Applying Bayesian networks to evaluate small-scale farmers’ perceptions of native reforestation practices in Brazil’s Caatinga biome

Abstract

In Brazil’s semi-arid Northeast, most rural dwellers derive income from the dry Caatinga forest through livestock farming, fruit collection, and firewood extraction. However, recurring droughts and inadequate land use practices jeopardize farmers’ livelihoods. The drought-resistant, endemic Umbuzeiro tree provides fruit for direct consumption and allows for the creation of transformed products. The planting of this native species can enhance the well-being of the ecosystem and establish future benefits for smallholdings. However, it is crucial that when taking up innovative practices to cope with environmental change, a willingness to apply them should be fostered among local farmers. We used constellation analysis as a transdisciplinary approach to identify elements of current land management which subsequently defined the nodes of a Bayesian network (BN). We developed probabilities of practice uptake that strengthen success, namely the conservation of natural resources while securing the incomes of smallholders. In collaboration with stakeholders and experts, 25 identified nodes for the BN were tested under various scenarios. Adopting all suggested innovative practices secures the final objectives—ecosystem health and farmer benefits (approx. 90%). The analysis quantified the relevance of single issues that may impede farmers to adopt the practices, such as having to cultivate seedlings or avoiding long-term investments. Further crucial actions include the fencing-off of livestock and marketing pathways. Affordable credit, research, and supportive farmers’ institutions can encourage farmers to implement innovative practices. The use of modeled scenarios can provide evidence, which might encourage sustainable land management.