Relative importance for six variables as predictors of tree mortality, based on Bayesian Information Criterion (BIC) values of single-variable models.

ΔBIC values are shown for each predictor variable for each species, which is the difference between the best-supported model and the model of interest. The “variable ranking” column ranks the models in order of statistical support (i.e. BIC value); models considered much less strongly supported than a higher ranking model (i.e. with ΔBIC >10 compared with the model ranked immediately above it) are indicated with “>>” while similarly supported models are indicated by “>”. Sample size (N) and ΔBIC for each species and each predictor is shown: diameter at breast height (D), basal area of larger trees (BL), species dominance index (sdi), mean annual temperature (mat), annual precipitation (ap) and organic matter content (omc).