Lakehead University Knowledge Commons

Knowledge Commons is an open access repository for scholarship and research produced at Lakehead University. It is a free and secure repository for LU faculty, students, staff, and researchers to preserve and present their scholarship.

Communities in KnowledgeCommons

Select a community to browse its collections.

Now showing 1 - 3 of 3

Recent Submissions

Long-term relationship between soil carbon dynamics, hydrology, and microbiome in peatlands around North America’s largest historical point source pollutant of toxic metals, metalloids, and sulfur (Sudbury, Ontario, Canada)
(2026) Samantha, Mitchell; Pendea, Florin; Basiliko, Nathan; McCarter , Colin; Diochon , Amanda
Peatlands are major long-term carbon (C) sinks whose stability depends on tightly coupled soil carbon dynamics, hydrology, and microbial communities, yet these relationships can be profoundly disrupted by industrial pollution. In the 1970s, Sudbury (Ontario) was North America’s largest point source of sulfur dioxide (SO₂) and toxic metal and metalloid (TMM) emissions, generating a deposition gradient that provides a natural experiment for examining the long-term consequences of atmospheric contamination on peatland ecosystems. Using radiocarbon-constrained peat cores from three poor fen sites - two proximal to the smelter centre (Laurentian, Transplant), one distal (Cartier) - this thesis integrates high-resolution geochemical analyses and paleoecological reconstruction to evaluate how industrial disturbance altered carbon accumulation and restructured microbial assemblages. Geochemical profiles (Corg, Cinorg, N, S, Ca, P, and nine key metals/metalloids) reveal distinct smelter-derived signatures, including pronounced enrichments of Cu, Ni, Pb, As, Cd, Zn, and S at or below the Industrial Isochron (1880–1975 CE), accompanied by Ca depletion and coincident increases in N and P. Long-term apparent rates of carbon accumulation (LARCA) show a paradoxical response: heavily polluted fens exhibit industrial-era peaks in apparent C accumulation but lower Holocene-scale mean LARCA relative to a minimally impacted site with intact Sphagnum cover, indicating a cumulative long-term carbon deficit attributable to enhanced decomposition and export from older catotelm strata. Stratigraphic and geochemical evidence suggests vertically divergent effects of pollution, including suppressed microbial decomposition and enhanced apparent C preservation in shallow horizons under extreme metal–acid stress, coupled with enhanced decomposition and C loss in deeper peat driven by acidification, sulfate migration, and destabilization of humic-Fe-S complexes. Stratigraphically constrained cluster analysis (CONISS), canonical correspondence analysis (CCA), variation partitioning, permutational multivariate analysis of variance (PERMANOVA), and indicator species analysis were used to characterise microbial community change relative to geochemical gradients. Pre-industrial testate amoebae (TA) assemblages at all three sites were dominated by sphagnophilous taxa, particularly Hyalosphenia subflava, indicating long-term hydrological stability. Pronounced community restructuring coincident with the Industrial Isochron was evident at the two smelter-proximal sites, where disturbance-tolerant TA taxa (Cyclopyxis arcelloides type, Centropyxis cassis type, Phryganella acropodia type) displaced wet-affinity forms. The distal reference site exhibited comparatively muted change. Geochemical variables explained a significant proportion of total community variance (20%; CCA p < 0.001), with carbon composition emerging as the strongest unique predictor (adj. R² = 0.057), exceeding the independent contribution of toxic trace metals (adj. R² = 0.019). Critically, no discrete post-industrial recovery assemblage was detected at either proximal site: industrial-era taxa persist in surface samples and CONISS does not resolve a recovery zone distinct from the disturbance interval. These findings indicate that passive recovery following emission reductions has not reversed the microbial legacy of industrial contamination and that active restoration intervention may be required to re-establish pre-disturbance ecological conditions. Keywords: Peatland degradation, Carbon dynamics, Industrial pollution, Testate amoebae, Community ecology, Paleoecology, Sudbury, Heavy metals, Ecosystem recovery, Smelter emissions
Examining the Increase of AI Tool usage among pre-service teachers: a comparative analysis of spring 2024 and spring 2025 results
(2026) Ahmed, Abubakr; van Barneveld, Christina; Kaefer, Tanya
AI tool usage has increased significantly in the past few years, but very few researchers have used cross-sectional surveys to measure the increase. This study sought to evaluate the extent to which the frequency of AI tool usage has increased among first-year pre-service teachers at a rural public university in Ontario, Canada. Data were collected in the spring of 2024 and the spring of 2025 through a self-report survey. A concurrent embedded mixed-methods design was used with a sample size of 158 participants. Findings indicate that there was a 57% increase year-over-year in AI tool use, going from 54% in 2024 to 85% in 2025. Further statistical analysis indicated a correlation between gender and AI tool use, with a greater percentage of females adopting these tools. The qualitative findings revealed various themes, including using AI tools for idea generation and editing, as well as the negative stigma associated with its use. These findings underscore the importance of policymakers adopting new strategies to address AI tool use in teacher education programs, with the hope of sufficiently preparing pre-service teachers to teach in this new digital era. Keywords: AI tool use, mixed-methods research, pre-service teachers (PSTs), gender differences, frequency of use
A critical discourse analysis of the Instagram account of a tradwife
(2026) Manisha; Chambers, Lori; Walton, Gerald; Parker, Barbara; Brady, Miranda
This thesis investigates how Instagram tradwife content constructs love through aesthetics of submission, dependence, and self-erasure. The research explores how tradwife discourse on Instagram aestheticizes and circulates ideals of control, coercion, and emotional violence as love. Drawing on Feminist Critical Discourse Analysis (FCDA) and visual analysis, the study analyzes 62 reels and captions posted by white tradwife influencer Aria Lewis between January 2 and March 30, 2024. Interpretation is guided by Foucault’s theory of power, Butler’s theory of performativity, and Bourdieu’s concept of symbolic violence, which helps trace how submission is produced and made desirable through discourse, embodiment, and visual style. Seven themes structure the findings. They are economic dependence as love, domestic servitude as devotion, illness and failure, spiritualized patriarchy and courtship, aestheticized modesty and historical femininity, scripted femininity, and tradwife discourse as harmless choice. Unequal power is softened through religious language, gratitude, nostalgia, humour, and routine-based formats that frame women’s accommodation, emotional containment, and one-income reliance as moral, safe, and chosen. Rather than showing overt conflict, the account builds a romantic common sense in which hierarchy appears as peace and protection. This creates vulnerability because women’s security depends on a husband’s kindness. The thesis locates coercive control not only in private couple dynamics but also in cultural and digital infrastructures that teach followers what love should look like. It also contributes to tradwife scholarship by focusing on a young woman in a pre-motherhood phase and the scripts she circulates to younger audiences. Keywords: Instagram, tradwife discourse, feminist critical discourse analysis, coercion, influencer culture
Petrological and geochemical constraints on the origin and nature of the Eagle’s Nest intrusion, McFaulds Lake Greenstone Belt, Ontario
(2026) Sheshnev, Vladislav; Hollings, Peter
The Eagle’s Nest intrusion is a mafic-ultramafic, blade-shaped dike that hosts the only known economically significant orthomagmatic Ni-Cu-(PGE) mineralization in the Ring of Fire region of Ontario. It is part of the Koper Lake subsuite of the more voluminous Ring of Fire Intrusive Suite (~2736–2732 Ma) and occurs within the Meso- to Neoarchean in age McFaulds Lake Greenstone Belt. The Ring of Fire Intrusive Suite is host to chromite, Fe-Ti-V, and Ni-Cu-(PGE) mineralization. Previous studies investigated the orebody and mineralization hosted by the Eagle’s Nest intrusion, with limited attention to the unmineralized parts of the system. This study applied multidisciplinary petrological and geochemical techniques to evaluate the petrogenetic controls on the formation of the Eagle’s Nest intrusion through the examination of the less mineralized portions of the intrusion and genetically related mafic dikes. The Eagle’s Nest intrusion can be subdivided into the marginal and the inner zone. The marginal zone is composed of gabbroic rocks that exhibit the most evolved mineralogical and geochemical characteristics, with evidence of intense pseudomorphic alteration that often preserves primary magmatic textures. Contacts with the host tonalite vary, generally reflecting a prolonged high magma flux, but only rarely preserving evidence of rapid cooling and chilled margins. The marginal zone gradationally transitions into the inner zones, which consists of ultramafic ortho- to mesocumulate rocks. The inner zone is characterized by coherent linear geochemical trends that reflect olivine and chromite accumulation with variable proportions of intercumulus silicate phases and interstitial sulfides, consistent with petrographic observations. Most inner zone rocks are characterized by a strong positive correlation between MgO and Cr2O3, reflecting the crystallization of olivine and chromite in cotectic proportions. However, several of the mineralized peridotite samples deviate from this trend despite containing similar proportions of these minerals. Petrographic observations and intercumulus pyroxene mineral chemistry suggest that the deviation from the cotectic trend may be caused by sulfide percolation and displacement of a Cr-rich intercumulus silicate melt, rather than the presence of less than cotectic proportions of olivine and chromite. A new parental magma composition estimate was established using olivine and chromite mineral chemistry, as well as whole rock geochemistry of ultramafic cumulate rocks interpreted to reflect cotectic proportions olivine and chromite, with variable proportions of intercumulus silicate melt. The estimate yielded a parental magma composition that contained ~15 wt% MgO and ~11 wt% FeOt, consistent with a komatiitic basalt magma. The new composition is more evolved than previous estimates, however, it is in close agreement with the composition of identified chilled margins, associated mafic dikes, and olivine. Forward thermodynamic modeling simulations of the new parental magma, reproduce the petrographically determined crystallization sequence at low pressures, suggesting that the Eagle’s Nest intrusion formed at shallow crustal levels. Whole rock geochemistry and Sm-Nd isotopes show that the Eagle’s Nest magma was derived from a depleted mantle source, above the garnet stability field, which then underwent extensive crustal contamination from multiple sources that included both the host tonalite, and older supracrustal rocks. Crustal contamination by sulfur-bearing supracrustal rocks likely contributed to attaining sulfide saturation of the magma, as evidenced by Δ³³S values consistent with mass-independent fractionation. The distinctive petrological and metallogenic characteristics of the Eagle’s Nest intrusion in the Esker Intrusive Complex may be a result of several distinct processes involving both emplacement dynamics and parental magma composition, resulting in unique metal endowments relative to other intrusions in the McFaulds Lake Greenstone Belt.
Uncertainty-guided Transformer learning for trustworthy medical image classification
(2026) Sibhai, Mohammed Maaz; Bin Ahmed, Saad; Alkhateeb, Abedalrehman; Ghaffar, Farhan; Bajwa, Garima
Reliable medical image classification is fundamental for the safe use of deep learning in clinical decision support. The state-of-the-art deep learning models, such as medical vision Transformers performs well in medical image segmentation. These models often present unreliable probability estimates and do not have built-in ways to explicitly handle uncertainty or interpretability. These issues become especially problematic when inputs are ambiguous or datasets are not uniformly distributed, which are common in real-world clinical settings. This study contributes in extending the architecture of Medical Transformer (MedFormer), a hierarchical medical vision Transformer guided by uncertainty and prototypes, to improve trustworthiness without reducing feature representation. The model uses per-token evidential uncertainty estimation via a Dirichlet approach, enabling explicit measurement of uncertainty and spatial localization. Instead of just acting as a post-hoc diagnostic tool, uncertainty actively guides feature routing and refinement during training, decreasing unreliable updates in uncertain regions. Additionally, prototype-based learning is incorporated to maintain a structured, classspecific geometry in the embedding space and support similarity-based, interpretable decisions grounded in visual patterns. The proposed model has been tested on various medical imaging types, including mammography, breast ultrasound, brain tumor MRI, and breast histopathology, providing a thorough testing across different dataset contexts. Experiments show that, while classification accuracy improvements vary across datasets, the method reliably improves calibration, reduces overconfidence, and enhances selective prediction compared to the baseline MedFormer. These results indicate that integrating uncertainty estimation and prototype-based regularisation into Transformerbased representation learning can greatly boost the reliability and explainability of medical image classifiers, supporting the development of trustworthy AI systems for clinical use Experimental results show that the proposed model improves calibration, reduces overconfidence, and enhances selective prediction across all evaluated datasets compared to the baseline MedFormer.The accuracies are reported in the selected benchmark datasets, with larger improvements in modalities with clearer visual cues and more modest changes in mammography due to inherent ambiguity. Overall, uncertainty-guided routing and prototype-based learning improve trustworthiness without sacrificing discriminative performance.