2.2. Data collection and processing
We obtained the C. imicola presence points (n = 1045) from literature (Leta, Fetene, Mulatu, Amenu, Jaleta, Beyene, Negussie, Kriticos, et al., 2019; Ye, Liu, Li, Qiu, & Liu, 2019) and the Global Biodiversity Information Facility database (https://www.gbif.org/). The reported AHS outbreaks were extracted from the archive of Food and Agriculture Organization (FAO) of the United Nations from December 22, 2005 to September 1, 2020 (n = 203).
To determine the influence of the environmental variables on the AHS distribution and C.imicola, we considered 19 bioclimate variables, 12 land cover variables, and global horse distribution density as risk factors in the models (Table 1). Bioclimate variables were obtained from the WorldClim database (http://worldclim.org/version1). Global land cover characterizations were obtained from the archives of the United States Geological Survey Earth Resources Observation and Science center (https://www.usgs.gov/). The global horse distribution density was downloaded from the FAO database (http://www.fao.org/livestock-systems/).
All the occurrence data were rarefied at 10 km2(Radosavljevic & Anderson, 2014) to minimize the spatial autocorrelation using the SDM Toolbox (Brown, Bennett, & French, 2017). Thus, 728 spatially rarefied occurrence records of C. imicola and 110 locations of AHS were used in the MaxEnt modelling of this study (Figure 1).
To avoid the multi-collinearity of environmental variables, Pearson correlation analyses were performed using the SPSS 22.0 software. Each pair of variables had a correlation value |r| ≥ 0.80, and one of the variables was considered to remove from the final model (Ma et al., 2019; Wu, Sharp, Zhao, Shirato, & Jiang, 2007). Finally, two sets of variables were included in the AHS model and theC.imicola model (see Table 1). In ArcGIS 10.2, all the environmental variables were resampled to the ASCII raster grids at a resolution of 2.5 arcmin.