Modeling habitat suitability for endemic Grizzled leaf monkey (Presbytis comata) using geospatial machine learning approach

Abstract

Global biodiversity, which plays a significant role in maintaining the sustainability of Earth's ecosystems, is threatened by increasingly destructive and uncontrolled human activities. Indonesia's Grizzled leaf monkey or Javan Surili (Presbytis comata) is an endemic primate whose population continues to decline. As the high potential for reducing animal habitats on Java has implications for endemic Surili primates, it is necessary to study the characteristics of suitable habitats that can be used as the basis for strategic policies. This study aimed to develop a habitat suitability index (HSI) for Surili with a study of forest areas in Java Island, Indonesia, using a multi-machine learning approach. The novelty of this study is the integration of three machine learning methods, random forest (RF), support vector machines (SVM), and maximum entropy (MaxEnt), and the use of four parameters, climate, topography, ecology, and anthropogenic factors. This is expected to increase the accuracy of assessing the suitability of the developed Surili habitat. Of the 116 Surili encounter points in the region, 7 (6%) were in the suitable HSI class, 38 (33%) in the high HSI class, 49 (42%) in the moderate HSI class, and 22 (19%) in the low and unsuitable HSI classes. Meanwhile, of the 13 release points, 12 (92%) were dominated by areas with high-to-suitable HSI classes. Based on the receiver operating characteristic curve value, the integrated multi machine learning algorithms value was 0.8578. This study could help formulate potential strategic policies in the restoration program that has been pursued to conserve Surili species in Indonesia.