Machine Learning Prediction of Wildlife Habitat Suitability Under Land Use Change

Abstract

Suitability of wildlife habitat depends significantly on alterations in land use and land cover. These alterations can affect the availability of wildlife habitats, connectivity of habitats, food resources, and environmental conditions. Due to rapid urbanization, expansion of agriculture, infrastructure development, and deforestation, there is an increasing necessity for an effective method that will allow to assess and predict alterations in wildlife habitats. In this work, machine learning methods are discussed in order to predict suitability of wildlife habitats under altering land use conditions. Such environmental and spatial parameters as land use and land cover, vegetative factors, elevation, water resources availability, climatic conditions, and human activity can be used to reveal patterns associated with suitable wildlife habitats. It is possible to develop predictive models based on the relationships between these parameters and the presence of wildlife species or their habitats and provide spatial predictions about suitability of wildlife habitats. Comparison of habitat suitability depending on various land use scenarios allows revealing areas of habitat loss, degradation, fragmentation, and even conservation priority