Based on an unprecedented synthesis of field observations of tree mortality, we quantified the rate of tree mortality across the world’s forests. We found that it did not differ systematically across biomes, nor was it closely related to forest productivity. Tree mortality did, however, play a very important role in explaining the spatial variation in the forest carbon sink. Using observations and modelling we showed that this mortality can be broken down between disturbances which kill a whole stand of trees, such as fires, windstorms and clearcut harvest, and events where one or a few individuals die. By developing novel disturbance models we found that much of the disturbances in the temperate and tropical forests could be attributed to harvest events, rather than natural causes. Changes to the disturbance-regime, caused by harvest and land-use change actions have substantially reduced the length of time carbon remains in forest vegetation (by 32% in northern-hemisphere temperate forests over 2001-2014). Using a novel fusion of modelling with observations of forest age, we showed for the first time that legacies of past disturbance currently account for a quarter of the global forest carbon sink (0.53 Gt carbon per year), offsetting about 5% of annual emissions from fossil fuel burning. These results clearly indicate that considering the effect of disturbances past and future is crucial to be able to properly assess the role of the world’s forests in taking up carbon. Many of the techniques developed in this project can be applied more broadly across the emerging class of vegetation models which explicitly simulate the lifecycle of trees.
Despite their often-dramatic impact on the landscape, however, over 2001-2014 our modelling showed that disturbances were responsible for a relatively modest 12.5% of tree mortality. Thus, mortality of individual trees dominates in the vast majority of forests. Using a new representation of plant hydraulics in our vegetation modelling we see that increases in individual tree mortality in the Amazon rainforest over 1984-2010 can be clearly attributed to drought stress. However, increasing tree mortality observed in Europe over 1986-2010 could largely be explained by increasing rates of forest growth, caused by higher carbon dioxide in the atmosphere, enabling increased rates of harvest. Our results show the power of process-based models to simulate tree mortality across large scales and attribute it to its underlying causes when the underlying mechanisms are appropriately accounted for. Building these mechanisms into the ensemble of vegetation models that underlie carbon cycle and climate projections therefore stands to substantially strengthen the confidence in future projections from these models.
We published our results in major international scientific journals including Nature, Science, PNAS, Nature Geoscience, Global Ecology and Biogeography, Environmental Research Letters. We also presented them in 36 international scientific conference presentations and 6 popular science articles, as well as on national radio and television.
