Development of a smartphone-based decision support system for grain post-harvest management in Zimbabwe.
MetadataShow full item record
An information gap exists regarding current crop postharvest management practices and perceptions of smallholder farming communities who are having to face increasingly variable climatic conditions. To deepen understanding of this topic, two districts in Zimbabwe - Mbire and Hwedza - were selected based on their distinctly different agroecological profiles. A multi dimensional survey was used to explore the practices and perceptions of the communities and stakeholders in the selected districts regarding grain postharvest management in the context of climate change. The survey involved 601 household interviews, six focus group discussions with women and men, and interviews with 40 district stakeholders and 53 community key informants. Farmers and stakeholders articulated how their food security is threatened by weather and climate variability which were negatively affecting maize and sorghum yields to less than 0.5 t/ha. High grain storage insect pest pressure has led to misuse of chemical grain protectants thus potentially increasing associated health risks. Bulk grain storage in traditional outdoor free standing granaries is less popular with a shift towards the use of polypropylene bags stored in their living quarters for security reasons. Agricultural extension officers are the most common sources of agronomic and postharvest information for farmers with indigenous knowledge still considered useful. Stakeholders highlighted the Larger Grain Borer, Prostephanus truncatus as being a particularly notorious stored maize insect pest. Reported challenges concerning efficacy of chemical pesticides and the limited information about the grain storage pest profile for maize in Zimbabwe prompted an investigation on pest dynamics and the possibility of modelling grain damage and P. truncatus populations in the study sites as influenced by recorded key climatic variables (relative humidity and temperature). A strong relationship (r>0.45) was found between P. truncatus and its associated insect parasitoids over two storage seasons. The boring and feeding activity of P. truncatus produced copious amounts of dust which was closely associated with the level of insect grain damage, and in the process, encouraged the development of secondary pest infestation on the grain. Using machine learning techniques, it was determined that populations of P. truncatus and grain damage level were closely related to patterns in relative humidity more than temperature. A decision tree algorithm was used to produce a model for predicting insect grain damage and P. truncatus population in stored grain, respectively. Based on literature, anecdotal evidence and observations during the grain storage experiments, a further exploratory study was conducted using insect pheromone traps to detect and model the flight activity of P. truncatus in the natural environment using weather variables as key inputs in Mbire, Guruve and Hwedza districts. Flight activity of P. truncatus was shown to be maximum when relative humidity was above 60% and temperature between 20°C and 30°C, conditions which typically occur during the southern African rainy season from December to April. Harvesting is conducted between February and April hence the increased flight activity was deduced to be the likely reason why grain damage in stored structures tended to be high during the same periods. Machine learning was used to show that high relative humidity was more influential in initiating P. truncatus flight while higher temperatures alone seemed to deter flight activity. A model was developed using a decision tree algorithm to predict flight activity of P. truncatus using weather variables and trap site elevation as key variables. The models for grain damage, P. truncatus populations and P. truncatus flight activity were used to develop a mobile application to operate on an Android smartphone based on standard Android design principles. The application was designed as a decision-support prototype tool intended to encourage further development to help farmers, extension agents and stakeholders in planning their postharvest activities to protect their stored food. Decision support tools and early warning systems are pivotal in disaster risk mitigation and development of these tools using easily accessible platforms and tools can help facilitate better mitigation of seed and grain damage loss in smallholder farming communities and increase food and nutrition security
Additional Citation InformationNyabako, T. (2020). Development of a smartphone-based decision support system for grain post-harvest management in Zimbabwe. [Unpublished Masters thesis]. University of Zimbabwe.
University of Zimbabwe