Extended Data Table 2.

Estimates for the two main meta-regressions using randomized and reduced datasets

We randomized observations of tree status within each species, thus removing any relationship between mortality and predictors but retaining species-level mortality rates, and observations of local conspecific density within each species, thus removing the relationship between mortality and conspecific density but retaining the relationships between mortality and confounders (see ‘Robustness tests’ in Methods). For the reduced dataset, we removed n = 99 influential species-site-specific CNDD estimates with Cook’s distances larger than 0.005 to evaluate the possibility that a few observations were responsible for the observed patterns. Species-site-specific CNDD estimates and predictors are defined as in Table 1. Predictions of the meta-regressions are shown in Extended Data Figs. 2 and 3.