Adaptive, consensus-based control strategies for managing meta-populations of pests

Abstract

We consider management strategies for natural populations, spatially distributed in patchy landscapes. Such patchy landscapes arise naturally, due to habitat fragmentation, or by design, such as in rural farmlands. Populations disperse within these patchy landscapes resulting in meta-populations. Management of such meta-populations then involves two modes of control action – action local to each patch and coordinated control of action between patches. The challenge is two-fold: To account for uncertainty in the localised population dynamics on patches we use adaptive control approaches; To counter the effects of dispersal, we combine the localised adaptive control actions with sharing of information and actions between patches. Population dynamics on each patch are described by population projection matrices. Dispersal of populations between patches and information sharing between control actions on patches are modelled using directed graphs on the set of patches. The novelty lies in combining information sharing with output driven adaptive control. Information sharing acts to anticipate potential outbreaks and to coordinate this with the adaptive control localised to patches. We explore situations when information sharing is and is not matched with dispersal. Information sharing improves the outcomes in that the size and extent of a pest outbreak and the amount of pesticide sprayed is reduced. The results are shown to be robust to uncertainties in the demography of pests.