Abstract:
This research demonstrates the ability of integrating spatial and dynamic models with a Geographic Information System (GIS) to predict and simulate the dynamics of biological invasion in both the spatial and temporal dimensions. In this spatio-temporal framework, GIS serves as a shell to integrate the spatial, dynamic and stochastic perspectives in a coherent workflow. It links decision tree analysis with GIS to reveal the spatial association between the invasive species Lantana camara and the heterogeneous landscape of Northland, New Zealand. The decision tree analysis quantifies the vulnerability of the environment for Lantana invasion, and identifies the most significant variables that influence its likely occurrence. GIS-generated maps illustrate and classify the spatial variation in Lantana invasion probabilities and also provided the spatial dimension of Lantana invasion for the framework.
A stratified diffusion model quantifies the dynamics of Lantana invasion. Analysis of the model provides an estimate of the species’ spread and explains the nonlinear expansion observed in Lantana invasions. It proves that Lantana spreads with a relatively lower rate of neighbourhood diffusion and a relatively higher rate of long-distance dispersal. The spatial and temporal information derived in the two previous steps is combined to serve in simulation on a realistically heterogeneous percolation landscape. In the GIS implementation, the spread is related to the spatial probability information provided by a decision tree model. The stratified diffusion model determines the dynamics of dispersal. The pattern of the Lantana spread is simulated as a sequence of discrete steps, expressing the dynamic way that an individual may propagate under structurally heterogeneous conditions. In particular, new colonies of the species are limited to the dispersal ability of the species and the local variations in the landscape.
This research develops a new methodology for modelling biological invasions and other ecological processes. It greatly expands the analytical capabilities of GIS and intersects the domains of raster GIS, percolation, ecology, diffusion-reaction systems and habitat management. The developed research overcomes some of the current modeling technical restrictions and supplies a flexible simulation management system to explore the spatial dynamics of ecological processes.