Markov Decision Processes in Artificial Intelligence

Chapter 12

Conservation of Biodiversity 1

12.1. Introduction

Globally, biodiversity is undergoing a rapid decline as a result of human activities [PIM 95]. Conversion of native forests and grasslands for agriculture, livestock grazing and forestry, resource extraction, urban and rural development coupled with an ever growing population are resulting in a loss of plants, animals and ecosystems at a rate never before experienced [ANO 07]. In the face of these rapid changes, species must adapt or risk extinction. The field of Conservation Biology has developed out of Ecology in an effort to quantify and predict the impacts of human actions on biodiversity and most importantly develop strategies to conserve and protect these species. Increasing the challenge of this task is the fact that resources to undertake biodiversity conservation are limited [WIL 06]. In response, we have seen the development of cost-effective conservation decision-making [POS 01]. Whether making decisions about the recovery of endangered species or the management of invasive species [REG 09], Markov decision processes (MDP) allow a clear and eloquent formulation of the optimal conservation resource allocation problem. There remains many complex ecological problems without solutions and only through collaboration of computer scientists with conservation biologists can we hope to find good solutions [MAC 09]. The application of MDP methods to conservation problems is in its infancy. Yet, as we will demonstrate ...