HABITAT USE ANALYSIS USING THIESSEN POLYGON AND TRIANGULATED IRREGULAR NETWORK INTERPOLATION

Jim Casaer, Martin Hermy & Pol Coppin
Labortory for Forest, Nature and Landscape Research / K.U.Leuven
Vital Decosterstraat 102
B-3000 Leuven / Belgium

BACKGROUND Since Burt defined "home range" in 1949, a range of methods have been proposed to quantitatively appraise it, but the literature shows that no consensus has been reached. And even the concept of "home range" itself has been questioned now. Nevertheless, habitat use analysis requires some form of numeric assessment to facilitate a scientific evaluation of the interaction between animals and their environment. This paper proposes a new non-parametric model to estimate habitat use using Geographical Information Systems (GIS) technology.

MODEL DESCRIPTION Contrary to many other home range models, e.g. the Ellipse and Minimum Convex Polygon models, the proposed model is based on "internal" home range use. In order to evaluate this concept, the utilisation distribution or UD must be derived from the distribution of the observation locations themselves and can not take into account a priori assumptions concerning the nature of the UD distribution, nor are ancillary parameters derived as is the case with more sophisticated parametric models .

Thiessen polygon interpolation around the collected location points divides the continuous feature space into discrete units. The model subsequently assigns to each location point the inverse values of the respective Thiessen polygons. These values are a quantitative indicator of intensity of habitat use (number of observations per area unit), and serve as input to an triangulated irregular network or TIN interpolation surface. The resulting three-dimensional model effectively depicts habitat use, whereby the area representing 95% of the total volume of this 3-D model is defined as the home range.

The ability of the proposed model to accurately derive habitat use was tested using a dedicated software package to simulate home range use in consecutive iterations with different numbers of observations. Grid analysis (Arc/Info operator) and Kappa coefficients of agreement served as evaluation tools.

CONCLUSIONS Accuracy assessment with both the grid analysis procedure and Kappa coefficient computations revealed a high degree of correspondence between model and simulation. The fact that model construction was done in the Arc/Info GIS environment allows for easy integration with habitat feature collection. It must be stated, that the model facilitates in essence an appraisal of habitat use, which is not synonymous with home range definition.