Stomatal penetration by aqueous solutions – an update involving leaf surface particles


Author for correspondence:

Juergen Burkhardt

Tel: +49 228 732186



  • The recent visualization of stomatal nanoparticle uptake ended a 40-yr-old paradigm. Assuming clean, hydrophobic leaf surfaces, the paradigm considered stomatal liquid water transport to be impossible as a result of water surface tension. However, real leaves are not clean, and deposited aerosols may change hydrophobicity and water surface tension.
  • Droplets containing NaCl, NaClO3, (NH4)2SO4, glyphosate, an organosilicone surfactant or various combinations thereof were evaporated on stomatous abaxial and astomatous adaxial surfaces of apple (Malus domestica) leaves. The effects on photosynthesis, necrosis and biomass were determined. Observed using an environmental scanning electron microscope, NaCl and NaClO3 crystals on hydrophobic tomato (Solanum lycopersicum) cuticles underwent several humidity cycles, causing repeated deliquescence and efflorescence of the salts.
  • All physiological parameters were more strongly affected by abaxial than adaxial treatments. Spatial expansion and dendritic crystallization of the salts occurred and cuticular hydrophobicity was decreased more rapidly by NaClO3 than NaCl.
  • The results confirmed the stomatal uptake of aqueous solutions. Humidity fluctuations promote the spatial expansion of salts into the stomata. The ion-specific effects point to the Hofmeister series: chaotropic ions reduce surface tension, probably contributing to the defoliant action of NaClO3, whereas the salt spray tolerance of coastal plants is probably linked to the kosmotropic nature of chloride ions.


Most leaf surfaces are hydrophobic or only moderately hydrophilic as a result of leaf surface waxes, although some rare exceptions do exist (Holloway, 1969; Neinhuis & Barthlott, 1997; Koch et al., 2009). Forty years ago, in their paper entitled, ‘Penetration of stomata by liquids – dependence on surface tension, wettability, and stomatal morphology’, Schönherr & Bukovac (1972) concluded that the combination of cuticular hydrophobicity, water surface tension and stomatal geometry should prevent water droplets from infiltrating the stomata, unless pressure is applied or the equilibrium surface tension undergoes an extreme reduction, to < 30 mN m−1. Such a reduction is achievable only by a few surfactants, typically the organosilicones (Stevens et al., 1991). This concept became dominant (cf. Niederl et al., 1998; Fernandez & Eichert, 2009; Taiz & Zeiger, 2010) and its validity was increasingly assumed for all types of aqueous solutions and environments, turning it into a paradigm. The a priori exclusion of stomatal liquid water transfer restricted the uptake and loss of liquid water and aqueous solutions to the cuticular part of the leaf surface. However, there has also been a steadily increasing number of experimental reports from different backgrounds that have pointed to the stomatal uptake of water (Burgess & Dawson, 2004; Breshears et al., 2008; Limm et al., 2009; Simonin et al., 2009), nutrients (Fernandez et al., 2006; Will et al., 2012), ionic fluorescent dyes (Eichert et al., 1998; Eichert & Burkhardt, 2001) and abscisic acid (Wilkinson & Davies, 2008) without the use of surfactants. The main argument against such conflicting observations was the possibility of spatial, temporal or spatiotemporal cuticular heterogeneities (e.g. higher cuticular permeability in regions near the stomata, exclusively during periods of open stomata (Schönherr & Bukovac, 1978)), which has been challenging to disprove. Only the visualization of the stomatal uptake of fluorescent nanoparticles (specifically water-suspended hydrophilic polystyrene particles of 43 nm in diameter) via confocal laser scanning microscopy (Eichert et al., 2008) finally provided an unequivocal demonstration that the stomatal uptake of water, solutes and dispersed substances is indeed possible.

Stomata may be rendered susceptible to this type of transport by environmental factors that reduce the innate hydrophobicity of leaf surfaces. In a process that affects individual stomata, the ‘hydraulic activation of stomata’ (HAS) may be established once a threshold is exceeded (Burkhardt, 2010). HAS signifies the formation of continuous, thin (presumably < 100 nm) liquid water films on the stomatal walls that connect the apoplast and the leaf surface and enable the bidirectional transport of water, dissolved or dispersed substances, and hydraulic signals (Burkhardt, 2010). Depending on the history and age of the plant and the environmental conditions, HAS might be a common phenomenon. It provides a link between air pollution and plant drought stress, and could play a role in nocturnal transpiration or stomatal reaction to humidity (Burkhardt, 2010).

Suggestions addressing the development of HAS comprise fungal hyphae (Burgess & Dawson, 2004), bacteria (Eichert et al., 2008), stomatal mucilage (Zimmermann et al., 2007; Westhoff et al., 2009) and hygroscopic leaf surface particles (Burkhardt, 2010). Leaf surface particles may originate from the atmospheric deposition of aerosols or trace gases (e.g. SO2, which is oxidized within the water covering the leaf; Burkhardt & Drechsel, 1997), residues of evaporated rain, fog or agricultural sprays, plant exudations or phyllospheric organisms. The amount of leaf surface particles may reach several μg cm−2, a similar weight range as the cuticular leaf waxes (Burkhardt, 2010 and references therein, Sæbø et al., 2012). The deposition of submicrometre aerosols is almost independent of gravity, that is, deposition should happen equally on the abaxial and adaxial sides of the leaves as long as their microroughness is equal (Peters & Eiden, 1992; Burkhardt et al., 1995). Even sustained rainfall does not effectively remove deposited aerosols from deciduous or coniferous trees (Freer-Smith et al., 2005). Decreasing contact angles with increasing age are most probably caused by the accumulation of noncuticular material and its chemical constituents (Cape, 1983; Neinhuis & Barthlott, 1998; Bringe et al., 2006), although cuticular abrasion may also play a role (Pitcairn et al., 1986; Hoad et al., 1992). Leaf surface particles may influence plant–water relationships. The stomatal diffusive resistance of deciduous trees is reduced by particle pollution (Flückiger et al., 1977) and the transpiration of elder trees (Sambucus nigra) at the same degree of stomatal aperture increases on treatment with NaNO3 particles (Burkhardt et al., 2001). Approximately 20–40% lower stomatal conductance and transpiration were observed for bean (Vicia faba) and sunflower (Helianthus annuus) plants grown under the exclusion of aerosols compared with control plants grown in normal air (Pariyar et al., 2012).

Schönherr & Bukovac (1972) concentrated on static droplets of pure water on a flat surface, with droplet dimensions exceeding the stomatal aperture. Real leaf surfaces are more complex: epidermal roughness, trichomes or epicuticular waxes often create structural heterogeneities that influence leaf wetting (Quere, 2008; Fernandez et al., 2011). Phyllospheric organisms actively bind water on the surface through biofilm formation (Schreiber et al., 2005; Beattie, 2011). Water is also attracted by deposited aerosols, many of which consist partially or entirely of hygroscopic salts. Salts suddenly become deliquescent (liquid) at a salt-specific relative humidity (RH), the deliquescence relative humidity (DRH; c. 75% for NaCl or NaClO3, 40% for NH4HSO4 and 80% for (NH4)2SO4). The salts absorb exponentially more water with further increasing humidity (Pilinis et al., 1989; Zhao et al., 2008; Mauer & Taylor, 2010). This mechanism is similar to the activation of cloud condensation nuclei, although DRH is slightly different for deposited particles (Gao et al., 2007), and plant transpiration provides an additional water vapour source (Burkhardt & Eiden, 1994). Repeated deliquescence and efflorescence of single particles on the leaf surface can lead to the development of ‘thin water films’ or ‘filaments’, even on hydrophobic leaf surfaces and even during hot summer days (Burkhardt & Eiden, 1994; Burkhardt et al., 1999). The RH at the salt surface (water vapour phase) and the water activity of the hygroscopic particles (liquid water phase) are a physical identity (Pilinis et al., 1989). Once a saturated solution is formed at the deliquescence point, the corresponding water activities (e.g. 0.75 for NaCl or NaClO3, 0.4 for NH4HSO4 and 0.8 for (NH4)2SO4) increase, whereas the concentrations (6.1 M for NaCl, 8.6 M for NaClO3, 9.0 M for NH4HSO4 and 5.7 M for (NH4)2SO4 at 20°C, respectively; IFA, 2012) decrease with increasing RH.

The chemical, physical and physicochemical properties of such highly concentrated solutions are considerably different from those of dilute solutions. Ion activities, rather than ion concentrations, determine the chemical reactions (Pitzer, 1981). Physical effects include capillary condensation, capillary transport of substances, Marangoni flow (cyclic inward or outward movement within the droplet), the accumulation of dispersed substances at the edges (coffee-rings), the reduction of contact angles by preferential evaporation from droplet edges and ‘line-pinning’ of droplets during evaporation (Eiden et al., 1994; Deegan et al., 1997; Herminghaus et al., 2008; Xu et al., 2010; Hunsche & Noga, 2012).

The most important physicochemical effect of ions in the present context is related to water surface tension. This parameter reflects the dispersive forces across the phase boundary, as well as the specific forces within one phase, such as hydrogen bonding (Dutcher et al., 2010). Water surface tension was central to the reasoning of Schönherr & Bukovac (1972), but they referred to pure water. Ions change the water surface tension in a concentration-dependent and ion-specific manner. This is related to the respective distribution of the ions between the surface and the bulk of a water droplet and to the order of the Hofmeister (or lyotropic) series (Collins & Washabaugh, 1985; Bostrom et al., 2001; Pegram & Record, 2007; Liao et al., 2009; Dutcher et al., 2010; dos Santos et al., 2010; Zhang & Cremer, 2010). The Hofmeister series originally classified ions in electrolytic solutions according to their ability to dissolve proteins (Hofmeister, 1888; Kunz et al., 2004a,b). For anions, the Hofmeister series is

display math

and, for cations, it is

display math

(Zhang & Cremer, 2006; Miller & Hoseney, 2008). Ions that are considered to be kosmotropic are at the left of the series, and chaotropic ions are at the right, and anions have a much greater effect than cations. Thus, the sulfate anion, which is on the kosmotropic side of the series, decreases the solubility of nonpolar molecules, increases the hydrophobic interaction (‘salting out’) and increases surface tension. The chlorate ion belongs to the group of chaotropic ions, which increase the solubility of nonpolar molecules, weaken the hydrophobic interaction (‘salting in’) and decrease surface tension. The chloride ion is sometimes considered to be kosmotropic (dos Santos et al., 2010) and sometimes considered to represent the dividing line between kosmotropic ‘structure makers’ and chaotropic ‘structure breakers’.

NaCl is the most important aerosol in coastal regions and leaf surface accumulation of up to 300 μg cm−2 NaCl has been found (Cheplick & Demetri, 1999). Coastal dune plant adaptations to salt spray may be manifested at both the genetic and acclimation level. Tolerance decreases with high wind, low rainfall and the presence of surfactants (Boyce, 1954; Sykes & Wilson, 1988; Griffiths & Orians, 2003; Griffiths, 2006; Lowry et al., 2009; de Vos et al., 2010). In several coastal areas around the world, the release of surfactants by landfills has been found to be responsible for forest decline symptoms as a result of enhanced stomatal salt uptake (Dowden et al., 1978; Grieve & Pitman, 1978; Truman & Lambert, 1978; Bussotti & Ferretti, 1998; Paoletti et al., 2005). By contrast, NaClO3 is well known as a defoliant (‘defol’), and is one of several salts and acids that have been used as ‘chemical desiccants’ in practical agriculture. This technique was primarily used by farmers several decades ago to kill potato vines before harvest, but has also been used in plant breeding applications to rapidly evaluate drought-tolerant cultivars, given that spraying with chemical desiccants creates drought symptoms under irrigated conditions. Other successfully used chemicals are potassium iodide, magnesium chloride, sulfuric acid and sodium hydroxide; by contrast, sodium chloride and ammonium sulfate (AMS) are not suitable for this purpose (Wilson et al., 1947; Murphy, 1968; Blum et al., 1983a,b; Nicolas & Turner, 1993; Regan et al., 1993; Royo & Blanco, 1998; Sawhney & Singh, 2002; Bond & Bollich, 2007). AMS has repeatedly been reported to increase the bio-efficacy of glyphosate (N-(phosphonomethyl)glycine; Gly) (Hall et al., 2000; Pratt et al., 2003; Liu, 2004b; Nurse et al., 2008). Gly, the world's most widely used herbicide (Alister et al., 2005), is a highly water-soluble agrochemical with pronounced translocation characteristics, but variable foliar uptake (Field & Bishop, 1988). Improved Gly uptake by AMS may come from water conditioning, with AMS preventing the inactivation of Gly through a chemical reaction with calcium (Thelen et al., 1995), but positive effects of AMS have also been observed for Gly in distilled water (Young et al., 2003; Nurse et al., 2008).

Hygroscopicity and ion-specific effects enable leaf surface particles to change both the cuticular hydrophobicity and water surface tension, the major pillars of Schönherr & Bukovac's (1972) argument. We therefore re-assessed the question of stomatal uptake of aqueous solutions using anions from different sides of the Hofmeister series, and evaluating their impact on physiology. We visualized key processes using an environmental scanning electron microscope (ESEM).

We hypothesized that:

  • the stomatous abaxial side of apple leaves should take up more NaCl, NaClO3, AMS, Gly and combinations thereof than the astomatous adaxial side;
  • the chaotropic NaClO3 should be taken up by the leaf more readily than the more kosmotropic NaCl and AMS;
  • both NaClO3 and AMS may act as adjuvants for Gly uptake;
  • salts on hydrophobic cuticles should attract liquid water considerably below saturation (100% RH) and be able to overcome cuticular hydrophobicity by spatial expansion.

Materials and Methods

Expt 1

Apple (Malus domestica Borkh.) was used as a model system, having astomatous upper and stomatous lower leaf sides. Seedlings raised from seeds of the cultivar Golden Delicious (Eichenberg und Co. Gehölzsamen GmbH, Miltenberg, Germany) were grown in soil until the appearance of c. 12 leaf pairs. All plants remained in the growth chamber (20°C, 50% RH day : 18°C, 70% RH night; 200 μmol m−2 s−1 photosynthetically active radiation (PAR)) where they were fully watered and nourished. Stomatal density and dimensions were determined microscopically with a mean of 102 ± 9 stomata mm−2 on the abaxial (lower) side, whereas the adaxial (upper) side was completely astomatous. The low stomatal density might be caused by the climate chamber conditions with comparatively low light intensity and practically no UV radiation.

Two leaves of each plant were treated, either on the adaxial or abaxial side of opposing leaves (leaves 13 and 14). A 2-ml syringe was filled with treatment solution and homogeneous droplets were released to either the adaxial or abaxial surface. Thirty droplets were applied to each leaf. The total volume reduction was recorded and the droplet volume was calculated as 6 μl. Five replicate plants were treated with a 10 mM NaClO3 salt solution on one occasion. Ten replicate plants were treated with a 10 mM NaCl salt solution on days 1, 2, 7, 8, 9 and 10. Corresponding replicate numbers of control plants were treated with deionized water. NaCl and NaClO3 were chosen as two salts having approximately the same deliquescence point (75% RH; Pilinis et al., 1989; Römpp et al., 1999), with the anions from different sides of the Hofmeister series.

Gas exchange was measured on the day before the first and on the second day after the last treatment, using a photosynthesis system (Li-6400; Lincoln, NE, USA), with a CO2 concentration of 400 μmol mol−1 and PAR of 250 μmol m−2 s−1. A low humidity value of 25% RH was set to stimulate transpiration. At the time of the second measurement, some of the leaves treated abaxially by NaClO3 had already started to turn brownish in colour in the area in which the droplets had been applied.

Expt 2

Apple (Malus domestica Borkh.) seedlings were grown under the same conditions as in Expt 1. Treatments were begun when approximately six to seven leaf pairs had appeared. Stomatal densities were 105 ± 10 for the first batch of apple plants to which AMS was applied, and 72 ± 8 for the second batch to which NaClO3 was applied. In each batch, the control and two different concentrations of the respective salt were applied (0, 25 and 150 mM), with or without Gly (43 mM; 62% purity, nonsurfactant formulated isopropylamine salt; Monsanto Europe SA, Antwerp, Belgium), and with or without the organosilicone surfactant ‘Break-Thru S233’ (BT; 0.1% v/v; Evonik Goldschmidt GmbH, Essen, Germany). For each treatment, seven or eight plants were placed as replications in a completely randomized statistical design. Two droplets of 2 μl each were placed on the central part of the leaf lamina opposing the mid rib, either on the adaxial or abaxial side of opposing leaves (leaves 3 and 4).

Chlorophyll-a fluorescence parameters were recorded using a ‘pulse-amplitude-modulation-fluorometer’ (PAM) imaging chlorophyll fluorometer (Heinz-Walz GmbH, Effeltrich, Germany). The ground fluorescence (Fo) of dark-adapted leaves (20 min) was measured after illumination with blue light (470 nm) at an intensity of 0.5 μmol m−2 s−1 PAR. Consecutively, the maximum fluorescence (Fm) was measured after illumination with blue light at an intensity of 2400 μmol m−2 s−1 PAR for 800 ms. Fv/Fm (where Fv = Fm − Fo), as a measure of the maximal photosystem II (PSII) quantum yield of dark-adapted samples, was calculated from the recorded Fo and Fm values. All the measurements and calculations were conducted in the ImagingWin v2.21d (Heinz-Walz GmbH) software environment. Fluorescence readings were taken at the middle part of the adaxial surface of the leaf lamina across the midrib of the first fully grown leaf, on untreated control plants and plants at 18 d (batch with AMS) and 13 d (batch with NaClO3) after treatment.

Fresh biomass above the ground was harvested and weighed at 19 d (batch with AMS) and 14 d (batch with NaClO3) after treatment in parallel with the untreated control plants. The target parameter was the fresh weight, as commonly used in the weed science literature (Nalewaja & Matysiak, 1991; Alister et al., 2005). The bio-efficacy of herbicides is often manifested as the desiccation of some or all parts of a plant, which is not reflected by the dry weight. Concurrently, the necrotic area of the treated leaves was evaluated. The leaves were clamped for flattening, and photographs were taken using a Samsung Pro 815 digital camera (Seoul, South Korea), with the following parameters: F stop, f/3.2; exposure time, 1/10 s; ISO speed, ISO50; focal length, 18 mm; no flash and maintenance of constant lighting. The necrotic area was measured and related to the total leaf area using ImageJ software (v. 1.41; Abramoff et al., 2004).

Expt 3

Stomata-free cuticles from tomato fruits (Solanum lycopersicum L.) were isolated and prepared using a procedure described elsewhere (Hunsche & Noga, 2008). To test the dynamics of the salt solutions on hydrophobic leaf surfaces, the isolated tomato cuticles were examined with an ESEM (XL 30 FEI-Philips, Eindhoven, the Netherlands). A Peltier device attached to the microscope stage allowed the fine control of sample temperature. The gas pressure in the ESEM chamber was regulated by introducing water vapour, allowing RH control. Pure salts of NaCl and NaClO3 were ground to heterogeneous particles with approximate diameters between 1 and 50 μm, and dispersed on the isolated tomato cuticles. Using a coupled energy-dispersive X-ray microanalysis (EDX) system (EDAX Inc., Mahwah, NJ, USA) and Genesis 4000 software, characteristic X-ray spectra were collected and displayed. The samples were cooled in the ESEM to 7°C or 5°C to allow the RH to increase to high values with small amounts of water vapour. RH was calculated from the temperature measured directly below the cuticle, and the bulk vapour pressure was measured within the ESEM; this measurement was relatively accurate because the thermal capacity of the cuticles was low. The first wetting cycle usually started at 400 Pa; the humidity was then increased in steps of 13 Pa. The next step was triggered after all visual changes had ceased, which usually took between 30 s and 1 min. When most of the salt was deliquescent, but before the entire surface was covered by water, the humidity was again decreased slowly. With this procedure, a complete wetting/drying cycle took between 30 and 50 min.

The contact angles of 1-μl droplets of water and of several concentrated salt solutions (AMS, NaCl, NaClO3, 1–7 M) on the cuticles were measured by a goniometer (DSA 30E; Krüss GmbH, Hamburg, Germany). The surface tension of the solutions was determined by the same instrument, using the pendant drop method. To assess the influence of earlier salt deposition on the hydrophobicity of cuticles on a macroscopic scale, 1 M NaCl or 1 M NaClO3 droplets were applied to a cuticle until it was almost completely covered by the solution. Cuticles were then brought into an oven at 60°C to let the water evaporate, after which the cuticle was covered by salt crystals. Using the DSA 30E, a 1-μl droplet of distilled water was applied to the surface. As the water immediately moved to the sides, no contact angle could be measured. Videos were recorded to document the droplet behaviour. The application of the water droplet was repeated after several minutes.


In Expt 1, the means and standard errors were calculated, and a Student's t-test was applied. In Expt 2, the statistical tests were conducted in single factorial design. A Student's t-test was used to compare the effects on the abaxially and adaxially treated plants in cases of normal distributions. In cases of non-normally distributed data, a nonparametric Wilcoxon rank test was applied; this was especially necessary for the evaluation of necrotic areas. The data were tested for normality using the Kolmogorov–Smirnov test, and the homogeneity of variance was determined using the Levene test. ANOVA was conducted if appropriate, and the means were separated using the Duncan test at  0.05. For data that did not satisfy the Levene test, the Kruskal–Wallis test was conducted, and treatment average ranks were separated using the stepwise step-down multiple comparisons method (Campbell & Skillings, 1985) at  0.05. The true means of the data instead of ranked means are presented. All statistical analyses were conducted using IBM SPSS Statistics 20 (Armonk, NY, USA). Within the presented bar graphs, the results of the statistical analyses are indicated by letters; treatments not marked by the same letter are significantly different.


Expt 1

The photosynthesis of abaxially treated leaves was reduced by NaClO3, but no difference was found when droplets were applied to the adaxial sides of the leaves (Fig. 1a,b). Some of the abaxially treated leaves began to develop necrosis in areas in which droplets had been applied, but adaxially treated leaves remained visibly unaffected. Transpiration was not affected by NaClO3 treatment (Fig. 1c,d).

Figure 1.

Gas exchange of apple (Malus domestica) leaves before and after a single application of NaClO3 droplets (a–d), as well as before and after six applications of NaCl droplets (e–h), to either the astomatous adaxial (white bars) or the stomatous abaxial (grey bars) side. Black bars: control. (a, b) Net photosynthesis (Pn) before and 2 d after NaClO3 treatment; (c, d) transpiration (E) before and 2 d after NaClO3 treatment; (e, f) net photosynthesis before the first and 2 d after the last NaCl treatment; (g, h) transpiration before the first and 2 d after the last NaCl treatment. Error bars + SE, n = 5 replicate plants (a–d), n = 10 replicate plants (e–h). Treatments not marked by the same letter are statistically different.

Similarly, the photosynthesis of apple leaves was reduced after abaxial, but not after adaxial, treatment with NaCl droplets (Fig. 1e,f). In contrast with the NaClO3 treatment, this effect was not observed after the first treatment, but only following the sixth treatment of repeated application of droplets. After the sixth treatment, transpiration of the abaxially treated leaves was lower than that of the adaxially treated leaves, although not significantly lower than that of the control (Fig. 1g,h).

Expt 2

The necrosis of a leaf pair was generally higher for the abaxial treatment than for the respective adaxial treatment; this phenomenon was independent of the salt and of additional treatments with Gly or surfactants (Fig. 2a,b). Means were always higher for abaxial treatments than adaxial treatments, and these differences were significant for all treatments without the organosilicone BT (Fig. 2a,b, −Gly−BT, +Gly−BT). For treatments including BT, five of 12 comparisons showed significantly stronger damage in the abaxial relative to the adaxial treatment (Fig. 2a,b, −Gly+BT, +Gly+BT). In cases in which no additional substances were added, AMS caused slight necrosis only in the abaxial treatment (Fig. 2a, −Gly−BT), whereas NaClO3 alone caused increasing necrosis with increasing concentration on abaxial and adaxial treatment. BT increased damage at adaxial 150 mM NaClO3 treatment, whereas no difference was noted between the abaxial NaClO3 treatments with or without BT (Fig. 2b, −Gly−BT, −Gly+BT). When AMS was applied adaxially together with BT, damage was produced at 150 mM AMS, and increasing damage with increasing concentration was found for abaxial application (Fig. 2a, −Gly+BT). When AMS was applied with Gly, no significant necrosis was found for adaxial AMS application (Fig. 2a, +Gly−BT, +Gly+BT). However, the abaxial application of 25 mM AMS, with and without BT, caused greater damage relative to the respective AMS treatments without Gly. Gly alone also caused necrosis when applied abaxially, but not when applied adaxially (Fig. 2a,b, +Gly−BT, 0 mM). When NaClO3 and Gly were applied together, but without BT, no necrosis was found for adaxial application (Fig. 2b, +Gly−BT); for abaxial application, damage increased at 25 mM, but not at 150 mM, relative to the treatment without Gly (Fig. 2b, −Gly−BT, +Gly−BT). When NaClO3, Gly and BT were applied, there were no differences relative to treatments without BT for abaxial application. However, necrosis increased strongly at 150 mM adaxial application (+Gly+BT, +Gly−BT).

Figure 2.

Necrotic areas of apple (Malus domestica) leaves after treatment with different concentrations of salt solution ((a) ammonium sulfate, (b) sodium chlorate), glyphosate (± Gly) and an organosilicone surfactant (‘Break-Thru S233′; ± BT). White bars, adaxial treatment; grey bars, abaxial treatment. Error bars + SE,= 7 replicate plants. Upper-case letters refer to adaxial treatments; lower-case letters refer to abaxial treatments. The upper box compares treatments with control (= no necrosis). The four lower boxes (with letters marked by apostrophes) show within-group comparisons of increasing salt treatments. Treatments not marked by the same letter are statistically different ( 0.05; Kruskal–Wallis, stepwise step-down multiple comparison). The result of the comparison for each pair of adaxial/abaxial treatments is given by the horizontal lines with the respective following levels: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001 (Wilcoxon rank test).

Chlorophyll fluorescence measurements did not suggest strong damage to PSII, as long as no Gly was added (Fig. 3a,b). Two adaxial treatments (150 mM AMS, −Gly−BT and 0 mM AMS, −Gly+BT) showed significantly lower values than the corresponding abaxial treatments, but both were > 0.7 and not lower than the control (Fig. 3a). Five adaxial treatments (0 and 25 mM NaClO3, −Gly+BT, and 0, 25 and 150 mM NaClO3, +Gly−BT) were lower than the control, but were > 0.7, which is a ‘healthy’ value (Fig. 3b). For AMS, a significant depression of chlorophyll fluorescence relative to the control was observed only when Gly and BT were applied concomitantly, namely at 150 mM adaxial application and at 25 mM abaxial application (Fig. 3a, +Gly+BT). Gly with BT also caused a significant effect relative to the control when applied abaxially (0 mM, +Gly+BT). This treatment and 25 mM AMS also caused stronger damage than the same adaxial application (Fig. 3a, +Gly+BT).

Figure 3.

Fv/Fm maximum efficiency of photosystem II of dark-adapted apple (Malus domestica) leaves after treatment with different concentrations of salt solution ((a) ammonium sulfate, (b) sodium chlorate), glyphosate (± Gly) and an organosilicone surfactant (‘Break-Thru S233′; ± BT). Black bar, control; white bars, adaxial treatment; grey bars, abaxial treatment. Error bars + SE, n = 8 replicate plants. Upper-case letters refer to adaxial treatments; lower-case letters refer to abaxial treatments. The upper box compares treatments with control. The four lower boxes (with letters marked by apostrophes) show within-group comparisons of increasing salt treatments. Treatments not marked by the same letter are statistically different. The result of the comparison for each pair of adaxial/abaxial treatments is given by the horizontal lines with the respective following levels: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

For NaClO3, a significant depression of chlorophyll fluorescence was found for all abaxial and all adaxial applications together with Gly, with or without BT (Fig. 3b, +Gly−BT, +Gly+BT). Gly also had an effect when applied on its own. Adaxial applications without BT remained at ‘healthy’ values of > 0.7 and showed less intense effects than the respective abaxial treatments (Fig. 3b, +Gly−BT), whereas the strongest adaxial effect was found with BT at 0 mM NaClO3 (+Gly+BT).

Biomass reduction was only observed if the treatment included Gly, and no effects were found for adaxial treatments with AMS (Fig. 4a,b). A significant biomass reduction by abaxial treatments was only found for Gly without any other substances (Fig. 4a, +Gly−BT, 0M) and for Gly with BT at 25 and 150 mM AMS (+Gly+BT). Without BT, increasing AMS decreased the effect of Gly for abaxial treatments (+Gly−BT). Four of six Gly/AMS treatments showed stronger effects for abaxial relative to adaxial application (Fig. 4a, +Gly−BT, +Gly+BT).

Figure 4.

Biomass of apple (Malus domestica) seedlings after treatment with different concentrations of salt solution ((a) ammonium sulfate, (b) sodium chlorate), glyphosate (± Gly) and an organosilicone surfactant (‘Break-Thru S233′; ± BT). Black bar, control; white bars, adaxial treatment; grey bars, abaxial treatment. Error bars + SE,= 10 (a), = 7 (b) replicate plants. Upper-case letters refer to adaxial treatments; lower-case letters refer to abaxial treatments. The upper box compares treatments with control. The four lower boxes (with letters marked by apostrophes) show within-group comparisons of increasing salt treatments. Treatments not marked by the same letter are statistically different. The result of the comparison for each pair of adaxial/abaxial treatments is given by the horizontal lines with the respective following levels: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

For NaClO3 without surfactant, a significant reduction in biomass relative to the control was observed for abaxial Gly application at 25 mM (Fig. 4b, +Gly−BT). If Gly was applied together with surfactant (+Gly+BT), both abaxial and adaxial treatments led to significant reductions in biomass at 25 and 150 mM NaClO3. Significant differences between adaxial and abaxial applications of Gly were observed for 0 and 150 mM NaClO3 without surfactant (+Gly−BT), but not for NaClO3 with surfactant (Fig. 4b, +Gly+BT).

Expt 3

The mean surface tension of the water droplets was 72.8 mN m−1 at 20°C, which is consistent with the data in the literature. The surface tension values of concentrated salt solutions were lower, but remained between 65 and 71 mN m−1 for AMS and NaCl until saturation was almost reached at 5 M (Table 1). For NaClO3, surface tension decreased continuously until, near to saturation at 7 M, c. 50 mN m−1 was reached. The mean contact angle of the water droplets on the tomato cuticle was 81.7°, which was similar to the values observed for 1 M solutions of NaCl and NaClO3; AMS showed 101.7° at 1 M (Table 1). The contact angles increased to values between 100° and 105° with increasing concentration for all three salts.

Table 1. Surface tension γ (mN m−1) of solutions and contact angles (CAs) (°) on tomato (Solanum lycopersicum) cuticles of water and several concentrated solutions. Measured values for distilled water were 72.8 ± mN m−1 and 81.7° ± 3.4°
 1 M3 M5 M7 M
γ (mN m−1)CA (°)γ (mN m−1)CA (°)γ (mN m−1)CA (°)γ (mN m−1)CA (°)
  1. Mean values ± SD, = 10 (γ); n = 20 (CA).

  2. AMS, ammonium sulfate.

  3. a

    Saturated solution.

AMS67.5 ± 0.4101.7 ± 0.765.6 ± 0.9104.2 ± 1.569.5 ± 0.3  104.7 ± 2.0an.a.n.a.
NaCl70.8 ± 0.6 84.1 ± 2.968.8 ± 0.2101.9 ± 1.867.9 ± 0.3103.1 ± 1.6n.a.n.a.
NaClO369.1 ± 0.684.9 ± 4.461.7 ± 0.4  95.3 ± 2.755.8 ± 0.5  99.5 ± 3.150.7 ± 0.4103.1 ± 1.9

NaCl crystals on tomato cuticles became deliquescent when RH increased above the deliquescence point (Fig. 5a,b and Supporting Information Video S1). After completion of each of the first four wetting/drying cycles, few changes were observed compared with the initial situation, although some formation of larger crystals was obvious. During the drying phase of the fifth cycle, the salt started to form dendritic patterns that originated from the base of the large crystals (Fig. 5c–h). This phenomenon occurred c. 4 h after the start of the measurements. NaClO3 crystals on the tomato cuticles became deliquescent for the first time above 80% RH (Fig. 6a,b, Video S2). NaClO3 showed a marked hysteresis, that is, delayed crystallization, with efflorescence beginning at c. 60% RH (Fig. 6c,d). The second wetting began earlier, at 75% RH, and gave the impression of liquid layers covering large areas in regions far from the crystals (Fig. 6e). During the next drying cycle, the dendritic growth of crystals into almost all regions of the cuticular surface could be observed (Fig. 6f). The formation of dendritic crystals from large NaClO3 crystals was also visible from the EDX images of Na and Cl (Fig. 6 g,h). In this case, the formation of a dendritic structure covering large areas began as early as 70 min after the beginning of the experiment.

Figure 5.

NaCl salt on isolated tomato (Solanum lycopersicum) cuticles (scale in a). (a) Initial state of ground dry particles, 55% RH. (b) First wetting, 81% RH. (c–h) Fifth drying, c. 4 h after start, detailing at (c) 76%, (d) 72%, (e) 69%, (f) 69%, (g) 67% and (h) 66% RH, respectively. An animated series of photographs of all five cycles is available in Supporting Information Video S1.

Figure 6.

(a–f) Environmental scanning electron microscope (ESEM) pictures of NaClO3 salt on isolated tomato (Solanum lycopersicum) cuticles (scale in a). (a) Initial state of ground dry particles, 66% RH; (b) first wetting, 81% RH; (c) first drying, 60% RH; (d) 54% RH; (e) second wetting, 72% RH; (f) second drying, 60% RH, c. 70 min after starting the first cycle. (g) Low-vacuum SEM picture of NaClO3 on tomato cuticle after two wetting/drying cycles. (h) Same site and scale as (g) showing an overlay of a characteristic X-ray map of Na and Cl. An animated series of ESEM photographs of both cycles is available in Supporting Information Video S2.

Droplets were applied to cuticles that had previously been covered by 1 M salt solutions. The droplets did not show finite contact angles, but they immediately flattened and moved sidewards as they touched the surface. After repeated application, droplets eventually stayed in one place, forming low contact angles with the surface (Videos S3, S4).


The experiments focused on ‘leaf surface particles’ undergoing physical changes of deliquescence and efflorescence, which creates a definition problem. In the following sections, the expression ‘leaf surface particles’ refers to the particles in both crystalline and dissolved states.

The study used physiological damage as an indicator of successful foliar uptake. All of the applied substances (AMS, NaCl, NaClO3 and Gly) damaged the plants, although the mechanism and extent of the damage varied. There are no reports describing the phytotoxic effects of the organosilicone ‘Break-Thru S233′ (BT) when applied as a single substance at the recommended concentration, even when using highly sensitive methods, such as the kinetic chlorophyll fluorescence parameters as tested in our group (M. Hunsche, unpublished).

Differences between adaxial and abaxial treatments

Approximately one-half of the treatments (32 of 70, including photosynthesis, necrosis, PSII efficiency and biomass reduction) produced significant differences between the paired abaxial and adaxial treatments. The effects were typically stronger for abaxial applications (see next paragraph). There were two exceptions, in which the Fv/Fm maximum PSII efficiency was lower for the adaxial than for the abaxial application. However, in both cases (Fig. 3a, 150 mM AMS, −Gly−BT; 0 mM AMS, −Gly+BT), the abaxial values were ‘healthy’ (> 0.7), and were not decreased relative to the control. In one of the cases, the treatment was identical to the NaClO3 series (−Gly+BT, 0 mM NaClO3, Fig. 3b), in which no differences were observed. Although the zero levels (0 mM) of AMS (Figs 2a, 3a, 4a) and NaClO3 (Figs 2b, 3b, 4b) were identical treatments, some differences between them were observed. These differences cannot fully be explained, but they may have been caused partly by the higher stomatal density of the first batch of apple seedlings compared with the second.

A total of 30 treatments showed stronger effects for the abaxial relative to the same adaxial treatment, including a reduction in net photosynthesis by NaCl and NaClO3 (Fig. 1), the formation of necrosis by AMS and NaClO3 as either single salts or in combination with Gly and/or BT (Fig. 2) and strong depression of Fv/Fm maximum PSII efficiency and reduction in biomass by Gly activity (in different combinations with AMS, NaClO3 and BT; Figs 3, 4). Abaxially applied NaClO3 alone decreased photosynthesis in Expt 1 and caused necrosis in Expt 2. Thus, the substance overcame the hydrophobicity of the leaves and was taken up without the use of surfactants. NaCl only affected photosynthesis after six treatments, which may have been a result of less efficient uptake. Repeated treatment most probably decreased the resistance of the leaf. Gly alone also caused greater necrosis and stronger biomass reduction in abaxial relative to adaxial treatments, as recognizable by the zero concentrations of the salts (Figs 2, 4). The adaxial Gly effects were generally increased with BT, but usually remained smaller than the abaxial effects. The abaxial Gly effects were also visible in the absence of BT treatment. Liu (2004b) did not observe stomatal uptake of Gly without organosilicone surfactants, but focused on stomatal Gly uptake via infiltration. Liu took radiolabelled samples, 10 min after application, because stomatal infiltration in the presence of organosilicones is a mass flow phenomenon and occurs rapidly (Stock & Holloway, 1993). Our experiments focused on the much slower diffusional transfer along continuous water connections.

In summary, the observed effects on plant physiological parameters support the hypothesis that abaxial uptake is stronger than adaxial uptake in apple leaves, consistent with results from a recent study on lychee leaves that used boron isotopes as a tracer (Will et al., 2012). The difference in uptake is most probably related to the different structures of the two leaf sides. We did not observe any hairs on the adaxial side, but there were single hairs on the abaxial side of the leaves. Because these hairs were nonglandular, it is unlikely that they contributed to the uptake of the solutions. Although the presence or absence of stomata is the biggest structural difference between the adaxial and the abaxial sides, we cannot completely rule out cuticular heterogeneities as a major cause for the larger abaxial uptake. However, the cuticular pathway for the transport of ions (especially anions) is restricted to aqueous pores of 0.3–2.5 nm (Schönherr, 2006; Eichert & Goldbach, 2008), whereas size restrictions or electrostatic hindrances for diffusion along stomatal thin water films after HAS establishment are absent, as long as stomata are not completely closed (Burkhardt, 2010). Based on the paradigm of impossible stomatal transport, higher cuticular permeabilities in peristomatal regions surrounding the guard cells have been postulated, with further increases in permeability during times of open stomata (Schönherr & Bukovac, 1978). However, experimental evidence for such heterogeneity has been lacking. Given the recent proof of an existing stomatal pathway for nanoparticles (Eichert et al., 2008), stomatal transport is the most probable main explanation for the observed differences between adaxial and abaxial uptake.

Surfactants and adjuvants

AMS and other ammonium salts have repeatedly been reported to increase the uptake of Gly and other herbicides at concentrations similar to those used here (Liu, 2004a,b; Wang & Liu, 2007). An increasing effect with increasing AMS concentration was confirmed for adaxial application together with BT with regard to Fv/Fm maximum PSII efficiency (Fig. 3a, +Gly+BT). However, the opposite effect on the same parameter, that is, decreasing impact with increasing AMS concentration, was found for abaxial treatment with the same substances (Fig. 3a, +Gly+BT). Decreasing impact with increasing AMS concentration was also found for the effect of abaxial Gly treatment without BT on biomass (Fig. 4a, +Gly−BT). Therefore, the hypothesis that Gly uptake is improved by AMS was supported for adaxial uptake, but must be rejected for abaxial uptake. The hypothesis of improved Gly uptake by NaClO3 is only weakly supported by the increased necrosis found for adaxial and abaxial NaClO3 treatment together with BT (Fig. 2b, +Gly+BT). Necrosis is a local parameter for the treated leaf and was possibly affected by NaClO3 itself. Stronger Gly uptake, however, should result in significant differences of the systemic parameters PSII efficiency and biomass which was not observed. BT addition did not cause any significant changes for any of the abaxial applications of NaClO3, and very similar mean values were generally maintained. This was found regardless of whether Gly was involved and for all three physiological parameters (Figs 2b, 3b, 4b). Such behaviour was not found for AMS or for adaxial treatments by NaClO3. Because the organosilicone BT is normally required to enable the stomatal uptake of other substances, this observation can be taken as an indication of the ready stomatal uptake of NaClO3.

Water surface tension and the Hofmeister series

The abaxial application of AMS restricted the uptake of Gly, which supports the view of sulfate as a kosmotropic ion causing high surface tension that excludes stomatal uptake, as in the case of pure water hypothesized by Schönherr & Bukovac (1972). The physiological effects caused by NaClO3 were stronger than those of AMS, as can be judged on the basis of the abaxial treatments without Gly and without BT leading to significant necrosis relative to the control (Fig. 2a,b, −Gly−BT). This indicates preferential stomatal uptake of NaClO3 relative to AMS as a result of the reported difference in surface tension, and thus supports our hypothesis that the position of an anion in the Hofmeister series is important for stomatal uptake. The chaotropic action of the chlorate anion is underlined by the failure to respond to BT addition in Expt 2, and by the stronger effects on photosynthesis in Expt 1 when compared with NaCl. The physiological reactions, however, might also result from a stronger specific toxicity of the chlorate anion relative to the chloride or sulfate anion. Thus, even with equal amounts of the three salts entering the leaf, NaClO3 could have had stronger effects; this possibility cannot be ruled out. Chlorate is a known substrate of nitrate reductase (NR), which reduces chlorate to the toxic chlorite (Labrie et al., 1991). High levels of NaCl within the leaves can cause necrosis as a result of osmotic disturbance, but neither the metabolic pathways through which salt stress damages plants nor the adaptive components of salt tolerance are fully understood (Ashraf & Harris, 2004). Less is known about the specific toxicity of sulfate anions in the leaf. However, 150 mM AMS caused significant necrosis when it was applied abaxially with BT and without Gly (Fig. 2a, −Gly+BT). Differentiation between uptake and specific toxicity is generally difficult, but the forest damage caused by polluted sea salt (Bussotti & Ferretti, 1998) shows that sea spray tolerance may be a result of the restricted uptake rather than the low specific toxicity of chloride. More efficient uptake rather than specific toxicity might also explain the suitability of certain substances for chemical desiccation. Suitable substances usually contain chaotropic anions or acids, whereas those with kosmotropic anions, such as AMS or NaCl, cannot be used. Acids also decrease water surface tension (Jungwirth, 2009).

Cuticular hydrophobicity as affected by initial wetting and extension

The progressive expansion of NaCl and NaClO3 was observed within the ESEM, supporting the hypothesis that cuticular hydrophobicity can be decreased by the influence of hygroscopic particles. The macroscopic observations with the goniometer, showing the influence of previously deposited and evaporated salt particles, underline this view (Videos S3, S4). In the ESEM between 75% and 80% RH, the condensation of water vapour to NaCl and NaClO3 crystals became visible and turned the salts into highly concentrated solutions. These microscopic droplets necessarily covered larger areas with increasing RH. When RH decreased again, the area shrank and contact angles seemed to decrease, showing the commonly described hysteresis between proceeding and receding contact angles (Herminghaus et al., 2008). Hysteresis, that is, an uneven reaction to moistening and drying conditions, was also reflected in the difference between the deliquescence point and the efflorescence point. It was especially strong for NaClO3 between first deliquescence at c. 80% RH and subsequent efflorescence at c. 60% RH. The size of the initial particles should not have strongly influenced the process, as equilibration times are very rapid. The equilibration times of condensation and evaporation for airborne aerosols are c. 0.1 s for 10-μm particles and 0.0001 s for 0.1-μm particles (Pilinis et al., 1989), and rates for particles on leaf surfaces are likely to be similar.

Increases in RH within the ESEM showed stronger spatial extension for NaClO3 relative to NaCl. Contact angles could not be measured within the ESEM, but appeared generally low, especially for the second NaClO3 cycle. At a salt-dependent ‘switch’ point (second cycle for NaClO3 and fifth cycle for NaCl), the appearance of the system changed markedly as a result of the dendritic growth behaviour of both salts. Dendritic growth of normally cubic NaCl crystals has repeatedly been described to be caused by different inorganic or organic additives or by bacteria (Ploss, 1964; Sarig & Tartakovsky, 1975; Lopez-Cortes et al., 1994). The presence of bacteria might have affected our results, but we could not identify them within the ESEM. The bacterial populations might have been too small for detection as a result of the growing conditions of the seedlings, or possibly because bacteria or biofilms are not easily detected within the ESEM (cf. Monier & Lindow, 2004). Generally, the formation of stable spatial patterns, such as dendrites, is a common feature in nonlinear, nonequilibrium systems. The points at which the corners and edges stick out grow most rapidly, attracting ions from the centres of the faces, which causes these regions to grow much more slowly (Rodriguez-Navarro et al., 2002).

EDX images revealed the eventual existence of traces of salt outside the main crystals after efflorescence (Fig. 6g,h). Such traces and residues of evaporated water are probably the preferred condensation sites for the next cycle of increasing humidity. In this way, continuous salt bridges may develop, which become solutions once the DRH of the salt is exceeded. Humidity fluctuations, especially around the deliquescence point, may thus cause considerable leaf surface particle dynamics. These dynamics decrease the initial hydrophobicity of the cuticle in a gradual and iterative process.

Expt 3 represented the simplest system in this study, because there was no stomatal influence, and so the RH on the surface of the isolated cuticles was dependent only on the ESEM atmosphere. On real leaves, the RH at the leaf surface is determined not only by atmospheric RH, but also by leaf transpiration; at a specific location on the leaf surface, RH depends on the distance from the nearest stomatal pore (Roth-Nebelsick, 2007). At the deliquescence point, which was c. 75% RH for both salts, the solution is almost saturated, and the surface tension near to saturation was c. 68 mN m−1 for NaCl and c. 50 nN m−1 for NaClO3. This difference in surface tension might also account for differences in expansion between the two salts. RH fluctuations causing efflorescence and deliquescence are probably the most efficient drivers of expansion, but changes in RH that cause the droplet to increase in size will transfer ions within the droplets or films. These transfer processes will not always be reversible with evaporation, causing, for example, ‘line-pinning’ (Deegan et al., 1997), or the formation of crystals at the outmost limit of the droplet before shrinking again. Again, these microsites of crystallization are early condensation sites when the humidity rises. In this way, a net movement, leading to continuous expansion on the cuticles, may occur. Although the physicochemistry behind these processes is complex, and other factors, such as trichomes, surface waxes, bacteria or fungi, might complicate it even further, the basic process of expansion can be noted for both salts.

The processes observed in the ESEM might eventually also lead to HAS. Efflorescence and deliquescence, as observed in the ESEM, may occur in the outer parts of the pore, and may be especially promoted by stomatal opening and closing. The cuticle lining the stomatal walls (Pesacreta & Hasenstein, 1999) experiences a high RH and crystallization of salts may possibly not take place (but compare Roth-Nebelsick, 2007). However, the condensation of water vapour will be facilitated by the presence of ions. The movement of a thin (< 100 nm) liquid solution front along the surface of the cuticle may then be driven by fluctuations of RH, which may be caused by, for example, lightflecks changing the leaf temperature (Farquhar & Raschke, 1978; Tyree & Yianoulis, 1980; Vesala, 1998; Roth-Nebelsick, 2007). Reduced surface tension will support this (relative to crystallization/deliquescence) slower process, and might be partly responsible for the strong effects observed for NaClO3 in Expts 1 and 2. Eventually, a continuous connection may form between the surface and the microsites of evaporation within the leaf, where the hydraulic system of the plant normally ends. Once this continuous layer is formed and HAS is completed, diffusion gradients between the leaf surface and the apoplast will cause ions from the surface to move inside, whereas others present in the cell walls and apoplastic solution might diffuse outwards and contribute to leaching (cf. Tukey, 1970).


This study further supports the relevance of stomatal uptake of substances, as indicated by the different behaviours of adaxial/abaxial leaf sides. Several reasons that might enable the stomatal uptake of water and solutes have been suggested, including bacteria (Eichert et al., 2008), fungal hyphae (Burgess & Dawson, 2004) and mucilage within the stomata (Zimmermann et al., 2007), but these have not been systematically assessed. The present results show that hygroscopic leaf surface particles have the potential to develop HAS. Aerosol deposition is usually not considered to be a relevant environmental factor with direct interaction with plants, but it is ubiquitous and cumulative. Agricultural as well as natural aerosols will usually decrease the innate hydrophobicity of plants, and it is shown here that they may also reduce the water surface tension, with ion-specific effects according to the Hofmeister series. This represents a new hypothesis regarding stomatal uptake. Compared with the previous paradigm, this differentiated view broadens the perspective of plant–atmosphere interactions in natural, managed and polluted plant systems.


The authors thank Knut Wichterich for his support in the ESEM and SEM-EDX analysis. This work was supported by a stipend of Theodor-Brinkmann Graduate School of the Agricultural Faculty at Bonn University, and a research grant (BU 1099/7-1) from the Deutsche Forschungsgemeinschaft (DFG).