, growth. Regardless of life-history stage, colonies grew little or fragmented when stressed for longer (2 weeks). Adult colonies at one field location produced fewer larvae when stressed for a shorter period (1, 2 weeks), whilst colonies at a second field location were less likely to release offspring and produced fewer larvae when stressed for longer (2, 3 weeks). When the frequency of exposure differed, colonies altered the quality of larvae. Colonies stressed once (cf. multiply) for two weeks, produced smaller larvae with lower settlement success at one field location, whilst similarly sized, yet better settling, larvae at a second field location. S. plicata accumulated more metal after the second (cf. first) dissolved exposure, and more Cd in the presence of Zn than when exposed singly via sea water. This thesis demonstrates that the temporal nature of stress is important for determining an individual’s response – their growth and reproduction, and future success of their offspring. However, the effects of stress for a given individual may depend on the stress’ temporality – timing, duration, frequency – and conditions at a specific location and time.
-
ItemNo Preview AvailableExploring cis-regulatory convergence between the extinct thylacine and the gray wolf( 2021)Models of convergent evolution allow us to examine the evolutionary processes that drive phenotypic repetition in nature. Investigating the way convergent traits evolve at the genome level is important for identifying the molecular mechanisms that regulate this kind of adaptation. One of the best models for morphological convergent evolution in mammals is that of the marsupial Tasmanian tiger (thylacine; Thylacinus cynocephalus; Thylacinidae) and the eutherian group Canidae (Carnivora), in particular the gray wolf (Canis lupus). These species have evolved striking similarities in craniofacial morphology as a result of a shared ecological niche. However, the precise regions of the genome controlling the evolution and development of convergent skull shape in the thylacine and canid species are unknown. Clues can be found in the non-coding portion of the genome, where regions under accelerated evolution for both species have previously been described (thylacine-wolf ARs). It is hypothesised that the molecular landscape underpinning this remarkable convergence in skull shape can be found here, in the cis-regulatory elements of the non-coding genome. In this thesis, I use computational, developmental and molecular methods to investigate the role of candidate loci in convergent craniofacial evolution for the thylacine and the gray wolf. Because the thylacine is extinct, my first steps were to develop the fat-tailed dunnart (Sminthopsis crassicaudata) as an alternative model for investigating the evolution and development of craniofacial structures in carnivorous marsupials. Importantly, I showed that ossification has not yet begun for newborn dunnart pouch young, highlighting shifts in the developmental timing of craniofacial structures when compared to eutherian models such as the mouse. I then defined the epigenetic landscape of craniofacial tissue for dunnart pouch young at birth, for the purposes of exploring putative cis-regulatory elements as potential drivers of craniofacial development. Comparisons with epigenetic data for mouse embryonic craniofacial tissues showed that the dunnart overlapped to some extent with all murine stages, further emphasising the rapid development of craniofacial structures in the dunnart compared to the mouse. I next assessed whether thylacine-wolf ARs show the same active marks of cis-regulatory activity that I identified in the dunnart. Interestingly, many thylacine-wolf ARs can be seen to overlap with H3K4me3 and H3K27ac enriched regions in the dunnart, providing further evidence for their potential role in convergent craniofacial evolution. Taking a candidate loci approach, I tested a single thylacine-wolf AR with active markers in both the dunnart and mouse in a lacZ reporter study. This revealed that a putative thylacine enhancer of Satb2 drove strong reporter signalling in key craniofacial prominences throughout embryonic development in the mouse. Finally, in order to assess the regulatory capacity of all thylacine-wolf ARs, I designed and optimised a Massively Parallel Reporter Assay (MPRA). My pilot study informed the feasibility of using this high-throughput approach, but also revealed convergent changes in reporter expression for the thylacine-wolf AR upstream of Satb2. Altogether, this thesis deepens our understanding of the role of cis-regulatory convergence in driving phenotypic convergence and contributes fundamental resources for future comparative studies between marsupials and eutherians.
-
ItemAn exploration of the impact of drug resistance mutations on the development of Plasmodium berghei throughout the parasite life cycle( 2021)Drug resistance in Plasmodium spp. significantly impedes malaria control and eradication efforts. Recent emergence of resistance to front line antimalarial treatments, such as artemisinin, means new strategies are needed to counter the spread of resistance and overcome the evolutionary agility of the malaria parasite. Malaria parasites undergo immense changes in metabolic activity between their vertebrate and mosquito life cycle stages. Consequently, drug selection for resistance during the vertebrate stage can have dire consequences for parasite fitness when it transitions to insect stages. We investigated drugs with targets likely to be under markedly different selection pressures between the two hosts to identify resistance mutations that could drive parasite failure in the mosquito, thereby blocking transmission. This works with the mitochondrion-encoded cytochrome b inhibitor atovaquone, where resistance is trapped in the host. We hypothesised that the same trap would hold true for drug targets encoded by the apicoplast, a parasite compartment homologous to the plastids of plants and algae that is very likely under differential selection between mammalian and insect life cycle stages. We selected for strains of the mouse malaria model Plasmodium berghei resistant to antimalarial drugs inhibiting apicoplast translation. We generated clonal strains resistant to azithromycin (PbAZMR) and clindamycin (PbCLINDR) but were unsuccessful in generating strains resistant to doxycycline. PbAZMR parasites carried mutations in the apicoplast-encoded 50S ribosomal protein L4, which also confer resistance to this class of drugs in P. falciparum, bacteria, and algae. PbCLINDR parasites had novel, independent mutations in an apicoplast translation pathway not previously linked to clindamycin resistance. Phenotype analysis of PbAZMR and PbCLINDR parasites across the life cycle showed that resistant parasites largely produce viable gametes and can infect mosquitoes. However, both PbAZMR and PbCLINDR parasites produced fewer midgut oocysts, and their oocysts developed aberrantly. Both PbAZMR and PbCLINDR parasites produced fewer salivary gland sporozoites, and—surprisingly—PbAZMR sporozoites lacked an apicoplast. Both PbAZMR and PbCLINDR parasites had severe impairments in their ability to transmit to naive mice, suggesting the resistance mutations had imposed cumulative fitness deficits in both mosquito and liver stages that reduced transmissibility. Our results suggest that the resistance trap initially discovered for mitochondrial gene targets could also work for apicoplast drug targets. Although apicoplast translation inhibitors are little used in therapy because the delayed death phenomenon (where parasite killing only occurs after a full asexual intraerythrocytic developmental cycle) makes them unsuitable for curing seriously ill patients, our demonstration that resistance to these drugs is poorly transmitted makes them of renewed interest as partners to protect fast killing frontline drugs. Combination therapies are better able to constrain resistance, and the safe, cheap, apicoplast inhibiting antimalarials could protect the efficacy of the primary compound and alleviate the spectre of combinatorial use selecting for multi-drug resistance, as has happened previously with sulfadoxine/pyrimethamine and artemisinin-based combination therapies.
-
ItemLife in our shadow: the effects of anthropogenic light and noise on ecological communities( 2021)Life on earth has evolved under a relatively constant cycle of light and dark that varies annually. This variation in light triggers physiological and behavioural processes at the level of the individual, with cascading effects for ecosystem structure and function. Increasing urbanisation and associated anthropogenic stressors are challenging these natural processes. Artificial light at night (ALAN) and anthropogenic noise are powerful disruptors of biological processes at the individual and community scales. Both can mask natural signals and drive shifts in behaviour, physiology and phenology. In this thesis, I examined the impact of ALAN and, to a lesser extent, urban noise on individuals and their communities. I used a combination of observational and experimental approaches to examine these effects across four unique ecological systems. First, in a series of observational studies, I compared the effects of natural light, ALAN and noise on the spatial distribution and acoustic timing of animal communities in a major city. Focussing on shifts in human activity due to daylight, I examined explicitly the relative effect of traffic noise (shifts abruptly in time) and natural lighting (no abrupt shift) on dawn and dusk soundscapes in bird communities. I found that frequency or amplitude shifts were exploited more than temporal shifts as strategies to reduce interference from traffic noise. I then explored how an urban bird, the bell miner (Manorina melanophrys), responds to both ALAN and natural variation in nocturnal light. Birds delayed foraging activity as moonlight increased, but this response was masked under high levels of ALAN, which also increased total foraging effort. Finally, I explored how ALAN altered the spatial distribution and composition of invertebrate communities. Variation in the intensity, proximity and spectrum of lighting differentially impacted spatially distinct invertebrate assemblages. In my final three data chapters, I experimentally illuminated Eucalyptus mesocosms to explore the individual- and community-level effects of introducing ALAN into a naive ecological community consisting of the river red gum (E. camaldulensis), the red gum lerp psyllid (Glycaspis nr brimblecombei, a specialised herbivore of E. camaldulensis), and other non-target fauna. I found ALAN promoted night-time photosynthesis and shifted leaf/root investment and leaf morphology, but this did not result in photosystem adaptation, or compromise tree water status or drought response. ALAN also promoted the sequential colonisation by various invertebrate taxa and frogs, and disrupted anuran and avian choruses in nearby habitat. Experimental transplantation of psyllids onto red gum saplings revealed that nymphs exposed to ALAN may shift feeding sites more frequently, resulting in the production of more lerps (carbohydrate-rich crystalline secretions). ALAN, and to a lesser extent noise, impacted a range of individual-, species- and community-level processes in urban wildlife. Many of these effects are likely to cascade through trophic interactions or by increasing competition in already crowded spatial, acoustic and temporal niches. My work adds nuance to the growing evidence that ALAN and noise adversely impact individuals and communities, and expands our knowledge of ALAN effects on tree physiology and trophic webs in Eucalyptus woodlands. My findings have important consequences for how we measure and address the effects of artificial light, since ALAN impacts are likely to differ across spatially, functionally and taxonomically diverse natural communities.
-
ItemTrawling for the unknown: An investigation of the impacts of harvest and oceanic warming on the life-histories of fish( 2021)Fishing is globally important as humans rely on fish for 17% of their total protein intake and millions of people are directly or indirectly dependent on fisheries as a source of income. Fisheries harvests do, however, impose a number of impacts on fish populations. Harvest can cause demographic truncation (loss of large and old individuals), as well as reducing the reproductive capacity of populations and modify ecological interactions. Furthermore, fishing is inherently size-selective, where the preferential catch of certain phenotypes using different gears can combine with relatively high rates of fishing mortality to drive potentially rapid evolutionary responses in populations (called fisheries induced evolution or ‘FIE’). As a result of fishing, many harvested stocks are expressing a faster ‘pace-of-life’, characterised by faster growth, earlier maturation, and increased relative reproductive investment. Concurrent to fishing activity, the world’s oceans are warming at unprecedented rates which is having significant impact on the physical environment experienced by fish. Importantly, warming is thought to drive a similar increase in the pace-of-life expressed by fish via direct physiological and indirect temperature-dependent processes. This can be seen in almost ubiquitous shifts to faster growth, earlier maturity and smaller adult size, a phenomenon called the temperature size rule (TSR). Warming is also impacting fish populations through factors such as increased mortality, range shifts and increased incidence of hypoxic conditions. Our understanding of how the effects of harvest and warming may combine to impact on fishes, and influence their subsequent recovery after initial impact, is poor. An improved understanding of the combined impacts of fishing and warming oceans and the mechanisms underpinning these, as well as heuristic, cost effective ‘monitoring’ tools are needed if we are to properly manage fish stocks into the future. In this thesis, I tested a recently published biphasic growth model that attempts to estimate age at maturity from commonly collected size-at-age data (Chapter 2). Statistical techniques such as these could be utilised as cost and time-effective fishery monitoring tools. I used data from multiple North Sea stocks to parameterise the biphasic model and found that the method performed poorly given the data. This poor performance was primarily due to the lack of temporal continuity and representation of older individuals which decreased the ability of the technique to model growth and thus estimate maturity. I suggest that future fisheries monitoring should include improved data collection across all age and size classes so that we can adequately utilise potentially cost-effective statistical techniques to estimate key life-history parameters. I then ran a multi-generational selection experiment that applied interactive treatments of warming (26 C ‘control’ and 30 C ‘warmed’) and realistic fisheries selection (random harvest ‘control’, ‘sigmoidal’ and ‘Gaussian’ selection) to 18 populations of the tropical freshwater zebrafish (Danio rerio). My sigmoidal treatment was designed to mimic trawl fisheries, where the largest individuals are harvested, and my Gaussian selection treatment simulated gillnet or slot limited fishing, where the midrange of sizes were affected by fishing. Selection was applied to six generations, where I followed responses across these ‘F’ generations as well as two common garden generations where selection was relaxed. Studying responses under common garden conditions reveals information as to the nature of responses, where plastic, maternal and evolutionary processes can be teased apart. In Chapter 3, I tracked the expression of early life-history traits in populations exposed to fishing and warming and found that recruitment collapsed after four generations of warming treatment. Moreover, temperature interacted with fishing such that sigmoidal harvested populations had the lowest recruitment rate. This interaction indicates that the removal of large fish individuals through harvesting can exacerbate warming-induced recruitment collapse. Preserving size diversity in fish populations is important to help increase the resilience of fisheries to the impacts of warming. In Chapter 4, I tested two alternative explanations of the temperature size rule (TSR), which describes the phenomena of increased juvenile growth, early maturity and smaller adult sizes at higher temperatures in ectotherms. Debate centres around whether TSR is the result of warming-induced physiological constraints on growing to larger sizes, or an adaptive outcome stemming from temperature-induced energy allocations. I found evidence that the metabolic rates of fish held at elevated temperatures acclimated after three generations. This result indicates that the commonly proposed physiological limitation mechanism could not solely explain TSR. Instead, I found evidence that populations adjusted reproductive investment via earlier maturation and higher early investment in reproduction, suggesting that TSR could be explained by shifts in resource allocation. These results have implications for models used to predict warming impacts in our oceans. Finally, I analysed trends in the expression of multiple juvenile and adult life-history traits through generations (Chapter 5). Here I found that warming and sigmoidal fishing treatments led to the largest reductions in adult body sizes. In contrast, fisheries selectivity that preserved large individuals allowed warmed population to maintain body sizes similar to those in the control temperatures. Temperature impacts on life-history traits recovered rapidly in common garden generations, whereas the legacy of fisheries selection remained. These findings suggest that even five generations of realistic fishing pressure can have long term impacts on populations, affecting their response to management interventions. Together, my work advances our understanding of how harvested populations respond to significant contemporary and future stressors and provides valuable insights into the future sustainable management of fisheries resources.
-
ItemThe evolution of life-history, dispersal, and plasticity: expanding the Daphnia model( 2021)It is often noted that, in the face of accelerating environmental change, species must adapt or disperse in order to avoid extinction. Increasingly, however, it is recognised that we might need to refrain from seeing these concepts as independent. The integration of dispersal as a plastic, evolving trait within the framework of life-history is critical to our understanding of evolution in changing environments. Through dispersal an individual’s environment can change, but so too does the environment change an individual’s dispersal decisions. While the role of dispersal has gained appreciation, the drivers and genetic underpinnings of dynamic dispersal have only begun to be explored. Currently, there are few models which explore the spatial facet of life-history evolution. This thesis details the key first steps in the development of a spatially explicit model for life-history evolution using the Australian water flea, Daphnia carinata. This process began with the collection of Daphnia clones from sites across south-eastern Australia to establish a laboratory population for experimentation, phenotyped for life-history traits. The effect of natural selection on traits is heavily influenced by the genetic and environmental relationships of correlated traits and this is reflected in locally adapted life-history strategies. I therefore use a lens of habitat permanency to explore the life-history strategies of Daphnia from temporary and permanent habitats to test for evidence of local adaptation. Using classic life-table experimentation and analysis I find that D. carinata clearly demonstrate life-history strategies that covary with features of habitat permanence. I analyse the environmental robustness of these life-history strategies by using a multivariate reaction norm approach and phenotypic trajectory analysis of plastic responses to two different stressors: food restriction and predation threat. The plasticity employed by organisms to navigate the variation in their local habitat may play a critical role in further evolution of adaptive responses to novel or extreme environments. When plasticity is in the direction of selection this may have consequences for the rate and trajectory of evolution, but identifying if plasticity is adaptive is not straightforward. With a multi-environment, multivariate approach I assess whether plasticity in this population shows evidence of a stressor specific response, and whether the populations demonstrate genetic variation in multivariate plasticity. I find that this population does not demonstrate a general response to the two stressors and that nature of plasticity also varies between habitat groups. Finally, I establish methods for measuring passive and active dispersal in Daphnia. I calculate clonal repeatability as a measure of broad-sense heritability to determine the suitability of a suite of dispersal measures to the relationship of dispersal with other life-history traits. Daphnia display high levels of intraclonal variation in measures of active dispersal, indicative of the diverse environmental drivers of this complex trait. I find that the effect of the environment on aspects of dispersal, from decision to disperse to distance travelled, often leads to low repeatability estimates. However, this suggests that the tools I have developed in measuring adaptive plasticity will be useful in studying the spatial aspect of life-history in this system.
-
ItemPhylogeography of Box Eucalypts with Disjunct Distributions in South-East Australia( 2021)Despite Australia generally being considered a continent of ancient and stable landscapes, the south-east has experienced major environmental changes during the Pliocene and Pleistocene. In addition to the slow continent-wide decline in rainfall and several global glacial-interglacial cycles, there have been more localised events that have impacted the vegetation of the region including marine inundation of the Murray Basin, uplift of the Padthaway High that dammed the Murray River and formed the large Lake Bungunnia, and volcanic activity in the Newer Volcanics Province of western Victoria and south-eastern South Australia. A record of the impacts of these events on the distribution of plants has been left in the patterns of genetic diversity and relatedness across the landscape. The eucalypts, >850 species of shrubs and trees in the genera Angophora, Corymbia and Eucalyptus, are one of the most widespread and species rich groups of woody plants in Australia, and almost define what is thought of as the ‘bush’. They occur in habitats ranging from wet forests of the east coast and south-west corner of the continent all the way through the climatic gradients to the inland deserts. In my thesis I focus on E. sect. Adnataria, one of the most diverse groups within the eucalypts with ~120 species, and, in south-eastern Australia, the species it contains dominate large areas of open forest and woodlands. This makes members of E. sect. Adnataria good candidates on which to undertake phylogeographic studies to build understanding of the biogeographical history of southeastern Australia, as I have done in this thesis. In chapter 2 I explore the phylogeography of E. behriana using ddRADseq data, showing that the populations of the species west of the lower Murray Basin were isolated earlier than those to the east. This is hypothesised to be related to the marine inundation and formation of Lake Bungunnia in the Murray Basin over the last several million years. The isolation of the populations at Long Forest and around West Wyalong are shown to have recently experienced geneflow with the larger populations of the Victorian Goldfields, Wimmera, and Murray Mallee in the east. This more recent isolations of eastern populations are hypothesised to be due to climate changes under the global glacial cycle, with a possible contribution from recent volcanic activity in the Newer Volcanics Province. I set out to resolve the relationships and taxonomy within the E odorata complex in chapter 3, however I didn’t manage to achieve this using both ddRADseq and DArTseq, with many relationships remaining unresolved, and many questions remaining unanswered. What I show is that most E. viridis populations, E. aenea and E. castrensis form a lineage sister to the remained of the clade, which includes an E. viridis population previous described as E. viridis var. latiuscula from south-eastern Queensland. I also show hybridisation is common within the complex and with taxa outside it, especially the closely related Grey Boxes, which has contributed to the diversity of morphology in the group. Chapter 4 returns to focus on E. behriana, as in this chapter I present a case study on the pitfalls and flaws in using plastid DNA for phylogeography in the absence of substantial outgroup sampling. I show how, by analysing the ingroup only, a geographically sensible and well resolved pattern of population relationships can be established. However, by drastically expanding my sampling to include extensive representation of co-occurring outgroups, I show these patterns become meaningless due to the large cyto-nuclear discrepancy in the eucalypts. A further phylogeographic study is presented in chapter 5, that of E. baueriana. This taxon occurs in coastal regions of south-east Australia, and I show there is a deep divergence between a south-western lineage and a north-eastern one located in eastern Gippsland. While the cause of this disjunction is not clear, we hypothesise it is related to the occurrence of gallery rainforests in the region that prevent E. baueriana form colonising it’s preferred riparian habitat. The fragmentation of the distribution of the two main lineages is hypothesised to be related to glacial-interglacial climatic cycles; in particular, the changes they cause to sea-levels.
-
ItemGroup contests in social insects: contested resources, contest outcomes and escalation( 2021)Theoretical models reveal that animal contests are typically resolved by signalling displays rather than fatal and physical fights and will escalate with the value of contested resources. These predictions have been confirmed in numerous studies that focus primarily on dyadic contests, involving two individuals. However, relatively few studies have explored their relevance to group contests, which are not uncommon in group-living animals, including primates, birds and social insects. My PhD thesis focuses on group contests in social insects, using the Australian meat ant Iridomyrmex purpureus as a model system to investigate the nature of these group contests based on theoretical predictions. In Chapter 1, I provide a synthetic review of group contests. In particular I discuss two issues associated with group contests: 1) what determines the outcome of group contests; 2) when do group contests escalate. At the end, I discuss how I addressed these two questions in social insects and why I. purpureus is a great model system for these kinds of investigations. In Chapter 2, I reveal that display grounds do not delimit borders that define exclusive territories in I. purpureus. Rather, the proportion of workers from a focal colony found in a quadrat declines monotonically with distance from the nest. In addition, I documented collective displays around food trees, where workers congregated in greater densities and engaged in more aggressive behaviour. These results refute the assumption that colonies of I. purpureus establish territorial boundaries by collective displays. Rather these collective displays may be related to the defence of specific resources, including food trees and nest sites. In Chapter 3, I investigated whether the aggressive level of displaying behaviour of I. purpureus is associated with the alarm pheromone that is released by displaying workers. My field experiment, in which displaying workers were exposed to a synthesised alarm pheromone 6-methyl-5-hepten-2-one, provided no evidence that the level of aggression was moderated by alarm pheromone. Subsequent field experiments revealed that the pheromone functions as an attractant, thereby increasing the density of displaying workers. More densely populated workers also display more aggressively, indicating that the interaction rate of displaying workers may determine the level of aggression in collective displays. In Chapter 4, I draw on insights from Lanchester’s Linear law to design novel field experiments that investigate the outcome of group contests involving neighbouring colonies of I. purpureus. First, I find that body size, a taxonomically widespread proxy of fighting ability, does not influence the outcome of contests between pairs of workers. Second, I show that the outcome of paired colony-level contests over food resources is determined by colony size, not average worker body size, with the numerically larger colony gaining exclusive access to the food, or neither colony obtaining exclusive access if they are matched in colony size but differ in body size. The results help explain why colonies of I. purpureus persistently deploy numerous workers to display grounds that apparently functions to assess the colony size of their neighbouring colonies, which may represent the group RHP. In Chapter 5, I investigated whether neighbouring colonies of I. purpureus escalated group contests when competing over food resources that differed in value to the colony. I used insights from the Geometric Framework to design a field experiment that manipulates the value of the food resource to the colony. The initial field experiment revealed that colonies deployed more foragers to a protein food source if they had been previously provided with carbohydrates than with protein, suggesting that the value of protein to the colony depends on their dietary history. I then compared the level of escalation in contests when the value of the food was high (pre-feeding with carbohydrate) or low (pre-feeding with protein). Consistent with theory, I discovered that colonies deployed more workers when contesting high than low value food and, importantly, the individual workers contesting high value food were more likely to physically grapple, a behaviour that frequently results in a fatality to either or both of the workers. These data reveal that dyadic contest theory is transferrable to group contests, and, in social insects, the behaviour of individuals is apparently derived from colony-level decisions. To conclude, my thesis provides insights, derived from field experiments, into our understanding of the nature of group contests and the evolution of animal contests in social insects.
-
ItemApproximate Bayesian inference for large-scale hierarchical modelling( 2021)Bayesian hierarchical models are widely used in applied statistics. Their main benefit is the ability to borrow strength across similar but distinct units, such as competitors in sports or species in ecology. However, the methods used to fit them are computationally intensive, often limiting their use to small datasets. This is problematic because large datasets are becoming increasingly common. This thesis develops approximate Bayesian inference techniques to efficiently fit hierarchical models to large datasets. It is comprised of three studies. Two studies focus on species distribution modelling (SDM). Here, the aim is to predict the spatial distribution of species by relating suitability of locations to environmental covariates, such as climate variables. The third study develops a dynamic paired comparison model for accurately ranking competitors in a sport over time. I apply hierarchical models in all three studies, using them to borrow strength across species for SDM, and across time, players and playing surfaces in tennis. All models involve a large number of latent variables, and I develop approximations to the exact Bayesian posterior to fit them efficiently. In SDM, the two chapters focus on modelling non-linear functions and interactions (Chapter 2) and on accounting for the data collection process (Chapter 4), respectively. In both cases, I develop techniques to scale existing models -- multi-output Gaussian processes and multi-species occupancy detection models, respectively -- to datasets that contain orders of magnitude more records than previous work. In Chapter 3, I develop an approximate Bayesian rating system for tennis. This rating system generalises the popular Elo rating system and is able to fit tens of thousands of data points quickly. In all chapters, I evaluate the predictive performance of the models and show that they outperform competing approaches. As datasets grow in size, it is important that inference methods can make the most of them. The results in my thesis show that hierarchical models fitted using approximate inference can be both fast and make accurate predictions, demonstrating that they are a compelling approach for fitting large datasets.
-
ItemThe phenology of outbreaks of the Australian plague locust, Chortoicetes terminifera (Walker), affecting Victoria.( 2021)This thesis examines the phenology of outbreaks of the Australian plague locust, Chortoicetes terminifera, which have affected the state of Victoria over the period from the mid 1930s to the current time. Victoria is the southern-most limit of the species’ occurrence in eastern Australia but there are episodic outbreaks of it which may cause significant economic damage to agriculture. C. terminifera does not usually survive in high populations in Victoria for more than a few seasons and reinfestation is usually due to migration from the northern states of NSW and Queensland in eastern Australia where it persists more continuously. The key research question investigated in the project asks what determines whether C. terminifera outbreaks will, or will not, affect Victoria when source populations exist in other states? The project explores whether there are identifiable patterns of outbreaks and recessions of C. terminifera in Victoria and seeks to identify the influencing factors such as climate, weather and control interventions in northern source areas which can affect migration to Victoria. Chapter 1 provides a general review of the literature pertaining to locusts, summarising their biology, behaviour and effects and Chapter 2 provides a more detailed review of the Australian plague locust, C. terminifera. These provide necessary context for the data collection, analyses, and conclusions in later chapters. Chapter 3 examines the long-term temporal phenology of C. terminifera outbreaks and recessions affecting Victoria utilising recent and historic data. A long-term phenology of outbreaks of C. terminifera in Victoria is constructed and compared to the overall eastern Australian situation. Chapter 4 examines the phenology of an individual outbreak cycle affecting Victoria using data from the 2009/2010/20011 outbreak which is the most recent major outbreak affecting the state. It identifies influencing factors important to migration into Victoria and the subsequent development and decline of C. terminifera populations in Victoria. Chapter 5 develops methodologies to examine the effects of two critical influencing factors, temperature and wind, which affect locust migration to Victoria. It identifies available data sources which can be used to investigate these in the case studies examined in Chapter 6 which considers several migrations to Victoria and what influences whether C. terminifera outbreaks will, or will not, affect Victoria when source populations exist in other states. Chapter 7 summarises the findings, discusses their significance in achieving the project's objectives, and discusses opportunities for further research. It concludes that whilst climate change is likely to alter the suitability of southern NSW and Victoria for population development and survival as well as the frequency of plague events, there is a continuing need for coordinated surveillance, control capabilities, training, and research and development resources to be maintained.