Modeling Species Distribution

There are many methods for modeling species distribution in the landscape. In this chapter, the authors elaborate on the concepts of species modeling and present three popular techniques to generate species distribution: cartographic overlay, logistic multiple regression and maximum entropy (MAXENT). The cartographic overlay method is relevant to generate a habitat suitability index. Logistic multiple regression generates the probability of distribution based on presence and absence of data in relation to habitat factors. The authors use pseudo absence data selected randomly from low suitability classes, because real absence data were not available. The third technique, maximum entropy method (MAXENT), uses presence-only data. The Asian elephant (Elephas maximus) was selected as a proxy species for this study. The study was conducted in Bun Tharik-Yod Mon, a proposed wildlife sanctuary in northeast Thailand. The results show that among the three approaches, the potentially suitable habitats derived from cartographic overlay cover the largest area and are likely to overestimate existing occurrence areas. The logistic regression model predicts approximately 56% as suitable area, while maximum entropy results covers approximately 9% of the sanctuary. Although the results show large differences in the suitable areas, it should not be concluded that any one method always proves better than the others. Utilization of any method is dependent on the situation and available information. If species observations are limited, the cartographic overlay or habitat suitability is recommended. The logistic regression method is recommended when adequate presence and absence data are available. If presence-only data is available, a niche-based model or the maximum entropy method (MAXENT) is highly recommended. DOI: 10.4018/978-1-4666-2038-4.ch125