Global pandemic of avian influenza have become a threat of poultry production. Spatial risk assessment is one of the important topics in order to decide management system for pathogens. Risk maps for avian influenza have been created for various area in the world, however, the major effected factors for the infection did not coincide with each other. Therefore it is needed to clarify the effective factors for each area.

Originally, Anatidae and Charadriidae have been identified as the major reservoirs for avian influenza in nature. Since ducks carried avian influenza among migratory birds in Japan, we focused on the distribution of ducks to create a potential risk map of avian influenza.

Potential distribution models were generated, using the maximum entropy approach, from presence-only data. Infected wild birds were found at 86 locations during winter in 2010-2011, and they were used for the presence data. Environmental factors such as temperature and precipitation in winter, factors about their hosts such as duck density and the habitat area, artificial factors such as urban area, poultry density, and spatial filters which alleviate spatial autocorrelation were used for environmental variables. The reliability of the map was examined by comparing the occurrence data in 2004-2010.