Print PageAccurate estimation for the spatial distribution of protected species is necessary for effective species conservation planning because such information makes it possible to predict how species will respond to landscape alteration and environmental changes. Species distribution model (SDM) is a useful tool which can provide a measure of a species occupancy potential in area. The Ministry of Environment started the ‘National Ecosystem Survey’ that has been carried out nationwide every ten years since 1986 as the largest survey project in Korea. The second one (from 1997 to 2005) and the third one (from 2006 to 2009) produced the GIS-based inventory of whole species. However, three survey methods were different each other and there were no studies for species distribution using national survey data in Korea. The purposes of this study are to test species distribution models for finding the most suitable modeling methods for the National Ecosystem Survey data and to investigate the modeling results according to survey methods and regions. Occurrence data of 9 species, which is endangered species in Korea, were extracted from the National Ecosystem Survey by taxonomical group (plant, mammal, bird). Plants are Korean winter hazel (Corylopsis coreana), Iris (Iris odaesanensis), Berchemia (Berchemia berchemiaefolia). Mammals are Common Goral (Nemorhaedus goral), Yellosw-Throated Marten (Martes flavigula koreana), Leopard Cat (Prionailurus bengalensis). Birds are Black Woodpecker (Dryocopus martius), Eurasian Eagle Owl (Bubo bubo), Common Buzzard (Buteo buteo). Environmental variables consisted of climate, topography, soil and vegetation structure. Two modeling methods (GAM, Maxent) were tested across 9 species and validated it with field experts, and predictive species maps of target species were produced.
Results/Conclusions
The results of this study were as follows. Firstly, Maxent showed higher 5 cross-validated AUC than GAM overall. Maxent for plant and mammal showed higher AUC than GAM but GAM for birds showed higher AUC than Maxent due to species characteristics and surveying methods. Therefore, Maxent is more useful species distribution model for National Ecosystem Survey data, which has presence only data, in Korea. Secondly, significant variables to species habitat selection were different by region and analysis area. Thirdly, the modeling results between the second and third survey methods were sometimes different. Therefore, we need to combine two data for producing a reasonable result. Lastly, the participation of experts could enhance accuracy of prediction for endangered species’ potential habitat which is narrow range.
