Faculty of Science (ETDs)

Permanent URI for this communityhttps://hdl.handle.net/10539/37933

Browse

Search Results

Now showing 1 - 10 of 214

Quantity Discriminatory Capacity and Choice Preference Between Binary Rewards in African Elephants
(University of the Witwatersrand, Johannesburg, 2024) Mariotti, Elena; Parrini, Francesca; Ross, Don
The ability to quantitatively discriminate between different numbers of the same items is shared by most animals and pre-verbal humans, and it is a requirement for being able to do simple arithmetic operations. This thesis investigated African elephants’ (Loxodonta africana) ability to do simple arithmetic by testing their quantity discrimination ability, potential drivers of it, and their learning ability, with a special focus on individual differences. The results from this thesis provide informative priors representing a necessary foundation to quantitatively estimate the risk preferences of elephants in further choice tasks, which is the aim of the larger project this thesis is part of. Individual elephants were presented with binary choices over different quantities of the same food. All data analysis used a Bayesian framework, to allow informed inferences about individual elephants. First, the preference of the elephants for different fruits was measured. Subsequently, using game pellets and the preferred fruits, quantity discrimination choices were observed using olfactory-only or olfactory-and-sound as cues. Elephants were successful in discriminating between different quantities based on olfactory cues only, but only when they could smell directly on the top of the perforated buckets containing the rewards. A combination of olfactory information and sound reduced their discrimination ability. The absolute difference between the alternatives had a stronger effect on the choices made by the elephants when the difference between the alternatives was larger, while the ratio of the alternatives affected the choices mostly when the two alternatives were similar in size, regardless of the food used. This finding suggests that elephant olfactory quantity discrimination was driven mainly by the numerosity of the rewards rather than by the idiosyncratic characteristics of the food. Grouping the quantity discrimination evidence over each reward together, I was able to show that elephants can perform simple arithmetic operations, being more precise when the alternatives differ by at least two units. The learning ability of the elephants was affected by the order and difficulty of the tasks, with initial tasks, when elephants were not accustomed to the experimental set-up yet, showing clear indication of learning. The fact that the elephants in this study were able to successfully discriminate between different quantities of different fruits, that they were relatively precise when the difference between the alternatives was at least 2 units, and that they showed signs of learning over the course of novel tasks, allows one to infer that eliciting the risk preferences of elephants might be possible. In addition to representing the first step to a quantitative elicitation of the risk preferences of elephants, this thesis proves for the first time that African elephants can do simple arithmetic and also provides vital information to increase our understanding of how elephants make decisions in the wild.
3D Human pose estimation using geometric self-supervision with temporal methods
(University of the Witwatersrand, Johannesburg, 2024-09) Bau, Nandi; Klein, Richard
This dissertation explores the enhancement of 3D human pose estimation (HPE) through self-supervised learning methods that reduce reliance on heavily annotated datasets. Recognising the limitations of data acquired in controlled lab settings, the research investigates the potential of geometric self-supervision combined with temporal information to improve model performance in real-world scenarios. A Temporal Dilated Convolutional Network (TDCN) model, employing Kalman filter post-processing, is proposed and evaluated on both ground-truth and in-the-wild data from the Human3.6M dataset. The results demonstrate a competitive Mean Per Joint Position Error (MPJPE) of 62.09mm on unseen data, indicating a promising direction for self-supervised learning in 3D HPE and suggesting a viable pathway towards reducing the gap with fully supervised methods. This study underscores the value of self-supervised temporal dynamics in advancing pose estimation techniques, potentially making them more accessible and broadly applicable in real-world applications.
Detecting Disease in Citrus Trees using Multispectral UAV Data and Deep Learning Algorithm
(University of the Witwatersrand, Johannesburg, 2024-06) Woolfson, Logan Stefan; Adam, Elhadi
There is a high prevalence, in South Africa, of fruit tree related diseases infesting lemon trees, subsequently affecting overall crop yield and quality. Ultimately, the income for the farmers is significantly diminished and limits the supply of nutritional food crops for the South African population, who already suffer from a high incidence of malnutrition. Currently, there are various methods utilized to detect diseases in fruit trees, however they pose limitations in terms of efficiency and accuracy. By employing the use of drones and machine learning methods, fruit tree diseases could be detected at an earlier stage of development and with a much higher level of accuracy. Consequently, the chances of remedying the trees before the disease spreads is greatly improved, and the supply of nutritious fruit within South Africa is increased. This research report’s aim is to investigate the effectiveness of a deep learning algorithm for detecting and classifying diseases in lemon orchards using multispectral drone imagery. This entails assessing the performance of a pretrained ResNet-101 model, fine-tuned with additional sample images, in accurately identifying and classifying diseased lemon trees, specifically those affected by Phytophthora root rot. The methodology involves the utilization of a pretrained ResNet-101 model, a deep learning architecture, and the retraining of its layers with an augmented dataset from multispectral aerial drone images of a lemon orchard. The model is fine-tuned to enhance its ability to discern subtle spectral variations indicative of disease presence. The selection of ResNet-101 is grounded in its proven success in image recognition tasks and transfer learning capabilities. The results obtained demonstrated an impressive accuracy of 80%. The deep learning algorithm exhibited notable performance in distinguishing root rot-affected lemon trees from their healthy counterparts. The findings indicate the promise of utilizing advanced deep learning methods for timely and effective disease detection in agricultural farmlands, facilitating orchard management.
Knockdown of long non-coding RNA PANDA improves the cytotoxic effects of cisplatin in oesophageal squamous cell carcinoma cell lines
(University of the Witwatersrand, Johannesburg, 2024-11) Moonsamy, Sasha Sarasvathee Keshnee; Mavri-Damelin, Demetra; Jivan, Rupal
Oesophageal cancer is one of the leading causes of cancer death worldwide, of which oesophageal squamous cell carcinoma (OSCC) is the major subtype in southern and eastern Africa. Cisplatin is a well-established drug used to treat multiple cancers, including OSCC. Drug resistance is a major impediment to continued cisplatin therapy in numerous cancers. LncRNA P21-associated non-coding RNA DNA damaged activated RNA (PANDA) is known to function in cell cycle regulation in response to DNA damage and is upregulated in OSCC. We aim to determine lncRNA PANDA expression in South African-derived OSCC cells and establish whether down-regulation of this lncRNA can be used to supplement cisplatin therapy. In this study, MTT assays were performed to determine the EC50 concentrations of cisplatin in OSCC (WHCO1, WHCO5, and SNO) cells and HEK293 cells as a non-cancer control. The cytotoxic effects of cisplatin were exerted in all cell lines, with WHCO5 and SNO appearing more responsive to cisplatin than WHCO1 and HEK293. RT-PCR was used to detect if lncRNA PANDA is expressed in untreated and cisplatin-treated cells and was detected in all cell lines. Knockdown of lncRNA PANDA by siRNA was assessed with RT-PCR. Phase contrast microscopy was used to assess whether siRNA reagents altered cell morphology at 5, 24, and 48 hours post treatment. No significant alterations in cell morphology were observed in WHCO1, WHCO5, SNO, and HEK293 cells. MTT assay evaluation after 48 hours of cisplatin exposure, with or without siRNA for lncRNA PANDA, showed a significant reduction in EC50 concentrations in WHCO5, SNO, and HEK293 cell lines, suggesting that knockdown of lncRNA PANDA may improve cisplatin cytotoxicity in some cell lines. However, the EC50 values were higher with lncRNA PANDA knockdown in the WHCO1 cell line, suggesting that not all OSCC cell types may be responsive to this approach. In conclusion, lncRNA PANDA is expressed in response to cisplatin-induced DNA damage, and the down regulation of lncRNA PANDA improves the cytotoxic effects of cisplatin; however, further investigations are warranted in OSCC.
The Wind Energy Potential of South Africa’s Eastern Cape Province in a Changing Climate
(University of the Witwatersrand, Johannesburg, 2024-10) Landwehr, Gregory Brent; Engelbrecht, Francois; Lennard, Chris
Due to the abundance of wind and solar renewable energy resources across South Africa, and the comparative low cost of installation and operation of wind and solar energy infrastructure, it is inevitable that the country’s dependence on fossil fuels for energy will decline in the future. At a practical level, developing wind energy facilities entails a complex array of activities and the ~20-30 year life spans of such facilities intrinsically implies that they will experience climate change. However, insufficient research and related modelling have been undertaken in South Africa to quantify future variability and systematic changes in the wind resource as it relates to specific synoptic weather types and wind energy production. The aim of this thesis is to develop methodologies to understand the synoptic drivers of regional wind energy production potential and in turn assess how and why South Africa’s wind energy production potential may change as a function of changing circulation patterns in a changing climate. The wind energy potential of the Eastern Cape Province of South Africa is quantified using energy yield analysis techniques. These results are mapped onto commonly occurring synoptic types for the region to assign an energy potential to each. When the changing frequency of these synoptic weather types is calculated in a climate change impacted future using Global Climate Models, it is possible to quantify the change in wind energy potential in the long term. Results show that the synoptic-circulation pattern with the highest wind energy potential is the Atlantic Ocean ridging High with its centre at about 30 °S, behind a northward displaced mid-latitude cyclone. Global Climate Model projections of the frequency occurrence of these high energy synoptic states show a decrease in frequency at all global warming temperature thresholds and in turn a decrease in wind energy production. The likely cause of this being the poleward expansion of the descending limb of the Hadley circulation which shifts these synoptic systems southwards. The methodologies presented in this thesis provide South Africa with the necessary climate change risk assessment and mitigation capability to address these impacts on the wind energy sector in South Africa.
Symmetry reductions and approximate solutions for heat transfer in slabs and extended surfaces
(University of the Witwatersrand, Johannesburg, 2023-06) Nkwanazana, Daniel Mpho; Moitsheki, Raseelo Joel
In this study we analyse heat transfer models prescribed by reaction-diffusion equations. The focus and interest throughout the work is on models for heat transfer in solid slabs (hot bodies) and extended surface. Different phenomena of interest are heat transfer in slabs and through fins of different shapes and profiles. Furthermore, thermal conductivity and heat transfer coefficients are temperature dependent. As a result, the energy balance equations that are produced are nonlinear. Using the theory of Lie symmetry analysis of differential equations, we endeavor to construct exact solutions for these nonlinear models. We will employ a number of symmetry techniques such as the classical Lie point symmetry methods, the nonclassical symmetry, nonlocal and nonclassical potential symmetry approach to construct the group-invariant solutions. In order to identify the forms of the heat source term that appear in the considered equation for which the principal Lie algebra (PLA) is extended by one element, we first perform preliminary group classification of the transient state problem. Also, we consider the direct group classification method. Invariant solutions are constructed after some reductions have been performed. One-dimensional Differential Transform Method (1D DTM) will be used when it is impossible to determine an exact solution. The 1D DTM has been benchmarked using some exact solutions. To solve the transient/unsteady problem, we use the two-dimensional Differential Transform Method (2D DTM). Effects of parameters appearing in the equations on the temperature distribution will be studied.
Insights into silver(I) phosphine complexes in targeting cell death and metastatic mechanisms in malignant cell lines
(University of the Witwatersrand, Johannesburg, 2023-09) Roberts, Kim Elli; Engelbrecht, Zelinda; Cronjé, Marianne J.
Cancer is the leading cause of death worldwide, with 18.1 million new cases and 9.6 million deaths reported annually. Cisplatin, a popular chemotherapeutic drug, exhibits certain limitations in terms of selectivity and efficacy. This emphasizes the necessity for novel therapeutic approaches in addressing a variety of cancer types. Multiple studies have shown that silver-based compounds suppress cancer cell proliferation and induce apoptosis. Thirteen novel silver(I) mono-dentate phosphine complexes were investigated for their anticancer effects on seven different human malignant cell lines; A375 non-pigmented melanoma, A549 lung adenocarcinoma, HEP-G2 hepatocellular carcinoma, HT-29 colorectal adenocarcinoma, MCF-7 and MDA-MB-231 breast adenocarcinoma, and SNO oesophageal squamous cell carcinoma. Two non-malignant human cell lines, HEK-293 embryonic kidney cells and MRHF foreskin fibroblast cells, were used to assess the selectivity of the complexes. Cisplatin and the efficient silver(I) phosphine complexes were selected for dose-response experiments to determine IC50 concentrations for the respective cell lines. On the basis of these screening results (chapter two), five difficult-to-treat cancer cell lines, and their most efficient complexes were selected for further investigation. Various cellular characteristics were investigated in chapter three (A549, HEP-G2, HT-29); these included morphological changes, ATP levels, GAPDH levels, Ptd-L-Ser externalization, mitochondrial membrane potential, oxidative stress levels, and the activity of a metabolic enzyme, cytochrome P450 isoform CYP1B1. The antimetastatic activity of the selected complexes was assessed by evaluating their ability to impede the migration of A549 cells. The fourth chapter examines the anticancer effect of selected complexes on hormone-dependent (MCF-7) versus triple-negative (MDA-MB-231) breast cells. Changes in morphology, Ptd-L-Ser externalization, alterations in mitochondrial membrane potential, oxidative stress levels, cytochrome c release, and DNA damage were studied. Furthermore, in chapter five, molecular docking simulations were used to determine whether the most potent silver(I) phosphine complex across all cell lines bonds to estrogen receptor alpha (ER-α) and estrogen receptor beta (ER-β). Seven of the thirteen silver(I) phosphine complexes significantly reduced cell viability in malignant cell lines while being less toxic to non-malignant cells. Complex 4 best targeted all cancer types, with IC50 values ranging from 5.75 to 10.80 µM across malignant cell lines. In the malignant treated cells, morphological changes, reactive oxygen species production, mitochondrial membrane depolarization, and Ptd-L-Ser externalization were observed. Complexes 1 and 4 repressed cell migration in the A549 cells. The presence of damaged nuclei, metabolically inactive mitochondria and cytochrome c translocation from the mitochondria’ intermembrane to the cytosol in MCF-7 cells were observed. These findings suggest that complexes 2, 4 and 7 induced apoptotic cell death. Furthermore, in silico computational predictions suggested a promising interaction between complex 4, and ER-α and ER-β. Overall, this study demonstrates the potential of silver(I) phosphine complexes as anticancer agents, with promising effects on various cancer cell lines.
The application of machine learning methods to satellite data for the management of invasive water hyacinth
(University of the Witwatersrand, Johannesburg, 2023-06) Singh, Geethe; Reynolds, Chevonne; Byrne, Marcus; Rosman, Benjamin
Biological invasions are responsible for some of the most devastating impacts on the world’s ecosystems, with freshwater ecosystems among the worst affected. Invasions threaten not only freshwater biodiversity, but also the provision of ecosystem services. Tackling the impact of invasive aquatic alien plant (IAAP) species in freshwater systems is an ongoing challenge. In the case of water hyacinth (Pontederia crassipes, previously Eichhorniae crassipes), the worst IAAP presents a long-standing management challenge that requires detailed and frequently updated information on its distribution, the context that influences its occurrence, and a systematic way to identify effective biocontrol release events. This is particularly urgent in South Africa, where freshwater resources are scarce and under increasing pressure. This research employs recent advances in machine learning (ML), remote sensing, and cloud computing to improve the chances of successful water hyacinth management. This is achieved by (i) mapping the occurrence of water hyacinth across a large extent, (ii) identifying the factors that are likely driving the occurrence of the weed at multiple scales, from a waterbody level to a national extent, and (iii) finally identifying periods for effective biocontrol release. Consequently, the capacity of these tools demonstrates their potential to facilitate wide-scale, consistent, automated, pre-emptive, data-driven, and evidence-based decision making for managing water hyacinth. The first chapter is a general introduction to the research problem and research questions. In the second chapter, the research combines a novel image thresholding method for water detection with an unsupervised method for aquatic vegetation detection and a supervised random forest model in a hierarchical way to localise and discriminate water hyacinth from other IAAP’s at a national extent. The value of this work is marked by the comparison of the user (87%) and producer accuracy (93%) of the introduced method with previous small-scale studies. As part of this chapter, the results also show the sensor-agnostic and temporally consistent capability of the introduced hierarchical approach to monitor water and aquatic vegetation using Sentinel-2 and Landsat-8 for long periods (from 2013 - present). Lastly, this work demonstrates encouraging results when using a Deep Neural Network (DNN) to directly detect aquatic vegetation and circumvents the need for accurate water extent data. The two chapters that follow (Chapter 3 and 4 described below) introduce an application each that build off the South African water hyacinth distribution and aquatic vegetation time series (derived in Chapter 2). The third chapter uses a species distribution model (SDM) that links climatic, socio-economic, ecological, and hydrological conditions to the presence/absence of water hyacinth throughout South Africa at a waterbody level. Thereafter, explainable AI (xAI) methods (specifically SHapley Additive exPlanations or SHAP) are applied to better understand the factors that are likely driving the occurrence of water hyacinth. The analyses of 82 variables (of 140 considered) show that the most common group of drivers primarily associated with the occurrence of water hyacinth in South Africa are climatically related (41.4%). This is followed by natural land cover categories (32.9%) and socio-economic variables (10.7%), which include artificial land-cover. The two least influential groups are hydrological variables (10.4%) including water seasonality, runoff, and flood risk, and ecological variables (4.7%) including riparian soil conditions and interspecies competition. These results suggest the importance of considering landscape context when prioritising the type (mechanical, biological, chemical, or integrated) of weed management to use. To enable the prioritisation of suitable biocontrol release dates, the fourth chapter forecasts 70-day open water proportion post-release as a reward for effective biocontrol. This enabled the simulation of the effect of synthetic biocontrol release events under a multiarmed bandit framework for the identification of two effective biocontrol release periods (late spring/early summer (mid-November) and late summer (late February to mid-March)). The latter release period was estimated to result in an 8-27% higher average open-water cover post-release compared to actual biocontrol release events during the study period (May 2018 - July 2020). Hartbeespoort Dam, South Africa, is considered as a case study for improving the pre-existing management strategy used during the biocontrol of water hyacinth. The novel frameworks introduced in this work go a long way in advancing IAAP species management in the age of both ongoing drives towards the adoption of artificial intelligence and sustainability for a better future. It goes beyond (i) traditional small-scale and infrequent mapping, (ii) standard SDMs, to now include the benefits of spatially explicit model explainability, and (iii) introduces a semi-automated and widely applicable method to explore potential biocontrol release events. The direct benefit of this work, or indirect benefits from derivative work outweighs both the low production costs or equivalent field and lab work. To improve the adoption of modern ML and Earth Observation (EO) tools for invasive species management, some of the developed tools are publicly accessible. In addition, a human-AI symbiosis that combines strengths and compensates for weaknesses is strongly recommended. For each application, directions are provided for future research based on the drawbacks and limitations of the introduced systems. These future efforts will likely increase the adoption of EO-derived products by water managers and improve the reliability of these products.
Assessing aquifer vulnerability to landfill pollution using drastic method in Gauteng, South Africa
(University of the Witwatersrand, Johannesburg, 2023) Mphaphuli, Idah; Abiye, Tamiru
This study integrated the DRASTIC method and field investigations into mapping the degree of vulnerability of aquifers to landfill pollution in the Gauteng Province, which is one of the most populated provinces in South Africa. In order to investigate the aquifer vulnerability of Gauteng's heterogeneous and complex geology, the DRASTIC method was used to generate intrinsic and specific vulnerability maps. Three vulnerability classes were generated from the DRASTIC index, namely, low vulnerability, moderate vulnerability and high vulnerability, which covered 46%, 37% and 17% of the study area, respectively. The highly-vulnerable areas were associated with the karst aquifer of Malmani dolomite, permeable vadose zone, high hydraulic conductivity and loamy sand/sandy loam soil type, whilst moderately-vulnerable areas were associated with fractured/weathered aquifers, high recharge and low topography. The intrinsic vulnerability was validated using average NO3+NO2-N (nitrate + nitrite as nitrogen) and the results of water samples from field investigations conducted in Marie Louise and Robinson landfill sites. Elevated NO3+NO2-N concentration (9.85-16.03 mg/l) was observed in the highly-vulnerable areas. Water samples were collected, in order to analyse the water chemistry, stable isotopes and radioactive isotopes (tritium). Gibbs and Piper diagrams were used to evaluate the main mechanism controlling the groundwater chemistry and the dominant major ions that influence it. Pollution by leachate was detected in the Marie Louise landfill site, where the groundwater showed high tritium and ammonia concentration. The main hydrochemical facies detected in Marie Louise were Mg SO4, Ca-SO4, Na-SO4 and Na-Cl. The hydrochemical facies detected in Robinson were Na-SO4, Ca-HCO3, Na-Cl and Ca-Cl. The DRASTIC method was shown to be effective in assessing groundwater vulnerability on a regional scale, provided that there is adequate input data.
Antibacterial activity and susceptibility testing of bacterial isolates from nematodes (Cruznema spp.)
(University of the Witwatersrand, Johannesburg, 2023-09) Mothapo, Maletjema Magdeline; Lephoto, Tiisetso E.
Nematodes are unsegmented worms found in different niches associated with a diverse range of bacteria. Various types of nematodes exist including those that are parasitic to insects, known as entomopathogenic nematodes (EPNs). EPNS of genera Steinernema, Heterorhabditis and Oscheuis are symbiotically associated with Xenorhabdus, Photorhabdus and Serratia, respectively. The symbiotic bacteria of EPNs have been reported to produce a broad spectrum of antimicrobial compounds active against human pathogens. The aim of this study was to isolate and identify nematodes and their associated bacteria from soil samples collected from a vegetative farm in Lesotho and study their antimicrobial activity against four species of pathogenic bacteria (E. coli, S. aureus, E. faecalis and P. aeruginosa). An uncharacterized species of Cruznema was isolated and named Cruznema NTM-2021 (GenBank 18S rDNA accession number: OQ408141). Based on the BLASTN search incorporating the phylogenetic analysis of the 16S rDNA region, three genera of bacteria were identified as Alcaligenes sp., Enterobacter sp. and Elizabethkingia sp. The study revealed that all three bacterial isolates were pathogenic to Tenebrio molitor. Symbiosis tests, using lipid agar method demonstrated the ability of the host nematodes to develop and reproduce in the presence of their associated bacteria. Bacterial supernatants of Alcaligenes sp. and Enterobacter sp. showed some inhibitory activity against Escherichia coli and Enterococcus faecalis, by disk diffusion method. Staphylococcus aureus and Pseudomonas aeruginosa were the most resistant bacteria to supernatants of the three isolates. This study also showed that the Alcaligenes, Enterobacter, and Elizabethkingia species isolated from Cruznema NTM-2021 were resistant to ampicillin, amoxicillin, cefuroxime/sodium, vancomycin and cephalothin but susceptible to gentamicin.

Faculty of Science (ETDs)

Browse

Filters

Settings

Sort By

Results per page

Search Results