I build probabilistic models of agricultural and land-use decisions together with the people who make them. Most decisions in development and farming are made under deep uncertainty, with little data and many competing goals. My work makes that uncertainty explicit, so that farmers, organisations and policymakers can compare options on what they are likely to gain, risk and trade off.
I use Monte Carlo simulation, Bayesian networks and expert knowledge elicitation to model interventions before money is spent on them. Models are built with stakeholders, then used to identify the variables that matter most and where better information would change the decision.
Why farmers adopt, adapt or abandon trees on farms, and what restoration and carbon schemes deliver in practice. Much of this work is with smallholders in Northwest Vietnam and East Africa.
How homegardens, wild foods and school food environments contribute to diets, and how to weigh investments in them against nutrition, biodiversity and economic outcomes.
Documenting how communities know, use and value plants, from homegardens in Uganda to medicinal teas on the Silk Road, and building open tools to analyse that knowledge.
My models are written in R and shared openly, mainly through the decisionSupport and ethnobotanyR packages. Tutorials and teaching material are on the Software & Tools page.