Graduate Theses & Dissertations

Pages

Time Series Algorithms in Machine Learning - A Graph Approach to Multivariate Forecasting
Forecasting future values of time series has long been a field with many and varied applications, from climate and weather forecasting to stock prediction and economic planning to the control of industrial processes. Many of these problems involve not only a single time series but many simultaneous series which may influence each other. This thesis provides methods based on machine learning of handling such problems. We first consider single time series with both single and multiple features. We review the algorithms and unique challenges involved in applying machine learning to time series. Many machine learning algorithms when used for regression are designed to produce a single output value for each timestamp of interest with no measure of confidence; however, evaluating the uncertainty of the predictions is an important component for practical forecasting. We therefore discuss methods of constructing uncertainty estimates in the form of prediction intervals for each prediction. Stability over long time horizons is also a concern for these algorithms as recursion is a common method used to generate predictions over long time intervals. To address this, we present methods of maintaining stability in the forecast even over large time horizons. These methods are applied to an electricity forecasting problem where we demonstrate the effectiveness for support vector machines, neural networks and gradient boosted trees. We next consider spatiotemporal problems, which consist of multiple interlinked time series, each of which may contain multiple features. We represent these problems using graphs, allowing us to learn relationships using graph neural networks. Existing methods of doing this generally make use of separate time and spatial (graph) layers, or simply replace operations in temporal layers with graph operations. We show that these approaches have difficulty learning relationships that contain time lags of several time steps. To address this, we propose a new layer inspired by the long-short term memory (LSTM) recurrent neural network which adds a distinct memory state dedicated to learning graph relationships while keeping the original memory state. This allows the model to consider temporally distant events at other nodes without affecting its ability to model long-term relationships at a single node. We show that this model is capable of learning the long-term patterns that existing models struggle with. We then apply this model to a number of real-world bike-share and traffic datasets where we observe improved performance when compared to other models with similar numbers of parameters. Author Keywords: forecasting, graph neural network, LSTM, machine learning, neural network, time series
Population Genetics and Gut Microbiome Composition Reveal Subdivisions and Space Use in a Generalist and Specialist Ungulate
Natural populations are often difficult and costly to study, due to the plethora of confounding processes and variables present. This is of particular importance when dealing with managed species. Ungulates, for example, act as both consumers and prey sources; they also provide economic benefit through harvest, and as such, are of high ecological and economic value. I addressed conservation and management concerns by quantifying subdivision in wild populations and combined movement with non-invasive sampling to provide novel insight on the physiological drivers of space use in multiple species. This thesis explored biological patterns in ungulates using two distinct approaches: the first used molecular genetics to quantify gene flow, while the second examined the relationship between movement and the gut microbiome using high-throughput sequencing and GPS tracking. The goal of the first chapter was to quantify gene flow and assess the population structure of mountain goats (Oreamnos americanus) in northern British Columbia (BC) to inform management. I used microsatellites to generate genotype data and used a landscape genetics framework to evaluate the possible drivers behind genetic differentiation. The same analyses were performed at both a broad and fine scale, assessing genetic differentiation between populations in all of northern BC and in a case management study area northeast of Smithers BC. The results indicated panmixia among mountain goats regardless of scale, suggesting distance and landscape resistance were minimally inhibiting gene flow. Therefore, management at local scales can continue with little need for genetically informed boundaries, but regulations should be tailored to specific regions incorporating data on local access and harvest pressure. My second chapter aimed to determine the extent to which the gut microbiome drives space-use patterns in a specialist (mountain goat) and generalist (white-tailed deer, Odocoileus virginianus) ungulate. Using fecal samples, we generated genomic data using 16S rRNA high-throughput sequencing to evaluate gut diversity and gut microbiome characteristics. Additionally, individuals were fitted with GPS collars so that we could gain insight into movement patterns. Gut microbiome metrics were stronger predictors of space use and movement patterns with respect to home range size, whereas they were weaker predictors of habitat use. Notably, factors of both the gut microbiome and age of a given species were correlated with changes in space use and habitat use. Ultimately, this research linked high-throughput sequencing and GPS data to better understand ecological processes in wild ungulates. Author Keywords: gene flow, genomics, gut microbiome, home range, population genetic structure, ungulates
Comparative efficacy of eDNA and conventional methods for monitoring wetland anuran communities
Identifying population declines and mitigating biodiversity loss require reliable monitoring techniques, but complex life histories and cryptic characteristics of anuran species render conventional monitoring challenging and ineffective. Environmental DNA (eDNA) detection is a highly sensitive and minimally invasive alternative to conventional anuran monitoring. In this study, I conducted a field experiment in 30 natural wetlands to compare efficacy of eDNA detection via qPCR to three conventional methods (visual encounter, breeding call, and larval dipnet surveys) for nine anuran species. eDNA and visual encounter surveys detected the greatest species richness, with eDNA methods requiring the fewest sampling events. However, community composition results differed among methods, indicating that even top performing methods missed species detections. Overall, the most effective detection method varied by species, with some species requiring two to three methods to make all possible detections. Further, eDNA detection rates varied by sampling season for two species (A. americanus and H. versicolor), suggesting that species-specific ecology such as breeding and larval periods play an important role in eDNA presence. These findings suggest that optimized monitoring of complex anuran communities may require two or more monitoring methods selected based on the physiology and biology of all target species. Author Keywords: amphibian, anuran, conventional monitoring, eDNA, environmental DNA, species richness
Assessing the Cost of Reproduction between Male and Female Sex Functions in Hermaphroditic Plants
The cost of reproduction refers to the use of resources for the production of offspring that decreases the availability of resources for future reproductive events and other biological processes. Models of sex-allocation provide insights into optimal patterns of resource investment in male and female sex functions and have been extended to include other components of the life history, enabling assessment of the costs of reproduction. These models have shown that, in general, costs of reproduction through female function should usually exceed costs through male function. However, those previous models only considered allocations from a single pool of shared resources. Recent studies have indicated that the type of resource currency can differ for female and male sex functions, and that this might affect costs of reproduction via effects on other components of the life history. Using multiple invasibility analysis, this study examined resource allocation to male and female sex functions, while simultaneously considering allocations to survival and growth. Allocation patterns were modelled using both shared and separate resource pools. Under shared resources, allocation patterns to male and female sex function followed the results of earlier models. When resource pools were separate, however, allocations to male function often exceeded allocations to female function, even if fitness gains increased less strongly with investment in male function than with investment in female function. These results demonstrate that the costs of reproduction are affected by (1) the types of resources needed for reproduction via female or male function and (2) via trade-offs with other components of the life history. Future studies of the costs of reproduction should examine whether allocations to reproduction via female versus male function usually entail the use of different types of resources. Author Keywords: Cost of Reproduction, Gain Curve, Life History, Resource Allocation Patterns, Resource Currencies
Assessing Brook Trout (Salvelinus fontinalis) Seasonal Occupancy in Haliburton County, ON Using Environmental DNA
Brook trout (Salvelinus fontinalis) are declining across Ontario in both numbers and distribution, prompting concern for their future. Here, conventional, emerging, and predictive tools were combined to document brook trout occupation across seasons using streams in Haliburton County, ON as model systems. By using the Ontario Ministry of Natural Resources and Forestry’s (OMNRFs) Aquatic Ecosystem Classification (AEC) system variables with environmental DNA (eDNA) sampling and backpack electrofishing, my research supports the development of species occupancy models (SOMs) and eDNA as tools to document brook trout occurrence. To do this, eDNA sampling was validated in Canadian Shield stream environments by comparison with single-pass backpack electrofishing before seasonally sampling two river systems across their main channel and tributaries to assess occupancy. Streams were classified as potential high, moderate, and low-quality brook trout habitats using indicator variables within the AEC and sampled seasonally with eDNA to quantify occupancy and relate it to habitat potential at the county scale. Results showed eDNA to be an effective tool for monitoring fish across Canadian Shield landscapes and that brook trout occupancy varied seasonally within and across watersheds, suggesting that habitat and fish management strategies need to consider seasonal movement and spatial connectivity. Using these tools will enable biologists to efficiently predict and document brook trout occurrences and habitat use across the landscape. Author Keywords: Aquatic Ecosystem Classification, brook trout, Canadian Shield, connectivity, environmental DNA, seasonal occupation
Diversity, Biogeography, and Functional Traits of Native Bees from Ontario’s Far North and Akimiski Island, Nunavut
Bees (clade Anthophila), are poorly studied in northern Canada, as these regions can be difficult to access and have a short growing season. This study examined bees from two such regions: Ontario’s Far North, and Akimiski Island, Nunavut. I present this study as the largest biogeographical study of bees performed in these remote areas to enhance knowledge of northern native bees. I found 10 geographically unexpected species in Ontario and on Akimiski Island. Rarefaction and the Chao 1 Diversity Index showed that Akimiski is nearly as diverse as the Far North of Ontario, a significantly larger area. I also found, based on log femur length versus latitude, Bombus worker size was consistent with Bergmann’s rule, and there were no apparent statistical differences in the community weighted means of functional traits between the Far North’s Boreal Shield and Hudson Bay Lowlands ecozones. This work provides invaluable knowledge of the native bee species from these regions, which has implications for their future conservation. Author Keywords: Akimiski Island, Bergmann's rule, Chao 1, Community-weighted means, native bees, rarefaction
Enhanced weathering and carbonation of kimberlite residues from South African diamond mines
Mafic and ultramafic mine wastes have the potential to sequester atmospheric carbon dioxide (CO2) through enhanced weathering and CO2 mineralization. In this study, kimberlite residues from South African diamond mines were investigated to understand how weathering of these wastes leads to the formation of secondary carbonate minerals, a stable sink for CO2. Residues from Venetia Diamond Mine were fine-grained with high surface areas, and contained major abundances of lizardite, diopside, and clinochlore providing a maximum CO2 sequestration capacity of 3–6% of the mines emissions. Experiments utilized flux chambers to measure CO2 drawdown within residues and unweathered kimberlite exhibited greater negative fluxes (-790 g CO2/m2/year) compared to residues previously exposed to process waters (-190 g CO2/m2/year). Long-term weathering of kimberlite residues was explored using automated wet-dry cycles (4/day) over one year. Increases in the δ13C and δ18O values of carbonate minerals and unchanged amount of inorganic carbon indicate CO2 cycling as opposed to a net increase in carbon. Kimberlite collected at Voorspoed Diamond Mine contained twice as much carbonate in yellow ground (weathered) compared to blue ground, demonstrating the ability of kimberlite to store CO2 through prolonged weathering. This research is contributing towards the utilization of kimberlite residues and waste rock for CO2 sequestration. Author Keywords: CO2 fluxes, CO2 mineralization, CO2 sequestration, Enhanced weathering, Kimberlite, Passive carbonation
Fungal pathogen emergence
The emergence of fungal hybrid pathogens threatens sustainable crop production worldwide. To investigate hybridization, the related smut fungi, Ustilago maydis and Sporisorium reilianum, were selected because they infect a common host (Zea mays), can hybridize, and tools are available for their analysis. Hybrid dikaryons exhibited filamentous growth on plates but reduced virulence and limited colonization in Z. mays. Select virulence genes in the hybrid had similar transcript levels on plates and altered levels during infection of Z. mays relative to each parental dikaryon. Virulence genes were constitutively expressed in the hybrid to determine if its pathogenic development could be influenced. Little impact was observed in hybrids with increased expression of effectors known to modify host response and metabolism. However, increased expression of transcriptional regulators of stage specific pathogenic development increased the hybrid’s capacity to induce symptoms. These results establish a base for investigating molecular aspects of fungal hybrid pathogen emergence. Author Keywords: effectors, hybrid pathogenesis assays, Sporisorium reilianum, transcription factors, Ustilago maydis, virulence factors
Adoption of a Finite Element Model of Material Deformation Relevant to Studying Corneal Biomechanics
The human cornea is required to exhibit specific material properties to maintain its regular shape under typical intraocular pressures which then allow for its correct optical functionality. In this thesis, the basis of continuum solid mechanics and the finite element method are introduced. We use finite element modelling to simulate the extension of an effective-1d, linear-elastic bar, a cornea-like body governed by Poisson’s equation, and the deformation of a loaded, linear-elastic, cube. Preliminary results for the deformation of a simulated, linear-elastic, cornea have also been achieved using the finite element approach. Author Keywords: continuum solid mechanics, corneal biomechanics, finite element method, intraocular pressure
Modelling Depressive Symptoms in Emerging Adulthood
Depression during the transition into adulthood is a growing mental health concern, with overwhelming evidence linking the developmental risk for depressive symptoms with maternal depression. In addition, there is a lack of research on the protective role of socioemotional competencies in this context. This study examines independent and joint effects of maternal depression and trait emotional intelligence (TEI) on the longitudinal trajectory of depressive symptoms during emerging adulthood. A series of latent growth models was applied to three biennial cycles of data from a nationally representative sample (N=933) from the Canadian National Longitudinal Survey of Children and Youth. We assessed the trajectory of self-reported depressive symptoms from age 20 to 24 years, as well as whether it was moderated by maternal depression at age 10 to 11 and TEI at age 20, separately by gender. The results indicated that mean levels of depression declined during the emerging adulthood in females, but remained relatively stable in males. Maternal depressive symptoms significantly positively predicted depressive symptoms across the entire emerging adulthood in females, but only at age 20-21 for males. In addition, likelihood of developing depressive symptoms was attenuated by higher global TEI in both females and males, and additionally by higher interpersonal skills in males. Our findings suggest that interventions for depressive symptoms in emerging adulthood should consider development of socioemotional competencies. Author Keywords: Depression, Depressive Symptoms, Emerging Adulthood, Intergenerational Risk, Longitudinal, Trait Emotional Intelligence
Drink my beer, smoke my weed, my good friends are all I need
Past research has predominately supported an association between insecure attachment and substance use. More recent research has found that while individuals with secure attachment may use substances, they do so with less risk. The current study attempted to replicate the finding regarding secure attachment and substance use and examined the motivational drives behind why students use substances. A total of 318 undergraduate students completed self-report questionnaires examining peer attachment, alcohol and marijuana use, as well as motivations for use. Results indicated that students who reported low frequency use of alcohol or marijuana did not have significantly higher security ratings compared to students who reported increased use. Additionally, although hypothesized, secure attachment ratings were not associated with social facilitation or enhancement motivations. However, fearful and dismissing attachment ratings were both significantly associated with coping motives as predicted, while preoccupied and fearful attachment ratings were significantly associated with conformity motives. Lastly, results from multiple regressions suggest that coping and enhancement motivations are significant predictors of alcohol use, while enhancement motivations are marginally significant in predicting marijuana use. Author Keywords: alcohol, attachment, marijuana, motivation
Calcium Stress in Daphnia Pulicaria and Exposure to Predator-Derived Cues
In recent decades, declining calcium concentrations have been reported throughout lakes across the southern edge of the Canadian Shield. This raises concern as Daphnia populations have shown to be decreasing as they require calcium not only for survival but to mitigate predation risks. Therefore, the purpose of my thesis was to study the adaptability of Daphnia under calcium limitation and predation risk from Chaoborus. Firstly, I examined the effects of calcium limitation and Chaobours kairomones on daphniid life-history and population growth. I found that low calcium concentrations and Chaoborus kairomones affected Daphnia calcium content, life-history traits, and survival. Next, I focused on how calcium concentrations and Chaoborus abundance affected the calcium content and abundance of daphniids. During this study, I also examined the relationship between the abundance of Daphnia and a competitor Holopedium. I found that calcium concentrations and the abundance of Chaoborus affects daphniid abundance. Overall, results from this study show the importance of considering both predation risk and calcium declines to better determine daphniid losses. Author Keywords: anti-predator responses, Chaoborus, competition , Life-History traits, predator cues, Zooplankton

Pages

Search Our Digital Collections

Query

Enabled Filters

  • (-) ≠ History
  • (-) ≠ Physiology
  • (-) ≠ Analytical chemistry
  • (-) ≠ Southeast Asian studies
  • (-) = Master of Science

Filter Results

Date

1981 - 2031
(decades)
Specify date range: Show
Format: 2021/09/20