A spatially-heterogeneous impact of fencing on the African swine fever wavefront in the Korean wild boar population

Abstract

In October 2019, South Korea's first case of African swine fever (ASF) was reported in wild boar in the north of the country. Despite the implementation of a 2300 km-long fencing strategy, the ASF wavefront continued to invade southward. Our study aimed to investigate the ASF wavefront dynamics in different regions of South Korea, as well as to assess the effectiveness of the fencing measures on ASF dispersal and wavefront velocity. From the nationwide wild boar surveillance system, we extracted 2661 cases, starting from 2 October 2019 (first detection) to 15 September 2022. The cases were categorised into four main spatiotemporal clusters. The average wavefront velocity over the four clusters was estimated at 0.52 km/week, with the cluster in the eastern part of the Korean peninsula exhibiting the fastest velocity (0.99 km/week) compared to the other clusters (0.44, 0.31, and 0.15 km/week). We hypothesise that these differences are related to different wild boar densities due to heterogeneous habitat suitability. We also found that fencing significantly impacted ASF dispersal in only two of the four main clusters, with no evidence that fencing slowed down the spread of the wavefront in any of the clusters. We argue that this heterogeneity might result from fencing locations being misaligned with the true (and unobserved) wavefront.

Keywords: African swine fever; South Korea; intervention strategy; spatial modelling; wild boar.

Figure 1

Workflow to select wild boar-mediated ASF wavefront cases to estimate the wavefront velocity and assess how it was affected by fencing

Figure 2

The temporal patterns of the cumulative length (A) and spatial locations of the installed fences (B). Persistent barriers indicate the natural or anthropogenic barriers used for fencing such as cliffs, rivers, highways, etc.

Figure 3

Workflow for analysing national surveillance data. From data collection (A) to selecting index case (B), identifying spatiotemporal clusters (C) and filtering the ASF wavefront cases (D).

Figure 4

Time series of reported and wavefront cases in each cluster. In each row, time series of reported cases (light shade) and identified wavefront cases (dark shade) within specific cluster were plotted. The number at the end of each row indicates the number of wavefront (in bold and italic type) and reported cases (in normal type) within that respective cluster.

Figure 5

Distribution of estimated wavefront velocities for each cluster (top) and the spatial distribution of wavefront cases for their respective clusters (bottom).

Figure 6

The joint distribution of the estimated velocity as a function of wild boar habitat suitability in each cluster. Each polygon represents the joint distribution of velocity and habitat suitability, with dots indicating the median point of velocity and habitat suitability. Marginal distributions of each variable are plotted in the margins of the plot.