Fig. 3: Relative importance of different ecological processes in response to warming.

a Under control. b Under warming. c Changes of homogeneous selection under warming (orange bar) and control (aqua bar). d Changes of drift. a–d were estimated by iCAMP. e Stochasticity estimated by different methods in the later 3 years. One-side significance based on bootstrapping test was expressed as ***p < 0.01; **p < 0.05; *p < 0.1. p = 0.001, 0.818, 0.014, 0.346, 0.035 in Year 1–5 for homogeneous selection; p = 0.001, 0.657, 0.066, 0.207, 0.058 in Year 1–5 for drift; p = 0.000, 0.542, 0.500, 0.014, 0.567 for tNST, pNST, NP, iCAMP, and QPEN, respectively. L, M, S, and N represented large (|d| > 0.8), medium (0.5 < |d| ≤ 0.8), small (0.2 < |d| ≤ 0.5), and negligible (|d| ≤ 0.2) effect sizes of warming, based on Cohen’s d (the mean difference between warming and control divided by pooled standard deviation). Data are presented as mean values ± SD. Error bars represented standard deviations; c and d n = 6 comparisons among four biologically independent samples at each time point; e n = 18 comparisons = 6 comparisons in each of the 3 years. Source data are provided as a Source Data file.