The use of dynamic landscape metapopulation models for forest management: a case study of the red-backed salamander

Abstract

Spatial models of population dynamics have been proposed as a useful method for predicting the impacts of environmental change on biodiversity. Here, we demonstrate advances in dynamic landscape metapopulation modelling and its use as a decision support tool for evaluating the impacts of forest management scenarios. This novel modelling framework incorporates both landscape and metapopulation model stochasticity and allows their relative contributions to model output variance to be characterized. It includes a detailed sensitivity analysis, allowing defensible uncertainty bounds and the prioritization of future data gathering to reduce model uncertainties. We demonstrate this framework by modelling the landscape-level impacts of eight forest management scenarios on the red-backed salamander ( Plethodon cinereus (Green, 1818)) in the boreal forest of Ontario, Canada, using the RAMAS Landscape package. The 100 year forest management scenarios ranged in intensity of timber harvesting and fire suppression. All scenarios including harvesting predicted decreases in salamander population size and the current style of forest management is predicted to produce a 9%–17% decrease in expected minimum population size compared with scenarios without harvesting. This method is amenable to incorporating many forms of environmental change and allows a meaningful treatment of uncertainty.