State-and-transition modelling for Adaptive Management of native woodlands

Adaptive Management (AM) is widely advocated as an approach to dealing with uncertainty in natural resource management as it provides an explicit framework for motivating, designing and interpreting the results of monitoring. One of the major factors impeding implementation is the failure to use appropriate process models; a core element of AM. Process models represent beliefs about the properties and dynamics of an ecological system and ecosystem responses to management. Quantitative models of ecosystem response help resolve ambiguity about the efficacy of management and facilitate iterative updating of knowledge using monitoring data. We report on the use of a state-and-transition model (STM) in the Adaptive Management of native woodland vegetation in south-eastern Australia. The STM is implemented as a Bayesian network, making it simple to communicate and update with new data as they arise. Application of the model is demonstrated using case-study and simulation data. We show how the model may be used to predict the probability of achieving desirable state transitions at restoration sites and how monitoring of those sites can be used to update the model (learn) and adapt (review restoration strategies). After just one monitoring/learning cycle, 7 years after the first investments, we found that updated models predict markedly different transition probabilities compared with initial models based on expert opinion. This has strong implications for the apparent cost-efficiency of restoration strategies. The STM approach provides a sound theoretical basis for restoration decisions, while the Bayesian network implementation provides a workable framework for using the STM adaptively.