Abstract:
African swine fever (ASF) is a transcontinental, contagious, fatal virus disease of pig with devastating socioeconomic impacts. Interaction between infected wild boar and domestic pig may spread the virus. The disease is spreading fast from the west of Eurasia towards ASF‐free China. Consequently, prediction of the distribution of ASF along the Sino‐Russian‐Korean borders is urgent. Our area of interest is Northeast China. The reported ASF‐locations in 11 contiguous countries from the Baltic to the Russian Federation were extracted from the archive of the World Organization for Animal Health from July 19, 2007 to March 27, 2017. The locational records of the wild boar were obtained from literature. The environmental predictor variables were downloaded from the WorldClim website. Spatial rarefication and pair‐wise geographic distance comparison were applied to minimize spatial autocorrelation of presence points. Principal component analysis (PCA) was used to minimize multi‐collinearity among predictor variables. We selected the maximum entropy algorithm for spatial modelling of ASF and wild boar separately, combined the wild boar prediction with the domestic pig census in a single map of suids and overlaid the ASF with the suids map. The accuracy of the models was assessed by the AUC. PCA delivered five components accounting for 95.7% of the variance. Spatial autocorrelation was shown to be insignificant for both ASF and wild boar records. The spatial models showed high mean AUC (0.92 and 0.97) combined with low standard deviations (0.003 and 0.006) for ASF and wild boar, respectively. The overlay of the ASF and suids maps suggests that a relatively short sector of the Sino‐Russian border has a high probability entry point of ASF at current conditions. Two sectors of the Sino‐Korean border present an elevated risk.