Handling multiple levels in agent-based models of complex socio-environmental systems: A comprehensive review

Agent-based modeling (ABM) has been successfully used, since its emergence in the 1990s, to model and simulate dynamics at work complex socio-environmental systems, many domains applications where interactions between people their environments give rise emergent phenomena that are difficult study otherwise (urban planning, land-use change, adaptation environmental changes, biodiversity protection socio-ecosystems, pollution control, etc.). The inclusion of multiple levels analysis, abstraction, representation these models, however, is much more recent still subject proposals discussions within a relatively informal field, Multilevel Agent-Based Modeling (ML-ABM), which most often presented as an approach extends classical ABM paradigm include multilevel concepts. Over past decade, ML-ABM increasingly adopted explored by researchers effective for framing defining mechanisms underlying dynamics. However, due youth no single definition, methodology, or tool unifies studies this rapidly expanding area. This review will begin with introduction systems (SES) challenges approaches face representing them properly, especially regarding complexity human behaviors organizations. presents opportunities SESs respect challenges, including simulation individual social behavior ability provide descriptive generative simulated system. limited represent scales, concepts absent from metamodel. A complete literature be carried out, structured around continuum emerged during review: distribution (and thus, software engineering perspective, control) across levels, allow only one level active time, rely on simultaneous activity feedback loops several levels. Different design choices will, meet different needs multi-level representation, focusing interest modelers strengths limitations each. In particular, we highlight limitation shared all reviewed approaches, namely inability parallel hierarchies interactions, capability appears crucial finely SES. new perspective AGR could allows us conclude research perspectives open.