A meta-network analysis of methodological specifications for system dynamics modelling application in agricultural food systems

This paper aims to review existing applications of system dynamics modelling in agricultural food systems and draw insights from the various modelling procedures in order to highlight best-practice guidelines on methodological specifications for SD modelling. A meta-network analysis was used to identify existing studies that have applied SD modelling in agriculture. Using an algorithm that automatically clusters closely connected research articles based on Boolean search strings that look at the title, keywords, abstract, and digital object identifier (DOI) of the journal articles, 354 journal articles were selected for in-depth content analysis. Based on the synthesised trends, two criteria for determining the type of modelling process and model type to apply for the model conceptualisation step are: (i) the immediate end goal of the modelling process, and (ii) data availability. Participatory modelling is appropriate when there is limited data and model outputs will inform the implementation of interventions by stakeholders. For action research focusing on well-researched food systems with substantial data available, the semi-participatory modelling process can be adopted, and quantitative SD models can be solitarily used. A key contribution of this paper is the proposed procedure for emergent participatory scenario development within the system dynamics modelling process. [Display omitted] •The term ‘scenario’ is the most used keyword in system dynamics (SD) modelling.•Scenarios development are often researcher-driven.•Non- or semi-participatory modelling processes have been predominantly used in SD modelling.•Data triangulation is used as a mechanism to achieve stakeholder inclusion in SD modelling.