Family farmers and biodiesel production: Systems thinking and multi-level decisions in Northern Minas Gerais, Brazil

• Bayesian network modelling for farming systems assessment of biodiesel production in Brazil.
• Explored field, farm, industry and policy decisions impacting family farmer engagement.
• Decisions supporting successful family farmer engagement are not singular.
• Need to minimise trade-offs between biodiesel crop and animal production.
• Field-level management must be complemented by higher-level management.

This study focuses on family farmer engagement in the Brazilian national programme for Production and use of Biodiesel (PNPB). The Brazilian government has been promoting the role of family farmers as producers of biomass for biodiesel since 2004; however, fewer than expected family farmers have decided to produce biomass for biodiesel. The North of Minas Gerais is one region where a biodiesel plant has been strategically located to source castor beans grown by family farmers. The target family farm type in this region specializes in beef and/or dairy production with low input pasture (approximately 30 ha per farm), maize intercropped with beans (approximately 1 ha per farm) and sugarcane (approximately 1 ha per farm). We selected this region for a case study to explore management decisions of farmers, industry and policy makers that influence family farmer engagement with biodiesel production through cultivation of castor beans. To evaluate outcomes for family farmers engaging with the PNPB, we focused on how cultivation of castor beans impacts family farmers in terms of income levels, income stability and levels of milk production. We used an application of systems thinking known as Bayesian network modelling (BNM). BNM was chosen for its suitability to integrate different types of knowledge and to include quantitative and qualitative variables. The study was built on a body of scientific literature explaining why family farmers have not been cultivating castor beans for biodiesel production and a body of experiential knowledge of local actors (farmers, extension officers, policy makers, biodiesel manufacturers and researchers in North of Minas Gerais). The complete BNM consisted of a ‘cause and effect’ diagram where the strengths of the causal relationships were quantified with elicited opinions from surveyed local actors. We used the complete BNM to explore scenarios that could improve outcomes for family farmers and consequently increase their level of engagement. For example, we addressed subsidy structures of the PNPB, crop management, farm-level trade-offs and value-chain innovations. We demonstrate that decisions to support family farmer engagement with biodiesel are not singular. Engagement by family farmers requires simultaneously: improvements in technical crop management, reductions in farm-level cash constraints and innovations in the production chain such that engagement of family farmers goes beyond cultivation of one more low-value crop. Finally we discuss some methodological issues from this application of BNM to farming systems research.