Tree mortality in a warming world: causes, patterns, and implications

This ERL focus collection has published 17 papers that have advanced our understanding of different dimensions of warming-induced tree mortality. Here we summarize these focus collection papers, organized by four topics related to tree mortality: pathogens, droughts/heat waves, fire/bark beetles, and teleconnections/air pollution. This focus collection illustrates a variety of methods in measuring and modeling tree-mortality, and adds significant new research findings into the scientific literature on tree mortality from hotter droughts. Some of these results also are useful for policymakers and forest managers in addressing amplified forest stress and tree mortality as a result of increasingly severe warming-induced climate and weather extremes.


Introduction
This collection of research reports focuses on how warming-induced climate extremes are killing trees. Since the second half of the 19th century our global climate has warmed by over 1 • C and this warming trend is expected to continue. Furthermore, the frequency of warming-induced climate extremes in the recent decades has been increasing (IPCC 2021) so that what in the past was considered extreme climate has become normal today (Hansen and Sato 2016), and today's climate extremes are expected to become normal in the future. Both socialand environmental-ecosystems are vulnerable in the face of these extreme changes in climate condition; extreme droughts have contributed to extensive wildfires and the fact that warm air can hold more water vapor, as predicted by Clausius-Clapeyron equation 150 years ago, has contributed greatly to extreme flooding . Catastrophic damages caused by these temperature-induced climate extremes have become a growing global concern. Spatially aggregated climate data indicate a global expansion of areas subject to warmer climate and prolonged drought. Recently, tree and forest pathologists and other scientists have been studying connections between forest decline and the extremes in our climate system. In these drought-prone areas tree mortality and forest die-off have increased markedly in recent decades and climate warming appears to be the main driver for widespread tree mortality acting through various mechanisms including drought, fire, weakened resistance to pest attack, and changing community competition. As summarized in table 1, 11 of the 17 published papers in this collection discuss tree-mortality caused by drought or heat waves, three by pathogens, two by fire and bark beetles, and one each by aerosol pollutants and an ecoclimate teleconnection effect.
The methods developed by the authors in this collection are quite diverse, as can be seen in table 1: (a) spatial scale from leaf-level to regional; (b) temporal scale from hourly sampling to multi-decadal tree-ring derivation; (c) field experiments including manipulated heat waves and drought condition, data sources including eddy-covariance fluxes, forest inventory databases, remote sensing images More attention needs to be given to canopy structure in forest management and risk assessments in the future. Guha et al (2018) Temperate tree species of the southeastern United States were exposed to manipulative heat wave events to investigate ecophysiological responses and physiological recovery. The study showed clear difference in heat resilience among the studied tree species with differential damage to photosynthetic capacity.
Loss in photosynthetic capacity was supported by the losses in PSII maximal efficiency and electron transport rate.
Effects of extreme events will not be uniform across the co-occurring temperate tree species of the southeastern United States. Heat-induced damages to PSII may be a mechanistic trait that can be used to project how different species respond to extreme weather events. Future studies should focus on the potential feedbacks of drought-induced tree mortality with altered microbial communities on plant community composition, ecosystem processes, and interactions with disturbance events such as fire.
(Continued.) Failed to find the expected positive temperature-monoterpene-emissions relationship in this study.
Drought may override the effects of temperature on monoterpene emissions and tissue concentrations, and the influence of drought may occur through processes sensitive to overall needle carbon balance. Consequently, added warming does not worsen drought-induced suppression of defensive pine emissions-findings that need to be incorporated into global biogeochemical and biogeographic models.
Data analyses support hypotheses: (a) the probability of ageing-driven tree mortality increases with global change and (b) the mortality probability associated with global change is higher for faster-growing trees.
Tree longevity may further decline with expected increases of atmospheric carbon dioxide and warming-related decreasing water availability in the region.
Annual tree-mortality probability model. The major drivers of declining longevity are increasing atmospheric CO2 and temperature, and decreases in water availability.
Brodrick and Asner (2017) • The airborne canopy water content (CWC) derived from the HiFIS data, and the LiDAR data, • used Lidar images to map out live and dead trees, and then saw how this overlapped with the NDVI values, • built relationship of NDVI with tree-mortality.
Developed remotely sensed CWC model to predict coniferous forest drought-induced mortality.
Future work could also explore the link between species traits and the community-specific relationships between CWC change and tree mortality. Finally, further investigation into fundamental, remotely-sensed limits of CWC could make these findings more concrete and universal.
Adams et al (2017) • Planted pine seedlings in growth chambers, • transplanted into tree pots, • manipulated increasing temperature until all seedlings die, • determined the time to mortality. Collected evidence that the time to mortality declines linearly with increasing temperature.
That tree mortality can be expected to accelerate across a range of increased temperatures should be represented in models, and motivate policy to reduce the anthropogenic drivers of climate warming.
(Continued.) Rising temperatures and greater risk of drought will likely increase tree mortality from fires and bark beetles during coming decades in this region. Thus, sustained monitoring and mapping of tree mortality is necessary to inform forest and greenhouse-gas management.
Used machine learning algorithms to assess regional tree-mortality from satellite maps and US Forest Service reports.

Bark beetles
Hood et al (2018) Conceptual model development based upon reviewing many post-fire tree mortality models.
Address fire-pathogens interaction.
Model development of the fundamental processes of post-fire tree mortality coupled with the development of better management tools. The observation of substantial biological effects under these moderate levels of aerosol pollution suggests that potentially stronger effects and greater biological risk may be observed in areas of high concentrations and deposition rates of hygroscopic aerosols.
Sampled particle numbers and epidermal minimum conductance (g min ) for three species of tree-seedlings. Scanning electron microscopy. and maps; and (d) modeling approaches including conceptual framework and synthetic analysis, statistical models; machine learning algorithms; and Community Land Models and Community Earth System Models.

Droughts and heat waves
Warming-induced tree-mortality is currently driven by drought in combination with increasing temperature called 'hotter drought' by Allen et al (2010Allen et al ( ), (2015. Increasing temperature and drying usually occur together. Climate data show that land temperature has been rising significantly since at least 1980s, with the warm zone annual T > 16 • C, expanding and becoming dryer (Yi et al 2014) and tree-mortality caused by the combination of warming and drought is expected to become more extensive (Allen et al 2015).
Eleven of seventeen studies published in this collection have explored tree-mortality caused mainly by droughts but including other aspects of mortality as well.   (2018) focused on examining forest growth performance between forest edge and interior in Germany following the 2015 European heatwave event. They investigated five different Scots pine stands and sampled 152 trees including tree-ring widths (RWs) and related variables in earlywood and latewood. Five study sites were paired between forest edge and interior, and treering properties, individual growth patterns, and climate-growth relationships were analyzed. They also used close-range remote sensing data (tree height, canopy area, distance to nearest neighbor, and normalized difference vegetation index (NDVI)) to analyze tree vitality along a forestedge distance gradient. They concluded that treerings showed a stronger response to drought at the forest-edge, while NDVI also revealed lower vitality towards the forest edge. • Will forests with shorter or taller canopies be more likely to die during extreme drought? Xu et al (2018) focused on the drought event that occurred in the southwestern United States in 2002 and obtained tree-mortality data associated with the drought event from previous publications. They obtained location-matched canopy data from Lidar measurement and used location-matched MODIS NDVI (MOD13A3) to compare leaf growth. Additionally, they used the standardized precipitation evapotranspiration index (SPEI) as a drought indicator and the tree-RW index (RWI) to perform tree growth analyses. RWI data were derived from the International Tree-Ring Data Bank. They used the SPEI ⩽ −1.67, a forest drought tipping point previously identified by Huang et al (2015), as a severe-drought threshold condition to screen all the data. They found that for trees shorter than 18 m, trunk and leaf growth under drought conditions increased with canopy height but decreased with height for trees taller than 18 m. They concluded that medium-height trees survive drought best.
Heat waves and droughts are usually linked together because when the land surface is dry then evapotranspiration E is low, so most of the landavailable energy from net radiation R n ≈ LE + H (L is latent heat coefficient) is used increasing temperature by sensible heat H. However, heat and water stress can play different roles in killing trees. Guha et al (2018) demonstrated that heat can cause photo-system damage and reduced photosynthetic capacity, but the specific tree-mortality roles of heat and drought from a long-term perspective remained poorly defined.
To fill this knowledge gap, Matusick et al (2018) focused on a typical Mediterranean Jarrah forest in Southwestern Australia that experienced severe heatcompounded drought damage. There was a steady decline in SPEI from 1950 to present, with increased VPD after 2009 peaking during the heat wave of 2011. They selected 20 study sites and made a first survey during May-June 2011. They used standard forestry methods to sample tree properties for each of 20 sites and classified them into four groups: healthy, dying, recently killed, and long dead. Then they remeasured everything in April 2015 during a second survey. Then they used statistical models to link these survey samples to climatic data-using binary logistic regression modeling to determine the influence of short-or long-term changes at tree-level, and using beta regression analyses at stand-level. They found that while chronic historical drought had legacy effects on tree-mortality processes, heat waves acted as an acute stress playing a role in triggering mortality, Tree-mortality driven by hotter-drought has been clearly evidenced globally (Allen et al 2010(Allen et al , 2015. It remains unclear if forest longevity becomes shorter with increasing temperature. Searle and Chen (2018) addressed this knowledge gap using an annual treemortality probability model based on 539 permanent sample plots located in Alberta, Canada from 1960 to 2009. They found that tree longevity was declining with increasing temperature, and suggested a warminduced tree aging factor.
Theoretically, plants' monoterpene emissions will increase with increasing temperature in hotter-drier sites and consequently lower needle monoterpene concentrations, and this temperature effect would dominate the seasonal pattern of monoterpene concentrations regardless of drought. Trowbridge et al (2019) used a field transplant experiment to quantify monoterpene foliar concentrations and emission rates under temperature and moisture conditions that are consistent with global change projections across the vegetative growing season. However, they failed to find the expected positive temperaturemonoterpene-emissions relationship in this study. They speculated that an increase in emissions at the hotter-drier site from a 1.5 • C average temperature increase was offset by decreased emissions from greater plant water stress. Thus, there is an open question to clarify in the future.

Pathogens
Drought clearly is a leading factor for tree-mortality in part because it acts persistently like 'gravity' as a chronic stress on tree-growth, weakening plants' resilience by photo-system damage or reduction in photosynthetic capacity as evidenced by Guha et al (2018). Plants with the weakened resilience are more prone to attack by insects or infection by diseases, accelerating tree death.
Mistletoes are common and widely distributed tree parasites, considered by forest pathologists to be detrimental pathogens in both angiosperm and conifer forests as their commandeering of water and nutrients from host trees can contribute to tree stress and mortality. In principle, hotter-dryer climate conditions will increase ecophysiological stress on tree growth, potentially making trees more susceptible to mistletoe infection, which in turn can lead to reduced tree growth and higher tree mortality rates. Mistletoe infection generally does not directly cause tree mortality, but infection can affect a broad range of forest ecosystem processes including tree productivity, stand dynamics, physiological processes, water budgets, energy budgets, nutrient cycling, and biodiversity.
Can mistletoe parasitism amplify tree mortality? To address this knowledge gap, Griebel et al (2017) conducted a synthetic analytical review and formulated a conceptual framework with an interdisciplinary ecosystem perspective. Authors summarized both positive and negative impacts of mistletoe presence on tree physiology, stand health and successional dynamics, soil nutrient cycling, carbon, water and energy cycling as well as plant-animal interactions and biodiversity. The authors not only identify the knowledge gap of mistletoes infection interacting as a factor in tree-mortality but also highlight priorities for future research on this widespread agent of biotic disturbance, determining critical thresholds that cause large-scale tree mortality events. Wood et al (2018), conducted a data-driven analysis to answer the question: do drought-pathogen interactions affect tree-mortality? They focused on a recent drought-induced tree mortality event that occurred in 2013 around the Missouri Ozarks Ameri-Flux site. They used many different data sources, including (a) forest inventory data spanning 24 years, (b) 12 years (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) of ecosystem-scale carbon and water fluxes, and biological data including predawn leaf water potential (ψ pd ) and annual plot inventories, (c) tree-ring analyses of individual white oaks that were alive and ones that died in 2013, and (d) documentation of pathogen infection. Their data-driven analysis indicates that these droughtpathogen interactions could amplify mortality under future climate conditions and thus warrant further investigation.

Fire and bark beetles
Climate change has contributed to alterations in fire regimes globally in recent decades (Bowman et al 2020), including longer and more severe wildfire seasons in many geographic regions (Pausas and Keeley 2021). Fire directly kills trees and converts organic matter into atmospheric CO 2 , thereby contributing to global warming. Heat is transferred to diverse living tissues of trees during fire, resulting in injuries to different parts of trees after fire. Fire-caused tree mortality results from injuries to the crown, bole, and roots, and their fire resistance and post-fire recovery can strongly differ, strongly affecting resilience (see Yi and Jackson 2021). Changes in fire regime can dramatically change forest structures, stand dynamics, and functions of forest water and carbon cycling. Post-fire resilience of burned trees can be substantially reduced, with lower capacity to deal with further disturbances.
What are the causes and mechanisms of post-fire tree mortality? Hood et al (2018) conducted a literature review on this topic. Authors have given more attention to how fire affects tree defenses and ultimately influences the susceptibility of host trees to bark beetle attack and pathogen infection. Many postfire models were reviewed, assessing their advantages and disadvantages, and two future research directions in post-fire modeling were suggested: (a) continued improvement and evaluation of empirical models to quantify uncertainty and incorporate new regions and species, and (b) acceleration of basic, physiological research on the proximate and ultimate causes of fire-induced tree mortality to incorporate processes of tree death into models. Berner et al (2017) have advanced an empirical modeling approach to tree-mortality at regional scale. The authors used all potential data sources including maps, remote sensing images, to estimate the magnitude and relative contribution to mean annual tree mortality from fires, bark beetles, and timber harvest from 2003 to 2012, both regionally and among the 11 western US states. They quantified annual tree mortality from fires and bark beetles across regions using remote sensing estimates of tree aboveground biomass (AGB) together with information on the carbon content of AGB. Also included were disturbance extent and severity, and mean annual tree mortality from timber harvest for each state using harvest statistics from the US Forest Service. They integrated these different data sources by machine learning algorithms and used Monte Carlo analyses to track uncertainty associated with parameter error and temporal variability. Their results provide an empirical path to estimate annual tree mortality at regional scales in this big-data era.

Ecoclimatic teleconnection and air pollution effect
Tree-mortality will substantially change land cover properties and the functions of water, carbon, and energy exchanges with the atmosphere locally or regionally. The influence of land-cover change induced by tree-mortality will be transported to anywhere on earth's surface through atmospheric circulations. The question remains: will forest loss have climate impacts sufficient to affect ecosystem functioning elsewhere? Swann et al (2017) used the Community Land Model to simulate forest loss with C-3 grass replacement, and using the NCAR Community Earth System Model to simulate the long-distance climatic teleconnections of regional forest loss, they conducted experiments within 13 ecoregions, each of which included one of the domains of the US National Ecological Observatory Network (NEON). They found that for the US as a whole, loss of trees in the Pacific Southwest region, an area undergoing rapid forest die-off, had the largest negative remote impact on US GPP; in contrast, the loss of trees in the Mid-Atlantic region had the largest positive impact. They provided strong model-based evidence that forest loss in one region can alter climate far away as a result of significant ecoclimatic teleconnections to other regions. As macrosystems biology develops to address continental-scale ecology, future research should consider long-distance ecoclimate teleconnections from broad-scale tree mortality and associated climate feedback mechanisms.
While we know that heat waves, drought, fire, bark beetles and pathogens cause tree-mortality and forest decline, air pollution may be another factor. Foliar accumulation of hygroscopic aerosols can cause leaf wax degradation and hence affect stomatal conductance (g min ). Can aerosol deposition on tree leaves cause forest decline? To tackle this question, Burkhardt et al (2018) used laboratory experiments to identify the impact of hygroscopic aerosol on plant's function step-by-step. One greenhouse was ventilated with ambient air (AA) and the other with filtered air (FA). Sampled particle numbers and epidermal minimum conductance (g min ) for three species of treeseedlings were determined. They observed lower g min with FA plants than with AA plants. They concluded that aerosol pollution would cause reduced drought tolerance by making stomata leaky. Deliquescent aerosols make waxes appear 'degraded' . The observation of substantial biological effects under these moderate levels of aerosol pollution suggests that potentially stronger effects and greater biological risk may be observed in areas of high concentrations and deposition rates of hygroscopic aerosol.

Concluding remarks
Tree-mortality science has been a subject of inquiry for many decades, and although interdisciplinary research on this topic has intensified markedly in the past 20 years, quantitative understanding of causes, patterns, and mechanisms of tree mortality remains insufficient to securely project the fate of Earth's forests this century in response to diverse global change stressors (Allen et al 2015, Hartmann et al 2018, McDowell et al 2020. Still, with the development of long-term forest monitoring observations from both remote-sensing (Hansen et al 2013) and ground-based plots (Crowther et al 2015) around the world-and the systematic aggregation of data from these plots into 'big-data' through national and global databases, combined with advances in computational hardware, software, and process-modeling-our ability to document, analyze, and realistically project the complexities and dynamics of forest ecosystems at broad spatial scales is rapidly improving. Authors in this collection have developed and implemented diverse methodologies to observe and diagnose how warming climate causes trees to sicken and die or recover, ranging from laboratory manipulations of seedlings to field studies of mature trees in large plots and broad-scale climate-vegetation models to remote-sensing across large regions. The 17 papers in this collection demonstrate advances in measuring and modeling tree-mortality from local to global scales. Some of these papers have taken advantage of new databases available to develop empirical models to estimate tree and forest resilience, resistance, recovery, or mortality from local to regional scales. Other articles in this special issue formulated conceptual models and frameworks through synthetic analyses of broad literature reviews that identified knowledge gaps for further exploration, and one paper used popular global models to test how tree-mortality in one region affects ecological functions in far-distant regions. Overall, these wide-ranging examples of new research and observation capabilities illustrate ongoing advances in tree mortality research, with future research directions identified by our authors outlined in summary table 1. Despite these examples of progress, ongoing global climate change proceeds apace, and clearly large challenges remain in monitoring, measuring, and modeling tree mortality on Earthfurther addressing these research challenges remains an urgent need and opportunity for Earth System science.