Introduction / Background / Justification

Tree mortality is fundamental in determining fluxes and stocks of carbon in a forest. Its stochastic nature hampers our understanding of this process and as a consequence tree death is currently poorly constrained by vegetation models. The rates of tree mortality have increased in Amazonian forests, compromising the capacity of these forests to act as a carbon sink. The increase in the frequency and intensity of the dry season and the rise in atmospheric CO2 concentration are anticipated to be the main cause of the observed increase in tree death. However, the extent to which each of these drivers contributes to the basin wide increase in tree mortality is still unknown.



Objective(s)/Hypothesis(es)

We evaluate how and why tree death varies across the Amazon basin by assessing the spatial variation of the risk factors tree mortality in different Amazonian regions. Our analysis provides insights on how to better constrain tree death in vegetation models.



Methods

Using data from over 30 years of forest monitoring across the Amazon, we quantify mortality rates per census and use linear regressions to quantify the trends within each plot. We used a survival analysis approach to quantify the contribution of different risk factors (tree size, growth and species-level traits) on tree death.



Results

Our results show that species-level growth rate is the best predictor of tree death in Amazonia. Tree-level growth rates were positively related to the probability of death and the bioclimatic affiliation of species was an important risk factor in the Southern fringes of the Amazon.



Implications/Conclusions

Our results highlight the importance of combining of species and individual-level variables to model tree tropical tree mortality. Linking mortality in vegetation models to multi-annual mean woody growth rate could result in an improved ability of these models to capture the ongoing changes in Amazon forests.