2.2.1 Environmental factors
Using 22 environmental factors (including climate, soil and biology) to study the changes in the distribution of T. chinense (Table S1). In order to match the environmental factors with the current climate scenario, based on the monthly temperature and precipitation data provided by the Worldclimate database (v2.1), we calculated 19 climate data in the time range of 2000-2018 through the “biovar” package. The two soil factors (T_OC and T_pH_H2O) were downloaded from Harmonized World Soil Database (HWSD, http://www.fao.org). As a semi-parasitic plant, vegetation (host) is the prerequisite for the growth of T. chinense . Therefore, we included NDVI factors that can reflect vegetation growth in the ensemble species modeling. NDVI data was downloaded in MODIS (https://www.earthdata.nasa.gov). We averaged the NDVI data from 2000 to 2020 according to the month of the growth period (May to July) of T. chinense , and finally obtained layer was required for the species distribution model modeling.
The high correlation between environmental factors may lead to the species distribution model overfitting. Therefore, we first built the initial model (10 repeated modeling) with MAXENT 3.4.4 software without adjusting the parameters, and removed the environmental factors that contributed less than 1%. The correlation function in ENMTools software was then used to analyze the remaining environmental factors’ correlation (Warren et al., 2010). The two environmental factors with | r | ≥ 0.8 were screened, and the one with relatively small contribution rate was removed. After the above steps, we reserved 11 environmental factors for the final modeling.
To estimate the distribution changes of T. chinense in different periods in the future (2050s: 2041-2060, 2070s: 2061-2080, 2090s: 2081-2100), we selected three widely used atmospheric circulation models (MIROC-ES2L, CNRM-CM6-1, MRI-ESM2-0) to build species distribution model in the future. Each atmospheric circulation model includes four shared socio-economic pathways: SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5, with 12 climate scenario combinations. Among the four selected shared socio-economic pathways, the low to high radiative forcing scenarios range from SSP1-2.6 to SSP5-8.5 (Jiang et al., 2020). ArcMap10.5 software was used to average the data of three climate models in the same carbon emission scenario and year. The spatial resolution of environmental data was 2.5 min.