Adoption of solar dryers among smallholder farmers in developing countries remains low, despite their proven potential to enhance food security. While prior studies emphasize techno-socio-economic barriers, they often overlook individual psychological traits that likely shape adoption behaviour, assuming uniform adoption patterns and limiting targeted interventions. This study addresses this gap by examining farmers' technology readiness classes and assessing how psychological and sociodemographic factors influence solar dryer adoption. Using survey data from 447 horticultural smallholder farmers in northern Tanzania, the study was guided by the Technology Readiness Index (TRI) framework. Latent class analysis (LCA) was employed to segment farmers into distinct readiness classes, while categorical structural equation modelling (SEM) assessed the association of TRI traits, sociodemographic factors, and adoption probability. The results identified four technological readiness classes, with 95.5% of farmers in low-TR groups: hesitators (35.8%), sceptics (30.6%), and laggards (29.1%). Among TRI constructs, optimism (beta = 0.59) and innovativeness (beta = 0.32), along with education (beta = 0.13), significantly increased the likelihood of adoption from 50% to 64%, 58%, and 53%, respectively. Gender (beta = -0.323) indicated that male farmers' probability of adopting solar dryers drops to 42%. These findings highlight the importance of tailored interventions that foster positive technology attitudes, enhance innovativeness, empower women, and support education-based capacity building. By integrating LCA with SEM, the study provides a novel, empirical demonstration of how psychological and sociodemographic factors jointly shape the likelihood of solar dryer adoption, offering actionable insights for policymakers, practitioners, and researchers to design context-specific strategies.
Small and medium enterprises (SMEs) in developing countries are often resource-constrained, making the identification and effective utilization of critical internal resources essential for their survival and growth. Tanzania's leather sector illustrates this paradox, whereby despite abundant livestock resources and government-led initiatives, SMEs struggle to achieve sustainable market performance. Grounded in the Resource-Based View (RBV) and informed by Penrosean and Institutional perspectives, this study examines the effects of managerial competence and production capability on market performance and the moderating role of government support. Survey data from 145 leather manufacturing SMEs were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results show that both managerial competence and production capability significantly enhance market performance, with production capability exerting a stronger effect. Government support strengthens the relationship between managerial competence and firm performance, but does not significantly moderate the link between production capability and firm performance, suggesting a misalignment between public support mechanisms and firms' operational realities. The study advances RBV and Penrosean perspectives by demonstrating how internal capabilities interact differently with external institutional support in shaping SME performance within an under-researched African manufacturing context. Practically, the findings emphasize the need for managerial competence and production-focused capability development and more targeted policy interventions aligned with sector-specific operational needs.
Gigasiphon macrosiphon (Harms) Brenan is a rare and endangered tree endemic to the coastal forests of Kenya and Tanzania. Yet, its population status and spatial ecology remain poorly documented, particularly in Tanzania. We quantified population size, growth-stage composition, and spatial patterns of G. macrosiphon in two forest reserves in southeastern Tanzania: Kwediboma Forest Reserve and Rondo Nature Forest Reserve. Individuals across all growth stages were systematically surveyed, with diameter measurements used to assess size-class structure and regeneration status, and spatial mapping used to evaluate patterns of aggregation. Rondo supported a substantially larger and more demographically continuous population, with abundant early growth stages indicating ongoing recruitment. In contrast, Kwediboma was dominated by intermediate and mature individuals, with limited representation of germinants and seedlings, suggesting constrained or episodic regeneration. Spatial analyses revealed strong clustering consistent with localized recruitment near parent trees, with broader spatial distribution and more extensive recruitment at Rondo than at Kwediboma. Together, these patterns indicate that G. macrosiphon populations are shaped by site-specific habitat conditions and regeneration processes, underscoring the need for targeted conservation strategies to maintain recruitment and long-term population persistence.
Phishing attacks pose a growing threat to individuals and organizations globally, leveraging deceptive emails to trick users into divulging sensitive information or installing malware. Traditional anti-phishing systems struggle to keep pace with these dynamic attacks, making machine learning a promising alternative for effective detection. However, a key challenge is identifying the most influential features for accurate classification. To address this gap, we propose a hybrid approach that combines Exploratory Data Analysis (EDA) with Random Forest (RF)-based feature importance to identify the most discriminative features in phishing emails across diverse datasets. In our experiments, we analyzed feature distributions, correlations, engineering, and importance to rank the most discriminative features. Our results revealed that only 28 out of 145 features (19%) were identified as discriminative. This suggested that most features (81%) were redundant, irrelevant, or lacked predictive power, resulting in unnecessary computational complexity and hindering model performance. Using these discriminative features, the RF, XGB, and hybrid Kim’s Text CNN models were evaluated for validation. The RF and XGB models achieved accuracies of 97.6% and 97.4%, respectively, while the CNN model achieved superior performance with 99.9% accuracy, 0.9993 recall, and a perfect ROC-AUC of 1.000 when trained on the 12 features combined with text embeddings. Although this study focused only on URL-based, content-based, and domain-based features, it provides a comprehensive understanding of the distinguishing features required to build robust and efficient machine learning models for phishing email detection while substantially reducing feature redundancy.
Rangelands, covering almost 50% of our global land surface, provide essential natural resources for pastoralists and their livestock. Most pastoralists follow a nomadic or seminomadic lifestyle, which is increasingly hampered due to high human and livestock populations as well as environmental, social, and political challenges. In eastern Africa, rangeland health is additionally threatened by overgrazing, land erosion, and increasing climatic extremes. Little is known about how pastoralists perceive these challenges, what their adaptation strategies are, and whether the latter are fostering new risks and challenges. Our mixed-methods approach used semistructured interviews with 69 pastoralists in Longido and Monduli districts, northern Tanzania. We also conducted four focus group discussions of both men and women pastoralists, combined with secondary data and expert interviews on livestock populations and mortalities from governmental offices. We applied statistical analyses ( t test, analysis of variance, and Pearson's correlation) and mapped rangeland use and movements in a geographic information system. We found that most pastoralists are well aware of declining pasture quality and have adapted to climatic and environmental challenges. The most frequent response was moving further with cattle while small livestock (goats and sheep) stayed at homesteads during severe droughts. Both female and male pastoralists mentioned that these longer movement routes bore risks of conflict, diseases, and famine. These novel, drought-triggered migration routes were up to 644 km long, directed North into Kenya or South into central or coastal Tanzania. Most pastoralists additionally used supplemental feed for livestock during difficult times. We conclude that, as rangeland quality declined, routes and movement in search of pasture increased in Tanzania, leading to increasing challenges and risks associated with drought, conflicts, encountering dangerous animals, and exposure to zoonotic diseases while crossing landscapes. A holistic way of addressing these risks is urgently needed for long-term land use planning and sustainability of pastoral systems in eastern Africa. (c) 2025 The Author(s). Published by Elsevier Inc. on behalf of The Society for Range Management. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )
