The use of genetic algorithms and Bayesian classification to model species distributions

Author
Termansen Mette
McClean Colin J.
Preston Christopher D.
Keywords
Abstract

<p>This paper develops a method to model species’ spatial distributions from environmental variables. The method is based on a search for an optimal identification of environmental niches to match observed species presence/absence data. The identification is based on Bayesian classification and the optimisation is based on a Genetic Algorithm (GA). The algorithm is tested on an artificial “species” and is shown to perform well. We apply the approach to a random sample of 100 plant species native to the British Isles. This enables an identification of the environmental variables that are most important for capturing the species’ spatial distribution. We show that both climate and land use variables are important for modelling the spatial distribution patterns of the sampled species.</p>

Year of Publication
2006
Journal
Ecological Modelling
Volume
192
Issue
3-4
Number of Pages
410-424
Date Published
02/2006
ISBN Number
0304-3800
DOI
10.1016/j.ecolmodel.2005.07.009
Short Title
Ecological Modelling