Grounded in local knowledge: Smallholder farmers’ mental models of soil quality in Madagascar

Soil degradation poses a severe threat to smallholder farmers' livelihoods in Madagascar, yet farmers' own perceptions of the processes affecting soil quality remain poorly understood. This study analyzes the mental models of 1,007 smallholder farmers in south-eastern Madagascar, examining how environmental and socio-economic characteristics shape their understanding of soil quality processes. As a second objective we examine whether mental model elicitation influences agronomic knowledge and adoption intentions. Using a representative survey with randomized assignment to the mental model elicitation task, we find that Malagasy smallholder farmers hold moderately complex mental models, most often drawing direct connections between drivers and soil quality. Manure, rainfall, and heat are the most frequently included drivers, with manure perceived as most beneficial for soil quality. Education emerges as the strongest predictor of mental model complexity, with complexity increasing progressively across education levels. Sex, age, and local climatic conditions also shape both complexity and the inclusion of specific drivers. Mental model elicitation did not meaningfully improve agronomic knowledge test scores, but was positively associated with intentions to adopt manure application. These findings underscore the value of farmers' systems thinking and highlight the importance of accounting for socio-demographic and environmental heterogeneity when seeking to understand local agricultural knowledge systems. They further suggest that structured mental model elicitation holds potential as a tool for targeted agricultural extension, particularly for practices that are already salient in farmers' mental models and positively perceived.