A first assessment of the impact of the extreme 2018 summer drought on Central European forests

In 2018, Central Europe experienced one of the most severe and long-lasting summer drought and heat wave ever recorded. Before 2018, the 2003 millennial drought was often invoked as the example of a “hotter drought”, and was classified as the most severe event in Europe for the last 500 years. First insights now confirm that the 2018 drought event was climatically more extreme and had a greater impact on forest ecosystems of Austria, Germany and Switzerland than the 2003 drought. Across this region, mean growing season air temperature from April to October was more than 3.3°C above the long-term average, and 1.2°C warmer than in 2003. Here, we present a first impact assessment of the severe 2018 summer drought and heatwave on Central European forests. In response to the 2018 event, most ecologically and economically important tree species in temperate forests of Austria, Germany and Switzerland showed severe signs of drought stress. These symptoms included exceptionally low foliar water potentials crossing the threshold for xylem hydraulic failure in many species and observations of widespread leaf discoloration and premature leaf shedding. As a result of the extreme drought stress, the 2018 event caused unprecedented drought-induced tree mortality in many species throughout the region. Moreover, unexpectedly strong drought-legacy effects were detected in 2019. This implies that the physiological recovery of trees was impaired after the 2018 drought event, leaving them highly vulnerable to secondary drought impacts such as insect or fungal pathogen attacks. As a consequence, mortality of trees triggered by the 2018 events is likely to continue for several years. Our assessment indicates that many common temperate European forest tree species are more vulnerable to extreme summer drought and heat waves than previously thought. As drought and heat events are likely to occur more frequently with the progression of climate change, temperate European forests might approach the point for a substantial ecological and economic transition. Our assessment also highlights the urgent need for a pan-European ground-based monitoring network suited to track individual tree mortality, supported by remote sensing products with high spatial and temporal resolution to track, analyse and forecast these transitions. © 2020 The Author(s). Published by Elsevier GmbH on behalf of Gesellschaft für Ökologie. This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/)

The severity of the 2018 drought event from a climatic perspective Since the beginning of instrumental climate recording, Central Europe has experienced several extreme hot and dry summers (Sch€ ar et al., 2004;Orth, Vogel, Luterbacher, Pfister, & Seneviratne, 2016;Hanel et al., 2018), but the climatic conditions during the 2018 growing season were exceptional. Across Austria, Germany and Switzerland (named DACH region hereafter), the 2018 mean growing season air temperature (MGT; April to October) was the highest temperature ever recorded for that period (Fig. 1). It was 3.3°C higher than the long-term average from 1961 to 1990, and even 1.2°C higher than MGT of the record-breaking 2003 summer (Rebetez et al., 2006). The 2018 growing season is thus the latest peak in a trend of rapidly rising temperatures in the DACH region that commenced in the 1980s, with 10 out of the 12 hottest growing seasons in the past 120 years occurring since the year 2000 ( Fig. 1).
Mean growing season precipitation (MGP; April to October) for the DACH region in 2018 was as low as during previous record drought events in 1911in , 1947in , 1976. Although climate models predict a decline in MGP (e.g. Fischer, Sedl a cek, Hawkins, & Knutti, 2014), observational data cannot confirm this trend yet. However, the combination of increasing MGT with overall stable MGP patterns has resulted in a consistent decline of the climatic water balance (CWB; precipitation À potential evapotranspiration) from April to October in Central Europe since the 1960s (Fig. 1). In addition, the 2018 growing season had the highest mean vapour pressure deficit (MVPD; April to October) and second most negative CWB ever recorded in the DACH region, which was even more negative than in 2003 (-238 mm versus -200 mm, respectively). The regional hydroclimate pattern of the 2018 growing season is consistent with a European-wide assessment of the 2018 drought event (Buras, Rammig, & Zang, 2020).
The record-breaking drought of the 2018 growing season from April to October confirms projections from climate models that the future climate in Central Europe will be hotter, drier and will encompass more extreme climatic events Trenberth et al., 2014;Zhou, Zhang, Park Williams, & Gentine, 2019;Zscheischler & Seneviratne, 2017). Prior to the 2018 drought, the 2003 summer was classified as the most severe drought event in Europe for the last 500 years according to historical archives (Camenisch et al., 2020;Luterbacher, Dietrich, Xoplaki, Grosjean, & Wanner, 2004;Orth, Vogel, Luterbacher, Pfister, & Seneviratne, 2016), exceeding the droughts of the years 1947 and 1976 in severity (Raspe, Grimmeisen, & Schulze, 2004). Since then, , and climatic water balance (CWB; d), i.e. precipitation minus potential evapotranspiration, for the DACH region in Central Europe for the period from April to October. Horizontal grey lines demarcate the corresponding value of 2018 for comparison with other record drought years in the DACH-region (1911DACH-region ( , 1947DACH-region ( , 1976DACH-region ( , 2003, which are highlighted by dots in the corresponding colour. Data-source is the CRU version 4.03 gridded data-sets for minimum, mean, and maximum monthly air temperature, monthly precipitation sum, and monthly vapour pressure (Harris, Jones, Osborn, & Lister, 2014). To compute VPD, saturation vapour pressure was derived from monthly mean temperature using the approximation by Murray (1967), from which monthly vapour pressure was subtracted. For CWB, we computed PET by supplying minimum and maximum monthly temperature and precipitation to the Hargreaves equation (Hargreaves, 1994), and subtracting PET from monthly precipitation sum. several additional drought events have occurred in the region. For example, 2011 had the third driest spring in the preceding 150 years in Switzerland (Meteo Schweiz, 2012), and in 2015 parts of the DACH region experienced the most severe drought event since 2003 (Hanel et al., 2018;Ionita et al., 2017;Orth, Zscheischler, & Seneviratne, 2016). Likewise, parts of Northern and Southern Europe were strongly affected by local droughts in (García-Herrera et al., 2019. Thus, the 2018 drought was apparently not an isolated event but was consistent with a trend towards a different climate in the DACH region that includes a higher frequency of droughts at high temperatures. Importantly, "hotter droughts" (Allen, Breshears, & McDowell, 2015) might have more severe impacts on the terrestrial vegetation than "normal" droughts (i.e. droughts that occur at lower more typical temperatures). Since such hot droughts have previously not been documented in Central Europe, delineating the consequences of this climatic trend for the composition and functioning of Central European forest ecosystems will be crucial in the very near future (cf. Buras & Menzel, 2019).
Large-scale assessment of the 2018 drought impact on forest health A prominent stress response of trees during the 2018 drought event was the widespread early discoloration and premature leaf senescence of dominant deciduous tree species, in particular European beech (Fagus sylvatica), which occurred at various sites in Central Europe across the DACH region and began as early as late July (Fig. 2). Aggravating this situation, coniferous tree species, primarily Norway spruce (Picea abies) and Scots pine (Pinus sylvestris), also showed signs of needle discoloration in the following months. Although various spatially representative ground-based monitoring networks for the assessment of tree vitality are available across the DACH region, they are not suited to quantify drought-induced tree mortality at larger scales. Thus, the 2018 drought impacts on trees and forests mentioned here combine personal observations, reports of forest research institutes (e.g., ForstBW, 2018;LF-RP, 2018;NW-FVA, 2019) and the German report on forest conditions by the German Federal Ministry of Food and Agriculture (BMEL, 2019a). The BMEL report incorporates the pan-European International Cooperative Programme on assessment and monitoring of air pollution effects on Forests (ICP Forests) Level I plot network. The assessment presents an annual overview of crown conditions from repeatedly surveyed tree individuals since 1984 based on a 16 £ 16 km grid (Michel, Seidling, & Prescher, 2018), and it provides a field-based evidence for the severe impact of the 2018 drought on forest health.
As field-based monitoring activities only produce temporally and spatially limited snapshots of forest stands or landscapes, remote sensing can critically complement fieldbased observations by assessments at higher spatial scale (Hartmann et al., 2018a;McDowell et al., 2015). Since physiological drought symptoms also change the light reflectance of leaves in affected trees, several remote sensing approaches are suited to assess the impact of drought events on forest health (Anderegg, Anderegg, & Huang, 2019;Anyamba & Tucker, 2012;Buras, Rammig, & Zang, 2020;Damm et al., 2018;Deshayes et al., 2006;Peng, Wu, Zhang, Wang, & Gonsamo, 2019;West, Quinn, & Horswell, 2019). Here, we employ satellite-derived Normalized Difference Vegetation Index (NDVI) data obtained from the long-term MODerate resolution Imaging Spectrometer (MODIS) dataset to assess the reduction in vegetation greenness of forests during the 2018 drought across the DACH region. The NDVI, originally developed by Rouse, Haas, Schell,   , & Harlan (1974) and first applied to drought monitoring by Tucker & Choudhury (1987), is a bulk information source representing vegetation greenness (i.e. leaf area combined with leaf chlorophyll content) as a function of spectrally contrasting red and near-infrared reflectance. Although the spatial resolution of MODIS is likewise limited, this mission has been ongoing since 2000 and thus allows a direct comparison between the 2003 and 2018 drought event. We computed NDVI quantiles over the period from 2000 to 2018 in order to rank the vegetation greenness relatively among all years (Buras, Rammig, & Zang, 2020;Orth, Vogel, Luterbacher, Pfister, & Seneviratne, 2016). This approach represents a modified version of the standardized vegetation index, which has been used earlier for monitoring drought stress on vegetation (Anyamba & Tucker, 2012).
NDVI quantiles across the DACH region clearly demonstrate the severe and widespread impact of the 2018 event on forest health and are in agreement with ground-based reports on early discoloration and premature leaf-shedding (Fig. 3). Under drought, lowest quantiles and thus lowest values of greenness in mid-August are the result of early leaf discoloration, premature leaf senescence, and may even indicate partial or complete canopy dieback. Importantly, observed symptoms were more severe in 2018 than during the 2003 drought event. In fact, a detailed analysis of the NDVI data stratified into coniferous and deciduous forests empirically confirmed these observations (Fig. 4). In 2018, the area of deciduous trees that could be identified as experiencing the lowest NDVI quantiles and thus greenness was twice as large as in 2003, which is roughly 11,200 km 2 vs. 5600 km 2 , respectively. In addition, low NDVI quantiles were spatially more homogeneously distributed across the DACH region in 2018 compared to 2003. This satellite-based assessment of drought impact illustrates the exceptionally strong effect of the 2018 drought event on forests in the DACH region and corroborates a recent European-wide assessment (Buras, Rammig, & Zang, 2020).
The economic consequences of drought-induced tree mortality were severe for the forestry sector and have been estimated to amount to a loss of 3.5 billion Euros for the year 2018 in Germany (press release by the Association of German Foresters; BDF, 2019). These losses are due to the accumulation of meanwhile 160 million cubic meters of dead wood (BMEL, 2020). Aggravating this situation, millions of tree saplings planted over the past years in Germany have died due to the 2018 drought (BMEL, 2019b). The immediate and long-term consequences of the 2018 drought event on the forestry sector are substantial and are likely to increase under the predicted higher frequency of such "hotter droughts" in the near future.
Importantly, a comprehensive assessment of the extent of drought-induced mortality in forests of the DACH region is difficult because existing ground-based monitoring networks across Europe are largely not suited to consistently detect drought-induced damages at the level of individual trees over continuous areas due to their grid-based sample design. Data from existing networks therefore need to be complemented by long-term monitoring and inventory data that allow partitioning of drought-induced tree mortality and mortality rates depending on site conditions, species, tree ages or sizes, and other important parameters (cf. www.tree-mortality.net). The quantitative assessment of drought-induced tree mortality at the level of individual trees would yet be critical to detect mechanisms that influence tree survival or mortality at different temporal scales, and for characterizing the impact of forest management, sites and species on tree mortality during events such as the 2018 drought. Because such detailed assessments are highly labour demanding, ground-based plot networks need to be supported by remote sensing products with high spatial and temporal resolution (Buras, Rammig, & Zang, 2020;Damm et al., 2018;Hartmann et al., 2018a;Stovall, Shugart, & Yang, 2019). The recently released forest condition monitor (www.waldzustandsmonitor.de/en) may be considered a first step in the direction of a near-real time monitoring of the state of European forests to identify potential early-warning indicators of tree mortality (cf. Rogers et al., 2018).

Potential causes of drought-induced tree mortality
Global change-type drought-induced tree mortality is a complex syndrome with several interdependent processes involved (McDowell et al., 2011;Meir, Mencuccini, & Dewar, 2015). Various factors, both biotic and abiotic, might ultimately and species-specifically contribute to the death of an individual tree (Waring, 1986;Anderegg et al., 2015; Wang, Peng, Kneeshaw, Larocque, & Luo, 2012). Trees often have a legacy of reduced vigour triggered by past events evoked up to several decades in the past, increasing the individual mortality risk (Buras et al., 2018;Cailleret et al., 2017;Hauck, Leuschner, & Homeier, 2019;Timofeeva et al., 2017;Trugman et al., 2018). According to our current understanding, two tightly inter-related physiological mechanisms are involved in drought-induced mortality that may also predispose trees to subsequent pest attacks (Hartmann et al., 2018b;McDowell et al., 2008). These are assumed to be hydraulic failure through tissue dehydration and partial or complete loss of xylem functionality, and carbon starvation due to drought impacts on photosynthesis and mobilization of non-structural carbohydrates (Sevanto et al., 2014;Adams et al., 2017). In conifers, the lethal level of hydraulic failure is thought to correspond to the xylem pressure at 50% loss of hydraulic conductance, while in angiosperms the water potential at 88% loss of hydraulic conductance is believed to represent the point of no return (Barigah et al., 2013;Brodribb & Cochard, 2009;Brodribb, Bowman, Nichols, Delzon, & Burlett, 2010;Urli et al., 2013). Given a tree's demand in conduit capacity under severe drought drops below a few percent after stomatal closure, an alternative explanation is the interruption of the soil-root capillary continuum, with tissue dehydration and xylem embolism as the consequences (K€ orner, 2019). The fatal consequences of embolized xylem conduits only come into play when soil moisture returns and the conduits remain blocked.

Physiological integrity of trees during the 2018 drought event
It is likely that the exceptional climatic conditions experienced in 2018 in the DACH region directly impaired the physiological integrity of trees (Adams et al., 2009;Leuzinger, Bigler, Wolf, K€ orner, 2009;Williams et al., 2013). During the 2018 heat wave, leaf temperatures above 40°C were measured in four urban broad-leaved tree species in W€ urzburg, Germany (B€ oll et al., unpublished) and up to 40°C in four forest broad-leaved tree species at the Swiss Canopy Crane II site in H€ olstein, Switzerland (Hoch et al., unpublished). This might have resulted in substantial residual water loss as g min increases dramatically at such high leaf temperatures as mentioned above.
To assess the water status of trees during the 2018 drought event, midday leaf water potentials (P min ) were assessed for 20 coniferous and broad-leaved tree species at several sites in the DACH region (Table 1). For many of the investigated tree species, observed P min were in the range of the most negative values ever reported under field conditions. In eight of the twenty species studied, P min values indicate that the air entry point of xylem embolism formation had been reached by mid-August 2018. These data confirm that during the 2018 drought event many of the observed trees in the DACH region must have reached their physiological limits, beyond which fatal tissue dehydration takes place, accompanied by a loss of the hydraulic integrity of the trees (Choat et al., 2018;Delzon & Cochard, 2014;Tyree & Sperry, 1989). Direct measurements of xylem hydraulic conductivity loss in sun-exposed branches in early September at the Swiss Canopy Crane II site in H€ olstein, Switzerland, showed that European beech trees at this site were embolized on average by 56.7 § 3.2% (mean § SE; n = 32 branches, 12 trees). Most of these trees showed partial to complete canopy dieback but no obvious signs of mortality. Yet, mortality of European beech trees was reported from other locations in the DACH region as mentioned above. In contrast to European beech, some Norway spruce trees showed complete loss of hydraulic conductivity in upper-canopy branches at the Swiss Canopy Crane II site; these individuals all died off in the following weeks. Our data suggest that drought-induced hydraulic failure is a strong indicator of severe damage observed in temperate forests as a consequence of the 2018 drought event. Because the degree of drought-induced embolism formation was not explored in temperate tree species for the drought events in 2003 or 2015 (e.g. Breda The widely observed premature leaf senescence in July and August 2018, particularly for European beech, could have been a response to reduce leaf area in the attempt to protect the hydraulic system from severe and lethal dehydration. This phenomenon is indeed well described for temperate broad-leaved tree species (Breda, Huc, Granier, & Dreyer, 2006;Marchin, Zeng, Hoffmann, 2010;Munne-Bosch & Alegre, 2004). Early leaf senescence might either be induced by hydraulic failure of the petiole (Hochberg et al., 2017;Wolfe, Sperry, & Kursar, 2016), or by metabolic dysfunctions in foliar cells as a result of high tissue temperatures or photo-oxidative stress (Juvany, M€ uller, & Munn e-Bosch, 2013;Lange, 1959;MartinezÀVilalta, Anderegg, Sapes, & Sala, 2019;Oliver & Bewley, 1997;Pflug et al., 2018). Investigations at the Swiss Canopy Crane II site in H€ olstein, Switzerland, revealed that discoloured European beech leaves showing signs of early senescence in early August had foliar nitrogen (N) contents of 1.1 § 0.1% (n = 10 trees), which is similar to the N concentration of Table 1. Minimum leaf water potential (P min , MPa) measured in mid-August 2018, xylem pressure at 12% loss of hydraulic conductance (P 12 , MPa) and the P 12 hydraulic safety margin (HSM P min -P 12 ) for 18-20 temperate tree species at four sites in Germany (from North to South: Unterl€ uß, G€ ottingen, Kranzberg, Freiburg) and one site in Switzerland (H€ olstein). A negative HSM P min -P 12 (in bold) indicates that the air-entry point of embolism formation has been crossed. At all sites, leaf samples were collected from the uppermost canopy of mature trees with the exception of the site in Freiburg, which was only established in 2011. For the P min measurements, one to eight leaves were measured and averaged per individual tree. The xylem vulnerability curves were constructed according to common protocols with the flow centrifuge technique (Delzon, Douthe, Sala, & Choat 2010;Schuldt et al., 2016); missing data were taken from the literature except for the two ring-porous species F. excelsior and Q. robur that were excluded ( 1) Choat et al., 2012;2) L€ ubbe et al. unpublished; 3) Dietrich, Delzon, Hoch, Kahmen, 2019;4) Tomasella et al., 2018;5) Schuldt et al. unpublished). Values are mean § SE. Further given are sampling date for P min measurements, average tree height per species and site (H), and number of trees (n) and branches (in parentheses) for P min and P 12 measurements.

Species
Location Date H (m) P min (MPa) n P min P 12 (MPa) n P 12 HSM P min -P 12 leaf-litter collected in autumn 2017 (1.0 § 0.1%, n = 10 litter traps). This indicates that early senescence in European beech leaves most likely resulted from a controlled physiological process, initiated to reduce the risk of lethal tissue dehydration (cf. Breda, Huc, Granier, & Dreyer, 2006;Marchin, Zeng, & Hoffmann, 2010;Ruehr, Gast, Weber, Daub, & Arneth, 2016). Despite the protective measure of early senescence, we observed that trees -in particular European beech -suffered from substantial loss of hydraulic conductivity and massive canopy dieback ( Fig. 2; Fig. 6). Either premature leaf senescence occurred too late, or it did not guarantee sufficient protection from lethal tissue dehydration caused by residual water loss through the bark (e.g. Oren & Pataki, 2001).

Abies alba
To assess the recovery potential of canopy dieback observed in European beech canopies, 795 trees presenting signs of early leaf coloration during August and September 2018 were monitored and compared with 138 'healthy' nearby control trees displaying regular leaf senescence at three sites near Basel, Zurich and Schaffhausen in Switzerland. A survey in April 2019 revealed that leaf flushing of individuals, especially from stands with good soil water holding capacity, was often unaffected despite extraordinary dry winter and spring climatic conditions. However, one fourth of the drought-affected trees showed partial crown dieback in 2019, and in one third of these trees more than 50% of the tree crown was dead in April 2019 (Table 2). In contrast, almost all of the less affected control trees had an intact crown at that time. Preliminary data confirm that the water transport capacity of branches with legacy effects had not been restored in the following spring 2019 (Fig. 6). Extremely low precipitation rates during winter and spring might have aggravated this situation. Further investigations will reveal whether the recovery from xylem embolism is Fig. 6. Visualisation of the conductive xylem (a, b), the leaf-to-sapwood area ratio (A l :A s , c) and percentage loss of hydraulic conductance (PLC, d) of healthy and damaged sun-exposed upper-canopy branches of mature F. sylvatica trees at the Swiss Canopy Crane II site in H€ olstein, Switzerland. For the visualization of the conductive xylem area, branches were collected in spring 2019 and subsequently placed in a 2% methylene blue solution for 24 h in the greenhouse allowing for transpiration-driven water uptake of the staining solution. For determining the supported leaf area per cross-section and the native degree of embolism, branches were collected from 5 healthy and 6 damaged trees in early summer 2019; the percentage loss of conductance (PLC) was measured according to standard protocols (Torres-Ruiz et al., 2015). Both A l :A s (x 2 =7.67, df=1, P<0.01) and PLC (x 2 =7.50, df=1, P<0.01) differed significantly between healthy and damaged branches (KruskalÀWallis test). possible in the season after a drought, or if it requires the formation of new woody tissue to fully restore hydraulic functionality (McDowell, Brodribb, & Nardini, 2019;Ruehr, Grote, Mayr, & Arneth, 2019).

Expected mid-and long-term effects
Trees suffering from drought-induced mortality often have a legacy of reduced vigour (cf. Cailleret et al., 2017;Kannenberg et al., 2019). In addition to the immediate climatic conditions experienced in 2018, preceding drought events could thus have triggered the high mortality rates observed. Long-term effects of the 2018 drought might well extend beyond the currently observed damages, especially as the summer in 2019 was again particularly dry and hot in many areas across the DACH region, aggravating stress on already weakened trees. Impaired carbon dynamics can limit regrowth of new woody tissue and may be responsible for patterns of delayed tree mortality after drought exposure .
Because the defence metabolism can be impaired after a drought event, trees are vulnerable to insect or pathogen infestation in the following growing season (Bigler, Br€ aker, Bugmann, Dobbertin, & Rigling, 2006;Huang et al., 2020). Consequently, drought is reportedly the major cause of massive bark beetle outbreak in the actual and the following years (Christiansen, Waring, & Berryman, 1987;Hart, Veblen, Eisenhart, Jarvis, & Kulakowski, 2014). In Norway spruce, resin exudation and thus defence capability declines with a reduction in relative tissue water content (Netherer et al., 2015). This makes this economically important tree species highly susceptible to successive infestations by Eurasian spruce bark beetle (Ips typographus) during and after drought-exposure, especially at lower elevations (Biedermann et al., 2019;Hentschel et al., 2014;Kazda & Pichler, 1998;L evesque et al., 2013;Seidl, Schelhaas, & Lexer, 2011;Seidling, 2007;Vitali, B€ untgen, & Bauhus, 2017). Aggravating this situation, warmer winters are expected to increase bark beetle annual population levels and distribution (Goodsman et al., 2018). As a consequence, increased Norway spruce mortality will most likely continue in the coming years, especially at lower or mid-range elevations outside its natural range. In 2019, for example, the level of bark beetle-infested wood has for the first time reached over one million cubic metres since 2005 in Switzerland (Stroheker, Forster, & Queloz, 2020).
Weakened and damaged European beech trees, on the other hand, are vulnerable to pathogen attacks by microfungal infections, a phenomenon well-described for this species after severe drought (Jung, 2009;Lonsdale, 1980;Schindler, 1951), as well as to insect attacks by beech splendour beetle (Agrilus viridis; G€ oßwein & Lemme, 2017;von Vietinghoff-Riesch & Vit e, 1952). Although European beech would naturally cover more than 60% of the area of Germany (Bohn et al., 2003), the impact of the 2018 drought event suggests that a future hotter and drier climate could severely affect the vitality of European beech, particularly at lower elevations and even in the centre of its distribution (cf. Buras & Menzel, 2019;Knutzen, Dulamsuren, Meier, & Leuschner, 2017;Zimmermann, Hauck, Dulamsuren, & Leuschner, 2015).

Conclusion and outlook
The 2018 drought and heatwave has set a new benchmark for future drought events, and has marked a change in Central European forest dynamics. Current reports on droughtinduced tree mortality are reminiscent of the internationally acknowledged and historical discussions on the European beech decline ('Buchensterben') of the late 1940s and forest decline ('Waldsterben') of the early 1980s. In the late 1940s, several exceptionally dry years resulted in large-scale die-off of European beech that was explained by repeated extreme events of drought and late frost (Br€ uck-Dyckhoff, Petercord, & Schopf, 2019;Schindler, 1951). In the 1980s, the apparent widespread forest dieback across large areas of Central and Eastern Europe was associated with air pollution, both dry and wet deposition by acid rain, sulphur oxide, heavy metals, oxidized and reduced N compounds and high ozone levels (Schulze, 1989; Sch€ utt & Cowling, Table 2. Legacy effects for Fagus sylvatica at three sites near Basel, Zurich and Schaffhausen in Switzerland. In 2018, a total of 933 tree individuals were selected. Of these, 138 trees were categorized as healthy, while 795 showed signs of premature leaf senescence or partial canopy dieback. In April 2019, the same individuals were surveyed once more and the bark and crown condition documented. Given are tree individuals from both groups (healthy and damaged trees, in numbers or percentage) that show signs of bark damage, bark exudation, and/or crown mortality between 20% and 100%. 1985; Ulrich, 1990). Remarkably, the symptoms of dieback became epidemic after the centennial 1976 drought. In the following years, air pollution was reduced by desulfurizing emissions from industrial installations and the mandatory introduction of automobile catalysers. Because current damages to trees are also ubiquitous, media and stakeholders have meanwhile introduced the term 'Waldsterben 2.0'. We do not recommend using such terminology because today and back in the 1980s trees were dying, not forests. Today, selected species may be massively affected by climate change such as Norway spruce, which has extensively been planted outside its natural range. However, tree species showing very high mortality rates will be replaced by others over time. As a consequence, we will most likely experience tremendous forest changes but not a complete dieback. Such a conversion will nonetheless be harsh for the forestry sector and for the general population as the forests we have appreciated for generations will change, at least in given areas. The high mortality rates observed in 2018 illustrate that climate change is acting on the structure and composition of forests in the DACH region, even if quantification of the direct and indirect effects remains challenging. At the stand level, species-specific differences in drought sensitivity might further alter tree species dominance and species composition (Cavin, Mountford, Peterken, & Jump, 2013;Weber, Bugmann, Fonti, & Rigling, 2008), especially if it likewise affects tree regeneration (Wohlgemuth, Gallien, & Zimmermann, 2016). In this context, it is important to consider how the resilience of forests to disturbance in Central Europe can be enhanced by management practices. Changes in forestry practices initiated by singular or repeated drought events might have even stronger impacts on the future composition of Central European forests than the drought events themselves. The likelihood that such shifts in forest policy will be implemented also highlights the need for intensified research on the consequences of drought events on individual tree species, different site conditions and forest ecosystems. For instance, various sources advocate a paradigm shift from monocultures towards mixed forest stands in order to compensate for the potential loss of individual species. Although tree diversity is not systematically related to increased drought resistance of forests (Forrester et al., 2016;Grossiord, 2019), mixing appropriate species to increase the likelihood of complementary effects might indeed be able to mitigate drought impacts (Ammer, 2017;Gonzalez de Andres et al., 2018;Anderegg et al., 2018). In addition, non-native temperate tree species from Northern America, Eastern Europe or Central Asia are frequently referred to as an all-in-one solution. Given the longevity of trees, however, it remains questionable if such species should be introduced to our forests; they would not only need to be economically profitable and resistant against local pathogens but also need to be non-invasive and able to provide ecological functions including host-insect interactions.
The current controversial discussions on the suitability of Douglas fir (Pseudotsuga menziesii), a tree species imported from Northern America a century ago as supplement or substitute for Norway spruce, shows the difficulty of such suitability assessments (cf. Spiecker, Lindner, & Schuler, 2019). Nevertheless, the first evidence of the impact of the 2018 drought event on forests in the DACH region highlights the need to adapt current silvicultural measures in order to prepare Central European forests for future climatic conditions. Finally, our assessment revealed the limitation of current European ground-based measures for spatially continuously monitoring drought impacts on forest growth and mortality, which is vital for the mechanistic and ecological interpretation of the consequences of extreme climatic events. A core aim of the ICP Forests Level II plot network, for example, is the monitoring of atmospheric deposition as a response to the above mentioned air pollution in the 1980s (Michel et al. 2018). Although these plots continuously monitor relevant parameters for the assessment of drought impacts, their number is unfortunately limited. However, in order to understand how site characteristics, management practices, species identity and community composition interact in determining the drought vulnerability of trees or forest stands, physiologically meaningful stress variables are needed at high spatial and temporal resolution. Novel remote sensing-based techniques suited to assess the water status of trees as a cross-scale indicator of mortality risk are promising in this regard (cf. Damm et al., 2018;Hunt, Ustin, & Riaño, 2013;MartinezÀVilalta et al., 2019). These techniques include microwave systems (Konings, Rao, & Steel-eÀDunne, 2019;Rao, Anderegg, Sala, Martínez-Vilalta, & Konings, 2019) or optical indices based on infrared reflectance (Marusig et al., 2020;Sun et al., 2019). Although current spatial and spectral resolution often limit their applicability, future space-born remote sensing missions providing data free of charge will allow to monitor forest health at high resolution.

Acknowledgements
This assessment evolved from a meeting of the working group 'Ecosystem Research' of the Ecological Society of Germany, Austria and Switzerland, held at the University of Basel in June 2019. We thank Tanja Sanders for helpful discussions, Miriam Fischer for provision of leaf water potential data from Unterl€ uß, Silvan Pfefferkorn for constructing the xylem vulnerability curves for all trees at the site in G€ ottingen, Celine Aymon and Martin Kistler for field data on beech trees in Switzerland, Charles Nock for tree height data from Freiburg, and Roman Link for data on the native degree of embolism in spruce and beech. The authors thank one anonymous reviewer, Kris Verheyen and the editor for comments that helped improving the manuscript. Peter Hajek was funded by DFG grant number 384026712. Allan Buras, Anja Rammig and Christian Zang acknowledge funding by the Bavarian Ministry of Science and the Arts in the context of the Bavarian Climate Research Network (Bay-KliF), Mana Gharun acknowledges funding by the Swiss National Science Foundation project ICOS-CH Phase 2 (20FI20_173691), and Ansgar Kahmen financial support by the Swiss Federal Office for the Environment for the Swiss Canopy Crane II site.