Development of in-field diagnositc tool for soil nutrient screening.

The use of external inputs in the form of inorganic fertilisers is rising across the world. Rapidly growing crops such as vegetables necessitate high fertiliser inputs, whilst remaining an attractive cash-crop option for farmers, especially smallholders in the developing countries. For vegetable farming to be sustainable, these inputs should be monitored so the crop nutrient use efficiency is high and the potential for under- and over-fertilisation is low. Therefore, there is a need for the development of low-cost tools that can bring site-specific soil information to farmers who do not ordinarily have access to such knowledge. In recent years, smartphone technology has given rise to a number of advanced apps that aim to improve agronomic production, especially in the Southern Hemisphere. The work in this thesis centres around method development and appraisal for the application of a smartphone-mediated diagnostic tool for use in soil nutrient screening. A smartphone application marketed as Akvo Caddisfly, used together with nutrient-sensitive test strips was repurposed for the analysis of soil samples. The app was used alongside selected test strip types and underwent rigorous laboratory testing to evaluate its suitability for soil analysis and to identify its strengths and weaknesses. The laboratory-based experiments allowed for the development of soil extraction, filtration and analysis methodologies, through the utilisation of variable soil samples obtained from Indonesia, an approach subsequently employed in field conditions in other study sites. The field-based experiments were undertaken in the People’s Republic of China, Ghana and Kenya, allowing for a critical appraisal of smartphone-mediated soil analysis as an effective tool for fertiliser recommendations in smallholder vegetable production. In China, where frequent over-fertilisation of crops is the chief cause of soil acidification and heavy metal pollution as well as eutrophication of waterbodies and high N₂O emissions, smartphone-mediated soil analysis was employed successfully in identifying over-fertilised plots. In contrast, in Sub-Saharan Africa, where soil Nitrogen content was low, smartphone-mediated soil analysis encouraged farmers to apply organic fertilisers to improve their yields. Referencing the metadata, which was collected during laboratory and field-based experiments, a framework for designing and evaluating future in-field soil test kits was created. The data consisted of a collation of quantitative analyses and qualitative observations and these were synthesised into a step-by-step process that can be used at the test kit evaluation stage to reduce the time and costs associated with their development. Finally, a range of statistical approaches were employed to investigate the level of agreement between the in-field method and the accepted laboratory standard methods employed in agricultural soil analysis. They were described in detail to encourage their wider application in method comparison studies across environmental science.