Scanning electron microscopy images in Figure 1 show the cuticular and stomatal patterns on abaxial surfaces of C. camphora leaves, with clear differences in the microstructures of AA leaves (Figures 1A,C) compared to FA leaves (Figures 1B,D). On the surfaces of AA leaves, there are more particles deposited visibly, which are shown as non-transparent, brighter, and randomly distributed granules in the SEM images, compared to FA. Flat, amorphous areas are only observed on AA leaf surfaces (Figures 1A,C), and may indicate salt crusts resulting from hygroscopic aerosols after deliquescence. Around these flat areas, the wax crystals are faintly covered; additionally, the original wax structures of stomata and epidermal cells are changed in AA leaves. In Figure 1C, it is visible that the arrangement of wax on the stomata and surrounding cells is less neatly distributed than in Figure 1D. Their appearance supports the hypothesis of the hygroscopic layer formed by deliquescent aerosols, which resulted in the overall impression of more disturbed surfaces, less defined stomatal structures, and a less visible stomata distribution for AA compared to FA leaves.

Table 2 shows the concentration of not dissolvable particulate matter and the overall amount of dissolvable aerosols deposited on leaf surfaces in greenhouse AA and FA, each number referring to the total adaxial and abaxial leaf area. The weight of not dissolvable particulate matter deposited on AA leaves was higher than on FA leaves, with the comparable median value of 3.59 μg/cm² (n = 12) and 1.40 μg/cm² (n = 9), respectively. The total amount of dissolvable aerosols in AA was more than 9 times higher than in FA. The ratio of not dissolvable particulate matter to total deposited aerosol amount within a square centimeter in AA is 74%, and in FA is 93%. Figure 2A details the ionic composition of the dissolvable aerosols, respectively. Nitrate, sulfate, and chloride are the dominant compounds of aerosol deposition in AA, while Na, K, Mg, and ammonium are the subordinate ones. In FA, the concentration of chloride, sulfate, and K is relatively higher than the other ions. Figure 2B shows the concentration of dissolvable aerosols deposited on leaf surfaces from field sites in Taiwan. The dominant compounds in Pingtung are K and Cl, which are much higher than the concentration of nitrate, sulfate, and ammonium. Mg and Na show a value close to zero of the concentration in Pingtung. A different distribution pattern is found in Hualien, with Cl having the highest concentration, followed by Na, sulfate, K, Mg, nitrate, and ammonium. Although there are differences between compound species in Pingtung and Hualien, the total concentration of dissolvable deposited aerosols in Pingtung (1.54 ± 0.142 μg/cm², n = 17) is not significantly higher than in Hualien (1.26 ± 0.132 μg/cm², n = 12).

Table 2 shows the difference in contact angles on adaxial and abaxial leaves from greenhouse AA and FA, as well as of leaves from the fields. There is no significant difference in adaxial contact angles between AA (124.56 ± 2.62, n = 12) and FA (127.00 ± 2.47, n = 9) leaves, neither between Pingtung (65.96 ± 7.59, n = 14) and Hualien (53.16 ± 1.87, n = 16). With abaxial contact angles, AA leaves (140.13 ± 1.37, n = 12) and FA leaves (143.14 ± 1.18, n = 9) do not differ either. However, Pingtung leaves (119.33 ± 9.47, n = 8) have higher values than Hualien leaves (75.68 ± 5.13, n = 16).

In Table 2, the key photosynthetic parameters of leaves from AA and FA are presented (n = 4). There are no significant differences in A_sat, V_cmax, J, TPU, R_d, and g_m between the leaves from two greenhouses, nor between the two field sites.

Stomatal conductance (g_sw) shows a decreasing tendency as VPD increases, both in AA and FA (Figure 3; n = 3, respectively). In AA, the g_sw value decreases more moderately from 0.015 ± 0.001 mol m⁻² s⁻¹ (while VPD 0.50 kPa) to 0.009 ± 0.0005 mol m⁻² s⁻¹ (while VPD 1.50 kPa), with small SEs. However, in FA, the g_sw value falls more rapidly from 0.042 ± 0.005 mol m⁻² s⁻¹ (while VPD 0.50 kPa) to 0.017 ± 0.001 mol m⁻² s⁻¹ (while VPD 1.50 kPa). With each VPD set point, leaves in AA show a lower g_sw value than in FA (p < 0.01), especially when VPD is low (i.e., 0.50 kPa). Additionally, it is shown that in both AA and FA, the g_sw of leaves maintain a similar value instead of decreasing while VPD changes from 1.25 to 1.50 kPa. Subsequently, differences are also found for the stomatal model parameters g₀ and g₁, calculated from the Ball Index on the basis of assimilation (A_n), relative humidity (H_s), and CO₂ concentration at the leaf surface (C_s), and then further performed with the Ball-Berry model (Figure 4). The values of A_n in AA are generally lower than FA, causing a lower range of the Ball Index. The regression line of AA leaves is, therefore, extended to the full range of the x-axis by using the data points predicted with the linear model; and two regression lines are compared based on the actual data points. Both regression lines indicate a positive correlation between g_sw and Ball Index, representing the fitted data calculated from the leaf-scale measurements, where both of the R² values are higher than 0.90. The slope of the linear regression (g₁) for AA is about half the slope for FA (Figure 4), and the g_sw intercept (g₀) of the linear regression for AA is also smaller than g₀ for FA (P < 0.005).

The SEM images of C. camphora bring out comparable results with previous research regarding the relation of deposited aerosols and leaf morphology, and the formation of amorphous regions similar to so-called “wax degradation” on the cuticle or close to stomata (Burkhardt, 2010; Burkhardt and Grantz, 2017; Chen et al., 2017). The pattern of hypothetical aerosol layer and amorphous wax degradation have been as well found on the leaf surfaces of Cryptomeria japonica (Sase et al., 1998), Brassica oleracea (Gongylodes Group) (Burkhardt et al., 2001), Platanus orientalis L. (Pourkhabbaz et al., 2010), Pinus sylvestris L. (Burkhardt and Pariyar, 2014), Quercus variabilis (Mo et al., 2015), and Vigna radiata (L.) R. Wilczek (Shabnam et al., 2021). Moreover, research has indicated that identified wax degradation might be actually a mixture of deliquescent aerosols and disturbed wax crystallization; the development of amorphous wax appearance can result from deliquescent salts covering tubular wax fibrils, following the process of (i) the attraction of water vapor by hygroscopicity; (ii) the dissolution of hygroscopic aerosols; and (iii) the resulting mobility and distribution across the leaf surface, leading to the coverage of tubular waxes by amorphous crusts and consequently showing the typical appearance of wax degradation (Burkhardt, 2010; Burkhardt and Pariyar, 2014; Burkhardt et al., 2018). In this study, few larger deposited aerosols on AA leaf surfaces are observed as crystalline, but most of them appear to be amorphous crusts caused by the humidity cycle correlated with the deliquescence of salt and the transpiration of stomata. This phenomenon consists of the previous studies aforementioned.

As for the results of not dissolvable particulate matter and dissolvable aerosol concentration from leaves grown in the greenhouses, it is significant that AA leaves accumulated higher aerosol concentration than FA leaves, regardless of the total aerosol concentration and specific aerosol compounds. Compounds such as Na and Cl may come from sea salts, even though the greenhouses locate a bit distant from the coast (Burkhardt and Eiden, 1990). In general, the dominant aerosol compounds in AA are similar to previous research which was done in the same greenhouse environment (Burkhardt and Pariyar, 2016). Although the epicuticular wax may partially also contain aerosols (Dzierżanowski et al., 2011; Victório et al., 2021), it is neglectable in this study since the focus is on researching the aerosol effects within one species, instead of the quantification and classification of deposited aerosols. The water dissolvable ions contributed about 30% to the overall aerosol mass found on AA leaves, which is in agreement with the reported range of European aerosol composition (Putaud et al., 2010).

The ionic deposition load on leaves at Pingtung was 1.54 μg/cm², exceeding the amounts on Hualien leaves (1.26 μg/cm²) by 22%. This difference was less than expected from long term monitoring data and literature (Lin et al., 2008; Li et al., 2016; Lee et al., 2020). The ionic composition on Hualien leaves was dominated by sea salt (Na, Cl), reflecting the small distance to the sea (50 km in the main wind direction). Nitrate and sulfate are mainly composed of secondary ammonium sulfate and ammonium nitrate from industry (Yang et al., 2017; Shen et al., 2019, 2020). Nitrate, ammonium, and potassium strongly contributed to the composition of particles on Pingtung leaves, whereas the sulfate and magnesium concentrations were higher on Hualien leaves (Figure 2B). The daily monitoring data were extracted from Taiwan Air Quality Monitoring Network, Environmental Protection Administration, Taiwan, in order to inspect the environmental aerosol concentration with an accurate time range (Figure 5). Continuous torrential rain probably caused the strong decrease of PM_2.5 concentrations shortly before the experiment at the Pingtung site, and also the removal of particles from leaves (Wang et al., 2015), particularly from upper leaf surfaces. However, rainfall itself is also able to contribute to the ion concentrations besides washing off particles; consequently, rainfalls might affect aerosol retention and long term accumulation of ionic aerosols on leaf surfaces, and foliage traits are the more important factors related to these effects (Xu et al., 2017; Pariyar and Noga, 2018; Zhang et al., 2019; Zhou et al., 2020). As an evergreen tree species, C. camphora is likely subject to a higher wash off rate of fine aerosols at high rainfall intensities, causing the indistinct aerosol distribution and concentration on leaf surfaces in the Pingtung site (Xu et al., 2019; Zhou et al., 2021). Therefore, this inconsistency is challenging the accuracy of the other field measurements.

The high g_min values and the low abaxial contact angles of leaves at the Hualien site likely are connected effects of aerosol deposition. The g_min values were more than twice as high and the ratio between adaxial and abaxial contact angles differed compared to the Pingtung site. Although other, e.g., biotic factors cannot be excluded, both effects are likely linked to the relatively high, sea salt dominated deposition at the Hualien site. Normally, NaCl is a kosmotropic salt that does not easily extend on hydrophobic cuticles, so coastal plants are usually not affected too much by sea salt. However, this may considerably change in the presence of detergents, as shown by a strong g_min increase in a previous experiment, where pine seedlings were sprayed with different salt solutions (Burkhardt and Pariyar, 2014). The detergent reduces the contact angle and promotes stomatal penetration by the salt, i.e., HAS establishment. Several cases of this process in the environment were reported in Italy and Australia, where detergents from close-by landfills caused the coating of sea-spray aerosols leading to the decline of coastal forests (Bussotti et al., 1995). A similar process might actually have played a role at the Hualien site because a landfill in the major source region of NaCl aerosols had been eroded by the sea for several years (Taiwan News, 2018). The occurrence of such detergents on the leaves was not measured and the distance of 50 km is considerable, but still, there is a realistic chance that the high g_min values and low contact angles were connected with this incidence.

Low contact angles particularly on the lower (abaxial) leaf sides of Hualien leaves point to enhanced deposition of fine, sub-micrometer aerosols, which are less affected by gravity but more by molecular mechanisms. The g_min parameter describes the uncontrollable water loss of leaves with closed stomata. A g_min increase is indicative of reduced drought tolerance and reflects the cuticular permeance, but also the contribution of “malfunctioning stomata,” which are linked to aerosols and HAS (Kerstiens, 1996; Burkhardt, 2010). The higher g_min values indicate that in the case of extended droughts, aerosol deposition might possibly become problematic for the trees at the Hualien site. It is not possible to deduct further impacts of aerosols on the trees from the field measurements. The δ¹³C values between the two field sites cannot be meaningfully compared, as they are influenced by too many different environmental factors, particularly soil water availability, temperature, and VPD.

The greenhouse study with equal environmental conditions between AA and FA enables the comparison of single parameters like g_min or δ¹³C. Differences between the groups can be attributed to the differences between AA and FA aerosol concentrations, as long as the AA and FA plants are physiologically comparable. This requirement was met in the present case, as seen by the comparison of A_sat and the AC_i curves, from which the photosynthetic parameters were extracted. These parameters were very similar between AA and FA. The higher daytime respiration R_d was consistent between AA and FA but was several times higher than at the field sites and in an earlier field study with camphor trees (Kosugi and Matsuo, 2006); possibly due to the effects of the incomparable temperature differences between the greenhouses and the fields, or the inaccuracy caused by different calculators while fitting AC_i curve data (Sharkey, 2016). It was hypothesized that physiological responses to aerosols would include higher g_min, less negative δ¹³C value, lower leaf water potential at turgor loss (Bartlett et al., 2012a; Maréchaux et al., 2015), and higher proline concentration in the AA compared to the FA greenhouse. With a similar experimental approach, aerosols had caused higher g_min for Quercus petraea, Abies alba, Pinus sylvestris (Burkhardt and Pariyar, 2014; Burkhardt et al., 2018), and Vicia faba (L.) (Grantz et al., 2018), as well as less negative δ¹³C for second year Abies alba needles, (Burkhardt et al., 2018), while Helianthus annuus, Pinus sylvestris, and Fagus sylvatica were found to have more negative δ¹³C values (Burkhardt and Pariyar, 2016).

In this study, particularly the results of g_min and δ¹³C did not confirm the hypothesis. The g_min results were almost identical between AA and FA greenhouses, which was about half the Pingtung values and only about one-fifth of the Hualien value. A major reason for missing significant differences probably was the small number of repetitions (n = 4). This is particularly relevant for the g_min parameter, where due to high variances and small effects often about 20 repetitions are required to reach significant results. The high variability probably comes from the situation that the water loss by incompletely closed, ‘leaky' stomata is an individual process affecting single stomata, but often is the dominating pathway of water loss in the g_min measurement compared to water loss across the cuticle (Heinsoo and Koppel, 1998; Burkhardt, 2010; Duursma et al., 2019). A study of Hedera helix indicated that 35% of water loss occurred across the incompletely closed stomatal pores and 65% across the other part of the cuticle which is without stomata, and the cuticular transpiration of the stomatous leaf surface was about 11 times higher than the astomatous leaf surface (Šantrůček et al., 2004). Moreover, taking conifer species as research material, it is concluded that the percentage of water loss from stomatal pores of detached leaves might depend on species-specific strategies for conserving water during drought (Brodribb et al., 2014). Because only few studies have found significant correlations between g_min and environmental factors, other procedures may be more useful under certain conditions (Brodribb et al., 2014; Schuster et al., 2017; Duursma et al., 2019). Under less defined conditions, another possible reason for questioning the reliability of g_min is the acclimation of plants to the environment. In general, research has shown that plants change the chemical composition of the cuticle while facing water stress, leading to a decreased g_min value (Bengtson et al., 1978; Premachandra et al., 1992; Macková et al., 2013; Bi et al., 2017). The observation that older leaves have higher g_min values (Jordan and Brodribb, 2007), might however be caused by the damage of cuticle on old leaves or the increasing contribution of HAS establishment and induced water loss across the stomatal pore (Burkhardt, 2010).

The discrimination value of carbon isotope composition (δ¹³C) provides information on the long term transpiration efficiency of plants, and a lower δ¹³C value is often determined as lower WUE (Farquhar and Richards, 1984; Farquhar et al., 1989; Hubick and Farquhar, 1989; Condon et al., 1992; Cabrera-Bosquet et al., 2007), but requires equal environmental conditions between the compared groups. However, recent studies have focused on more comprehensive and practical conditions instead of an ideal growing environment such as breeding fully fertilized plants in the greenhouse (Conte et al., 2003; Cabrera-Bosquet et al., 2007; Burkhardt, 2010; Berriel et al., 2020; Vogado et al., 2020). Thus, the correlation between δ¹³C and WUE might be influenced by deposited aerosols and HAS, but also by soil water, the nutrient conditions, and the acclimation to stresses (Cabrera-Bosquet et al., 2007; Berriel et al., 2020; Tarin et al., 2020), which is why the field values cannot be compared.

The hypothesis of lower leaf water potential at wilting (i.e., turgor loss point, π_tlp) by aerosols was also not confirmed. π_tlp is considered another important determinant of ecological and physiological drought tolerance, which is also strongly correlated with the cell solute potential at full hydration (i.e., osmotic potential, π_o) (Bartlett et al., 2012a,b; Banks and Hirons, 2019). Previous research has focused on π_tlp of plant species such as woody species, crops, and herbaceous grassland species, concluding that this indicator of drought tolerance varied across species and environmental conditions; π_tlp is as well correlated slightly with several leaf functional traits such as leaf dry matter, leaf vulnerability to hydraulic failure, leaf toughness, and leaf thickness (Maréchaux et al., 2015; Griffin-Nolan et al., 2019). Normally, a more negative π_tlp increases the functional range of foliar water potential, showing a greater leaf-level drought tolerance (Mart et al., 2016; Banks and Hirons, 2019). Under defined conditions, a more negative π_tlp would thus mean that the plant had experienced drought stress by aerosols (Navarro et al., 2007; Burkhardt, 2010). This should be further evaluated using experiments with higher numbers of biological repetitions, including the evaluation of an eventual accumulation of proline. Proline is an additional indicator of osmotic adjustment, responding to environmental stress such as water deficit, salinity, heat, and pollutants (Bates et al., 1973; Dolatabadian et al., 2008; Acosta-Motos et al., 2017). In this study, the proline concentration of leaves did not differ with aerosol exposure, and concentrations in both AA and FA were relatively low.

Aerosols did not decrease WUE, as it originally had been expected. Contrariwise, the VPD curve of the FA plants had higher g_sw values than AA, which was highly significant. The subsequently calculated Ball-Berry g₁ parameter for FA was twice the value compared to AA. Because g_min (which can be considered the g₁ factor of the Ball-Berry equation; Duursma et al., 2019) was negligible compared to g_sw for both AA and FA, this means double WUE of AA compared to FA (Equation (4); Miner and Bauerle, 2017). The relationship between g₁ and WUE is originally linked to intrinsic WUE (A/g_s) but is also indicative of actual (“instantaneous”) WUE (A/E; Franks et al., 2017). According to the original HAS hypothesis (Burkhardt, 2010), AA leaves should have lost more water than FA at the same degree of stomatal opening; and because this additional water loss is not accounted for by CO₂ uptake, AA leaves should have lower WUE than FA. But probably this is not the full picture and there may be several independent responses to aerosols. In an AA/FA experiment with Vicia faba (L.), aerosol exposure (i.e., AA) had three effects (Grantz et al., 2018):

In the present study with camphor, the HAS effect of additional water loss at the equal aperture (effect ii) was likely overcompensated by the aperture reduction of AA stomata (effect i). A reduction of stomatal aperture, however, is known to increase the WUE of seed plants, e.g., in response to drought stress (Franks et al., 2015; Guerrieri et al., 2019; Xu et al., 2021; Yang et al., 2021). The measured increase of WUE_i by aerosols thus indicates a reduction of stomatal aperture, in agreement with the results of the Vicia faba (L.) experiment (Grantz et al., 2018, 2020). It is also in agreement with these earlier results that the error bars of the AA data points were smaller than for FA, indicating lower variation, higher coordination between stomatal apertures (effect iii), and less patchiness - a general susceptibility of the C. camphora to the stomatal patchiness phenomenon has earlier been reported (Takanashi et al., 2006). The aperture reduction was not directly measured but would have been independently supported if lower δ¹³C values of AA compared to FA leaves were observed. This was not the case, possibly because the results of the VPD curves and the δ¹³C signals were determined by different micro-climatological conditions: The VPD curves were measured within ventilated cuvettes. The δ¹³C values are a time integrated signal of gas exchange, produced under the calm greenhouse conditions with a thick leaf boundary layer surrounding the leaves most of the time; so stomatal responses are decoupled from the environmental VPD and its interaction with deposited aerosols.

Generally, the g₁ parameter represents a compromise between the costs and benefits of g_sw relative to the photosynthetic activity of the leaf (Ball et al., 1987; Miner and Bauerle, 2017). The g₀ is normally defined as either (i) a fit parameter extrapolated as the intercept of the least squares regression between g_sw and the Ball Index (Ball et al., 1987; Ball, 1988; Collatz et al., 1991), or (ii) the residual conductance when A_n ≤ 0 (Leuning, 1995). The g₁ values here were 1.71 (AA) and 3.41 (FA) and, thus, considerably lower than the value of 7.4 observed for C. camphora in a field study (Kosugi and Matsuo, 2006). Both g₀ and g₁ were at the lower end but still within the range of previously recorded values (Miner et al., 2017; Wolz et al., 2017). Drought affected plants, e.g., Eucalyptus, Quercus, Zea mays, and Helianthus, often have lower g₁ and g₀ values compared with well-watered plants of the same species (Cavender-Bares et al., 2007; Heroult et al., 2013; Zhou et al., 2013; Miner and Bauerle, 2017; Miner et al., 2017). The lower g₁ value of AA camphor leaves compared to FA can thus possibly be interpreted as aerosol induced drought stress. The reason for the involvement of H_s for plant transpiration in the original, semi-empirical Ball-Berry model has remained elusive and its relevance was questioned, compared to VPD which seems to be physiologically more meaningful (e.g., Monteith, 1995). The successful H_s use, however, might well be due to the direct interaction of hygroscopic, deposited aerosols with water vapor on the leaf surface. This kind of interaction is immediate and direct and the method to determine water absorption to specific salts has been used to determine the relative humidity in weather balloons (Wylie, 1955).