National Agricultural Research Organization, National Coffee Research Institute, P. O. Box 185, Kituza-Mukono, Uganda and Uganda Christian University, Department of Computing and Information Technology, P. O. Box 4, Mukono, Uganda
Auto-generated profile, based on publicly available information
Daphne Bitalo
National Agricultural Research Organization, National Coffee Research Institute, P. O. Box 185, Kituza-Mukono, Uganda and Uganda Christian University, Department of Computing and Information Technology, P. O. Box 4, Mukono, Uganda
The tope shark (Galeorhinus galeus Linnaeus, 1758) is a temperate, coastal hound shark found in the Atlantic and Indo-Pacific oceans. In this study, the population structure of Galeorhinus galeus was determined across the entire Southern Hemisphere, where the species is heavily targeted by commercial fisheries, as well as locally, along the South African coastline. Analysis was conducted on a total of 185 samples using 19 microsatellite markers and a 671 bp fragment of the NADH dehydrogenase subunit 2 (ND2) gene. Across the Southern Hemisphere, three geographically distinct clades were recovered, including one from South America (Argentina, Chile), one from Africa (all the South African collections) and an Australia-New Zealand clade. Nuclear data revealed significant population subdivisions (FST = 0.192 to 0.376, p<0.05) indicating limited gene flow for tope sharks across ocean basins. Marked population connectivity was however evident across the Indian Ocean based on Bayesian clustering analysis. More locally in South Africa, F-statistics and multivariate analysis supported moderate to high gene flow across the Atlantic/Indian Ocean boundary (FST = 0.035 to 0.044, p<0.05), with exception of samples from Struisbaai and Port Elizabeth which differed significantly from the rest. Discriminant and Bayesian clustering analysis indicated admixture in all sampling populations, decreasing from west to east, corroborating possible restriction to gene flow across regional oceanographic barriers. Mitochondrial sequence data recovered seven haplotypes (h = 0.216, π = 0.001) for South Africa, with one major haplotype shared by 87% of the individuals and at least one private haplotype for each sampling location except Port Elizabeth. As with many other coastal shark species with cosmopolitan distribution, this study confirms the lack of both historical dispersal and inter-oceanic gene flow while also implicating contemporary factors such as oceanic currents and thermal fronts to drive local genetic structure of G. galeus on a smaller spatial scale.
Little attention has been paid to on-farm cocoa fermentation practices, leading to poor-quality cocoa beans. A cross-sectional study was conducted in three agroecological zones of Uganda (Lake Victoria Crescent, Western Rangelands, and Lake Albert Crescent) to investigate on-farm fermentation practices, the association between farmer training and adoption of recommended fermentation practices as well as the impact of fermentation practices on cocoa quality. Data on fermentation practices and farmer training was collected using a questionnaire. The quality of cocoa beans collected from interviewed farmers was assessed for bean cut test, fermentation index, pH, and titratable acidity following recommended methods. Results revealed significant variation in fermentation practices used by farmers, with over 65% employing rudimentary methods such as sack, bucket, and jerrycan. Training of farmers did not increase the likelihood of farmers adopting the recommended fermentation practices. As a result, the quality of fermented cocoa beans was low with only 40-51% of the beans from all zones exhibiting the fully brown color, which is below the 60% threshold for commercial-grade cocoa beans. Furthermore, the fermentation index of cocoa beans from all zones ranged from 0.61-0.67, indicating a high percentage of under-fermented beans. Farmers who used the recommended practices (heap and box fermentation for 5-7 days while turning cocoa) in each zone, produced cocoa with a significantly (p<0.05) higher percentage of fully fermented beans. The continued use of rudimentary postharvest handling practices among farmers leads to the production of poor-quality cocoa beans. It is important to strengthen the enforcement of recommended fermentation practices to improve the quality of cocoa beans and ensure a sustainable market.
High-throughput drought phenotyping in perennial crops is limited by slow, costly, destructive, and labour-intensive methods that restrict genetic gain. Here, we developed and validated a portable near-infrared spectroscopy (NIRS) framework for drought phenotyping in Coffea canephora. The framework integrates variance decomposition, ensemble machine learning, the Drought Factor Index (DFI), and temporal trajectory modelling to distinguish drought responses among 22 genotypes under controlled water deficit. Leaf spectra were collected from both adaxial and abaxial surfaces using a handheld MicroNIR sensor (908–1676 nm), alongside predawn leaf water potential and chlorophyll fluorescence measured using conventional methods. Abaxial measurements reduced residual variance and improved the signal-to-noise ratio compared with adaxial measurements, identifying the abaxial surface as the preferred target for sensor deployment. Reduction of Repeatability Error–ASCA (REP-ASCA) partitioned spectral variance into biologically structured components. Drought treatment, genotype, and their interaction each explained significant proportions of spectral variance (p < 0.001). Ensemble machine learning models (Random Forest and XGBoost) accurately predicted predawn leaf water potential (R² = 0.82, RPD = 2.42, CCC = 0.90). In contrast, chlorophyll fluorescence and gas exchange traits were poorly predicted (R² < 0.15), revealing trait-specific sensing limitations imposed by the device’s restricted spectral range. Temporal modelling of NIRS-predicted water potential trajectories produced a slope-based classification metric that grouped genotypes as drought-tolerant, intermediate, or sensitive. Drought-tolerant classifications showed 75% concordance with DFI, demonstrating that NIRS can serve as an interpretable and scalable genotype-selection pipeline for climate-resilient coffee breeding, rather than merely as a static trait predictor.