22962 research outputs found
Sort by
Investigating the Role of the Environment on Antimicrobial Resistance in Canada
Full text linkAntimicrobial resistance (AMR) is a complex global challenge affecting human
and animal health. This thesis applied multiple analytical approaches to explore the role of the environment in AMR dissemination across Canada. A scoping review identified water and soil as major AMR transmission pathways studied and revealed gaps in AMR studies related to air. Using factor analysis of mixed data and hierarchical clustering on data from the scoping review and 2 other data sources (the Canadian Integrated Program for Antimicrobial Resistance Surveillance and the Antimicrobial Resistance Genes in Bioaerosols Project), this work revealed similar AMR profiles across livestock farms and other environmental settings. Additionally, source attribution models using both frequentist and Bayesian approaches, identified chicken and pig farms as major contributors of AMR subtypes to other animal production systems and environmental settings. Finally, an integrated assessment model was used to integrate and analyze AMR proportions across environmental and agricultural sources. This model suggests that interventions targeting a single source result in limited reductions in overall AMR exposure, whereas coordinated reductions across multiple sources lead to substantially greater decreases in human exposure to AMR. Collectively, these findings demonstrate that AMR cannot be effectively understood or mitigated by examining individual sectors in isolation. Instead, the results underscore the importance of integrated, multi-sectoral strategies that
explicitly include environmental pathways alongside animal production systems. This thesis provides a methodological framework for integrating heterogeneous data sources to inform coordinated AMR mitigation strategies through a One Health approach
Cover Crops and 4R Nutrient Management Strategies to Mitigate GHG Emissions
No full textThis two-year study (2023-2024) at Elora, Ontario, evaluated tillage (conservation vs. no-tillage), with or without fall-seeded cereal rye cover crop, and urea ammonium nitrate ± Tribune, (dual urease + nitrification inhibitor) under 4R nutrient management in a corn-soybean rotation. It measured nitrous oxide (N2O) emissions, ammonia volatilization and corn yields. Cumulative N2O emissions increased from 0.93 kg N ha-1 (mean) in 2023 to 1.06 kg N ha-1 in 2024, driven by higher soil mineral N and drought stress despite a 17% yield reduction in 2024 (up to 73% in some combinations). Tribune reduced N2O emissions by 57% in 2023 and 30% in 2024, effective across wet and dry conditions. Ammonia losses decreased by 24% in 2023 and 10% in 2024 with Tribune. Tillage and cover crop effects were inconsistent and weather-dependant. Tribune proved a robust strategy for reducing N losses while sustaining yields, supporting climate-smart N management in southern Ontario.Agriculture and Agri-Food CanadaGrain Farmers of OntarioNatural Sciences and Engineering Research Council of Canad
Evalutating Multi-Agent Policy Gradient Methods for Resource Allocation in C-V2X Networks
Full text linkCellular Vehicle-to-Everything (C-V2X) communication is a key enabler of cooperative intelligent transportation systems, yet radio resource allocation (RRA) remains challenging due to high mobility, strict latency constraints, and non-stationary interference. While deep reinforcement learning has shown promise for adaptive RRA, existing studies are largely value-based and fragmented across scenarios and evaluation protocols, limiting reproducibility and comparative insight.
This thesis develops a unified evaluation framework for multi-agent policy gradient methods in C-V2X highway environments. A realistic system model is constructed with standardized channel dynamics, queueing behavior, and time-scale control. Three benchmark task families, NFIG, SIG, and POSIG, are introduced to vary temporal horizon, observability, and coordination complexity. Independent and centralized-training policy gradient algorithms are evaluated alongside value-based baselines. Results demonstrate improved stability and coordination of policy gradient methods in long-horizon and partially observable settings, while value-based learners remain competitive in simpler tasks
Using eDNA to Detect Native Lamprey Species in the Laurentian Great Lakes and Upper St. Lawrence River Basins
Full text linkSupplemental data for Chapter 3: Zeinstra, Nathan, 2026, "Supplemental data for: Using eDNA to detect native lamprey species in the Laurentian Great Lakes and Upper St. Lawrence River Basins", https://doi.org/10.5683/SP3/UORCDY, Borealis, V1.Native lamprey species in Ontario and the Great Lakes have experienced significant population declines over the last decades, warranting their conservation. This thesis explores the use of environmental DNA (eDNA) to detect native lampreys in this region to inform conservation decisions. First, the widely used traditional survey method of rapid assessment electrofishing was compared to eDNA during routine larval lamprey assessment monitoring. The eDNA method had a higher detection probability and lower false negative rate than electrofishing, demonstrating its utility in detecting native lampreys. Second, an Ontario-wide eDNA survey was conducted at 164 stations across 58 waterways, which resulted in the detection of silver (Ichthyomyzon unicuspis) or northern brook lamprey (Ichthyomyzon fossor) eDNA in 13 streams with historical distribution and five streams without any previous occurrence records. These findings demonstrate the efficacy of eDNA for detecting native lampreys and provide novel species occurrence information to aid in conservation efforts.Great Lakes Fishery Commissio
Optimizing Differentially Private Federated Learning for Medical Image Segmentation
Full text linkCollecting medical data is constrained by privacy regulations and institutional policies. Moreover, models trained on centralized datasets often exhibit limited generalizability to unseen or heterogeneous samples from different institutions due to variations in imaging protocols and acquisition settings. Such limitations necessitate repeated data collection and retraining, which are time-consuming and clinically critical, as degraded model performance may compromise patient safety. Differentially Private Federated Learning can enable collaborative model training and periodic fine-tuning without sharing raw data while providing formal privacy guarantees but can suffer from substantial utility degradation. This study presents a novel variant of Adaptive DP-FL for medical image segmentation, integrating adaptive mechanisms with enhanced training strategies to mitigate privacy-utility trade-offs. ADP-FL was evaluated on segmentation tasks including skin lesions in dermoscopic images, 3D kidney tumors in CT scan, and brain tumors in multiparametric MRI under simulated and real multi-institutional heterogeneity. Benchmarking against conventional FL and DP-FL demonstrates that ADP-FL significantly improves accuracy, convergence, and generalization compared with DP-FL, achieving performance approaching non-private FL
Photosynthetic Photon Flux Density and Carbon Dioxide Concentration Regulate Spearmint (Mentha spicata) Biomass and Terpene Yield in Controlled Environments
Full text linkSpearmint (Mentha spicata) is cultivated for its terpene-rich essential oil, but field production is challenged by disease pressure and climatic variability. As an alternative to field production, the viability of controlled environment (CE) spearmint cultivation is evaluated by characterizing physiological, biosynthetic, and morphological responses to photosynthetic photon flux density (PPFD) and carbon dioxide (CO2) concentration. Dry biomass and terpene yield (mg/plant) increased proportionally with average PPFD (APPFD) and responded nonlinearly to CO2 concentration. In both instances, terpene yield was more related to changes in leaf dry mass than changes in terpene concentration (mg g-1 leaf dry mass). Higher PPFD and CO2 concentration independently led to physiological and morphological changes at the leaf level but did not affect the overall plant dimensions. The relationships and protocols described in this thesis can be used as a framework to maximize spearmint biomass and essential oil in CEs.Natural Sciences and Engineering Research Council of Canad
Mitogenomes of Adeleid Coccidia (Apicomplexa: Coccidia: Adeleorina: Adeleidae) Infecting Invertebrates
Full text linkMitochondrial genome sequences were generated from two adeleid species: Obvallatus mesnili, infecting Indian meal moths (Plodia interpunctella) in Türkiye; and, Adelea ovata, infecting garden centipedes (Lithobius forficatus) locally. The circular-mapping mitochondrial genome of Obvallatu mesnili (6760 bp) possessed 3 CDS regions (cytochrome c oxidase subunits I and III [COI, COIII], cytochrome B) and 36 fragmented rDNAs comprising 82% of this compact extrachromosomal genome. Genome content and organization agreed with those of the mitochondrial genomes of Klossia species (Adeleidae). A 2828 bp portion of the mitochondrial genome of Adelea ovata contained the COIII CDS region and 18 fragmented rDNAs in an organization that matched O. mesnili. These sequences represent the first mitochondrial genome data from any member of either genus, Obvallatus or Adelea. The mitochondrial genome of O. mesnili can act as the genotypic reference sequence and help resolve the taxonomic affinities of these understudied parasites.Natural Sciences and Engineering Research Council of Canad
Electrochemical and Spectroscopic Studies of Carbon Dioxide Reduction at Bismuth-based Bimetallic Nanostructured Catalysts
No full textThe advancement of nanostructured catalysts for the electrochemical reduction of CO2 to value-added chemicals has attracted significant interest. However, progress remains limited by the complexity of the catalyst synthesis methods and the lack of fundamental understanding of the reduction reaction mechanisms. In this thesis, bimetallic bismuth-based catalysts with nanostructured morphology were synthesized using relatively simple methods, including galvanic replacement reactions and co-electrodeposition. The effects of bismuth incorporation on catalytic activity and Faradaic efficiency were investigated and compared with those of pure Zn, Sn, and Cu catalysts. High Faradaic efficiencies of the bimetallic ZnBi, SnBi, and CuBi catalysts were mainly attributed to their increased electrochemically active sites and facilitated charge transfer. A three-compartment membrane electrode assembly (MEA) cell was employed to achieve higher current densities compared with a conventional H-type cell. A high faradaic efficiency of 84% and 98% towards formic acid formation was obtained at -90 mA cm-2 for the SnBi and CuBi catalysts, respectively. In situ electrochemical attenuated total reflection Fourier transform spectroscopy (ATR–FTIR) was further employed to elucidate the reduction reaction mechanism. This study revealed that formate and carbon monoxide formation proceeded through the carbon-bound adsorbed CO2 on the ZnBi catalyst. However, the oxygen-bound adsorbed CO2 and the adsorbed
bidentate formate were detected during the electrochemical CO2 reduction using CuBi and SnBi catalysts, respectively. The cost-effective synthetic procedures, the formic acid MEA cell design, and the CO2 reduction mechanistic studies in this thesis can help pave the way toward the development and optimization of catalysts for formic acid generation.2027-01-0
Application of Bentonite Adsorption for the Removal and Fractionation of Protein from Acid Whey
Full text linkThe recovery of valuable components from dairy waste presents an opportunity to generate economic value for producers while reducing the environmental burden associated with treatment and disposal. Gaps in cheese whey utilization persist due to the lack of accessible treatment options for small-scale producers to valorize their waste, necessitating investigation into alternative technologies. To address this, bentonite clay was evaluated as an adsorbent for recovering and fractionating proteins from acid whey. The objectives of this investigation were to optimize the adsorption process, gain insights into the mechanisms responsible for adsorption and identify suitable applications of the process. The process was evaluated through a series of batch adsorption experiments, in which key operational parameters, including contact time, adsorbent dosage, pH, and temperature, were systematically varied. Adsorption performance was quantified, with a focus on the removal and separation of the major whey proteins, alpha-lactalbumin (α-LA) and beta-lactoglobulin (β-LG). Adsorption of major whey proteins using bentonite reached equilibrium after 10 hours at 20 g/L and pH 4.7, with α-LA exhibiting a higher rate constant under pseudo-second-order kinetics. Increasing bentonite dosage to 30 g/L enabled complete protein removal at equilibrium. Elevated temperatures enhanced removal efficiency, while pH variation from 3 to 9 revealed the potential for selective adsorption, with α-LA removal favoured at low pH, and β-LG at high pH. The selective removal of the major whey proteins was optimized using factorial experiments. Variation of pH and temperature was used to determine an optimal region for the selective removal of β-LG. Further optimization of the adsorbent dosage and contact time resulted in a condition where β-LG removal reached 88%, while α-LA removal was only 26%. This investigation establishes this process as an effective means of whey processing, utilizing a low-cost, scalable strategy. Bulk removal of whey proteins via adsorption is a viable technique for producers unable to implement conventional membrane filtration systems, providing a valuable unit operation for separating protein from whey. Furthermore, the method’s potential to fractionate major whey proteins enables the generation of high-value, pure protein fractions, thereby presenting an opportunity to reduce the cost and complexity of traditional processing approaches.Dairy Farmers of OntarioOntario Ministry of Farming, Agriculture and Agribusiness and Ministry of Rural AffairsNational Research Council Canad
Structure and Stability Across Scales in Changing Ecological Systems
Full text linkEcological systems are undergoing rapid changes as a result of human activities (i.e., land-use, harvesting, climate warming), driving generally widespread declines in biodiversity globally. To understand the functional consequences of these changes, I use conceptual ideas from food webs, the complex networks of species interactions that structure whole ecosystems, as a framework for linking underlying structures within food webs to stability. There are key properties related to biodiversity, such as species traits, differential species responses (i.e., compensatory dynamics), and the distribution of species within food webs, that play an integral role in structuring the flow of energy, with important consequences for whole system functioning and stability (i.e., variability). To examine how these structures and stability respond to global changes, I integrate ecological theory with long-term data across multiple spatial, temporal, and taxonomic scales. In Chapter 1, I synthesize consumer-resource and life-history theory to argue that a species’ intrinsic growth potential (rmax) represents its maximum interaction strength potential or capacity for energy flux, with predictable consequences for stability (e.g., CV). Using data on mammals, I show that stability scales allometrically with body size and mass-independent slow-fast traits, producing a weak-to-strong interaction strength gradient with stable-to-unstable-dynamics. My results suggest that global shifts towards smaller and faster species may be assembling ‘faster’ food webs with stronger interactions and reduced stability. In Chapter 2, I test how biodiversity influences stability through its effect on asynchrony to stabilize a zooplankton metacommunity over 30 years in a large temperate lake. Seasonal stability arises primarily from temporal asynchrony among species, and its decline in contemporary, smaller-bodied assemblages likely reduces energy transfer to upper trophic levels. In Chapter 3, I revisit questions of scale dependence in trophic structure using trophic position and biodiversity data from 54 large marine ecosystems (LMEs) globally. While trophic structure varies across a global biodiversity gradient, it remains temporally invariant within three Northwest Atlantic LMEs despite major ecosystem shifts. Instead, it is the variance around the mean trophic structure that is affected, driven by harvesting and large-scale climatic signals. My results suggest that measuring the variability in trophic structure may serve as a structural early warning signal of ecosystem state shifts in large ecosystems. Collectively, this body of research links changing traits, synchrony, and biodiversity to stability across ecological scales, illustrating how global changes are reshaping key structures that promote stability and functioning in complex ecological systems