Each year, many millions of euros are spent on interventions aiming to introduce new crop varieties to help improve food security in developing countries. The aim of these new varieties is to increase the crop yield and/or their resilience against extreme climate conditions, such as drought. Only recently, bottom-up farmer evaluations of new varieties have reached a scale in which they become viable as an alternative to inform variety introduction decisions, thanks to crowd-sourced citizen science (van Etten et al., 2019). Farmers test varieties on their own farm and report their findings so that they can be analysed to devise variety recommendations over larger regions and under several abiotic conditions. An example of such crowdsourced recommendations is depicted in the maps above which were taken from Brown et al. (under review).
An important question that remains is: What is the right combination of interventions to generate a learning system (bottom-up individual learning, peer-to-peer learning, top-down push marketing) to help farmers to adopt new seed technology that is suitable for their own conditions? In other domains, the influence of policy interventions on the interactions between actors has been studies with Agent-Based Models (ABMs) (e.g. Moncada et al. 2018). ABMs are fit for this purpose, because they simulate how macro-level patterns (here seed technology adoption levels) emerge from the interaction of micro-level entities (here farmers). In addition, ABMs can capture the non-linearity of these interactions (e.g. because of learning), feedback loops, and the path-dependence of the system (e.g. one variety becoming dominant because it performed well in one specific year’s weather conditions).
In this thesis, you will build a simple ABM of farmers that cultivate crops and interact with and learn from their neighbors to find the best-performing seed technology. With this ABM, you will analyze a set of scenarios that represent different individual or combinations of policy interventions in order to answer the question above.
- Determine relevant farm typologies
- Spatially analyse crop variety adoption by farmers using agent-based modelling
- Compare adoptive bottom-up variety adoption via peer-to-peer learning (and policies stimulating this) with conventional top-down intervention approaches
- Add complexity by considering other agents such as seed vendors (optional)
- Model learning amongst farmers in a more sophisticated manner with algorithms like reinforcement learning within the context of seed adoption (optional)
- Brown, D. et al., under review. Rank-based data synthesis of common bean on-farm trials across four Central American countries.
- Moncada J.A., Verstegen J.A., Posada J.A., Junginger M., Lukszo Z., Faaij A.P.C., Weijnen M. (2018). Exploring policy options to spur the expansion of ethanol production and consumption in Brazil: An agent-based modeling approach. Energy Policy 123, 619-641. DOI: 10.1016/j.enpol.2018.09.015.
- Pautasso, M., et al., 2013. Seed exchange networks for agrobiodiversity conservation - A review. Agronomy for Sustainable Development, 33 (1), pp. 151-175.
- Richards, P., et al., 2009. Seed systems for African food security: linking molecular genetic analysis and cultivator knowledge in West Africa. Int. J. Technology Management, Vol. 45, (1/2), pp.196–214.
- van Etten, J., et al., 2019. First experiences with a novel farmer citizen science approach: crowdsourcing participatory variety selection through on-farm triadic comparisons of technologies (tricot). Experimental Agriculture, Vol 55(S1), pp. 275–296
- We look for a motivated student with strong analytical skills and research interests in agent-based modelling.
- Ideally, the student should have followed the course Spatial Modelling & Statistics or another course in which agent-based modelling was taught.
Theme(s): Modelling & visualisation; Human - space interaction