Camouflaged as degraded wax: hygroscopic aerosols contribute to leaf desiccation, tree mortality, and forest decline

Some 40 years ago, air pollution caused widespread forest decline in Central Europe and eastern North America. More recently, high levels of tree mortality worldwide are thought to be driven by rising temperatures and increasing atmospheric drought. A neglected factor, possibly contributing to both phenomena, is the foliar accumulation of hygroscopic aerosols. Recent experiments with experimentally added aerosols revealed that foliar aerosol accumulation can (i) create the microscopic impression of ‘wax degradation’, considered an important proxy of forest decline associated with air pollution, though the mechanism remains unexplained; and (ii) increase epidermal minimum conductance (gmin), a measure of cuticular permeability and completeness of stomatal closure—both could lead to reduced drought tolerance. Here, those studies with applied aerosol are extended by addressing plant responses to reduction of ambient aerosol. Scots pine, silver fir, and common oak seedlings were grown for 2 years in greenhouses ventilated with ambient air (AA) or air filtered to remove nearly all aerosol particles (FA). Removal of ambient aerosol prevented the development of amorphous structures viewed in the electron microscope that have typically been interpreted as degraded waxes. Lower gmin values suggested that FA plants had better stomatal control and therefore greater drought tolerance than AA plants. The co-occurrence of apparent wax degradation and reduced drought tolerance in AA plants suggests a common cause. This may be mediated by the deliquescence and spreading of hygroscopic aerosols across the leaf surface. The liquid film produced may penetrate the stomata and facilitate unproductive stomatal transpiration. In this way, aerosol pollution may enhance the impacts of atmospheric drought, and may damage trees and forests on large spatial scales.


Introduction
Two large-scale forest diebacks have been attributed to atmospheric factors. Both have affected remote regions, in both cases biotic factors have been ruled out as primary drivers, and yet these dieback events appear to be independent. Starting from the 1970s, forest decline by air pollution (FDAP) affected trees in Europe and eastern North America, which was linked to industrial and combustion emissions of SO 2 , NO x , and resulting acids (Johnson and Siccama 1983, McLaughlin 1985, Schütt and Cowling 1985. However, while the phenomenon was intensively studied for about 20 years, no coherent mechanism for the decline was identified (Bussotti and Ferretti 1998, Schulze 1989, Ulrich 1990). Although regulatory action improved air quality, forest health was not completely restored. Recurrent die-off events have tended to follow drought events (Klap et al 2000, Vacek et al 2015. Although drought has not been considered a major contributor to FDAP, seemingly anomalous drought-related symptoms are observed as a component of responses to pollutants. A welldocumented example is the reduced drought tolerance of silver fir (Abies alba Mill.), associated with relatively low levels of SO 2 from distant sources. Drought-like symptoms such as needle loss, reduced expansion of growth rings, and early dieback were correlated with the sulfur content of the needles-and improved when SO 2 emissions decreased (Elling et al 2009). Thus, there may be unrecognized interaction between FDAP and drought.
Tree mortality by hotter drought (TMHD), the second attributed cause of large-scale dieback, has been discussed since the beginning of this century, in arid and semi-arid regions of western North America and worldwide (Allen et al 2015, Bigler et al 2006, Breshears et al 2005 al 2018, van Mantgem et al 2009, Patrut et al 2018. It is a manifestation of global warming, that exposes plants to linear increases in temperature but exponential increases in atmospheric drought (evaporative demand; vapor pressure deficit, VPD) (Novick et al 2016, Williams et al 2013. This accelerates transpiration and challenges plant hydraulic systems. Hydraulic failure is associated with drought-induced mortality across multiple tree taxa (Adams et al 2017).
Air pollution has not generally been considered a contributing factor to TMHD. 'Carbon starvation' was suggested to increase mortality during drought by restriction of CO 2 uptake by stomatal closure (McDowell et al 2008). This appeared to explain the anomalous observation that species reacting most sensitively to drought by stomatal closure (i.e. drought resistant) were most affected by TMHD (i.e. drought sensitive), as in the classic example of the juniper-pinon pine system (McDowell et al 2008). However, the 'carbon starvation' hypothesis is not consistent with observed patterns of non-structural carbohydrate reserves associated with mortality (Adams et al 2017).
Here we suggest a novel reconciliation of these diverse observations. Both types of forest decline may be linked mechanistically by the accumulation and hygroscopic action of aerosols deposited on leaves. The global increase of atmospheric aerosol concentrations is an indicator of the Anthropocene (Andreae 2007, Steffen et al 2015. Aerosols are ubiquitous and increasingly hygroscopic, and are readily deposited on trees (Burkhardt and Grantz 2017). In situ measurements of electrical leaf surface conductance (Burkhardt and Eiden 1994) and direct observation (Burkhardt et al 2001, Grantz et al 2018 indicated the existence of thin water films on needles and leaves caused by deliquescence (absorption of water by hygroscopic aerosol). These solutions overcome cuticular hydrophobicity and creep into individual stomata (Burkhardt et al 2012, Eichert et al 2008, Huang and Huang 1976, van Enckevort and Los 2013. By this 'hydraulic activation of stomata' (HAS; Burkhardt 2010), a thin liquid film (< 100 nm) can form a connection between the leaf surface and the leaf apoplast that normally forms the end of the plant hydraulic system (figure 1). We hypothesize that by the establishment of this liquid connection, the plant hydraulic system is extended to the leaf surface, enabling bidirectional liquid water transport through the stomatal pores, according to water potential gradients. In daytime, HAS may generate additional transpiration with water evaporating from the outer leaf surface as well as through the stomatal pores (figure 1). Partial stomatal closure has no effect on evaporation from the leaf surface, and little impact on transport of liquid water through the pores. This may reduce stomatal control of water loss (Burkhardt 2010) and contribute to hydraulic failure in TMHD. Furthermore, this deposition may lead to the appearance of 'degraded wax' that has been considered diagnostic of FDAP (Burkhardt and Pariyar 2014, Trimbacher and Weiss 1999, Turunen and Huttunen 1990.
These effects of aerosol deposition have been demonstrated experimentally by application of aerosol to diverse plant species (Burkhardt et al 2001, Burkhardt and Pariyar 2014, 2016. Similarly, exclusion of ambient aerosol reduced the epidermal minimum conductance (g min ) of short-lived faba bean leaves, and reduced water flux per directly measured stomatal aperture (Burkhardt and Pariyar 2014, 2016, Grantz et al 2018. Here we test these concepts using the exclusion of ambient aerosol from long-lived tree needles.

Plants and greenhouses
Potted individuals of Scots pine (Pinus sylvestris L.), Silver fir (Abies alba Mill.), and Common Oak (Quercus robur L.) seedlings were used for the study which ran from 2015 to 2017. In 2015, pine seedlings were 6 years and 110 cm tall, fir 4 years and 30 cm tall, and oak 3 years and 70 cm tall. Oak and fir were planted in 15 cm (dia.) × 20 cm (ht.) pots. Pine was planted in 25 cm × 33 cm pots. All pots contained potting mix (Profi Substrat; Einheitserde, Typ Topf). From March to October, plants were fertilized every other week with complete nutrient solution including micronutrients. Plants were irrigated as needed. Plants were randomly assigned to one of two adjacent research greenhouses at the University of Bonn, Germany (figure S1 available at stacks.iop.org/ERL/13/085001/mmedia Figure 1. Hydraulic activation of stomata (HAS): liquid water forms by condensation to hygroscopic particles on the leaf surface, from where it creeps into the substomatal cavity (blue line; thickness probably < 100 nm) and connects with apoplastic liquid water (green line). In this way, the plant's hydraulic system is extended to the leaf surface. Liquid water reaches the leaf surface from where it evaporates (solid arrow), mostly independent of stomatal aperture. Dashed arrows, gas exchange. C a and C i : external and internal CO 2 concentrations. for photo). Measurements were obtained discontinuously throughout 2 years, only using leaves and needles that had developed within the greenhouses.
One greenhouse was ventilated with ambient air (AA) and the other with filtered air (FA) from which nearly all particles were excluded, both with 2 air exchanges min −1 , as described previously (Pariyar et al 2013, Burkhardt and Pariyar 2016, Grantz et al 2018. Filtration of FA to HEPA (high efficiency particulate air) standards was achieved with a cloth bag followed by high efficiency filter pad (No. 13; ACS; Essen, Germany). Within the greenhouses, total aerosol number concentrations (> 10 nm) were measured by a cloud chamber condensation nuclei counter (TSI 3783; TSI, Shoreview, MN, USA). Total aerosol number concentration in AA was 6-7 × 10 9 particles m −3 . In FA, by opening the door which briefly admitted ambient air, and by movement of personnel within the FA greenhouse, a few particles were admitted or resuspended from the floor and bench. Nevertheless particle number concentration was reduced to 5−10 × 10 6 m −3 , i.e. reduced by > 98% as previously reported (Grantz et al 2018).
Elemental aerosol and trace gas concentrations in the greenhouses were collected throughout the entire year 2015 with monthly resolution by CEH DELTA low-volume denuder sampling systems (Tang et al 2009). Tubing and filters were removed monthly and sent to the Croatian Meteorological Service, Zagreb, Croatia for chemical analysis. The mean mass concentration of all ionic aerosol components was 4.90 g m −3 in the AA greenhouse (table 1, figure S2), a reduction to 13.7%. Filtration of aerosol mass was less effective than of particle number (Grantz et al 2018), probably reflecting the cut-off of aerodynamic diameter of the Delta filter (4−5 m), thus excluding coarse particles with heavy mass, and the unknown rates of volatilization of NH 4 NO 3 and NH 4 Cl aerosols, and resuspension of coarse particles as mentioned above. The three most abundant ions were NO 3 − , contributing 45% (AA) and 43% (FA) to ionic aerosol mass, respectively; NH 4 + (21% and 29%); and SO 4 2− (24% and 10%) (table 1). The aerosol concentration in the AA greenhouse was about half of that in the the atmosphere that caused FDAP in a remote forest site in 1988 (Ludwig and Klemm 1990, (table 1). For gas phase NH 3 and SO 2 , mean AA and FA concentrations differed by less than 10% (table 2(a)); monthly AA and FA concentrations were highly correlated (r 2 > 0.9) and close to the 1:1 line (figures 2 and S3). For gas phase HNO 3 and HCl, mean AA and FA concentrations differed by less than 20% (table 2(a)). AA and FA concentrations of HNO 3 were highly correlated (r 2 = 0.84). Over the year, and particularly in summer, FA HNO 3 concentrations exceeded AA values by 19% (p = 0.02), resulting in a steep slope of the correlation (figures 2 and S3). Mean FA HCl concentrations also exceeded AA values by 19% but were less well correlated and not significantly enhanced (r 2 = 0.31; figure 2). Ozone (O 3 ) concentrations in the greenhouses were determined from April to October Table 1. Mean ionic aerosol mass concentrations, measured in the AA and FA greenhouses throughout 2015 (see table S1 for concentrations in ppb). FDAP: reported ionic aerosol mass concentrations within a remote area affected by forest decline from air pollution, measured 1988 in northeastern Bavaria, Germany (Ludwig and Klemm 1990, rural site WF, all size classes combined). Ratios for Σ FDAP are calculated without K + . Except calcium and magnesium, ions were statistically different between AA and FA, p(Na) = 0.03, others p < 0.001. Magnesium was just at the limit of detection. 2016. Hourly and annual trends were similar in the two greenhouses (Grantz et al 2018). Aerosol filtration caused a slightly warmer and drier environment in the FA compared to the AA greenhouse. Over the period of 2 years, the FA greenhouse was significantly warmer (ΔT = 0.67 • C), drier (ΔRH = 3.32%,), and with higher vapor pressure deficit (ΔVPD = 0.05 kPa), respectively (table 2(b)).

Foliar rinsing
Fir twig segments with needles that had developed in the years 2016 (year 1) and 2015 (year 2) were cut from the seedlings. The segments were sealed at the cut ends with paraffin to prevent leaching. Each twig segment was transferred to a plastic beaker containing 60 ml of deionized water and shaken for 10 s by hand. Twigs were removed and the water samples were frozen (−28 • C). Needles were separated and the projected leaf area determined photographically (ImageJ). The projected leaf area was doubled to obtain effective needle surface area. Ion concentrations were related to needle surface area. Thawed wash water was filtered (0.45 m) before chemical analysis. The NH 4 + was analyzed with a continuous flow analyzer. The anions Cl − , NO 3 − , SO 4 2− , and PO 4 3− were determined using High Performance Liquid Chromatography (HPLC; Dionex). The cations Ca, K, and Na were determined by Atomic Absorption Spectrometry. Manganese was analyzed by Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES).

Minimum epidermal conductance (g min )
The g min was determined from oak leaves in September 2015, from the youngest fir needles in November 2015, and from the youngest pine needles in November 2016. The cut ends were sealed with paraffin, and g min determined from leaf drying curves as previously described (Burkhardt andPariyar 2014, Grantz et al 2018).

Foliar carbon isotopes
Fir needle samples (1 and 2 years old) were obtained in April 2017 for determination of 13 C values by mass spectrometry, following the procedure described earlier (Burkhardt and Pariyar 2016).

Aerosol deposition on fir needles
Ionic mass on the needle surface depended both on the aerosol environment as well as on needle age (table 3; figure S4). The overall amount of ionic mass was between 1.66 g cm −2 (1 year old FA needles, FA1) and 9.01 g cm −2 (2 year old AA needles, AA 2). The overall value of FA1 was 29.9% of the AA1 value, while FA2 was 56% of AA2. On all needle classes, nitrate was the ion contributing most to the aerosol mass, contributing about half in FA1, and about one third for all other needle classes (table 3). On a molar basis, potassium was the most abundant ion on 2 year old needles ( figure S4). The strongest reduction from AA to FA for single ions were NH 4 + (6.5% of AA) and Na + (6.6%) for 1 year old needles, while for 2 year old needles, PO 4 3− (32.0%) and SO 4 2− (32.9%) were most strongly reduced. Under FA compared with AA, the smallest reduction for 1 year old needles was found for nitrate (51.5% of AA) with the smallest reduction for 2 year old needles found for potassium (82.6% of AA; table 3; figure S4). Several cations and anions increased with leaf age, likely reflecting leaching as well as deposition (Tukey 1970). Manganese concentrations were higher on needles in the AA greenhouse and strongly increased with needle age (table 3, Figure S5).

Minimum epidermal conductance (g min )
The highest g min was measured for AA oak leaves (0.475 mmol m −2 s −1 , table 4). The g min of AA fir needles was 0.388 mmol m −2 s −1 and of AA pine leaves, 0.130 mmol m −2 s −1 . The g min decreased in all cases when aerosols were excluded. FA g min was 78% of AA for oak, 81% for fir, and 77% for pine, respectively (table 4).

Foliar carbon isotopes
Carbon isotope discrimination differed significantly between AA and FA leaves of second year fir needles. AA samples had a less negative value than FA, indicating lower values of intercellular CO 2 . Carbon isotopes for first year fir needles did not differ (table 5).

Scanning electron microscopy
There were few crystalline particles > 10 m on the leaves and needles of silver fir. Some of these appeared to be the source, by deliquescence, of amorphous structures covering the tubular waxes ( figure 3(a)). These were distinct from the native tubular wax structure in some areas, but were visually mixed in other areas as the salt solutions permeated the tubular wax structures (figure 3(b)). The identification of such deliquescent salts became possible only by following the decomposition Table 3. Leaf surface ion concentrations per leaf area (mean ± SE), from silver fir needles of different age classes, determined by rinsing and subsequent chemical analysis. AA: ambient air; FA: filtered air; 1 and 2: needle age in years. Within rows, numbers followed by the same letter are not statistically different (p > 0.05). Sample size n = 10. * : manganese values are medians, given in ng cm −2 . FA1 ( g cm −2 ) AA1 ( g cm −2 ) F A 2( g cm −2 ) AA2 ( g cm −2 ) FA1/AA1 (%) FA2/AA2 (%) and spreading of the larger salt structures, illustrating the difficulties of distinguishing waxes from salts in SEM images. In fir, the comparison between AA and FA needles showed clear differences. The FA needles lacked the amorphous structures that were found on AA. At low magnification, AA needles had more contrast and crystalline particles (figure 3(c)), whereas deposits on the FA needles appeared soft and were difficult to focus (figure 3(d)). However, at higher magnification, amorphous, crust-like structures were apparent in the stomatal regions of the AA needles, sometimes together with micron-sized crystalline particles (figure 3(e)) that were not observed on FA needles (figure 3(f)). Further magnification revealed details about the penetration and coverage of tubular waxes by the crust-like material (figures 3 (g)(AA), (h)(FA)). These spreading crusts appeared inside the stomatal antechamber of the AA needles, and gave the extremely fine tubular structures in this region a coarser appearance. Some of the fine voids in this area became blocked by crusts (figures 3 (g)−(h)).
Crust-like structures could also be recognized in the stomatal antechambers of AA needles of Scots pine (figure 4(a)) that were not found on FA needles ( figure  4(b)). Less distinct differences were observed between FA leaves and AA leaves of oak (figures 4(c)−(f)). The surfaces of AA leaves (figure 4(c)) appeared more cluttered and with less contrast than FA leaves ( figure 4(d)) in the SEM. At greater magnification (squares and arrowheads in figures 4(c) and (d)), crust-like structures were observed that produced a hazy appearance of the needle-like waxes in AA (figures 4(e): AA, (f): FA) and were responsible for low contrast in the AA SEM images (figures 4(c) and (e)). There were obvious differences between those oak leaves that had been sprayed with additional submicron aerosols (figure 4(g)) and untreated FA leaves (figure 4(h)). Waxes on the sprayed leaves appeared 'eroded' (figure 4(g)) compared to FA waxes ( figure 4(h)). The rapidly developed symptoms demonstrated that the traditional interpretation of this morphology as wax degradation is not correct. The waxes had been covered by crusts of the deliquescent aerosols, giving the appearance of degraded wax.

Discussion
Scanning electron microscopy is the only technique by which 'wax degradation' has been identified and described. Air filtration prevented the formation of 'wax degradation' symptoms and was found to be associated with lower g min of FA fir, pine, and oak seedlings. In a previous study with amendment by ammonium sulfate aerosol, the change from fine tubular waxes to an amorphous surface on pine needles was observed within a few minutes, induced by a short humidity cycle causing the salt to deliquesce, expand across the surface, and cover the tubular waxes (Burkhardt and Pariyar 2014). Here, the ambient, chemically diverse aerosol produced amorphous surfaces on AA leaves over the course of months, by a similar process: (i) the attraction of water vapor by hygroscopicity; (ii) the dissolution of the hygroscopic aerosol fraction; and (iii) their resulting mobility and distribution across the leaf surface, leading to coverage of tubular waxes by amorphous crusts with the typical appearance of 'wax degradation'.
Most aerosols are hygroscopic, attracting atmospheric humidity and enhancing condensation, similar to their function as cloud-condensation nuclei in the atmosphere (Petters and Kreidenweis 2007). On leaf surfaces, stomatal transpiration is an additional source of water vapor. Few deposited aerosols appear crystalline, separate, and large enough to be clearly recognized in SEM images; many others are too small or poorly delimited to image clearly. Our experimental approach enabled the identification of deposited aerosols by the divergent appearance of AA and FA leaves. Aerosols appeared amorphous, as on pine and fir needles, or hazy as on younger oak leaves. The hazy, transparent appearance (figures 4(c) and (e)) reflected areas of moderate deposition of aerosols, while heavier deposition turned oak leaf surfaces to amorphous, opaque crusts ( figure 4(g)).
Salts, which are the most hygroscopic component of aerosols, usually deliquesce below 80% RH, doubling their volume with absorbed water (Pilinis et al 1989). This process solubilizes and mobilizes salt ions, creates concentration gradients across the leaf surface, and enables the dissolution of trace gases. Assuming homogeneous needle coverage with 5 g cm −2 ionic aerosol (table 3), the absorbed water mass at deliquescence yields a hypothetically homogeneous water film thickness of about 50 nm. Minute amounts of liquid water are sufficient for the dissolution of highly soluble trace gases such as HNO 3 and HCl. Enhanced dissolution in the AA surface solution of HNO 3 likely increased the FA/AA nitrate ratio and acted as an additional nitrate source on fir needles, while a limiting concentration of hygroscopic aerosols in the FA greenhouse may have resulted in higher atmospheric concentrations of HNO 3 and HCl.
SO 2 is only moderately soluble in water. Efficient aqueous oxidation in the atmosphere of SO 2 is generally by H 2 O 2 and O 3 (Martin 1984, Seidl 1989). However, in leaf surface water the oxidation of SO 2 by O 2 can be significant, effectively catalyzed by Mn 2+ ions present in leaf surface water due to substantial leaching from leaves (Burkhardt and Drechsel 1997,  Martin 1984, Brandt andvan Eldik 1995). Manganese is among the cations most enhanced in throughfall of forest canopies, relative to precipitation (Kreutzer et al 1998, Seufert 1990). The sulfuric acid formed on the leaf surface by SO 2 oxidation in turn increases cation leaching; leading to a positive feedback. This may contribute to the previously reported strong correlations of needle Mn content with forest decline symptoms and needle wax quality (Burkhardt and Drechsel 1997, Gärtner et al 1990, Seufert 1990, Trimbacher et al 1995. Potassium also increased with needle age, and is also strongly leached from leaves (Tukey 1970), contributing to the nearly equal differences between needle age classes observed in AA and FA trees in the present study. Differently from Mn, K was neither depleted nor enhanced below a forest canopy in the eastern US (Lovett and Lindberg 1992), suggesting that leaching and deposition were approximately balanced.
The second, concurrent, aerosol effect was enhanced g min . The g min parameter describes the residual water loss of leaves when stomata are fully closed. It represents the water loss the plant cannot effectively control, and is thus linked to drought tolerance. Increased g min suggests a desiccating impact of ambient aerosols on all three species, which is in line with the recently reported ambient aerosol effects on faba beans (Grantz et al 2018). With stomata closed during determination of g min , residual transpiration happens largely by trans-cuticular (TC) or trans-stomatal (TS) liquid water movement. TC liquid water connections are tortuous, and their hydraulic conductance strongly declines with increasing VPD (Fernandez et al 2017, van Gardingen andGrace 1992). In contrast, the existence of TS liquid water connections has been demonstrated by stomatal uptake of nanoparticles (Eichert et al 2008). Compared to pure water (Schönherr and Bukovac 1972), deposited aerosols reduce hydrophobicity, and concentrated solutions from deliquescent aerosols can reduce water surface tension (Burkhardt et al 2012). Videos obtained by ESEM show that stomatal geometry does not prevent chaotropic ions that disorder water from spreading into stomatal pores . When these films join with apoplastic water, the hydraulic system of the plant is extended to the leaf surface (HAS, figure 1). Under complete stomatal closure, thin salt crusts with absorbed water vapor may persist and remain active as thin wicks, carrying liquid water between nearly closed guard cells to sites of evaporation at the surface. Increased g min with increasing age or air pollution has often been observed and attributed to leaky (defective or incompletely closed) stomata (Anfodillo et al 2002, Burghardt and Riederer 2003, Cape and Fowler 1981, Cape and Percy 1996, Kerstiens 1996, Jordan and Brodribb 2007, Mengel et al 1989, Reich 1984, Sase et al 1998. This leakiness has been considered as residual transport of water vapor rather than liquid water. While peristomatal TC transport may contribute to g min , an important role for TS flux associated with HAS has fewer chemical and structural constraints. However, while the present results (table 4) together with other studies (Burkhardt and Pariyar 2014, Chappelka and Neufeld 2018, Grantz et al 2018, Maier-Maercker and Koch 1992 show a direct impact of particulate air pollutants on plant drought tolerance, the exact pathway of additional water loss remains unclear.
Conditions varied modestly between the two greenhouses. Temperature and VPD were slightly higher in the FA compared to the AA greenhouse, while RH was marginally lower. Lower RH may lead to reduced deliquescence of deposited aerosol in the FA greenhouse. This was minor and dominated by the reduced aerosol deposition in the filtered treatment. Humidity at the leaf surface contributed by stomatal transpiration was also substantial in both treatments. Measurements of g min were made on detached leaves and needles held in darkness in the laboratory, unaffected by VPD differences between the greenhouses. Gaseous concentrations of HNO 3 and HCl were higher in the FA greenhouse, and therefore did not contribute to the greater 'wax degradation' on AA needles.
The microroughness of waxes on the lower surface of silver fir needles collects large amounts of fine aerosols (Burkhardt et al 1995). Under natural conditions these are not efficiently removed by rain (Freer-Smith et al 2005). This might be an important reason why FDAP started as 'fir decline' in the mid-1970s (Elling et al 2007, Schütt 1977, Ulrich 1990). The species lost its well-established drought tolerance and exhibited needle loss, reduced growth, and reduced or completely absent tree rings (Elling et al 2009). These symptoms were highly correlated with needle sulfate content, with the symptoms reducing when the air became cleaner by regulatory action (Elling et al 2009, Hauck et al 2012. Fir suffered more than other species by relatively low concentrations from distant sources, but was relatively tolerant to acute, high SO 2 concentrations (Brenninger andTranquillini 1983, Elling et al 2007). This suggests a reduced role for SO 2 toxicity and a greater role in this species for sulfate aerosols. The formation of particulate material on the leaf surface from gaseous precursors may be an important contributor to the effects documented in the present study.
In the Pyrenees mountains, near the xeric edge of its distribution area, silver fir is currently affected by TMHD (Linares andCamarero 2012, Peguero-Pina et al 2011). The species is isohydric, i.e. relying on sensitive stomatal closure in relation to drought. Carbon isotope measurements indicate that fir trees on declining sites had lower intercellular CO 2 concentrations (C i ) and higher intrinsic water use efficiency (WUE i ) than trees on non-declining sites (Linares and Camarero 2012). For plants grown under similar environmental conditions, foliar 13 C content can be interpreted as a time integrated measure of C i and WUE i (Farqhuar and Richards 1984). In the present study, 13 C values of second year fir needles in the AA greenhouse were less negative than in the FA treatment, indicating reduced C i and higher WUE i (see Farquhar and Richards 1984). In contrast, greater g min values indicated increased water loss in AA than FA needles. These conflicting results suggest that HAS (figure 1) may lead to liquid phase transpiration that uncouples 13 C values from actual WUEi. The isohydric strategy, relying on sensitive stomatal response to VPD and total water loss may thus become less effective under high aerosol loading.

Conclusions
The following conclusions result from this study in the context of other publications: Hygroscopic action is a primary factor for aerosol impact on plants. The hygroscopic aerosol fraction is mobilized by deliquescence. Creeping salt solutions may expand into the stomata, causing reduced stomatal control and desiccation. This can be recognized in the present study by SEM images (figure 3(g), figure 4(a)) and by the demonstrated effect on g min (table 4). It is anticipated that the importance of this process increases with (i) increasing aerosol hygroscopicity, which fosters deliquescence and mobility, (ii) increasing chaotropicity, which reduces surface tension, and (iii) the number of deposited aerosols.
Hygroscopic action produces 'wax degradation'. Notwithstanding its well established role as an indicator of forest decline by air pollution (FDAP), the explanatory power of 'wax degradation' as a mechanism of FDAP has remained low. The multitude of chemical compounds that have been found to produce the symptom (Turunen and Huttunen 1990) suggests that simple chemical reactions are not responsible. The present study shows that aerosol accumulation produces the same SEM images as hypothetical 'wax erosion' or 'wax degradation'. Aerosol accumulation on leaves may be the actual proxy for FDAP.
Hygroscopic action enhances VPD effects on plants. The present study shows that there are consequences of ambient aerosol loading on plant water relations, demonstrated here on g min and previously on g min and flux per unit of stomatal opening (Grantz et al 2018). Hygroscopic deposited aerosols react to leaf surface RH. Hydraulically activated stomata (figure 1) cause trans-stomatal liquid water flow and may decrease leaf water potential. Leaf water deficit reduces stomatal aperture, which may reduce CO 2 uptake, but is ineffective in reducing liquid water loss. In isohydric plant species, relying on stomatal closure for protection from atmospheric drought, increased VPD may enhance liquid and vapor phase transpiration, further reducing stomatal conductance and CO 2 uptake. These species may be most vulnerable to tree mortality by hotter drought (TMHD).
Generality of aerosol impacts on plants. In the present and previous studies we have observed similar effects of aerosol deposition, whether ambient or experimentally applied, on a broad range of plant taxa. The measured concentrations of total aerosol and of major ions in the ambient (AA) greenhouse used in this and previous studies were below those observed in many developed and developing countries. 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.