Since 2019, African Swine Fever (ASF) has caused significant damage to pig farms in South Korea. This study aimed to enhance the reliability of habitat suitability models for preventing the spread of African Swine Fever (ASF) by optimizing the parameters of the ST-DBSCAN clustering algorithm using wild boar ( Sus scrofa ) spatial data. Key ST-DBSCAN parameters were optimized using temporal and spatial information, and DNA-derived genetic relatedness was used as an independent criterion to validate and tune the clustering results, enabling clusters to better reflect biologically related individuals rather than purely proximity-driven groupings. The Maxent model was used to compare habitat suitability performance across three types of location data: raw (unclustered) occurrences (Type 0), occurrences clustered using literature-based ST-DBSCAN parameters (Type 1), and occurrences clustered using DNA-validated optimized ST-DBSCAN parameters (Type 2). The analysis revealed that the reliability of 'Type 2 (AUC = 0.883)' was significantly higher compared to 'Type 0 (AUC = 0.777)' and 'Type 1 (AUC = 0.814)'. Notably, 'Type 2' demonstrated a strong ability to identify areas with a wild boar occurrence probability of 0.8 or greater. These habitat suitability outputs provide a basis for ASF prevention by prioritizing surveillance and guiding targeted placement of barriers (e.g., fencing) or buffer measures near high-risk wildlife–livestock interfaces.
Baek et al. (Mon,) studied this question.