A Decision-Support Tool for Managing Tree-Cover in Smallholder Cocoa Agroforestry Systems

Abstract In cocoa agroforestry systems (AFS), tree-cover is a fundamental driver of ecosystem functioning and crop productivity. However, for smallholders, implementing specific cover targets remains challenging because tree-cover is difficult to measure without specialized equipment and technical skills. Consequently, sustainability standards and technical guidelines typically rely on simplified structural metrics, such as tree density or basal area, as proxies for cover. This study addresses the technical disconnect between these simplified metrics and actual canopy outcomes. Using data from 150 plots in Côte d’Ivoire, we developed a Beta regression model within a Bayesian framework to estimate tree-cover from standard inventory data. Our results show that tree density or basal area alone are unreliable predictors of cover, whereas a model integrating both variables achieves high accuracy (R2 = 0.89). Our results also demonstrate that existing guidelines can lead to highly variable canopy conditions that deviate markedly from intended cover targets. We propose a state-space representation as a decision-support tool to help smallholders estimate tree-cover and to provide a scientific basis for updating existing policies and standards. This framework allows smallholders to visualize management trajectories, driven by recruitment, tree growth, and mortality, to steer their systems toward ecologically meaningful cover targets.