Exploring on the climate regionalization of Qinling-Daba mountains based on Geodetector-SVM model

according to the number of Pinus tabuliformis forests. Second, the boundaries of previous studies were used as a reference for the real subtropical/temperate zone boundary. Compare the differences between them and our result. The values of each factors in the subtropical zone, transition zone, and temperate zone are calculated respectively, including the coldest month average air temperature (meantem01), the warmest month average air temperature (meantem07), DEM, the annual average precipitation (pa), the accumulated temperature with �0˚C (aat0dem), and the accumulated temperature with �10˚C (aat10dem). To analyze the characteristics of the transition zone, we calculate the q-statistic values for evaluating spatial stratified heterogeneity was calculated based on Geode- tector model. Results and discussion Establishment of the typical vegetation index system The basis of comprehensive physical regionalization is spatial stratified heterogeneity [36]; a higher q represents a stronger spatial stratification. The q-statistics of covariates for every paired vegetation were calculated, as shown in Table 1. With the exception of the sand content and the silt content with q-statistics of 0.471834 and 0.3547604, respectively, which were iden- tified as the driving factors of Group I vegetation (Quercus wutaishanica forest, Quercus fabri forest), no statistically significant spatial heterogeneity of soil-related explanatory factors is noted in the study area, and the q-statistic is generally less than 0.3. This finding indicates that

the spatial distribution of vegetation is weakly affected by soil elements. In contrast, the annual average precipitation and the coldest month average air temperature, contributed a remarkable impact, generating q values from 0.47 to 0.63. This finding indicates that both of these factors are the determining factors and significantly affect the spatial distribution of vegetation. Statis- tically significant spatial heterogeneity of some factors, including the variables of accumulated temperature with �10˚C, DEM and warmest month average air temperature, is noted in Groups I and II, with a q-statistic greater than 0.3. However, significant heterogeneity is not Table 1. Interpretation degree of each explanatory variable. Variable Group I Quercus wutaishanica forest, Quercus fabri forest Group II Pinus tabuliformis forest, Pinus massoniana forest Group III Populus davidiana forest, Clobalanopsis glauca forest aat0 0.39281 0.53895 0.300672 aat10 0.422326 0.496743 0.284727 pa 0.578824 0.471186 0.637127 tadem 0.396613 0.542032 0.314665 DEM 0.45271 0.409616 0.217703 meantem01 0.471868 0.639598 0.490474 meantem07 0.45276 0.507524 0.287343 sand content 0.471834 0.124422 0.119135 silt content 0.347604 0.216484 0.141976 clay content 0.273074 0.178613 0.076621 soil organic content 0.005412 0.021158 0.00498 soil phosphorus 0.040388 0.070519 0.096646 soil nitrogen 0.009881 0.020509 0.008813 soil type 0.026083 0.03357 0.003346 https://doi.org/10.1371/journal.pone.0241047.t001 PLOS ONE Climate regionalization based on Geodetector-SVM model

PLOS ONE | https://doi.org/10.1371/journal.pone.0241047 November 2, 2020 9 / 16