Can public-domain datasets be leveraged to identify factors associated with the occurrence of African swine fever in europe?

TL;DR

This study explores how public data can help identify factors linked to African swine fever outbreaks in Europe, focusing on differences between domestic pigs and wild boars.

Contribution

The study introduces a novel use of public-domain data and spatial modeling to identify environmental and human factors influencing African swine fever in Europe.

Findings

Land cover types like industrial areas and wetlands are significant for ASF in both domestic pigs and wild boars.

Human population density is a key factor for domestic pigs but not wild boars.

Finer spatial resolution reduces the number of associated variables for domestic pigs but not for wild boars.

Abstract

African swine fever (ASF) is a highly contagious and deadly viral disease affecting domestic pigs and wild boars. This study uses public domain datasets to identify the association between (a)biotic variables and occurrences of ASF in domestic pigs and wild boars in Europe. The public domain databases WOAH-WAHIS and EMPRES-i were used to obtain data about ASF cases in domestic pigs and wild boars from 2018 to 2023. Several (a)biotic variables were considered as potential drivers for ASF: precipitation, temperature, human-animal interface density, and type of land cover. A Shiny app was created to offer an interactive platform for data analysis and visualisation.

Uni- and multivariable mixed negative-binomial models were used to assess the association between (a)biotic variables and ASF occurrence. The statistically significant associations between the (a)biotic variable ‘land cover’ and ASF cases differ between domestic pigs and wild boars. The land cover types ‘industrial, commercial, and transport units’ and ‘inland wetlands’ were identified as significant factors associated with ASF in both domestic pigs and wild boars. However, pig density, temperature, and human density were statistically associated with ASF occurrence only in domestic pigs. When a finer spatial resolution (5 × 5 km) was applied for all (a)biotic variables around the reported ASF cases compared to the coarser resolution of 10 × 10 km, the associations with (a)biotic variables for wild boars remained consistent. In contrast, for domestic pigs, only human population density remained significantly associated with ASF occurrence at this finer scale.

The model showed high accuracy for ASF prediction in domestic pigs but low accuracy for wild boars, highlighting the limitations of public domain (a)biotic factors alone. Integrating restricted data on animal movements, migration, and carcass interactions could enhance future predictions and improve disease control strategies. The change in spatial resolution did not affect the associations between (a)biotic factors and ASF occurrence in wild boars but reduced the number of associated variables in domestic pigs, suggesting that ASF in wild boars is driven by broader-scale factors, while in domestic pigs it is influenced by more localised conditions.

The online version contains supplementary material available at 10.1186/s13028-025-00832-7.

Figures