Global field observations of tree die-off reveal hotter-drought fingerprint for Earth’s forests

Earth’s forests imperiled by further warming

We quantified a global-scale hotter-drought fingerprint, representing a global climate signal for years with documented site-specific tree mortality. Climate-induced tree mortality in recent decades under hotter-drought conditions has been documented across forests from a diverse array of boundary conditions, spanning from the tropics to the boreal, from sea level to 3,500 m, and across a four-meter precipitation gradient and 30 °C of mean annual temperature. One reason that the hotter-drought fingerprint is similarly evident in the year prior to reported mortality onset (Fig. 3), as well as largely echoed in the year after, may be due to the imprecise nature of identifying the “onset” and duration of mortality (e.g., visual indications of mortality may lag significantly behind environmental drivers¹⁶). In addition, chronic drought conditions commonly span multiple years, cumulatively predisposing eventual, lagged mortality events^13,26,27—consistent with our observed “3-year hotter-drier window,” centered on the nominal mortality year (Fig. 3).

Our global-scale hotter-drought fingerprint, focused on acute hotter-drought extremes, represents a cohesive signal for climatic drivers of tree die-off in many of Earth’s forests. Other approaches could consider other temporal dimensions of climate signals (e.g., shorter-term heat-wave stress, longer-term chronic drought, changes in seasonal drought duration or timing), which may further improve our understanding of climatic drivers of tree mortality. Ideally, future efforts to harmonize global forest inventory and monitoring methodologies, including their currently-disparate documentation of tree mortality, will reduce the inherent sampling biases (typically favoring northern hemisphere and/or areas adjacent to well-funded research institutions) and presence-only limitations of our present database¹¹.

Additionally, we found that many of Earth’s forests may become increasingly imperiled by further warming and drought, as the frequency of lethal climate conditions observed with recently documented global mortality events will accelerate with further warming (Fig. 6d). Although our approach does not reveal the particular detailed mechanistic ecophysiological responses to the hotter drought that are driving mortality for each specific site, it exemplifies the powerful utility and practical potential of empirical approaches that link direct observations of tree mortality from diverse precisely georeferenced locations to observed climate drivers. While multiple emerging lines of evidence indicate that warming puts trees at greater risk under drought conditions^{9,14,15,19,24,35}, the quantitative hotter-drought fingerprint we identified here suggests that further warming may accelerate global forest die-off across many biomes. The impact of this hotter-drought fingerprint is acting on Earth’s forests already, with nearly half a billion trees having died from hotter-drought events in Texas and California alone since 2010^36,37. In central Europe, hotter drought starting in 2018 has led to extensive dieback of forests that is ongoing—and of yet undetermined magnitude and extent—which could lead to significant ecological transitions³⁸. Other notable global tree mortality events documented during hotter-drought episodes include three pulses of large-tree mortality since 2005 across Amazon basin tropical moist forests^39,40, and historically unprecedented hotter-drought-triggered dieback in Jarrah forests of southwest Australia in 2011^8,19.

Individual trees and forest ecosystems may benefit in various ways (e.g., increased water-use efficiency, stored non-structural carbon, etc.) from productivity gains under elevated atmospheric CO₂²²—when soil nutrients and water are not limiting. However, the net effects of increasing CO₂ in combination with a changing climate on the mortality of global forests during hotter drought are uncertain^4,9,35. In particular, during hotter-drought events, plant uptake of CO₂ is limited by the initial closing of stomata—with CO₂ uptake eventually blocked as leaves lose turgor, followed by failure of the coupled plant water-and-carbon transport system which may ultimately result in death^16,28. Thus, potential amelioration of tree mortality risk by the ~85 ppm atmospheric CO₂ increase during the timeframe in our database (1970–2018) might have been overwhelmed by the concurrent increases in temperature during mortality-event years (Fig. 5). This warming presents a triple threat to tree survival in the form of amplified soil drought, atmospheric drought, and heat stress, and our results are consistent with experimental findings that drought and warming can negate or overcome the effects of elevated CO₂^17,18.

Earth’s historical forests are especially vulnerable

As the longest-lived organisms on Earth, trees routinely are imbued with historical and cultural significance by human societies, while also persistently sequestering carbon and amplifying local biodiversity for centuries, sometimes millennia. In contrast, extreme climate stress events occur on the scale of days to months to a few years, and in these relatively brief periods, large old trees—exemplars of Earth’s historical forests⁶—can be especially susceptible to mortality^{5,41,42,43,44}. Forests will certainly persist and thrive over large areas into Earth’s future, but increasingly they will have to rapidly shift in physiological function, morphology, genetics, species composition, structure, and geographic distribution in response to anticipated climate changes. Where the pace of climate change outruns the adaptive or acclimation capacities of historically-dominant tree individuals and species, additional die-off events are likely to occur and some forests may even cease to exist. In particular, the current tree communities of Earth’s historical old-growth forests—which took centuries, sometimes millennia, to grow to structural dominance under now locally-shifted climate conditions—may continue to often be most negatively affected by continued warming and drying^4,43, as novel hotter-drought extremes increasingly exceed their range of survivable climate across diverse forested biomes. The expected near-term outcome is simplified tree communities, where more drought- and heat-tolerant species survive, and less tolerant species diminish or perish. In many cases, this may lead to lasting changes in vegetation composition, stature, and spacing, where surviving woody plants in these communities do not maintain or develop the complex canopy structure typical of historical old-growth forests^4,9,35,45.

Underestimation of tree mortality from hotter droughts

While our projections for an increase by up to 140% in the frequency of climate conditions associated with recent forest die-off under +4 °C may seem severe, they are modest in comparison to some current empirical and mechanistic process-based model predictions for catastrophic forest die-off at continental scales under hotter droughts^12,14. Our projections for increasing die-offs under further warming are consistent with projections showing the potential for large increases in mortality under future hotter drought^12,14,46, although these projections are often limited to single species or single biomes. Even continental-scale projections for up to 40% increases in the frequency of mortality-inducing hotter droughts under ~+2.5 °C since pre-industrial²⁰ are in general agreement with our global analysis’s 20% under +2 °C (Fig. 6d). Further, our projections of increasingly frequent, historically lethal climate conditions for Earth’s forests may be conservative for several reasons:

1. (1)

   Requiring that all six climate variables meet or exceed mortality year conditions, concurrently in the same year, is a strong filter. For example, TMAX, VPD, and PDSI all exceed mortality-year conditions under +4 °C in about 4 out of every 5 years (Supplementary Fig. S3), whereas under the same warming scenario, all six metrics exceeded the hotter-drought fingerprint only half as often.

2. (2)

   Tree mortality involves diverse disturbance processes that amplify forest die-off in the presence of global warming and hotter droughts^4,24,35 but these were excluded in our analysis, including insects^44,47, pathogens⁴⁸, wind^40,49, and lightning⁵⁰. Additionally, anthropogenic warming promotes greater wildfire activity, particularly fire extent and severity in many forests worldwide^7,51, driving further declines in some of Earth’s forests. We also have not considered disturbance interactions among these many amplifying and synergistic agents of tree mortality^49,52—but conversely, we also acknowledge that thinning from either climate-triggered mortality or these associated synergistic agents, may partially buffer against future losses^35,45.

3. (3)

   Our findings indicate that climate anomalies of tree mortality event years are trending towards ever hotter and drier conditions (Fig. 5, Supplementary Fig. S7), concurrent with any potential ongoing forest acclimation to temperature and/or CO₂ fertilization^15,22. Yet the potential for tree species to acclimate to ongoing climate warming, even with increasing atmospheric CO₂ concentrations, is not unlimited—and when exhausted—forest die-off may rapidly accelerate^9,35,53. Since projected warmer climate conditions include unprecedented extremes of hotter drought for which there are no observed analogs, the potential for crossing historically unknown tipping-point climatic stress thresholds may increase, further amplifying tree mortality³⁵.

4. (4)

   Our analysis of mortality-year frequency uses monthly climate data, yet important drivers can occur on longer (e.g., drought²⁶), and shorter (e.g., heatwave^8,19) timescales. For example, the 4-year-prior signal of cooler/wetter climate (Fig. 3) may reflect favorable pre-drought conditions promoting structural overshoot of trees, which could amplify dieback and mortality risk during subsequent years of hotter drought⁴⁵.

Roadmap for research enabled by a quantitative ground-based global database

The widespread global coherence of our empirically quantified hotter-drought fingerprint may provide immediate opportunities to validate projections of tree mortality in existing models of the Earth system, while also enabling diverse future analyses. Although global in geographic extent, our database is limited by the availability of peer-reviewed, ground-based empirical studies of climate-induced tree mortality, and thus only sparsely covers some regions, particularly large portions of boreal and tropical forests. For example, our hotter-drought fingerprint was consistent across all biomes except the tropical rainforest (Fig. 4)—despite published direct observations of hotter drought as a driver of tree mortality at these tropical rainforest sites^39,40. Additionally, this biome may experience pulses of tree mortality in response to different climate fingerprints, particularly involving longer-duration dry seasons—not just intensified single monthly extremes.

Despite this and some other limitations, our database represents a globally-distributed dataset with precisely geo-referenced sites where ground-based heat- and drought-induced tree mortality has been documented. Our use of this database to quantify a global hotter-drought fingerprint of tree mortality illustrates the potential for rapid progress in empirical modeling of forest mortality drivers and thresholds at spatial scales from local to global, where direct observations of forest responses to climate stress can help identify and quantify mortality drivers. Toward the goal of fostering further rapid community development of many more such direct observational records of climate-induced forest stress and tree mortality worldwide—with methods ranging from local ground-based sites to synoptic remote-sensing—this database immediately will be served as an open-access resource at the International Tree Mortality Network (https://www.tree-mortality.net), an academic networking initiative associated with the International Union of Forest Research Organizations’ (IUFRO) task force on monitoring global tree mortality patterns and trends (https://www.iufro.org/science/task-forces/tree-mortality-patterns). The complete database—along with an interactive version of Fig. 1 from this paper—will allow users to zoom in on dense plot networks, with direct links to the supporting literature for each point. This online database includes the reference for each plot, its precise coordinates, dominant species, associated biotic agents, and the year of mortality onset. To further update and rapidly increase the quantity and spatial representativeness of global tree mortality observations, ongoing online contributions from diverse observer groups, ranging from practicing foresters and field ecologists to remote-sensing scientists, can be integrated into the website in near-real-time via a user-friendly entry form.

As the only global set of ground-truthed observations of drought- and heat-induced tree mortality, this database can immediately aid in validating remote-sensing technologies for eventual synoptic monitoring in near-real-time of tree mortality (which could then feedback into the database). Additional groups to benefit from the database are those interested in climate and physiological mechanisms of tree mortality, including the connected fates of all forest-dependent life^5,19, with an aim toward improving the representation of climate-induced tree mortality representations in Earth system models. Related future research opportunities associated with this initial online database include:

1. (1)

   Identify additional chronic (e.g., seasonal to decadal) and acute (daily to weekly) climatic signals of tree mortality, including thorough analyses that quantitatively consider antecedent and lagging factors, and duration and seasonality of drought stress;

2. (2)

   Synthesize mortality observations from extensive forestry plot inventory networks, to increase spatial representation for the global climate signal of tree mortality, and to identify where during these events trees did not die-off;

3. (3)

   Apply remote-sensing approaches to mortality detection using this spatially precise (and in some places plot-dense) database for ground-truthing, to determine the full spatial extent of known mortality events, and aid in ongoing monitoring of forest stress and tree mortality events in near-real-time;

4. (4)

   Benchmark state-of-the-art Earth system models via hindcasting, to assess the accuracy of tree mortality event representation—and to do so across spatial resolutions (as in Supplementary Fig. S4) at which these planetary models operate;

5. (5)

   Develop approaches to understand potentially unique features and drivers of hotter-drought mortality in tropical rainforests (differing climate signals, e.g., extended dry seasons, where warming/drying of typically moderate shoulder seasons may matter more than intensified single-month extremes), the single biome in which our global approach did not reveal a strong hotter-drought fingerprint;

6. (6)

   Investigate how the severity of forest die-off events will respond to further warming; and

7. (7)

   Invest in monitoring, documenting, and gathering mortality data for forests under-represented in this initial global database—especially in the extensive critical carbon sinks of boreal forests and tropical rainforests.

Future challenges for Earth’s forests and societies under hotter drought

In conclusion, our findings reveal the emergence of a global acceleration of lethal climate conditions, associated with recent forest mortality events, under further warming. Earth’s historical forests in particular face a challenging future, including dramatic changes in the extent, composition, age, and structure of these unique and irreplaceable forests, with planetary-scale consequences for biodiversity and the cycling of water and carbon. Our findings both corroborate earlier studies of hotter-drought driven mortality at local to regional scales^{8,13,19,20,24,36,38} and extend these findings by quantifying the commonality in climate anomalies across this planetary-scale observation-based database of tree die-off. Although forests often are invoked as an important part of the solution to the present global climate crisis, their role as reliable carbon sinks in mitigating climate change depends upon their ability to survive further warming^10,22,52—which our global hotter-drought fingerprint identifies as an imminent threat. Our findings show that limiting warming to +2 °C over pre-industrial levels could reduce the frequency of these climate conditions associated with observed tree mortality events to less than half that predicted at +4 °C. Efforts to protect the world’s climate from excessive warming likely will be decisive in determining the future persistence of many of Earth’s forests.

Source: Ecology - nature.com