Challenges, constraints and solutions for modeling regional-scale dispersal of invasive organisms: from practice to policy

Invasive organisms can cause dramatic impacts to a wide range of stakeholders. Once established within a country their management is most often resourced and implemented at local to regional scales. Predicting the dispersal potential of these key organisms is central to effective management. For example, management actions include early detection and eradication of new outbreaks, and the establishment and maintenance of containment lines, both of which require a good spatial understanding of invasion risk. We present modeling work which seeks to predict the spread of the invasive pasture weed parthenium (Asteraceae; Parthenium hysterophorus) in the Queensland Murray Darling Basin, with the aim of helping natural resource management agencies to best target their resources. This is part of a broader modeling program aimed at developing generic spatial modeling approaches for invasive organisms. A Bayesian Belief Network capturing the key drivers of distribution and dispersal was developed during an expert elicitation workshop and subsequently linked to relevant environmental layers within GIS. The models combined predictions of habitat suitability (where the plant will do well given the opportunity) and introduction risk (the risk of arrival of propagules from known sources) to probabilistically predict habitat susceptibility. Models predicted dispersal and spread within a 10 year time-frame to account for inter-annual stochasticity in some key environmental variables and vectors (e. g. major flood events). Multiple vectors were important and these were grouped as local "incidental" dispersal, water dispersal and long-distance transport. However, process knowledge of dispersal typically outstripped the spatial data on vector movement required to implement the models. We discuss how these and other challenges and constraints were dealt with, and how the resulting model predictions helped inform policy and management.