Selection of movement rules to simulate species dispersal in a mosaic landscape model

Dispersal is an ecological process central to population dynamics, describing one of the most important movement behaviours between populations and across landscapes. In spatial population models for terrestrial vertebrates, capturing and portraying plausible dispersal behaviour is of particular importance when considering the spread of disease or invasive species. The distribution of distances travelled by dispersers, or the dispersal kernel, is typically highly skewed, with most individuals remaining close to their origin but some travelling substantially further. Using mechanistic models to simulate individual dispersal behaviour, the dispersal kernel can be generated as an emergent property. Through stepwise simulation of the entire movement path, models can also account for the influence of the local environment, and contacts during the dispersal event which may spread disease. In this study, we explore a range of simple rules to emulate individual dispersal behaviour within a mosaic model generated using irregular geometry. Movement rules illustrate a limited range of behavioural assumptions and when applied across these simple synthetic landscapes generated a wide range of emergent kernels. We establish that naturalistic kernels can emerge when simulating dispersal across irregular mosaic landscapes. Given the variability in dispersal distances observed within species, our results highlight the importance of considering landscape heterogeneity and individual-level variation in movement, with simpler rules approximating random walks providing less plausible emergent kernels. As a case study, we demonstrate how rule sets can be selected by comparison to an empirical kernel for a study species (red fox; Vulpes vulpes). These results provide a foundation for the selection of movement rules to represent dispersal in spatial agent-based models, however, we also emphasise the need to corroborate rules against the behaviour of specific species and within chosen landscapes to avoid the potential for these rules to bias predictions.