Predicting aquaponics adoption among smallholder fish farmers in Bhutan using machine learning

Aquaponics has significant potential as a circular economy technology for sustainable food production, particularly for resource-constrained smallholder farmers in developing countries. However, its adoption remains largely limited to a handful of developed countries. A major challenge is the lack of empirical insights needed to inform effective government policies and business development that can promote broader adoption. Using nationally representative survey data on smallholder fish farmers in Bhutan, this study developed a random forests classification model to identify personal, farm-level, and institutional factors that predict the propensity of farmers to adopt aquaponics. Results suggest that adoption propensity is strongly associated with farmers’ perception of its benefit, access to institutional support and markets, and existing fish production skills. These results provide practical guidance for policies and programs that aim to promote aquaponics as a sustainable food production technology. The findings highlight the need to raise awareness, expand training opportunities and support, and strengthen market linkages to make adoption more feasible and attractive for smallholder farmers. By addressing these areas, governments and development partners can design more effective interventions to accelerate the diffusion of aquaponics and promote sustainable food production systems in resource-constrained environments.