Abstract

Simulation of animal movement as an Ornstein-Uhlenbeck process and modifications thereof to account for some type of external attractor or repeller has proved useful in constructing movement paths over homogeneous landscapes. Over heterogeneous landscapes, particularly those that are complicated by the existence of barriers and other types of sharp transitions, different ways are needed for constructing movement pathways. This is especially true once individuals exhibit different modes of behavior that reflect the biology and ecology of diurnal, lunar, and seasonal cycles, as well as the presence of other individuals (whether competitors, predators, or prey) on the landscape. Here we move away from unifying stochastic process paradigms to a more general individual-based approach that allows us to incorporate both external landscape drivers (e.g., location of known sources of water, resources, or home locations) and individual internal state drivers (e.g. thirst—time since last drink, hunger—time since last intake of food or water, current physiological state, etc.) of behavioral change.   We have found that movement model implementations benefit from an individual-based model (IBM) design informed by the history of an individual's prior states. Our approach provides a two-tiered platform that on its primary level manages an IBM simulation. The individual agents are implemented using internal components combined with algorithms and data structures customized to particular applications. Moreover, the resulting package is easily distributed and deployed, and provides unique features that enhance it usability. The latter include interoperability with external platforms such as Python and the R statistical language; and safe access to elements of the system's internal business logic to permit modification and experimentation.   A demonstration of a hierarchical, multimodal, movement model for assembling the tracks of animal diel activity routines will be presented.