Forecasting the adoption and spread of a community-based marine management initiative using agent-based models

Abstract

While many successful initiatives for conserving nature exist, efforts to take them to scale have been inadequate. Moreover, conservation science currently lacks a systematic methodology for determining if or when interventions will reach effective scales and how programmatic decisions will affect the scaling process. This paper presents a modelling framework that aims to address both issues by operationalizing Diffusion of Innovations theory and local knowledge using agent-based modelling and Bayesian inference. By applying our framework to existing data on the spatiotemporal adoption of a community-based marine management initiative in Fiji, we demonstrate that our approach can identify the mechanisms that govern the observed adoption patterns. In this case, the relative advantage of the intervention, village social networks, and perceived knowledge stand out as important drivers of adoption. Using the identified causal processes, our approach can forecast business-as-usual and counterfactual future scenarios and hence inform conservation policy. Finally, we highlight the importance of spatiotemporal data for making detailed scaling predictions. We structure the paper as a step-by-step guide, highlighting our modelling decisions and possible limitations. Thus, besides presenting a case study, this paper serves as a template for practitioners and researchers to better model the scaling process of other conservation interventions.