Abstract
The mountain hawk-eagle (Spizaetus nipalensis) is an endangered and an umbrella species ranking high in the food chain of the forest ecosystem. The objective of this study was to estimate the potential habitat of the mountain hawk-eagle in Yamagata prefecture, northern Japan. Habitat suitability of each grid was estimated using values of indices corresponding to terrain and vegetation features of the surroundings. Previous studies have computed terrain indices such as minimum and mean altitude and mean slope gradient within each grid-square of approximately 5km in size. To estimate potential habitat more accurately, we used 13 terrain and vegetation indices, including the valley distribution index that is important for the mountain hawk-eagle, and set a grid resolution of 50m to describe potential habitat map at fine scale. Since these indices would be sensitive to the analyzing window size placed around each grid, we examined this effect by changing radii as 0.5, 1, 2 and 3km. Values of these indices were computed in each grid with a 50m digital elevation model (50m DEM) and a rasterized 1/50,000 digital vegetation map. By comparing Akaike’s Information Criterion (AIC) values and accuracies among logistic regression models with different combinations from a set of 13 indices, an optimal model with the lowest AIC value and high accuracy was determined. The model used the distribution of valleys within 1km radius and the area of forest within 3km radius, which indicates the importance of valleys and forest for the mountain hawk-eagle’s habitat. The potential habitat areas estimated by this model substantially corresponded (89.0%) to the areas where the mountain hawk-eagle has been observed, with the overall accuracy of 91.8%. We concluded that this model could accurately estimate the potential habitat of the mountain hawk-eagle.
