关键词: 松材线虫病, 时空热点分析, 数据挖掘, 逻辑回归, 驱动因素

Abstract:

Objective: With integrating national county-level epidemic data of China from 1982 to 2023, spatiotemporal scan statistics and regression models were employed to to systematically reconstruct the transmission pathway of pine wilt disease (PWD) in China over past 42 years, identify key evolutionary stages, quantify the heterogeneous effects of drivers of recent hotspot outbreaks, and reveal differences in driving modes among regional hotspots, thereby providing a scientific basis for formulating differentiated and precise prevention and control strategies. Method: SaTScan was used to perform spatiotemporal scan statistics based on the discrete Poisson model. The temporal unit was year, and both the maximum spatial and temporal scanning windows were set to 30%. A total of 999 Monte Carlo simulations were conducted to identify significant hotspots (P<0.05). A binary logistic regression model was separately constructed for the three temporally overlapping spatiotemporal clusters identified during 2015—2023 (located in central-southern China, the Shandong Peninsula, and eastern Liaoning, respectively). The dependent variable was whether PWD occurred at the county level (1=occurred, 0=not occurred). The independent variables were the number of spatially neighboring epidemic counties (NEIGHBOR), mean temperature and precipitation from June to September (TEMP_MEAN, PREC_MEAN), normalized difference vegetation index (NDVI), tree cover (TC), and regional gross domestic product (GDP). The Hosmer–Lemeshow test was used to evaluate model goodness-of-fit. Result: 1) Spatiotemporal evolution pathway: a total of 9 significant hotspots were detected, exhibiting a four-stage transition pattern. During 1982—1988, only one significant hotspot was identified at the Jiangsu–Anhui border (RR=3.39), reflecting early local clustering of the epidemic. During 1989—1999, two clustering centers formed in the Yangtze River Delta (RR=12.22) and the Pearl River Delta (RR=8.66). During 2000—2014, the epidemic expanded into the central-western inland, with a new significant hotspot in the Wuling Mountains. During 2015—2023, there was a trend of a trend of “northward expansion and westward advancement”. A significant hotspot identified in the Hunan–Hubei–Jiangxi border region (RR=14.15), and northern epidemic areas were established for the first time in the Shandong Peninsula and eastern Liaoning, showing a multi-point outbreak pattern. 2) Common driving mode: the spatial neighbor effect was significant in all three hotspot regions (P<0.05), but its intensity varied: the NEIGHBOR odds ratios for central-southern China, the Shandong Peninsula, and eastern Liaoning were 2.606, 2.359, and 1.627, respectively. This indicates that for each additional epidemic county in surrounding areas, the odds of local epidemic occurrence were 1.627–2.606 times the reference level, suggesting that local diffusion is a common driver across the three hotspot regions. 3) Regionally heterogeneous dominant factors: precipitation in the central-southern hotspot showed a significant negative effect, suggesting that high-humidity environments might reduce the risk of PWD transmission by inhibiting vector insect activity. In the Shandong Peninsula hotspot, NDVI was the strongest positive factor, with a standardized odds ratio of 9.33, indicating that host vegetation density might be a key habitat screening factor for colonization and outbreak in this region. In the eastern Liaoning hotspot, the June–September mean temperature was the dominant limiting factor, indicating that temperature suitability is a fundamental prerequisite for colonization at the high-latitude frontier. Conclusion: This study constructs, for the first time, a national-scale long-term analytical framework, revealed the four-stage invasion and diffusion pathway of PWD (coastal introduction-regional spread-inland jump-northward outbreak), and clarifies the driving mode of “spatial neighbor effect as a common fundamental driver, with regional habitat factor heterogeneity dominating outbreak intensity”. This research provides a key scientific basis for formulating precise prevention and control strategies. In particular, emerging northern epidemic regions need to strengthen climate early warning and source interception, while established epidemic regions should optimize monitoring layouts according to dominant factors.

Key words: pine wilt disease, spatiotemporal hotspot analysis, data mining, logistic regression, driving factor