Student: 

Steve Wangen, Lincoln University

Staff: 

Professor Richard Duncan

Invasive weeds in the South Island's Craigieburn Forest Park will be nipped in the bud if a Bio-Protection Research Centre PhD student has his way.

Steven Wangen is developing a computer model that not only simulates seed dispersal, but also the likelihood of those seeds reproducing and establishing populations. Focusing on Hieracium lepidulum, a problematic hawkweed (somewhat akin to a dandelion), the ultimate aim is to identify effective control strategies.

By combining dispersal models with population dynamics and real geographic landscapes, Steve's simulations have an additional level of realism not often found in predictions of spread. They will be used to predict the invasive plant's advance and determine which combinations of factors are most influential in the spread. Steve will also look at which inherited traits play a part in the weed's spread.

"Together these approaches represent a novel approach to invasive species modelling, and a step forward in how we approach the interactions between an introduced organism and its new environment," he says.

"There are plenty of instances where a single one of these approaches has been applied to invasive spread in isolation. Their integration, however, still remains a relatively rare occurrence."

Steve's project builds on two previous Centre PhD projects; in one student Joel Pitt used computer simulations to model and predict the spread of invasive insects in New Zealand.

To this, Steve is adding a population modelling component that simulates the life-cycle of individual organisms, enabling him to predict birth, growth, and death rates. Much of the data used to develop the population models stems from another previous Centre student, Alice Miller. In her project, Alice established plantings in the Craigieburn valley at a range of densities in different habitats. Steve will continue to monitor them to see how the populations respond to the different habitats found in Craigieburn. Once these behaviours are quantified, the computer simulation can be applied to other areas, predicting where the weeds are likely to spread and establish.

"We can't predict the exact spots that weeds will grow but these techniques let us boil it down to typical pattern."

Also noteworthy in this project is Steve's use of hierarchical Bayesian statistics for estimation of model parameters, which helps to facilitate incorporating data that was collected at various scales. This approach helps determine certainty that the data represents what happens in the real world.

"With this approach we can actually quantify this uncertainty and use that information to define variation in behaviour between individuals in the simulation, instead of writing off that variation as a type of error."

Project name: Simulating the spread of Hieracium lepidulum in heterogeneous landscapes.