Risk of African Swine Fever Virus Transmission among Wild Boar and Domestic Pigs in Poland

Authors

Kim M. Pepin, USDA APHIS WS NWRC ColoradoFollow
Tomasz Borowik, Mammal Research Institute, Polish Academy of Sciences
Maciej Frant, Polish National Veterinary Research Institute
Kamila Plis, Mammal Research Institute, Polish Academy of Sciences
Tomasz Podgórski, Czech University of Life Sciences

Date of this Version

2023

Citation

Frontiers in Veterinary Science (2023) 10: 1295127

doi: 10.3389/fvets.2023.1295127

Edited by Ferran Jori

Reviewed by Federica Loi and Brandon Hayes

Comments

United States government work

and

License: CC BY 4.0

Abstract

Introduction

African swine fever (ASF) is a notifiable disease of swine that impacts global pork trade and food security. In several countries across the globe, the disease persists in wild boar (WB) populations sympatric to domestic pig (DP)operations, with continued detections in both sectors. While there is evidence of spillover and spillback between the sectors, the frequency of occurrence and relative importance of different risk factors for transmission at the wildlife-livestock interface remain unclear.

Methods

To address this gap, we leveraged ASF surveillance data from WB and DP across Eastern Poland from 2014–2019 in an analysis that quantified the relative importance of different risk factors for explaining variation in each of the ASF surveillance data from WB and DP.

Results

ASF prevalence exhibited different seasonal trends across the sectors: apparent prevalence was much higher in summer (84% of detections) in DP, but more consistent throughout the year in WB (highest in winter with 45%, lowest in summer at 15%). Only 21.8% of DP-positive surveillance data included surveillance in WB nearby (within 5 km of the grid cell within the last 4 weeks), while 41.9% of WB-positive surveillance samples included any DP surveillance samples nearby. Thus, the surveillance design afforded twice as much opportunity to find DP-positive samples in the recent vicinity of WB-positive samples compared to the opposite, yet the rate of positive WB samples in the recent vicinity of a positive DP sample was 48 times as likely than the rate of positive DP samples in the recent vicinity of a positive WB sample. Our machine learning analyses found that positive samples in WB were predicted by WB-related risk factors, but not to DP-related risk factors. In contrast, WB risk factors were important for predicting detections in DP on a few spatial and temporal scales of data aggregation.

Discussion

Our results highlight that spillover from WB to DP might be more frequent than the reverse, but that the structure of current surveillance systems challenge quantification of spillover frequency and risk factors. Our results emphasize the importance of, and provide guidance for, improving cross-sector surveillance designs.