Fig 1.
(a) Annual mortality rates for the climatic types. Significant differences between the annual mortality rates of the climatic types are labelled with different lowercase letters (P < 0.05). Values represent the mean ± SE. (b) Relationship between annual mortality rate and mean annual precipitation (MAP).
Table 1.
Mean tree mortality rates and wood densities by tree group, forest type, and stand age.
Significant differences between biotic factors (angiosperm or gymnosperm, deciduous or evergreen, sapling or adult) are labelled with different lowercase letters (P < 0.05).
Fig 2.
Relationships between annual mortality rate and (a) wood density and (b) tree density.
Table 2.
Multi-way analysis of variance of tree mortality rate for mean annual precipitation (MAP), mean annual temperature (MAT), elevation, group, forest type, and wood density.
Group includes angiosperm and gymnosperm species; forest type includes evergreen and deciduous species.
Fig 3.
Sensitivity of mortality to drought for all sites with available data.
(a) and (b) showed relationships between tree mortality metrics and MAP. (c) and (d) showed relationships between tree mortality metrics and SPEI. Dark-grey dots indicate droughts. The best-fit models for each drought index and mortality-rate metric are shown, with the 95% bootstrapped confidence intervals. MAP, mean annual precipitation; SPEI, standardized precipitation-evapotranspiration index. The pink ovals indicate the lower tree mortality rate for tropical rainforests.
Table 3.
Stepwise regression to identify factors (elevation, mean annual precipitation, mean annual temperature, standardized precipitation-evapotranspiration index, and wood density) determining the annual mortality rate during droughts.
M, annual mortality rate during droughts; E, elevation; P, mean annual precipitation; T, mean annual temperature; S, standardized precipitation-evapotranspiration index; W, wood density.
Fig 4.
Sensitivity of mortality to drought for angiosperms.
(a) and (b) showed relationships between tree mortality metrics and MAP. (c) and (d) showed relationships between tree mortality metrics and SPEI. Dark-grey dots indicate droughts. Dark-grey dots indicate droughts. The best-fit models for each drought index and mortality-rate metric are shown, with the 95% bootstrapped confidence intervals. MAP, mean annual precipitation; SPEI, standardized precipitation-evapotranspiration index. The blue ovals indicate the lower tree mortality rate for tropical rainforests.
Fig 5.
Sensitivity of mortality to drought for gymnosperms.
(a) and (b) showed relationships between tree mortality metrics and MAP. (c) and (d) showed relationships between tree mortality metrics and SPEI. Dark-grey dots indicate droughts. Dark-grey dots indicate droughts. The best-fit models for each drought index and mortality-rate metric are shown, with the 95% bootstrapped confidence intervals. MAP, mean annual precipitation; SPEI, standardized precipitation-evapotranspiration index.
Table 4.
Model fits for the global response of tree mortality to drought.
Data sets varied with tree group (angiosperm and gymnosperm), tree mortality metric (annual mortality rate (% y-1) and mortality rate during droughts (%)), and drought metric (MAP and SPEI). Best-fit models are highlighted in bold and are displayed graphically in Figs 3, 4 and 5. For polynomial models, we fitted all possible two- and three-factor models and only selected a model with cubic terms when it had an AIC lower than all other models and an R2 higher than all other models. MAP, mean annual precipitation; SPEI, standardized precipitation-evapotranspiration index; AIC, Akaike’s information criterion.