Monday, August 2, 2010 - 4:00 PM

COS 8-8: State-and-transition modelling for adaptive management of native woodlands in south-eastern Australia

Libby Rumpff1, David H. Duncan2, Peter A. Vesk1, David A. Keith3, and Brendan A. Wintle1. (1) The University of Melbourne, (2) Arthur Rylah Institute for Environmental Research, (3) Department of Environment and Climate Change New South Wales


There are large investments in native vegetation restoration being carried out worldwide, but there is high uncertainty about how, where, and under what circumstances vegetation restoration is a good investment.  This is a common feature of natural resource management, because biological systems are highly complex, and actions may be carried out over broad temporal and spatial scales.  The issue is of great importance to governments at all levels as vegetation restoration is likely to be a popular climate adaptation investment option.  Adaptive Management (AM) is commonly advocated as a way to deal with this issue.   However, despite a long rhetorical commitment to AM, there still remains a need to demonstrate its efficacy.  One of the major factors impeding implementation is the failure to develop and use appropriate process models, which are a core element of AM.  Process models are a representation of the belief about the properties and dynamics of an ecological system, and the system response to management intervention.  In this study we present the first example of AM underpinned by a vegetation condition state and transition model (STM) for native woodland vegetation dynamics.   We use Bayesian Belief Networks to structure the process model.    


Quantifying system response allows us to resolve ambiguity about the efficacy of management intervention, and iteratively update knowledge using monitoring data, improving management decisions.  In order to close the adaptive management cycle the model needs to be validated and updated with appropriate data and we illustrate some important data collection considerations. This includes the identification of appropriate measures and environmental variables to add to a monitoring strategy, how we can use the model to identify which management interventions are required at a given site, and how we can use the subsequent data collected in a monitoring strategy to improve our confidence in the relationships and thresholds used in the model.  We see our work as a major step toward routine application of Adaptive Management approaches, because STMs provide an essential means for assisting management decisions and continuous improvement. The ability of STMs to incorporate ecological realism, combined with the flexibility and accessibility of Bayes Nets provides a sound, practical underpinning for Adaptive Management of native vegetation.