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

Explainable data-driven approach for smart crop yield prediction in Sub-Saharan Africa: Performance and interpretability analysis

(2026-04) Olatinwo, DD; Myburgh, HC; De Freitas,A; Abu-Mahfouz, Adnan MI

The increasing demand for innovative strategies in sustainable food production—driven by rapid global population growth, particularly in sub-Saharan Africa (SSA)—necessitates urgent attention to agricultural resilience. Recent technological advancements have enhanced crop productivity, post-harvest preservation, and environmentally sustainable farming practices. However, three critical bottlenecks remain: (i) the lack of accurate, maize-specific yield prediction methods tailored to SSA; (ii) limited multimodal modeling approaches capable of capturing complex, nonlinear interactions among heterogeneous data sources; and (iii) a lack of explainability mechanisms, which render high-performing models “black boxes” and hinder stakeholder trust. To address these gaps, this study presents an explainable machine learning framework for smart maize yield prediction. We integrate multimodal SSA-specific soil, crop, and weather data to capture the multi-dimensional drivers of maize productivity. Six diverse algorithms—including extreme gradient boosting (XGBoost), light gradient boosting machine (LGBM), categorical boosting (CatBoost), support vector machine (SVM), random forest (RF), and an artificial neural network (ANN) combined with a k-nearest neighbors (kNN)—were benchmarked to evaluate predictive performance. To ensure robustness against spatial heterogeneity, we employed a Leave-One-Plot-Out (LOPO) cross-validation strategy. Empirical results on unseen test data identify CatBoost as the best-performing model, achieving a coefficient of determination of (𝑅2 =~76%), demonstrating its ability to capture complex, nonlinear relationships in agricultural data. To enhance transparency and stakeholder trust, we integrated Local Interpretable Model-agnostic Explanations (LIME), providing plot-level insights into the physiological and environmental drivers of maize yield. Together, these contributions establish a scalable and interpretable modeling framework capable of supporting data-driven agricultural decision-making in SSA.

Item

Trait–based adjustments: Key to improving model representation of the bloom seasonal cycle in the Subantarctic Zone

(2025-11) Mashifane, Thulwaneng B; Tagliabue, A; Thomalla, Sandy J

Southern Ocean phytoplankton have unique traits that allow them to survive extreme environmental conditions of low iron, light, and temperature. These traits are not well represented in Earth System Models and may lead to misrepresentation of the bloom seasonal cycle in this critical oceanic region. This has implications for understanding the role of the biological carbon pump in driving ocean carbon uptake and storage. We conduct a sensitivity analysis using a 1D model in the Subantarctic Zone to identify (a) the most sensitive parameters for improving representation of the seasonal cycle and its mechanisms and (b) the impact of these changes on carbon export. Lowering the phytoplankton iron quota or increasing zooplankton iron stoichiometry and both improve the seasonal cycle timing and amplitude, aligning chlorophyll and carbon seasonal cycle. However, the resulting increases in bloom amplitude translates into minimal change in carbon export due to enhanced cycling through the dissolved organic carbon pool.

Item

Isolation and characterisation of phage-displayed scFv antibodies targeting PfHSP70 and PfLDH of plasmodium falciparum

(2026-07) Gasa, NL; Zuma, LK; Chiliza, TE; Kwezi, Lusisizwe; Luthuli, SD; Pooe, OJ

Plasmodium falciparum is the most virulent human malaria parasite and is responsible for numerous deaths annually. The increasing resistance of P. falciparum to antimalarial drugs necessitates the development of improved diagnostic tools for timely malaria detection. Malaria biomarkers such as PfHSP70 and PfLDH are highly valuable for malaria detection because they are essential for parasite survival and are consistently expressed during infection. PfHSP70 is associated with the parasite’s stress adaptation and proteostasis mechanisms under febrile and drug-induced conditions, while PfLDH plays a central role in glycolytic metabolism and redox balance. Their functional importance, parasite specificity and elevated expression during active infection make these proteins reliable molecular indicators for the sensitive and specific detection of P. falciparum, thereby supporting their potential application in rapid diagnostic and biosensing platforms for malaria surveillance and disease management. In this study, an M13 phage-displayed single-chain variable fragment (scFv) antibody library was used to screen for antibodies against recombinant PfHSP70 and PfLDH. Four rounds of biopanning were conducted to enrich high-affinity binders, followed by ELISA, UV-visible spectroscopy and microscale thermophoresis (MST) to evaluate specificity and binding affinity. Functional interactions were assessed in Escherichia coli expressing PfHSP70 or PfLDH. Recombinant PfHSP70 and PfLDH were successfully expressed and purified from E. coli, with approximately 50% of selected colonies demonstrating significant binding to both targets, confirming the enrichment of antigen-specific phages. Specific phages were validated using ELISA and transmission electron microscopy (TEM). Selected scFvs exhibited strong target binding, with MST-determined dissociation constants (Kd) of 7.47 μM for PfHSP70 and 3.64 μM for PfLDH. Exposure to scFv-displaying phages impaired the survival of PfHSP70- and PfLDH-expressing E. coli and induced spectral shifts consistent with protein–antibody interactions. This study validates phage display as a robust platform for isolating high-affinity scFvs against P. falciparum targets. These binders have the potential to develop into low-cost diagnostic tools, addressing the urgent need for novel diagnostic interventions. Nevertheless, further studies are required to confirm binding specificity and assess translational applicability.

Item

Practical deployment of a private NB-IoT network with open-source platforms

(2026-08) Makhalanyane, Thapelo; Hlabishi, Kobo I

Narrowband Internet of Things (NB-IoT) has emerged as a Low Power Wide Area Network (LPWAN) technology for industrial and enterprise IoT applications, offering licensed spectrum operation with guaranteed Quality of Service (QoS). This work presents a comprehensive, step-by-step guide to deploying a private NB-IoT network using open-source platforms. Existing literature on open-source NBIoT implementations is reviewed, and candidate platforms, including OpenAirInterface (OAI) and srsRAN, are evaluated against commercial solutions. A suitable open-source NB-IoT prototyping platform is then selected based on its demonstrated compatibility with commercial offthe-shelf (COTS) user equipment and robust support for 3GPP Release 13/14 features. The complete system architecture is detailed, including the eNodeB implementation on USRP B210 software-defined radio hardware and its integration with a lightweight Evolved Packet Core (EPC) optimised for IoT applications. A comprehensive testing methodology is employed to progressively validate the system, from basic signal generation and synchronisation through broadcast channels and random access procedures to end-to-end data connectivity. Experimental results using commercial off-the-shelf equipment demonstrate successful deployment, achieving a measured uplink throughput of 11 kbps, a downlink throughput of 17 kbps, and latency ranging from 300 ms to 15 s, depending on signal conditions. This work provides researchers and practitioners with a fully reproducible framework for establishing cost-effective private NB-IoT networks suitable for industrial automation, smart agriculture, environmental monitoring, and enterprise IoT deployments.

Item

Dry sliding wear behaviour of laser cladded altisicrco high entropy alloy coatings on Ti6Al4V: influence of CR/CO and AL/TI enrichment

(2026-08) Raselabe, KM; Makhatha, ME; Makoana, Nkutwane W; Skhosane, Besabakhe S

This study investigated how compositional variation within the AlTiSiCrCo high-entropy alloy (HEA) system affects the microstructure, hardness, and dry sliding wear behaviour of laser-cladded coatings on Ti6Al4V. Three coatings, namely, equiatomic (HEA 1), Cr/Co-enriched (HEA 2), and Al/Ti-enriched (HEA 3), were characterized by SEM, EDS, XRD, and Vickers microhardness and tested for dry sliding wear using a ball-on-disc tribometer at 5 N and 15 N. All coatings comprise a BCC solid solution matrix reinforced by intermetallic precipitates. HEA 2 and HEA 3 gave the highest hardness (755 HV and 754 HV, respectively) against 705 HV for HEA 1 and 345 HV for the Ti6Al4V substrate. All HEA coatings reduced wear rate relative to Ti6Al4V; HEA 2 recorded the lowest rate (4.570×10−5 mm3/N.m and 2.744×10−4 mm3/N.m at 5 N and 15 N, respectively), well below the substrate (2.257×10−4 mm3/N.m and 0.0014 mm3/N.m at 5 N and 15 N, respectively). Worn surface analysis showed abrasive/delamination at 5 N transitioning to more severe abrasive, adhesive, and delamination wear at 15 N. The enhanced wear resistance of HEA 2 stems from the BCC solid solution strengthening, intermetallic reinforcement, and high chromium content, which aids in the formation of the Cr2O3 protective oxide film. Overall, enriching the coating with chromium and cobalt proved to be the most effective approach for improving the tribological performance of laser-cladded AlTiSiCrCo HEA coatings on Ti6Al4V.