**2. Results**

## *2.1. Standard Deviational Ellipse Analysis*

A standard deviational ellipse analysis was applied to describe the directional trend and dispersion of CSF notifications in the study area throughout the study period. The study covered the period between September 2018 and June 2019, which was divided into three stages (September–December, January–March, and April–June). Figure 1 illustrates standard deviational ellipses and CSF notifications between September 2018 and June 2019 (Figure 1). To indicate the potential explanation for the directional trend of the CSF outbreaks, the ellipses were overlaid on a map of snowfall area in Gifu Prefecture obtained from the National Land Information Division, Ministry of Land, Infrastructure, Transport and Tourism [27]. The findings showed that CSF notifications appeared to move northeast while spreading along the border of the snowfall area.

**Figure 1.** Directional distribution of classical swine fever (CSF) notifications from September 2018 to June 2019. Standard deviational ellipses (SDEs) identified between September and December 2018, between January and March 2019, and between April and June 2019. Ellipses were overlaid with CSF notifications distinguishing domestic pig (DP) (square) and wild boar (WB) (circle). Ellipses with centroids were combined to indicate the directional trend of the CSF outbreaks.

## *2.2. Multi-Distance Spatial Cluster Analysis*

The multi-distance spatial cluster analysis was applied to explore the maximum distance between cases of CSF notifications. The results indicated that 23 km was the maximum distance of the significant spatial association between CSF notifications in Gifu Prefecture. The obtained maximum distance was used in the subsequent analyses.

#### *2.3. Kernel Density Estimation Analysis*

The kernel density estimation analysis was applied to describe the spatial distribution of the CSF notifications. The analysis showed that the highest density of CSF notifications was located in the southern part of Gifu Prefecture (Figure 2) with further expansion to the east. Among the 16 CSF-positive farms, 37.5% were located in areas with very high or high density of notifications, 31.25% in areas of medium density and 31.25% in areas of low density. Moreover, most of the non-affected domestic farms were located in areas with very low density of notifications (80%), followed by areas with low density (20%). The analysis revealed that CSF-positive farms were located in areas with higher density of notifications, whereas the non-affected farms tended to locate in areas with low density.

**Figure 2.** Density of CSF notifications in Gifu Prefecture. The heat map illustrates the estimated kernel density of CSF notifications (notifications/km2) from very high (red) to very low (transparent). Each coloured area indicates the density of CSF notifications per square kilometer: very high (>0.400), high (0.300–0.399), medium (0.200–0.299), low (0.100–0.199), and very low (<0.100). The highest density of CSF notifications was located in the southern part of Gifu Prefecture. A very low density of CSF notifications was located in other areas of the prefecture. Locations of pig farms not affected by CSF are represented by crosses.

#### *2.4. Space-Time Cluster Analysis*

The space-time permutation analysis was applied to analyze the space-time patterns of the CSF notifications. The analysis identified two significant space-time clusters (*P* < 0.05) in Gifu Prefecture during the study period. Cluster 1, which had a radius of 12.12 km, covered 9 September 2018 to 13 January 2019, and contained 83 notifications, including 4 outbreaks on domestic pig farms. Cluster 2 had a radius of 19.79 km, spanning the period from 11 February 2019 to 19 May 2019, and contained 198 notifications, including three outbreaks in domestic pigs (Table 1).


**Table 1.** Observed and expected notifications, duration, start and end dates, and radius of each space-time cluster detected (*P* < 0.05) in CSF notifications in Gifu Prefecture.
