2008 Volume 13 Issue 1 Pages 37-45
To maintain viable populations of Northern Goshawk (Accipiter gentilis), it is essential to have knowledge of its abundance and distribution. We created a habitat model to predict the number of Northern Goshawk nests in 5×5 km cells established in and around the Kanto district, central Japan. From systematic nest surveys throughout 88 cells, we located 182 nests. Using the number of nests in these cells as a response variable, we generated a Poisson regression model with area of flatland, area of forest, area of urban land, and area of open land < 200 m from forest edge as explanatory variables. Multimodel inference based on Akaike's information criterion was used" to obtain parameter estimates. The model explained 57% of the deviance, indicating a good fit of the model to the data. When we tested the model prediction with validation data from two prefectures, the model successfully predicted which cells would have nests (predicted number of nests > 0.5) except for four cells in mountain regions in Shizuoka prefecture. These results suggest that the model can predict regional-scale distribution of nests in the Kanto district. We produced a map representing the number of nests across the Kanto district, in which the average number of nests in a cell was estimated as 1.25 and the total number as 2,909 (95% confidence interval: 1,699 - 5,196). The large confidence interval and a possibility of overestimation in these estimates should be taken into account when the estimates are used in Northern Goshawk conservation.
