关键词: GTWR模型, 松材线虫病, 时空异质性, 台州市

Abstract: Clarifying the spatiotemporal variations and influencing factors of pine wilt disease epidemic provides a reference for the spatiotemporal dynamic monitoring and regionally differentiated prevention and the control of pine wilt disease. The Mann-Kendall trend test, Moran’s I index, and LISA index were employed to analyze the spatiotemporal distribution of pine wilt disease from 2021 to 2024. OLS, GWR, and GTWR models were constructed based on variables such as elevation, summer mean temperature, precipitation, wetland area, and population density to explore the spatiotemporal heterogeneity of their relationships with pine wilt disease severity and to compare model performance. The results showed that the northern, western, and southeastern regions of Taizhou experienced higher infection severity, while the eastern region showed lower severity. The average infection severity for Taizhou from 2021 to 2024 was 17.76%, 16.96%, 13.53%, and 8.72%, respectively. Over this period, 60 towns/subdistricts showed a significant decline in infection severity, while 7 towns showed a significant increase. Moran’s I index was significantly greater than 0, indicating spatial clustering. The GTWR model outperformed the OLS and GWR models, with an R² of 0.543, an AICc of 4093.23, and an RSS of 75995.8. The regression coefficients for each variable were as follows: elevation (0.019), precipitation (0.060), summer mean temperature (1.880), wetland area (0.30), and population density (0.003). These findings suggest that high-infection areas are mainly distributed in mountainous and hilly regions with high-density forests while low-infection areas are concentrated in flat, densely populated urban centers, showing a “high-high” and “low-low” clustering pattern. From 2021 to 2024, the overall infected area showed a declining trend, with an increasing rate of decrease each year, while the spatial clustering pattern remained relatively stable. The GTWR model more accurately captures the spatiotemporal non-stationary relationships between influencing factors and pine wilt disease severity. Summer mean temperature and elevation positively contribute to pine wilt disease outbreaks, whereas the effects of precipitation and wetland area exhibit fluctuations and complexity.

Key words: GTWR model, pine wilt disease, spatiotemporal heterogeneity, Taizhou