Toggle Main Menu Toggle Search

Open Access padlockePrints

Transferability of Predictive Species Distribution Model: A Case Study of Tropical Palms of Sulawesi

Lookup NU author(s): Wiske Rotinsulu, Dr David Fairbairn, Dr Meredith Williams


Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


The palm flora of Indonesia is one of the richest in the world. Palm flora plays an important role in the functioning of tropical forest ecosystems by enhancing leaf and fruit productivity, maintaining animal relationships, and providing special components of forest structure. In addition to its ecological importance, this family provides a wide array of useful products such as food, fibre, medicines and construction materials for humans. As such, conservation and sustainable use of the palm family is crucial. Although the exploitation of palm species for commercial and subsistence purposes appears to be a major factor which could place them at risk, habitat loss due to forest conversion to other uses is identified as the largest threat to the survival of many palm species in the tropical forests of North Sulawesi (Rotinsulu, 2003). Considering these conditions, there is an urgent need to reduce threats to the survival of palms and promote conservation in this region. This can be partly addressed by improving knowledge of the geographical distribution, biology and preferred habitats of native palms. Basic species distribution mapping is a useful conservation tool for predicting patterns of biodiversity, or identifying geographical areas of conservation significance. It also can improve our understanding of the appropriateness of habitat areas for individual species. More advanced predictive models of species distribution based on habitat can also be used to assess the likely effects of changes in land-use on a species. A Bayesian approach to decision making will be used as an inductive modelling process for analysis of patterns within native palm distributions. This approach has become a powerful alternative method to the traditional ones for building predictive relationships between species and their environment. Integrated with a GIS, it has been widely used in wildlife studies for modelling habitat (Pereira and Itami, 1991; Aspinall, 1992, Aspinall and Veitch 1993; Hepinstall and Sader, 1997; Tucker et al.,1997; Fleishman et al., 2000). The lack of previous studies on tropical plant species modelling using GIS and Bayesian approaches provided the main impetus for this work. The main goal of the predictive habitat models is to produce a map for assessment of management and conservation actions. As such, a validation of the model has become a central issue of the species distribution modelling (Corsi, et al., 2000). Validation of the species distribution model in time and space is crucial to assess the efficacy and transferability of models. At present, no well-established framework exists for testing transferability of GIS-based habitat models (Leftwich, et al. 1997). In this study we will examine the predictive power and general applicability of our predictive model to other areas which are ecologically similar.

Publication metadata

Author(s): Rotinsulu W, Fairbairn D, Williams M, Carlisle B

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Unknown

Conference Name: Proceedings of 13th Annual GIS Research UK Conference (GISRUK)

Year of Conference: 2005