Extended Data Fig. 5: The diversity − stability relationships (DSRs) across multiple spatial scales, using sites with ≥ 5 years records (N = 24 sites).
From: Consistent stabilizing effects of plant diversity across spatial scales and climatic gradients
Shown are the log−log slopes between α diversity and α stability at the quadrat level (A, F1,22 = 12.34, R2 = 0.36, P = 0.002), between γ diversity and γ stability at the plot level (B, F1,22 = 9.79, R2 = 0.31, P = 0.005), between τ diversity and τ stability at the site level (C, F1,22 = 16.99, R2 = 0.44, P = 0.001), between \(\beta _D^{\alpha \to \gamma }\) and \(\beta _S^{\alpha \to \gamma }\) across quadrats (D, F1,22 = 11.13, R2 = 0.34, P = 0.003), between \(\beta _D^{\gamma \to \tau }\) and \(\beta _S^{\gamma \to \tau }\) across plots (E, F1,22 = 8.51, R2 = 0.28, P = 0.008), and a comparison of regression slopes across scales using ANCOVA (F, F2,66 = 0.889, P = 0.416 among quadrat-, plot-, and site levels; F1,44 = 0.233, P = 0.632 between across-quadrats and across-plots levels). In (A − E), lines represent the overall significant relationships between biodiversity and community stability from the best-fit linear regression models (LMs), and shaded areas are the error bands and denote 95% confidence intervals. Significance levels are indicated as follows: ‘*’: P ≤ 0.05 and ‘**’: P ≤ 0.001. R2 is the explained variance in LMs. In (F), bars and error bars are regression coefficients and standard errors from p-LMs (n = 36 for all) in (A − E). Note that in (F), pairwise comparisons between quadrat, plot, and site levels are non-significant (P > 0.1 for all). More information about the fitted models and partial regression models is provided in Supplementary Tables 1–2.
