1722 research outputs found
Sort by
Climate envelope models for three endangered skipper butterflies at their northern range margins in Manitoba, Canada
No full textClimate change is accelerating biodiversity loss worldwide, intensifying pressure on already imperiled species. In Canada, several butterfly species are federally listed as endangered due in part to their narrow ranges, small and isolated populations, and dependence on rare or declining habitats, traits that heighten their vulnerability to climate change. Yet, the effects of future climate shifts on their persistence remain poorly understood. This study attempts to model the potential effects of climate change on the future extent of climatically suitable habitat for three at-risk species occurring at their northern range margins in Manitoba, Canada: the Dakota skipper (Hesperia dacotae), Poweshiek skipperling (Oarisma poweshiek), and Mottled duskywing (Erynnis martialis). Endangered species with few occurrences and restricted distributions pose unique modeling challenges, but understanding how climate change may alter their climatically suitable habitat is vital for guiding conservation efforts. To achieve these aims, ensemble climate envelope models (CEMs) were developed using six commonly used algorithms. These models were projected to future conditions using an ensemble of eight high-resolution CMIP6 climate projections for mid- (2041–2070) and late-century (2071–2100) periods, representing two shared socioeconomic pathways (SSP2-4.5 and SSP3-7.0). All ensemble models achieved strong predictive performance based on commonly used evaluation metrics. The Dakota skipper model performed best overall, likely due to more numerous and geographically spread occurrence records. Projections for all species revealed significant declines in climatic suitability across currently occupied areas under all scenarios. Only the Dakota skipper model showed newly suitable regions under future conditions, some within protected areas, offering opportunities for assisted colonisation or targeted habitat assessments. In contrast, no newly suitable areas were identified for the Poweshiek skipperling or Mottled duskywing, likely due to sparse and clustered occurrence data, which limited model generalizability. A key finding is that limited and highly localized species occurrence data can substantially influence model outputs, underscoring the need to interpret CEM results within the context of the quality and quantity of input data. The reliability of ensemble projections depends heavily on input data quality; overlooking this can distort estimates of species’ future suitability even when evaluation metrics indicate strong model performance.Master of Science in Environmental and Social Chang
Assessing the Precision of Magnetic Resonance Imaging Axon Diameter Inferences using Oscillating Gradient Spin Echo Pulse Sequences in a 15 T System
No full textPrevious research has linked numerous neurological disorders post-mortem to abnormalities in axon distribution and integrity within white matter tracts. Therefore, it is of high interest to investigate methods that will eventually be able to measure axon diameters in white matter tracts in vivo. Diffusion Magnetic Resonance Imaging is a method with the potential to infer microstructure in vivo using temporal diffusion spectroscopy. Temporal diffusion spectroscopy, when used with certain pulse sequences, such as Oscillating Gradient Spin Echo, can be used to infer micron-scale axon diameters. The most common geometric model used to fit the diffusion signals assumes that axons are long, parallel, straight, cylinders, where only the transverse intra-axonal diffusion coefficient is sensitive to the cylinder’s inner diameter. However, previous research has demonstrated that this geometric model tends to overestimate the intra-axonal diameter of axonal fibers. Due to recent advances in hardware, high-gradient strengths can be used to achieve shorter diffusion times and probe smaller restriction sizes than previously possible. To calibrate temporal diffusion spectroscopy with Oscillating Gradient Spin Echo pulse sequences in this project ex vivo mouse brains were imaged, and the genu substructure of the corpus callosum was analyzed. The images were collected using a 15.2 T Bruker NMR system located at the Vanderbilt University of Institute of Imaging Science. Data were collected in two experiments; first an initial calibration experiment was performed to test the proposed parameters. Then a second experiment was conducted to validate the inferred magnetic resonance axon diameters using transmission electron microscopy in a sample of 6 mice (3 male, 3 female)."I would like to acknowledge the financial support provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Research Manitoba."Master of Science in Bioscience, Technology and Public Polic
MMLTC: A novel Tolerance-Based Clustering Framework for Multimodal Sentiment and Harmful Meme Classification in Multilingual Settings
No full textThis thesis introduces a novel Tolerance-Based Clustering Framework (MMLTC) framework for affective analytics of multimodal/multilingual content in social media. A key feature of the MMLTC framework is its ability to overcome limitations of prior tolerance-based classifiers through the construction of label specific pure tolerance classes. Unlike traditional global partitioning methods, the proposed MMLTC employs a local, label aware grouping strategy that captures fine grained intra class variations while reducing the risk of label impurity. The proposed framework was tested on diverse multimodal datasets consisting of four English and three Bengali languages, using accuracy and weighted F1 metrics complemented by statistical significance testing via the Wilcoxon signed rank test and effect size analysis using Cohen’s d measure. MMLTC demonstrates strong performance across the seven diverse benchmark datasets, each involving classification tasks such as sentiment analysis, hate speech detection, offensive language identification, and multilingual content categorization. MMLTC outperforms state-of-the-art deep neural classifiers in six out of seven datasets with the weighted F1 score. Additionally, MMLTC consistently achieves performance that is either superior to or on par with five baseline classifiers: Random Forest, Support Vector Machines, Logistic Regression, K-Nearest Neighbors, and XGBoost. To assess the tolerance class clustering ability of MMLTC, t-SNE interpretability technique was applied.Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery GrantMaster of Applied Computer Scienc
Signal-to-Noise Ratio and Contrast-to-Noise Ratio Comparisons Between a Wireless Volume Radiofrequency Coil and a Commercially Available Wired Radiofrequency Coil
No full textThe design of magnetic resonance imaging (MRI) radiofrequency (RF) coils plays a critical role in determining image quality, with signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) serving as key metrics for evaluation. This study systematically compares the imaging performance of a novel wireless RF coil against a conventional wired coil across three experimental conditions: phantom imaging, fruit imaging, and human brain imaging. Using a phantom, results revealed that the wireless coil exhibited higher SNR, lower CNR, and the images had better quality both in terms of high signal and uniformity. In fruit imaging (watermelon, banana, and pineapple), SNR differences varied by fruit type and imaging protocol, with the wired and wireless coils performing similarly and in more cases, the wireless coil significantly outperforming the wired coil than underperforming. For human brain imaging, the two coils once again performed similarly under most conditions with the wireless coil significantly improving the SNR, particularly in T2-weighted imaging. The wireless coil eliminates bulky structures, reduces costs, and simultaneously produces similar SNR or enhances SNR quality compared to wired coils which provides motivation for our industry partner to continue the development of wireless coils. Future work should further explore the wireless coil’s performance in clinical imaging scenarios, including diffusion-weighted imaging and functional MRI, to better assess its diagnostic potential.Master of Science in Bioscience, Technology and Public Polic
Environmental DNA as a tool to supplement sampling for fish community monitoring
No full textEnvironmental DNA (eDNA), the genetic material shed by organisms into the environment, is a promising tool for fisheries monitoring due to its non-lethal, and efficient sampling. My thesis evaluates the use of eDNA as a complementary approach to gillnet surveys within the Coordinated Aquatic Monitoring Program (CAMP) in Manitoba. The primary objectives of my thesis were to assess (1) whether seasonal variation and site differences influence eDNA detectability of target species, and (2) whether eDNA concentration is associated with traditional measures of relative abundance, including Catch Per Unit Effort (CPUE) and Biomass Per Unit Effort (BPUE). Water samples were collected at various time points from Lac du Bonnet and Pointe du Bois, Winnipeg River system, and analyzed with species-specific assays for five fishes: Walleye (Sander vitreus), Spottail Shiner (Hudsonius hudsonius), Yellow Perch, (Perca flavescens), Trout-perch (Percopsis omiscomaycus), and Burbot (Lota lota). Seasonal eDNA sampling detected all species on both waterbodies during spring and fall, except for Burbot, which was not detected at one of the three sites in the fall. eDNA and gillnetting results showed consistent detections for four of the five species, with Burbot detected more often by eDNA than gillnets. While most species showed weak association between eDNA and CPUE/BPUE, Burbot displayed a significant positive relationship. These findings demonstrate that eDNA has the potential to expand detection of underrepresented species and enhance monitoring programs.Mitacs Accelerate Program, Manitoba Hydro, GEN-FISH programMaster in Environmental & Social Chang
Investigating the Impacts of Climate Change-Induced Drought on Dissolved Organic Carbon in Boreal Forest Soils and Streams
No full textThe boreal region is a vast storage of carbon, particularly dissolved organic matter (DOM), which governs the mobilization of carbon from watersheds. Under a changing climate, fluctuations of temperature and precipitation are expected, with variability in the frequency and intensity of hydrological events such as droughts. Using a combination of a long-term data study on boreal streams and a laboratory study on boreal soils from IISD-Experimental Lakes Area, the impacts of drought events on export of dissolved organic carbon (DOC) from boreal forest soils and streams can be investigated. Event characteristics of duration and severity of dry and wet events was revealed using long-term streamflow from 1971-2023 at the IISD-ELA, by applying a standardized streamflow index (SSI) at varying timescales. The associated long term stream chemistry of DOC concentration and flux to each dry and wet event was evaluated. DOC concentration had significant relationships for dry events but only at the shorter timescales, indicating there was limited mobility under dry conditions. DOC fluxes were significantly higher during wetness events, and increases with longer timescales, this could be explained by the enhanced hydrological connectivity. Landscape specific controls was also observed where wetland dominated catchments displayed higher DOC concentration during dry periods. The findings from the laboratory study reveals the importance of considering antecedent soil moisture conditions for soil carbon dynamics. Soils were collected and separated at two different soil depths at the IISD-ELA, where soils were subjected to drying or rewetting treatment at varying intensities. The quantity (DOC concentration) and quality (SUVA254, humification index HIX, biological index BIX) was evaluated post treatments to characterize DOM composition. There was more DOC released in the drying treatment compared to rewetting treatments, specifically in the shallower and most intense treatments. In terms of DOM quality, there was more limited variation, with shifts in HIX and BIXat the deeper soils. The role of antecedent soil moisture acts as first-order control on carbon dynamics, with increasing drought severity intensifying the shifts in DOM concentration and quality. Overall, this highlights how prolonged drought conditions can play a major role for the fate of carbon in both quantity and quality found within streams and soils, and how it may impact the downstream ecosystems within boreal catchments.NSERCMaster of Science in Environmental and Social Chang
Data-Driven Methodologies for Intelligent Systems
No full textThis thesis presents a unified investigation into data-driven intelligent systems through three major contributions: (i) a research study introducing a hierarchical two-level genetic algorithm for automated feature engineering, (ii) Query Weave, a conversational structured-data analysis system, and (iii) ImmiAI, a retrieval-augmented immigration assistance chatbot grounded in authoritative Canadian sources. The first contribution develops a novel genetic algorithm that balances predictive performance and interpretability through multi-objective optimization, ensemble evaluation, and non-linear feature transformations. The second contribution addresses the limitations of large language models in tabular reasoning by proposing a layered architecture for schema profiling, statistical discovery, and service routing. The third contribution integrates web scraping, data-lake pipelines, vector retrieval, and grounded LLM reasoning to support accurate and auditable immigration guidance. Together, these components demonstrate how automation, metadata-driven analytics, and grounded natural-language interfaces can reduce technical barriers and expand access to AI systems. The thesis highlights the importance of verifiable computation, modular reasoning pipelines, and tool-augmented conversational design in the development of trustworthy intelligent systems.MitacsMaster of Science in Applied Computer Science and Societ
Milk-derived extracellular vesicles may influence metabolic fuel usage in Rattus norvegicus neonates
No full textPerinatal exposure to a maternal high-fat diet (mHFD) is associated with hypothalamic inflammation, dysregulation of metabolic pathways, signal transduction, and increased adiposity in offspring. Peroxisome proliferator-activated receptors (PPARs) are a family of nuclear receptor proteins that regulate cellular differentiation, development, and metabolism. PPARs regulate the function of pyruvate dehydrogenase (PDH) primarily by controlling the expression of PDH kinases (PDK), enzymes that inactivates PDH by phosphorylating three serine residues on the E1α subunit. PDH is a central enzymatic hub that controls glucose and lipid flux and play a key role in maintaining metabolic flexibility and balance. Milk-derived extracellular vesicles (MEVs) are lipid-coated nanovesicles present in mammalian milk that have been shown to attenuate inflammation and metabolic dysregulation in offspring during critical periods of postnatal life. However, whether MEVs can influence metabolic flexibility in offspring with perinatal mHFD exposure with respect to PPAR-mediated regulation of PDH signalling remains largely unknown. The objective of this study was to determine the effects of perinatal mHFD exposure on PPAR and PDK gene expression and PDH function in offspring, and to assess whether postnatal treatment with human MEVs (HMEVs) modulates these outcomes. PPAR and PDK expression and PDH function were examined in the liver and hypothalamus of male and female rodent neonates using RT-qPCR and western immunoblotting, respectively. Two developmental time points were examined: postnatal day (PND) 11, corresponding to stress-hyporesponsive and peak lactation period, and PND21, representing the post-weaning, pre-pubertal period. Perinatal mHFD exposure resulted in significant, sex-specific alterations in PPAR and PDK gene expression in both the liver and hypothalamus, consistent with shifts towards altered metabolic fuel utilization. These effects were most pronounced at PND11 and diminished by PND21. Postnatal HMEV treatment produced tissue- and sex-specific responses in offspring with perinatal mHFD exposure, where Ppara transcript levels increased in the liver of male neonates and in the hypothalamus of female neonates. An inverse regulation was also seen in PDH function in the liver and hypothalamus. Overall, perinatal mHFD exposure exerted a stronger influence on PPARs and PDC function than HMEV treatment. These findings identify the early postnatal period as a critical window for metabolic programming and suggest that HMEV treatment may partially mitigate, but not fully reverse, mHFD-associated metabolic dysregulation.This work was supported by a Discovery Grant awarded to Dr. Sanoji Wijenayake from the Natural Sciences and Engineering Research Council of Canada.Master of Science in Bioscience, Technology and Public Polic
An Analysis of Indigenous Engagement in Forest Management Planning in Northern Ontario
No full textA forest management plan outlines strategies and measures to promote sustainable use of forest resources while balancing environmental, social, and economic interests within a specific area. In accordance with the planning process, when a forest management plan overlaps with Indigenous territories, communities have a legitimate right to engage in the planning process. However, there are significant barriers to meaningful engagement for many Indigenous communities. This study investigates the challenges and opportunities for increasing Indigenous engagement. A capacity-based analytical framework was developed based on three perspectives: 1) tools, techniques, and technologies affect for engagement; 2) the capabilities/ characteristics of peoples involved in engagement activities; and 3) characteristics of the engagement process and surrounding conditions. A case study approach was used to examine Indigenous engagement in Forest Management Planning (FMP) in northern Ontario, where focused attention is needed to enhance meaningful engagement. Using a qualitative research design, semi-structured interviews were conducted with five Indigenous community representatives and sixteen forestry professionals, researchers and stakeholders who have expertise or experience in FMP from Ontario and across Canada. This research recognized several key factors that influence Indigenous engagement. Findings indicate that community capacity (i.e. funding and skilled human resources) remains a considerable challenge. Additionally, relationships, timelines, language, training programs, scholarship opportunities and the planning process are significant areas that need attention to enhance Indigenous engagement. Identifying challenges and facilitators is crucial for fostering meaningful Indigenous engagement in creating effective forest management plans. Addressing these concerns will also help build respectful, trust-based relationships among actors and contribute to long-term collaborative and reconciliation efforts.Social Sciences and Humanities Research Council of Canada; The University of WinnipegMaster of Environment in Environmental and Social Chang
Unearthing trends in environmental impact assessments for mines and quarries across Canada
No full textIn Canada, at least 200 activemines and 6500 quarries produce 60 commodities. Many proposedmines and quarries undergo impact assessment (IA) to predict potential impacts and inform final decision-making, alongside conditions formitigating risk. We (1) present the first complete list of IA laws that apply to mining in Canada; (2) provide an open-source database of mine and quarry projects subject to IA; (3) quantify mining and IA trends; and (4) assess availability of information by jurisdiction. Our database includes 266 assessments of 227 projects under 13 jurisdictions proposed from 1974 to 2023. Over time, target commodities shifted from coal, oil sands, and peat, to metals (e.g., gold, copper, nickel) and production sizes increased. For many projects assessed across multiple jurisdictions, we could not explain differences in reported key metrics (production size, lifespan, footprint). Our database’s comprehensiveness is limited by a lack of publicly available data. We recommend regulators adopt the findable, accessible, interoperable, and reusable principles for information sharing by standardizing key metrics, publishing project documentation, and improving registry functionality.We encourage enhanced inter-jurisdictional IA cooperation to ensure the public and decision-makers have access to fulsome information about new projects during a modern mineral rush in Canada."We sincerely thank the Wilburforce Foundation for funding the completion of this work. We also acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC) and Leaders in Energy Sustainability program, funding reference number 555306."https://www.facetsjournal.com/doi/10.1139/facets-2025-011