Habitat Suitability Model Statistical Test

settlement features are taken from Indonesia National Coordination Agency for Surveys and Mapping with 1: 50.000 of scale. While the forest edge features are taken from land cover map derived from Landsat ETM 7+ acquisition date October 6 th , 2005. Mustari 2003 also explain that like other species in the genus Bubalus, Anoa is water dependent animal. They need water for drinking everyday and they were frequently observed wallowing. Distribution of Anoa is significantly associated with water source during the dry season. This statement clearly explains that Anoa have high dependencies in hydrological resource. Based on this fact the hydrological condition will be the one of the parameters in estimating Anoa ’s habitat suitability and presence. Hydrological factor was taken from river feature in Peta Rupa Bumi Bakosurtanal with 1: 50.000 of scale. The stream line feature was processed using Euclidian distance to develop distance from river value in each grid observation. Hydrological data were consists of river stream line, both for all year round rivers and seasonal river.

3.3. Habitat Suitability Model

Habitat model is not a definitive attempt to predict presence or absence, but it is more an attempt to identify areas where conservation and forest habitat enhancement could be prioritized. The deductive modeling approach allows the utilization of statistically significant quantitative data to build a habitat model that describes the similar areas to those used by species. The absence of Anoa in predicted suitable area will not indicate unsuitable Anoa habitat. By identifying similarities, in habitat features across the Lore Lindu National Park, the improvement in Anoa’s preservation could be increased suitable habitat patch size and connect neighboring patches. The habitat suitability map was constructed by coupling field data with geospatial information derived from satellite imagery and other parameter spatial resources. This study is compiling field data indicating the presence or absence of Anoa from suitable habitat parameter. The logistic regression approach uses the environmental parameter layers to characterize the habitat of known Anoa locations as well as those areas with no sign of Anoa. Areas throughout the Lore Lindu National Park exhibiting land cover, vegetation, and biophysics characteristics similar to locations where Anoa was observed in the field are associated with a higher metric in the derived map. Areas exhibiting characteristics similar to locations where field data indicated the absence of Anoa are associated with a lower metric on the map. Figure 3.2 Habitat Suitability Model Workflow

3.4. Statistical Test

Spatial distribution model will build the model based on ecological parameter and testing it against current Anoa locations. To assess fit and relative strength of the selected model, we used the Hosmer and Lemeshow goodness of fit test and Nagelkerke’s rescaled R 2 respectively. Probabilities from the logistic regression are used to derive a predictive habitat map from the significant habitat characteristics across the Lore Lindu National Park Presence Absence Data Field Survey Environment Variable Fit Threshold Logistic Regression Distance to Settlement Altitude Slope Distance to River Fit Logistic Regression Model Habitat Suitability Map Best Logistic Regression Model Distance to Forest Edge landscape at threshold values. The results of the regression produced variable coefficients that were then applied to the area and the entire study area ranging from zero to one, where values greater than 0.5 were considered to be presence of Anoa.

3.5. Vegetation Analysis