Options for Sustainable Agricultural Intensification in Maize Mixed Farming Systems: Explorative ex-ante assessment using Multi-Agent System Simulation

Abstract

Nutrient depletion is a major limiting factor to agricultural sustainability in cereal dominated smallholder farming systems in Africa where over 80% of arable land is unsuitable to support primary productivity. This constrains food and nutritional security of rural communities. For appropriate design of interventions, there is need for empirical evidence on drivers of change. A common sampling frame is used to integrate social-ecological data from farm surveys of soil, biomass and crop yield, nutrient inputs and outputs, and their determinants. The nutrient distributions are predicted using randomForest machine learning algorithm in R with remotely sensed reflectance for topography (30 m STRM-DEM), vegetation and soil (10 m Sentinel2 imagery) as co-variates. We use behavioural economics to unravel farm-type specific drivers of human induced nutrient inputs and a mixed model for crop yield function for outputs. Further, existing nutrient stoichiometry and transfer functions based on NUTMON, FarmDESIGN models with parameters from the study region are used to capture dynamic stocks and flows. Lastly, we build a multi-agent system for simulating sustainable agricultural intensification (MASSAI) in NetLogo and piloted to explore, ex ante, the agentic behaviours of farmers when faced with ambiguity in fertilizer subsidy regimes and its implications on nutrient budgets, human decision making and land productivity. Though soil management in smallholder farming systems aims at addressing the most critical nutrient(s), the results from this study show that the soils are deficient in all three major nutrients (NPK) and structurally unstable due to low soil organic carbon (SOC). Farmers strive to utilise the commonly available soil fertility management: nine in every ten households used inorganic fertilizers, a third integrated legumes and almost half applied manures of various forms. From the empirical and simulated results, it is indicative that the maize mixed smallholder farming system in Malawi has become inelastic to changes in input policies. Much as improvement in contribution of women in decision-making widens the scope for legume cropping, it negatively affects manuring. Ther efore, addressing challenges that women face in manuring could offer greater opportunities for integrated soil fertility management. After 15 years of fertilizer subsidy program, farmers have internalized it in their expenditure plan: some exclusively relying on subsidy while others source increasing amounts from the market and are becoming self-reliant. Those that rely on limited fertilizer acquired through subsidy proactively reduce the nutrient gap by increasing manuring. These behaviors have implications on nutrient management and sustainability of the farming systems. Although subsidy alone might not significantly shift the nutrient and productivity trajectories for the next 20 simulated years, increased subsidy could relatively accelerate nitrogen and phosphorus losses.