Predicting Ecosystem Wide Impacts of Wallaby Management Using a Fuzzy Cognitive Map

At Booderee National Park, south-eastern Australia, the intensive control of the introduced red fox (Vulpes vulpes) resulted in a major increase in the abundance of a browsing macropod, the swamp wallaby (Wallabia bicolor). This has led to a major decrease in the abundance and biomass of a range of palatable plant species. Fox control has also started a trophic cascade that has resulted in a decline in the abundance of the greater glider (Petauroides volans) a folivorous arboreal marsupial, mediated either through increased predation by owls or increased competition with common brushtail possums (Trichosurus vulpecula). We identified five potential scenarios for managing the effects of over-abundant swamp wallabies on the ecosystem as a whole. These were (1) the present scenario of continued intensive fox control and four possible scenarios to redress the problem: (2) ceasing fox control; (3) intensive fox control and intensive wallaby control; (4) introducing dingoes and ceasing fox control; and (5) introducing dingoes and maintaining fox control. We used an ecosystem modelling approach based on a fuzzy cognitive map (FCM) to predict relative estimates of abundance for each scenario for a wide range of taxa in the Booderee National Park ecosystem likely to be affected by each scenario. We addressed uncertainty in our knowledge of the interactions between species by creating alternative models of the system by removing one or more of the uncertain links between species and varying the strength of the remaining interactions in the FCM and aggregated predictions from 100,000 models to estimate the effect of uncertainty on the predictions from our FCM model. In comparison with the current scenario of intensive fox control, scenario 3 had the greatest likelihood of improving the status of palatable plants. Scenarios 2 and 4 reduced the abundance of a range of medium-sized mammals but improved the status of greater gliders, whereas the predicted effects of scenario 5 were uncertain. The FCM modelling approach ...