ResearchSpace

Communities in ResearchSpace

Welcome to ResearchSpace, the institutional repository of the CSIR. ResearchSpace is an open access electronic archive collecting, preserving and distributing scholarly digital materials created by the CSIR.

Recent Submissions

Item

The Evolution of Southern Ocean Net Primary Production in a Changing Climate: Challenges and Opportunities

(2025-12) Tagliabue, A; Thomas Ryan-Keogh, Thomas; Baker, A; Bibby, TS; Follett, C; Lohan, MC; Naveira-Garabato, A; Mayor, DJ; Milne, A; Moore, CM; Ussher, S

Net primary production in the Southern Ocean plays a critical role in governing ecosystem production, the biological carbon pump, and global biogeochemical cycles. Recent work has advanced our understanding of novel factors regulating Southern Ocean net primary production and the regional physiological adaptations employed by Southern Ocean phytoplankton in terms of their photosynthetic strategies and resource acquisition. Here we assess trends in Southern Ocean net primary production from different remote sensing algorithms and bgc-Argo floats to compare them to the latest Earth System Models used to forecast future changes under three different future climate scenarios. Overall, remote sensing and bgc-Argo floats indicate net primary productivity in the Southern Ocean is declining at basin scale. This contrasts with the Earth System Models that display muted contemporary trends and consistent increases in net primary production that are relatively robust across SSP2-45, SSP3-70, and SS5-85. This mismatch in trends suggests low confidence in these projected net primary production changes, with implications for assessments of changes in ecosystem services. Despite their coherence in terms of net primary production trends, Earth System Models show large disagreement in the relative role of different drivers, suggesting we lack sufficient mechanistic understanding. Improved knowledge of the role of manganese alongside iron and the coupled responses of phytoplankton and zooplankton through the integration of observations and experiments into a new generation of models is necessary to deliver confident forecasts of Southern Ocean ecosystem change. Advancing knowledge in these areas is an important priority for future research in the region and provides context for policy discussions around the marine protection of Antarctic ecosystems that depend on sufficiently confident projections of climate change impacts.

Item

Evaluating the reliability of LLMs in OSINT investigations: A friend or foe?

(2026-06) Baloyi, Errol; Siphambili, Nokuthaba; Ntshangase, Ntomfuthi L; Letshwenyo, Mpho; Makharamedzha, Fhatuwani; Mmbodi, Rendani; Hlongwane, Ndabezinhle E

This study investigates the use of Large Language Models (LLMs), including GPT, Claude 3, Gemini, Meta, DeepSeek, Qwen 2.5, Mistral Large, and Grok, in Open-Source Intelligence (OSINT) investigations, focusing on their capabilities, limitations, and practical implications. Using a controlled fictional organization, the CtrlZ Society, to simulate a plausible online footprint, the LLMs were provided with a synthetic dataset purportedly linked to the organization. The dataset included social media posts, a Reddit thread, a Pastebin document, a GitHub repository, a blog post, a Telegram broadcast, a WHOIS record, and a news article. Based on this dataset, the models were evaluated on accuracy, timeline reconstruction, account attribution, evidence traceability, susceptibility to hallucination, and handling of ambiguity and incomplete information. Results revealed substantial variation among models: Claude 3, GPT, and Qwen 2.5 demonstrated strong analytical performance and reliable synthesis of investigative outputs, while Gemini and DeepSeek exhibited weaker capabilities. Some models, including Meta were also prone to forced narrative construction when prompted adversarially, highlighting risks of misinterpretation or overreach. Despite these limitations, all LLMs provided valuable support for structuring and summarising complex data, demonstrating their potential as efficiency multipliers in OSINT workflows. Based on these findings, the study provides recommendations for practitioners, including rigorous human oversight, multi-model validation, adherence to verification protocols, and careful evaluation of outputs to mitigate risks and maximise the reliability of LLM-assisted investigations.

Item

A global comparison of net-zero carbon building practices

(2026-07) Senaratne, S; Nadeeshani, M; Perera, S; Domingo, N; Lombardi, P; Moghadam, ST; Van Reenen, Coralie A; Lai, JHK; Chan, DWM; Mallawaarachchi, H

Despite extensive research on individual net-zero carbon building (NZCB) practices, prior studies fail to systematically map global adoption patterns across diverse regulatory and market contexts. This study addresses this gap by investigating current NZCB practices worldwide. A comprehensive list of global NZCB practices, with associated enablers and barriers, was first identified through a literature review. A global survey was then employed to explore the practical adoption of the practices. Six countries/ regions participated in the survey: Australia, New Zealand, Italy, South Africa, Sri Lanka and Hong Kong. Survey data were analysed using descriptive and frequency analysis. The study identifies three empirically derived adoption patterns: common, varied, and low across countries/regions. Use of solar photovoltaics, energy-efficient HVAC and modular systems were found to be commonly adopted, while advanced technologies like carbon-infused materials were least adopted. Lack of direct financial incentives emerged as the key global barrier, while robust government policies were the main enabler for NZCB adoption worldwide. These findings provide empirical benchmarks for cross-country NZCB adoption and actionable strategies to accelerate low-adoption practices globally.

Item

From description to implementation: Key takeaways from the 3rd African Microbiome Symposium

(2025-12) Marsh, CC; Nel Van Zyl, K; Babalola, OO; Böhmer, R; Cowan, DA; Moganedi, KLM; Moroenyane, I; Naidoo, Jerolen; Delgado, AN; Posma, JM

The 3rd African Microbiome Symposium was held in Cape Town, South Africa, from 20 to 22 November 2024. The symposium featured a diverse range of local and international microbiome research and provided a platform for 79 researchers, students, and industry members to engage in discussions on the microbiome within an African context and focusing on translational research. This meeting review shares highlights, findings, and recommendations derived from the event. Insights from two panel discussions revealed key barriers to microbiome research in Africa, including limited funding, infrastructure gaps, and a shortage of trained local scientists. Recommendations centered on increased investment, institutional training, adherence to ethical guidelines, and the fostering of equitable global partnerships.

Item

Improved diabetic classification using chaotic grey wolf random forest optimiser for hyperparameter tuning

(2026-06) Chukwujekwu, CK; Bello-Salau, H; Mu’azu, MB; Aliyu, HA; Onumanyi, Adeiza J; Ibrahim, OM

The application of machine learning in medical diagnostics has significantly advanced the early detection and classification of diabetes. However, classifiers often suffer from challenges related to high-dimensional data, overfitting, and ineffective hyperparameter tuning, which compromise their predictive performance and generalisation capability. To address these limitations, this study proposes an improved diabetic classification framework using a chaotic Grey Wolf Optimisation (cGWO) algorithm for hyperparameter tuning of a Random Forest (RF) classifier. The cGWO enhances the standard GWO by improving its exploration and exploitation abilities, thus reducing the risk of local optima and premature convergence. Using the BRFSS dataset, the study evaluated the proposed cGWO-RF model and compared its performance against an RF and a standard RF-GWO model. The cGWO-RF outperformed the RF-GWO method, achieving a 0.03% increase in accuracy, 0.51% in sensitivity, 0.31% in precision, 0.51% in F1score, 0.2% in AUC, and 1.46% in average precision-recall because of the balance in exploration and exploitation introduced by the cGWO . These results show the efficacy of the cGWO in optimising RF hyperparameters, leading to more accurate, efficient, and generalisable models for diabetes classification. The study contributes an optimisation strategy that enhances predictive modelling in healthcare and supports more reliable decision-making for early diabetes diagnosis.