Extended Data Fig. 8: Impacts of climate data resolution and spatial autocorrelation on vegetation thermophilization.

a, Relationship between thermophilization rates estimated using fine (30 arc-sec, around 1 × 1 km) and coarse (10 arc-min, around 20 × 20 km) resolution climate data. Each point represents a plot (n = 457). The dashed lines represent the 1:1 relationship, while solid black lines show the fitted linear regression, with shaded bands indicating the 95% confidence interval around the fitted relationship. b, Overall thermophilziation rates estimated from Bayesian mixed-effects models using fine and coarse resolution climate data. Circles represent means (n = 457 plots) with 80% (thick line) and 95% (thin line) credible intervals (CIs) and posterior distributions obtained from Bayesian mixed-effects models. Thermophilization rates were quantified as the declines of cold-adapted species (5^th percentile shift), overall (median), and increases of warmth-demanding species (95^th percentile shift) using vegetation abundance data excluding rare species. c, Comparison of estimated thermophilization rates (°C per decade) with and without accounting for spatial autocorrelation using a Gaussian Process (GP). Circles represent means (n = 4372, 1209, and 457 plots for forests, grasslands, and alpine summits, respectively) with 80% (thick line) and 95% (thin line) CIs and posterior distributions obtained from Bayesian mixed-effects models. Thermophilization rates were quantified as the declines of cold-adapted species (5^th percentile shift), overall (median), and increases of warmth-demanding species (95^th percentile shift) using vegetation abundance data excluding rare species. The overlapping 95% CIs across models suggests that spatial autocorrelation does not confound the observed ecosystem differences in thermophilization, and that the ecosystem effects are robust to spatial structure.