Skip to main content

Redefining the carbon sink capacity of global forests: The driving role of tree mortality

Periodic Reporting for period 2 - TreeMort (Redefining the carbon sink capacity of global forests: The driving role of tree mortality)

Reporting period: 2019-08-01 to 2020-09-30

Evidence increasingly suggests widespread increases in tree mortality rates across many of the world's forests. The local impacts of increased tree mortality can be substantial and varied, but they also have a global implication; taken together they may threaten the enormous uptake of carbon dioxide by the world's forests, which currently offsets a net 20% of anthropogenic emissions. Yet understanding tree mortality remains a challenge. Compared to growth, about which we understand a great deal, tree death is much more difficult to observe because it happens relatively infrequently. This observational challenge is magnified when trying to understand the behaviour of forests at the continental or global level. Yet such large-scale understanding is necessary in order to quantify the impact of forests on the carbon cycle and on climate. Without this, the computational models used to understand how terrestrial ecosystems function within the Earth system have had very limited information on which to base estimates of mortality rates. As a result, projections of present and future tree mortality rates from these models contain a striking lack of consensus. This has a very serious implication: Depending on the rates of tree mortality, the world's forests may continue to take up 20% or more of humanity's carbon dioxide emissions, this uptake may diminish, or forests may even become a net source of carbon dioxide. The size of this uptake is central to assessments of how much carbon dioxide humans can continue to emit with an expectation of keeping global temperature change within defined boundaries, such as the 1.5° and 2° targets set out in the Paris Climate Change Agreement. A large reduction in forest carbon uptake would result in smaller carbon dioxide emission budgets and thus require more rapid action to reach net zero emissions.

The TreeMort project aims to provide a solution to this problem. By combining newly available sources of data with appropriate conceptualisation of processes and state-of-the-art ecosystem modelling, it seeks to:
- Provide observation-based constraints on recent tree mortality rates across the world's major forest biomes.
- Identify the contributions of different forms of tree mortality and formulate a mechanistic understanding that is appropriate for application at continental-to-global scales.
- Integrate this new understanding into global terrestrial ecosystem models to reduce substantially the uncertainty in the role of forests in the present and future carbon cycle.
Work in the first half of the project has concentrated on (1) assembling tropical, temperate, boreal and global forest observational datasets which can be applied to constrain tree mortality rates and drivers over the last 2-4 decades, (2) analysis of these datasets and (3) propagation of the results into a state-of-the-art global vegetation model.

We have produced a first consistent estimate of occurrence frequencies of stand-replacing disturbance events across the world's forests. These events include large patches of tree death from causes such as wildfire, storm damage, biotic outbreaks and harvest. Applying this information in a vegetation model, we were able to identify that these disturbances contribute about 12% of total carbon turnover from tree mortality globally, but with large regional variation (Figure 1). We were also able to identify those regions in which forest biomass is strongly controlled by such disturbances and those in which it plays a minor role. By assimilating information on forest age structure, based on forest inventories, into the same vegetation model, we were able to calculate a new estimate of the impact of past disturbances on the current uptake of carbon by the world's forests, finding that a quarter of forest carbon uptake is due to disturbance legacies.

As the type of disturbance has major implications for both estimating possible changes in disturbance rate and the fate of the carbon in dead trees, we have further worked to breakdown the disturbance rate by cause. We have individually delineated the more than 400 million such disturbance patches that occurred globally between 2000 and 2018, broken down rates by size (Figure 2) and assessed the contributions of fire, land-use change and human activities for different forest biomes between 2002-2016. These results provide the basis for a more nuanced assessment of the current role of stand-replacing disturbances in global forest carbon cycling, providing the baseline from which future assessments can be made.

We have assembled and standardised a global forest dynamics dataset, following over time 10 million individual trees in 45 different countries, from tropical, temperate and boreal biomes. For the tropical forests, 627 long-term plots, managed and quality-controlled by ForestPlots.net and contributed by many hundreds of tropical ecologists, especially from Latin America, have provided a large observational dataset of tree mortality and carbon turnover. This has been complemented by data from a further 29 networks across all biomes, including both research plots and national forest inventories. This dataset provides the basis for quantifying the rates and understanding the drivers of the remaining 88% of carbon turnover from tree mortality that happens below stand-scale. Two analyses on subsets of this dataset for Europe and the Amazon have shown that tree mortality in most European forests is dominated by harvest, whilst in remote Amazon forests there is a relatively even split between those trees that die broken or fallen from physical disturbances versus those which die standing, albeit with significant regional differences in both cases. These results provide hard quantification of the contribution of different drivers of tree death at the continental and finer scales, as well as establishing the groundwork for the analysis of the full global dataset.

Complementary to this, we have assembled and standardised a global database of 19 plant hydraulic traits and developed a methodology to quantitatively identify different functional strategies employed by trees to withstand drought stress. These functional strategies are designed to be employed in a novel model of tree hydraulics suitable for application at the global scale. We have contributed to the development of this model and combined with the functional strategies it forms a strong basis for a global scale estimate of the role of hydraulic failure in tree mortality - believed to be an increasingly important contributor to tree death.
Our published assessments of rates of stand-replacing disturbances and their interactions with the carbon cycle provide the first pan-global quantification of an individual component of tree mortality and represent a significant step forward in understanding of the drivers of forest carbon turnover. Likewise, our calculations of the contribution of past disturbances to the carbon sink showcase a novel approach to the quantification of this important component of the global carbon cycle, drawing heavily on ground-based observations. They also provide the first breakdown of the global carbon sink by forest age. Moving forward, the completion of analyses on the other assembled datasets is expected to provide hard quantification of the remaining components of tree mortality for the recent-historical period. With this information we will be able to much more tightly constrain global vegetation model simulations to reality by implementing new observation-derived process representations for individual tree mortality and for major causes of disturbance. This will enable future projections of global forest carbon cycling with substantially higher confidence than has previously been possible.
Percentage of carbon turnover due to mortality that results from stand-replacing disturbances
Global average rates of forest disturbance as a function of minimum patch size considered, 2001-2016