Spatial scale of species community assembly
To quantify species β-diversity in relation to spatial distance, regional differences, and elevation gradient, the grouped plot-level β-diversity matrix was calculated and then partitioned into various independent spatial components that reflect various β-diversity levels (β-dissimilarity between sampling plots within regions) and δ-diversity (δ-dissimilarity between regions). We thus calculated insect species dissimilarity of (1) plots within transects (β1: 40-100 m scale), (2) plots between two neighboring transects within a region (β2: 200-300 m scale), (3) plots between two transects covering the highest elevation gradient within a region (β3: 1-3 km scale), (4) plots between two neighboring regions (δ1: 250-300 km scale), and (5) plots between two regions covering the highest spatial distance (δ2: > 500 km scale). We firstly performed two-way ANOVAs to test for differences in Horn similarity values between insects and plants at various spatial scales, and the checking of residuals’ normality and homoscedasticity caused doubt about the robustness of the anova result, so we choose Nonparametric Kruskal–Wallis ANOVA, followed by Scheffe’s post hoc tests. The spatial component of turnover in tree species composition was investigated in an identical fashion. Next, Wilcoxon paired tests were used to assess whetherβ-diversity was similar for trees and beetles for each respective separation distance (i.e., β1, β2, β3, δ1andδ2). P-values were adjusted accordingly (Bonferroni corrections).