Testing a Susceptible Population Density Among Other Explanatory Factors of African Swine Fever Spread in Wild Boar...

This study aims to identify the role of various natural, socioeconomic, and demographic factors in the development of the African swine fever (ASF) epizootic among wild boar in the Russian Federation (RF) from 2007 to 2023. In this study, particular emphasis was placed on testing the significance of wild boar population density as a key factor contributing to the spread of ASF within this population. During the study period, 1711 outbreaks in wild boars were reported in the RF, accounting for 41.7% of all ASF outbreaks in the country. We tested two regression approaches to model the dependance of the total number of ASF outbreaks in second-level municipal units (districts) on a range of potential explanatory factors, including the dynamically changing annual population density of wild boar. We employed negative binomial regression (NBR) and, as an alternative approach, classification and regression trees (CARTs). The predictive capabilities of both models were evaluated using 10-fold cross-validation. One of the most significant identified factors was the number of ASF outbreaks in domestic populations, which may indicate a close coexistence of both domestic and wild ASF cycles. Population density showed limited significance in the negative binomial model (p=0.05). The CART model demonstrated high significance for this factor in the Far Eastern regions of the country, where the highest number of outbreaks occurred at density values above 0.120 individuals/km². For the European part of the RF, the threshold density value was 0.026 individuals/km², which closely corresponds to the threshold established by country's authorities for managing wild boar populations to prevent the spread of ASF. The results demonstrated a complex and nonlinear influence of wild boar population density and ASF outbreaks among domestic pigs on the likelihood of new infection foci emerging in the wild fauna. The modeling results indicated that although both types of models had comparable predictive capabilities, the CART approach provided better visualization and understanding of the analysis results. These findings can be used to optimize population management activities to regulate wild boar numbers in infection hotspots across different geographical areas delineated by the risk level of infection spread.