Foliar water uptake (FWU) is a common water acquisition mechanism for plants inhabiting temperate fog-affected ecosystems, but the prevalence and consequences of this process for the water and carbon balance of tropical cloud forest species are unknown.
We performed a series of experiments under field and glasshouse conditions using a combination of methods (sap flow, fluorescent apoplastic tracers and stable isotopes) to trace fog water movement from foliage to belowground components of Drimys brasiliensis. In addition, we measured leaf water potential, leaf gas exchange, leaf water repellency and growth of plants under contrasting soil water availabilities and fog exposure in glasshouse experiments to evaluate FWU effects on the water and carbon balance of D. brasiliensis saplings.
Fog water diffused directly through leaf cuticles and contributed up to 42% of total foliar water content. FWU caused reversals in sap flow in stems and roots of up to 26% of daily maximum transpiration. Fog water transported through the xylem reached belowground pools and enhanced leaf water potential, photosynthesis, stomatal conductance and growth relative to plants sheltered from fog.
Foliar uptake of fog water is an important water acquisition mechanism that can mitigate the deleterious effects of soil water deficits for D. brasiliensis.
The occurrence of frequent fog events is a defining climatic attribute of tropical montane cloud forests (TMCFs), with multiple and yet poorly understood ecological effects (Bruijnzeel & Veneklaas, 1998). The direct contact of fog water droplets with the surface of stems and leaves causes water to drip to the soil, and this additional precipitation is considered to be a major hydrological input in TMCFs (Bruijnzeel et al., 2011; Cavelier et al., 1996; Holder, 2006). In addition, a substantial proportion of the fog intercepted by tree crowns may be retained by the foliage and subsequently evaporate back to the atmosphere, or potentially be absorbed by leaves. Foliar water uptake (FWU) of fog has been widely reported in several temperate ecosystems (Burgess & Dawson, 2004; Breshears et al., 2008; Limm et al., 2009; Simonin et al., 2009), and, with the exception of a recent study in Costa Rican TMCFs (Goldsmith et al., 2013), the prevalence and ecophysiological consequences of this water acquisition mechanism in TMCF tree species remain largely unexplored.
Leaf wetting events can increase the water status of plants when water is absorbed by leaves (Grammatikopoulos & Manetas, 1994; Yates & Hutley, 1995; Breshears et al., 2008), but may be detrimental to photosynthesis as the presence of a water film over a leaf reduces the diffusion velocity of CO2 by c. 104 times (Smith & McClean, 1989; Brewer & Smith, 1997). As a consequence, a strong selective pressure for high leaf surface hydrophobicity should be expected in plants inhabiting very wet environments (Smith & McClean, 1989; Brewer & Smith, 1997; Feild et al., 1998). However, empirical data have shown that leaf water repellency is, in fact, lower in environments often subject to leaf wetting events (Holder, 2007; Rosado et al., 2010). Furthermore, a recent study in redwood forests has demonstrated that fog interception and FWU can decouple leaf-level gas exchange from the soil and have a positive effect on carbon balance when plants are under soil water deficits (Simonin et al., 2009).
TMCFs occur under a wide range of annual rainfall regimes (from 600 to 4500 mm), and some forests occur under climates with a marked seasonal drought ( Jarvis & Mulligan, 2011). Moreover, during periods when fog is absent, high-altitude environments in tropical regions are considered to be arid environments for plants because of their high potential evapotranspiration caused by the high radiation load and atmospheric demand (Leuschner, 2000). On top of this, climate models suggest that global warming is causing an upward shift of the cloud basis for most of the world's TMCFs, which will impose drier conditions for TMCF species in the near future (Still et al., 1999). Therefore, experimental research on how fog affects plant performance is crucial for our understanding of how TMCF tree species will respond to future drier climates.
In the present study, we assessed the ecological importance of FWU for Drimys brasiliensis Miers. (Winteraceae), an abundant and widely distributed woody species in the Atlantic cloud forests of Brazil. We hypothesized that the foliar uptake of fog water is an important water acquisition mechanism for D. brasiliensis which can improve plant water status and favor the plant's carbon balance. We conducted a suite of experiments using a combination of sap flow, stable isotopes, apoplastic tracers and ecophysiological measurements under field and glasshouse conditions to investigate how fog and FWU affect the water and carbon balance of D. brasiliensis. In addition, our study provides information on the anatomical pathways involved in the process of FWU and novel insights into plant water dynamics. We demonstrate, for the first time, that fog water is not only absorbed by leaves, but also transported through the xylem to plant belowground components (plant roots and, probably, mycorrhizal hyphae and rhizosphere soil) when plants are simultaneously exposed to fog events and dry soil.
Materials and Methods
Field observations were carried out in a cloud forest stand, at Campos do Jordão State Park (CJSP, 22°69′S, 45°52′W), located in the Mantiqueira Range, 15 km east of the town of Campos do Jordão, São Paulo, Brazil. The forest stand is located at an altitude of c. 2000 m, characterized by a low canopy height (maximum of 12 m) and a contrasting floristic assemblage compared with lowland evergreen forests (Bertoncello et al., 2011). Drimys brasiliensis Miers. is an overstory and evergreen tree that is extensively distributed in Atlantic cloud forests (Safford, 1999; Bertoncello et al., 2011) and is the most abundant tree species in our site.
Mean annual rainfall at the site is 1705 mm with a distinct dry period of 3 months ( June–August), in which < 8% of total annual precipitation falls, indicating pronounced seasonality of rainfall (Center for Meteorological and Climate Research Applied to Agriculture, 2013). Dry spells are frequent in the Mantiqueira region and, in some years, can be very long and severe, especially in years under El Niño-Southern Oscillation influence (Safford, 1999). The mean annual temperature is 14.9°C and July is the coldest month (mean of 10.3°C), with minimum temperatures frequently falling to zero (CEPAGRI, 2012). Fog occurs on 65–90% of the days in cloud forests of the Mantiqueira range (Segadas-Vianna & Dau, 1965), as they are under the influence of subtropical, temperate and polar air masses (Safford, 1999). These polar masses are more frequent and intense during the dry season (June–September), and are responsible for intense fog and weak rain events (Safford, 1999).
To investigate how fog affects the direction of water flow in the vascular system of D. brasiliensis, we monitored sap flow in stems and roots of five adult trees (c. 8–12 m height) in the cloud forest stand using the heat ratio method (HRM; Burgess et al., 2001). Reversals in the direction of sap flow during fog events were used as evidence of FWU. The sensors (ICT International, Armidale, NSW, Australia) were installed at a height of c. 60–90 cm in the stems and at c. 10–40 cm below the soil surface in the roots. The sensors emitted heat pulses every 30 min and the measurements were stored in a data logger (model SL5 Smart logger; ICT International). We calculated the vapor pressure deficit (VPD) at the site using data collected by air humidity and temperature sensors (model U23-001; Onset Computer Corporation, Bourne, MA, USA). Leaf wetness and precipitation data were also measured using leaf wetness sensors and pluviometers (models S-LWA-M003 and RG3-M, respectively; Onset Computer Corp.). All these sensors were connected to a data logger (model H21-002; Onset Computer Corp.). We used these micrometeorological data to infer fog events: when VPD was close to zero, leaves were wet and no rain was being collected in the pluviometers, this was considered a period of fog. At the end of the study, we interrupted sap flow in two stems by drilling holes to form a 3-cm-wide cut immediately above and below the probes. We used the values collected by the sensors after the flow was interrupted as our reference zero, so that we could correct the data using the procedures described in Burgess et al. (2001). Based on these data, we established the micrometeorological conditions when zero flow occurred, which allowed us to determine the baseline for the sensors that did not have their flow interrupted. More details on this approach can be found in Burgess & Dawson (2004) and Rosado et al. (2012).
Branch spraying experiment
To evaluate the short-term effects of FWU on leaf water relations, we conducted an experiment by spraying water onto cut branches at CJSP, based on the protocol used by Breshears et al. (2008). Two branches under similar micrometeorological conditions were collected from 12 individuals. One branch of each pair was sprayed with water (SP: spray treatment) and the other was kept dry (CT: control); both were kept in separate dark plastic bags for 1–2 h. Foliar water content (FWC) was determined for 10 mature leaves after the treatments, and leaf water potential (ΨL) was measured on two to three samples before and after the treatment. We calculated the final FWC (%) as: ((weight of fresh leaf − weight of dry leaf)/weight of fresh leaf) × 100. We used a Scholander pressure bomb (Model 1000; PMS, Corvallis, OR, USA) to measure ΨL.
Apoplastic tracer experiment
To evaluate the anatomical pathways involved in foliar uptake, we performed a laboratory experiment exposing leaves of D. brasiliensis to a fluorescent apoplastic tracer solution. For this experiment, we used fresh, mature and detached leaves from D. brasiliensis saplings collected at CJSP and kept in a glasshouse. The cut portion of the leaves was sealed with parafilm and maintained in a dark moist chamber (Mastroberti & Mariath, 2008), where it remained in contact with 100 μl of 1% Lucifer Yellow carbohydrazide dilithium salt aqueous solution (LY; Sigma-Aldrich) for 24 h. This tracer is nontoxic to plant tissues, and can only move through apoplastic pathways (Oparka & Read, 1994). Leaves were then washed in distilled water, carefully dried with filter paper, hand sectioned and prepared for microscopic observation in a 90% glycerol-phosphate buffer (Mastroberti & Mariath, 2008). Sections were observed using epifluorescence (Leica DFC500M-R; Wetzlar, Germany), under intense blue excitation of 450–490 nm with a 515-nm barrier filter (Oparka & Read, 1994). Classical anatomical assays were also conducted to determine the chemical composition of leaf structures and possible hydrophilic pathways in the leaves. Mature leaves collected at CJSP and fixed in 50% ethyl alcohol–formaldehyde–acetic acid (FAA; Johansen, 1940) were embedded in plastic resin (Historesin®; Leica Biosystems, Wetzlar, Germany), and then sectioned. These sections were then stained with Sudan Black, to identify lipophilic cuticular structures (Pearse, 1980), Periodic Acid-Schiff reaction (PAS) to identify hydrophilic polysaccharide compounds, such as mucilages, glycogen, glycolipids and glycoproteins (McManus, 1948), and Red Ruthenium to identify pectins (Johansen, 1940).
We performed a series of four experiments in a glasshouse at the University of Campinas (Campinas, Brazil) from June 2009 to June 2011 to evaluate how fog affects the water and carbon relations, growth, survival and leaf water repellency of D. brasiliensis, and also to trace fog water movement from foliage to belowground components of plants. Plants in the glasshouse were subjected to natural daily and seasonal cycles of solar radiation (with a daily peak of photosynthetically active radiation (PAR) of 470–840 μmolphotons m−2 s−1), natural temperature (8–39°C) and relative humidity (10–100%).
E1. Ecophysiological performance experiment
To assess the effects of regular fog exposure on the ecophysiological performance of D. brasiliensis, we conducted a glasshouse experiment with saplings of D. brasiliensis (20–60 cm in height) collected at CJSP and planted in 34-l pots. One month after being transplanted, plants were randomly assigned to control (well watered), fog and drought treatments. Before the start of the experiment, all plants were watered until the soil reached field capacity. The total number of plants was 24 for the control, 25 for the fog and 24 for the drought treatments. Within this pool of plants, two subsets were assigned for destructive and nondestructive measurements, as described in the sections below. In the fog treatment, we exposed the shoots of all plants to 8 h of artificial fog inside fog chambers made of PVC and plastic. We submitted the plants to nocturnal artificial fog three times per week for 2 months during the spring of 2010 (September–October). Fog was generated by an ultrasonic device (model Waterclear Premium; Soniclear, São Paulo, Brazil), which produces a water aerosol with droplets smaller than 20 μm. The soil in pots was completely sealed with plastic bags and parafilm to prevent fog water from reaching the soil. Watered plants received water on the soil until saturation every day, whereas drought-affected plants were not irrigated during the experiment. We measured the ecophysiological parameters seven times during the 2 months of the experiment, and also monitored environmental variables (VPD and relative water content (RWC)) inside the fog chambers. VPD was calculated using the data collected by air humidity and temperature sensors (model U23-001; Onset Computer Corp.); RWC was measured gravimetrically on three soil samples randomly chosen within each treatment. To avoid excessive removal of soil in some pots, different experimental units were used during the experiment. Soil samples were collected weekly using a PVC cylinder (10 mm) from the surface to the bottom of the pot, so that every sample contained a full depth profile of the soil. All samples were weighed immediately after collection and RWC was calculated as: ((weight of fresh soil − weight of dry soil)/weight of fresh soil) × 100.
We measured leaf water potential at predawn (ΨPD) and at midday (ΨMD) for three to eight plants per treatment, randomly selected from a group of 13 individuals per treatment. This group of 13 plants per treatment was used only for destructive measurements with the Scholander pressure bomb. Several saplings had only three to five leaves, and so we used this sampling method for water potential measurements to minimize drastic reductions in the leaf area of the saplings.
We measured the instantaneous maximum net CO2 assimilation rate (Amax) and stomatal conductance (gs) using an infrared gas analyzer (ADC BioScientific LCpro+; Analytical Development Company, Hoddesdon, Hertfordshire, UK) in mature, fully expanded leaves from a group of five plants for the fog treatment and four plants for the drought and control treatments. Measurements were taken between 08:00 and 09:30 h, the period of greatest photosynthetic activity, as determined by a daily photosynthesis curve for this species performed just before the beginning of the experiment. Leaf chamber PAR was controlled at 900 μmol photons m−2 s−1, which was sufficient to light saturate the species A, according to a light response curve that was obtained before the experiment. VPD, CO2 and the temperature of the IRGA cuvette were kept at ambient levels.
We measured the stem diameter (θst), height (h) and estimated total leaf area at the beginning and end of the experiment on five control plants and six drought and fog plants. None of these plants was submitted to destructive measurements. The stem diameter was measured with a digital caliper (Mitutoyo Sul Americana Ltda, Suzano, Brazil) at the base of the stem, and plant height was measured from the base of the stem to the highest leaf with a metric tape. To calculate the total leaf area, we measured the area of 25 leaves with ImageJ 1.42 and used a linear regression between the measured leaf area (y) and the product between each leaf length and width (x). We used the equation derived from the linear model (y =0.909x + 0.7504; R2 = 0.98) to predict the area of the other leaves and obtained the total leaf area by summing the areas of every leaf in the plant.
To obtain a proportional rate of growth or loss in total leaf area, we divided the final values by the initial values of these variables. Mortality rates of the plants that were not submitted to destructive measurements (12 plants in the fog treatment and 11 plants in the control and drought treatments) were determined at the end of the experiment.
E2. Sap flow and leaf hydrophobicity experiment
To evaluate how fog affects the direction of water flow in the vascular system of D. brasiliensis, we monitored sap flow in stems of 12 saplings (70–140 cm in height) collected at CJSP and planted in 88-l pots, in a second glasshouse experiment. We installed HRM sensors at the base of the stems, and exposed four of each of these plants to the same treatments as described in the previous experiment (control, drought and fog). We calculated the reference zero flow value for these plants using the VPD data collected with air humidity and temperature sensors (model U23-001; Onset Computer Corp.) kept inside the fog chamber. We assumed that the sap flow was zero during nights with low VPD and no leaf wetness (Rosado et al., 2012). Sap flow velocity was calculated according to Burgess et al. (2001). Sap flow data are expressed as the percentage of the maximum value reached by each sensor during the experiment.
In addition, we measured leaf hydrophobicity for three leaves from each plant, before and after the treatments, to assess how fog affects leaf surface wettability. We compared the changes in leaf hydrophobicity of the four plants exposed to fog with the group of plants not exposed to fog (control treatment, N =4). We measured the contact angle of a 5-μl droplet of distilled water on both sides of the leaf surfaces. We fixed leaves on a horizontal flat surface and photographed the droplet resting on each side of the leaf surface. We used the software ImageJ 1.44 to analyze the contact angle of the leaf and the water droplet, using the same principle as described by Aryal & Neuner (2010).
E3. Deuterium labeling experiments
To evaluate the amount of water absorbed by FWU, we exposed the shoots of the saplings to deuterium-enriched fog during the night in a third experiment in the fog chamber. Soil irrigation was suppressed 1 wk before the beginning of the fog treatment. The isotopic composition was expressed in delta notation (δD‰) relative to the V-SMOW standard. Labeled water (c. 668‰ δD) used in the fog treatment was composed of a mixture of tap water (c. −44‰ δD) and water enriched in deuterium oxides (99.8%; from Cambridge Isotope Laboratory, Andover, MA, USA). The isotopic enrichment of fogged leaves compared with control leaves was used as evidence of foliar uptake, as soils and roots were completely isolated from the fog by plastic bags and parafilm. We used the IsoError mixing model to estimate the contribution of fog to leaf water content (Phillips & Gregg, 2001).
Before the beginning of each fog treatment, leaves from control and treatment groups were carefully collected at 17:30 h, washed with tap water, dried with paper towels and kept sealed in vials with parafilm at −20°C. The same procedure was performed the following morning at 07:30 h, after the plants had been exposed to nocturnal fog. Leaf wetness was monitored using a leaf wetness sensor (S-LWA-M003; Onset Computer Corp.) to ensure that the fog chamber reached saturation as soon as nebulization started, to avoid the potential occurrence of water diffusion back to the atmosphere via stomata (Limm et al., 2009).
Because of the magnitude of FWU and sap flow reversals observed in D. brasiliensis and the effect of the fog treatment on soil RWC (in the above-mentioned experiments), we hypothesized that D. brasiliensis might transport a significant amount of water absorbed by the shoots to belowground pools. To test whether fog water was actually being transported belowground, we performed a fourth experiment, exposing six D. brasiliensis saplings to fog water enriched in deuterium (c. 295‰ δD). We then compared the isotopic enrichment of soil water samples (including fine roots) collected from rhizosphere soil in sealed pots after three sequential nocturnal nebulizations, between plants exposed to enriched fog and plants exposed to nonenriched fog (control). We opted to use plants exposed to nonenriched fog as the control in order to impose similar nocturnal micrometeorological conditions for both groups and to minimize the occurrence of night-time transpiration or soil evaporation events that could cause evaporative enrichment of the stable isotope composition of soil water. Tap water was used to generate fog for control plants and to irrigate plants before the experiment. Because soil and roots were isolated from the fog, we considered the isotopic enrichment of soil water (δD) after the nebulizations as evidence that water was being transported belowground. Soil samples from the pot of each plant were collected from three different spots in the pot using a metal pipe of 10 mm in diameter. The samples were collected from the surface to the bottom of the pot (c. 35 cm), so that every sample contained a full depth profile of the soil. The most superficial layer of soil (c. 2–3 cm) was discarded to ensure that the sample was not contaminated by any dripping or subjected to evaporative enrichment. The samples were kept sealed in vials with parafilm and frozen before analysis. Water from leaf and soil samples was extracted by cryogenic distillation at Laboratório de Ecologia Isotópica (CENA/USP, Piracicaba, Brazil). Subsequently, these water samples were analyzed in a DeltaPlus Advantage mass spectrometer (intrinsic error c. −0.5‰ ; Thermo Finningan, San Jose, CA, USA) at CPGeo/USP, São Paulo, Brazil.
Responses were analyzed for the ecophysiological experiment using a repeated-measures mixed-model ANOVA using PROC MIXED (SAS v9.3; SAS Institute, Cary, NC, USA), with treatment and time as fixed factors. We used a one-way ANOVA to compare the effects of treatments on growth data using Systat 11 (Systat Software Inc., Richmond, CA, USA). Data were tested for normality and homogeneity of variance and, when necessary, log (base 10) transformed before the analysis. Bonferroni post-hoc tests were conducted when the effects were significant (P <0.05), to assess the differences between treatments.
Responses for the isotope, branch spraying and leaf wettability experiments were analyzed using t-tests, adjusted for dependent samples and unequal variances when necessary. We used a pairwise Fisher's exact test (with Bonferroni correction for multiple comparisons) to assess the effects of treatments on plant mortality. These tests were performed using R v.2.15.1 (R Development Core Team, 2012).
Foliar water uptake
We demonstrate by several lines of evidence that foliar uptake of fog water is an important water acquisition mechanism for D. brasiliensis. First, we observed that the LY solution diffused directly through leaf cuticles of both surfaces (Fig. 1a–d), moved through apoplastic routes and was retained by parenchymatic cells (Fig. 1c,d). We found that leaves of D. brasiliensis were rich in hydrophilic compounds (Fig. 1f–h), which may have contributed to water absorption and the retention of apoplastic tracers in the mesophyll of this species. Leaf surfaces were rich in polysaccharides (Fig. 1f ), and cell walls of palisade parenchyma closer to the midrib and leaf edges were rich in acidic pectins (Fig. 1g,h).
These hydrophilic compounds probably influenced the low leaf surface hydrophobicity observed. Leaves of D. brasiliensis had a mean contact angle of c. 44° (adaxial surface) and c. 71° (abaxial surface), being considered highly wettable (contact angles < 90°), according to the classification used by Aryal & Neuner (2010). The leaf adaxial surface was significantly more wettable than the abaxial surface (paired t-test: t =−6.879, df = 15, P <0.001; Fig. 2). Leaves of plants exposed to regular fog events showed a substantial decrease in their abaxial surface hydrophobicity, whereas leaves not exposed to fog became less wettable (t-test: t =4.230, df = 6, P =0.005; Fig. 2a). Fog had no effect on the adaxial leaf surface hydrophobicity (t-test: t =−0.285, df = 6, P =0.785; Fig. 2b).
Results from our branch scale experiments in the field and in the glasshouse provided further evidence that water was directly absorbed by D. brasiliensis leaves. Significant amounts of deuterium (D) were traceable in leaves exposed to deuterium-enriched fog: c. 42% of fog water D was found in the leaves within 8 h (Fig. 3a). FWU increased the FWC of sprayed D. brasiliensis leaves by 2.1% (absolute value) in comparison with control nonsprayed branches (t =18.454, P <0.001), and this resulted in an increase in ΨL of 0.39 MPa (Fig. 3b).
Fog internal hydraulic redistribution
We observed reversals of sap flow in stems and roots in every adult plant monitored in the field during fog events (Fig. 4a,b); reversals of sap flow caused by fog happened not only during the night-time, but also during the day (Fig. 4a,b). Similar patterns of sap flow reversals were observed in plants exposed to fog in the glasshouse experiments (Fig. 4c,d). In this experiment, paired plants that were not exposed to fog did not show any sap flow reversals. The magnitude of sap flow reversals was higher in plants in the glasshouse experiment (Fig. 4c), reaching up to 26% of the diurnal sap velocity, compared with 10% in adult plants in the field. Most of the negative sap velocity values observed during the fog events were higher than the resolution limit of the HRM sensors (0.5 cm h−1).
We confirmed that fog water reached plant belowground components (roots and probably the rhizosphere) with our isotopic tracing experiment. δD of soil water collected close to D. brasiliensis roots increased more significantly in plants exposed to deuterium-enriched fog than in control plants (Fig. 5). Fog water contributed an average of 6.8 ± 3.1% (mean ± SE) to the rhizosphere soil water.
Ecophysiological performance experiment
At the beginning of the experiment, there was no difference in ΨPD between the treatment groups and the means of all treatments ranged from −0.16 to −0.21 MPa (Fig. 6a). After 45 d, plants regularly exposed to fog maintained ΨPD similar to control levels, with mean values of −0.12 and −0.17 MPa in the water and fog treatments, respectively. At this time, plants under the drought treatment had a ΨPD significantly more negative than that of the control group (P =0.012), with a mean value of −0.6 MPa. At the end of the experiment (after 59 d), fogged plants still maintained similar ΨPD to control plants. The treatments did not affect ΨMD significantly (Table 1, Fig. 6b).
Table 1. Mixed-model repeated measures analyses of the effects of treatment through time on Drimys brasiliensis ecophysiological and environmental response variables
P values in bold are significant at α = 0.05.
Treatment × time
Gas exchange parameters (Amax and gs) were affected by the treatments and by the interaction between treatments and time (gs was only marginally affected by the interaction between treatment and time at P =0.07; Table 1, Fig. 6c,d). All the treatments started with similar Amax (Fig. 6c,d), with mean values ranging from 4.5 to 5.7 μmol CO2 m−2 s−1, and, at the end of the experiment, plants in the drought treatment had a lower Amax (mean of 0.22 μmol CO2 m−2 s−1) compared with the other treatments (P < 0.05). Amax values for the fog and control groups were not significantly different at the end of the experiment (means of 2.9 and 5.2 μmol CO2 m−2 s−1, respectively). The gs value of plants subjected to the fog treatment was not significantly different from the control (means of 0.07 and 0.12 mol H2O m−2 s−1 in the fog and control treatments, respectively). Towards the end of the experiment, gs values of plants exposed to fog remained similar to those of the control group, with means of 0.04 and 0.07 mol H2O m−2 s−1, respectively. The drought treatment was the only treatment that showed significantly lower gs values at the end of the experiment (P =0.019), beginning the experiment with a mean of 0.10 mol H2O m−2 s−1 and decreasing to 0.01 mol H2O m−2 s−1.
Soil RWC was also affected by the water treatments and by the interactions between treatments and time (Table 1, Fig. 6e). At the beginning of the experiment, RWC was not significantly different among the treatments (P >0.1), and, at the end of the experiment, only RWC of fogged and control treatments remained not significantly different. In the drought treatment, RWC decreased after the start of the imposed drought, and remained lower for that treatment (P <0.001), being 14.6% at the end of the experiment.
Finally, fog positively affected the growth and survival of D. brasiliensis (Fig. 7). Fogged plants had similar stem and height growth to control plants, whereas plants from the drought treatment showed less growth than plants in the other treatments (P <0.05; Fig. 7a). Plants in the drought treatment had completely lost their leaves by the end of the experiment, whereas fogged plants decreased their leaf area (but maintained some leaf area) and control plants increased their leaf area (Fig. 7c). The survival of D. brasiliensis plants under fog and control treatments was not significantly different, with mortalities of 9.1% and 25% in the control and fog treatments, respectively. Plants under drought showed a lower survival rate than that of the control treatment (P =0.022; Fig. 7d), with a mortality of 74% by the end of the experiment.
Foliar water uptake
In this study, we have demonstrated, by several lines of evidence, that foliar uptake of fog water is an important water uptake mechanism that positively affects the water and carbon balance of D. brasiliensis. In addition, our study provides information on the anatomical pathways involved in the process of FWU. Our apoplastic tracer experiment suggested that large portions of the abaxial and adaxial surfaces of leaves were involved in water absorption, instead of a single specialized leaf structure (Fig. 1a–d). Drimys brasiliensis has a mesophyll rich in hydrophilic phenolic compounds, polysaccharides and mucilage cells that have already been suggested to be involved in the foliar uptake and retention of water in leaves of Araucaria angustifolia (Mastroberti & Mariath, 2008). We also observed a high concentration of polysaccharides in the epidermis of both leaf surfaces and the presence of pectins in the palisade parenchyma cell walls – structures that retained the fluorescent salt solution (Fig. 1c,d).
The abaxial surfaces of leaves have a papillose epidermis covered by a thick cuticle and lipophilic stomatal plugs that fill the cavity above the guard cells (Feild et al., 1998; Fig. 1e). The function of these structures was first thought to be related to adaptation to drought and to contribute to an alleged inefficient hydraulic system composed of tracheids (Bailey, 1953). Later, Feild et al. (1998) proposed that the function of stomatal plugs was to repel water lamina formation in leaves of Drimys winteri. However, in contrast with D. winteri, our leaf hydrophobicity results suggest that both sides of D. brasiliensis leaves are highly wettable, and that the hydrophobicity of the abaxial leaf surfaces decreases even more after being exposed to regular fog events (Fig. 2). This reinforces the view that cuticles are complex and dynamic structures that can modify their permeability in response to changing environmental conditions (Schreiber et al., 2001; Schönherr et al., 2005; Rosado et al., 2010). The difference between leaf wettability on the adaxial and abaxial surfaces of D. brasiliensis also suggests that the adaxial surface might contribute more to FWU, whereas the abaxial surface (where most stomata are located), being more hydrophobic, might facilitate gas exchange during and immediately after fog events.
FWU can occur in distinct plant taxa through a variety of leaf mechanisms and structures, such as stomatal aperture (Eichert et al., 2008; Burkhardt et al., 2012), guard cells (Schlegel et al., 2005), trichomes (Schreiber et al., 2001; Schönherr, 2006), hydathodes (Martin & von Willert, 2000) and, possibly, even by stomatal plugs (Westhoff et al., 2009). FWU through the cuticle can occur via polar pathways (aqueous pores; Schönherr, 1976; Kerstiens, 2006) and ectodesmata (Franke, 1961; Schönherr, 2006). The existence of multiple ‘water entry pathways’ in leaves of different plant taxa suggests that FWU might be selected for in trees subjected to large soil-to-leaf water potential gradients; (e.g. tall trees) and/or in environments in which leaf wetting events and dry soils are common.
Fog internal hydraulic redistribution
We observed that fog water was not only absorbed by the leaves of D. brasiliensis, but also internally redistributed through the plant (Figs 4, 5). Reversal in stem sap flow during fog events has been reported for two temperate gymnosperm species and for Prosopis tamarugo in the Atacama desert (Sudzuki, 1969; Burgess & Dawson, 2004; Nadezhdina et al., 2010), and, as far as we know, this is the first evidence of fog hydraulic redistribution in stems of a tropical cloud forest species. The observed sap flow reversal caused by FWU, not only in stems, but also in roots (Fig. 4a), reinforces the idea that absorbed fog water can be transported to the soil by the plant. The deuterium enrichment of the soil close to D. brasiliensis roots (Fig. 5) provides further evidence that FWU-absorbed water is probably released into the rhizosphere, and not used solely to refill the plant's internal capacitance, as observed for Sequoia sempervirens (Simonin et al., 2009). However, as we did not sieve the soil samples before the analysis, we cannot disregard the possibility that fine roots were present in our soil samples. Hence, our isotope data do not provide definitive evidence that fog water was actually exuded from the roots to the soil, and further experiments using root-excluding mesh screens would be necessary to confirm this exudation. However, considering that most plants do not have mechanisms that prevent water from flowing out of their roots towards the soil, at least on a short time scale (Caldwell et al., 1998), and given the existence of a water potential gradient sufficiently large to allow water movement from leaves to the roots, it is likely that water moved out of the roots into the soil.
The amount of water redistributed was proportionally higher in the stems of D. brasiliensis when compared with other studies. The reversal rates of sap flow peaked at c. 25% of maximum daily transpiration rates in our glasshouse experiment, whereas it peaked at 5–7% in S. sempervirens (Burgess & Dawson, 2004). This difference in magnitude might even explain why fog decouples S. sempervirens water and carbon relations from soil water deficit (Simonin et al., 2009), whereas, in D. brasiliensis, it slows down the soil drying process (Fig. 6e).
In summary, our findings demonstrate a novel mechanism by which plants use fog water; D. brasiliensis is not only able to absorb water directly through its leaves (Figs 1, 3a), but can also redistribute this water through its xylem (Fig. 4), rehydrating plant tissues (Fig. 3b) and moving water to plant belowground components (Fig. 5). This alternative water acquisition strategy provided enough water to allow D. brasiliensis saplings to keep the soil water content in their pots relatively stable for 2 months without any additional water input (Fig. 6e), and allowed the plants to keep a positive carbon gain and growth at higher levels than the drought treatment (Figs 6, 7). We expect that FWU and internal hydraulic redistribution might happen not only during fog events, but also during any leaf wetting event that does not significantly wet the soil, such as dew, light rain or drizzle.
Drimys brasiliensis is very sensitive to soil drought and significantly reduced gas exchange, ΨPD, growth and survival in response to decreases in soil water content (Figs 6, 7; Table 1). By contrast, fog maintained the plant's hydration (fog contributed as much as 42% of leaf water content) and allowed D. brasiliensis to maintain its ecophysiological performance close to the control treatment during the 2 months of the experiment. The impact of nocturnal nebulizations on the soil water budget of our glasshouse experiment may be explained by three mechanisms: (1) suppression of night-time transpiration during nebulization, which reduces the plant's total water use; (2) fog internal hydraulic redistribution which rehydrates internal tissues (Figs 3b, 4, 5) and might be used for transpiration, reducing the uptake of soil water; and (3) fog hydraulic redistribution was probably a consequence of the transport of fog water to the roots (Figs 4, 5), such that the direct release of water into the soil could also affect soil RWC.
There is some controversy about the consequences of wet leaves on plant performance. We propose that, in some environments, the benefits to plant performance granted by FWU should surpass the performance detriment caused by the reduction in gas exchange when the leaf is wet. Considering that, during most leaf wetting events (fog, drizzle and light rain), the photon flux density is greatly reduced, any reduction in gas exchange in leaves caused by the formation of a water film during a foggy period should not greatly reduce a plant's total carbon balance. Thus, FWU might provide an ecological advantage in environments in which plant performance is limited by water, at least seasonally, such as at our study site and in redwood forests in California, where leaf water uptake is quite common (Limm et al., 2009). In environments that are not limited by water during any season, predictions about high leaf hydrophobicity in foggy environments (Smith & McClean, 1989) should still be the norm.
The ecological benefits granted by FWU are probably not limited to the effects shown in our study (maintenance of gas exchange, growth and survival; Figs 6, 7). Considering that the water absorbed by FWU reaches plant's roots and also probably their rhizosphere, all the effects associated with the moistening of roots by hydraulic redistribution could also be happening here, including the minimization of root embolism and prolongation of root life span (Domec et al., 2004, 2006; Bauerle et al., 2008), benefits to rhizosphere fungal associations (Querejeta et al., 2007) and even increasing nutrient availability (Dawson, 1997; Pang et al., 2013).
As our experiment was performed on young isolated plants under glasshouse conditions, our results on the effects of FWU on plant performance might be hard to extrapolate to adult individuals under field conditions. However, drought can be especially detrimental to younger plants, because of the smaller volume of soil explored by their roots. Therefore, FWU probably plays an important role in seedling establishment and survival during seasonal droughts.
In our study, we have shown that FWU promotes fog water redistribution, a process that plays an important role in plant survival and growth during low rainfall periods in tropical environments. Our results strengthen the soil–plant–atmosphere continuum (SPAC) view proposed by Simonin et al. (2009) that water movement in plants should be seen as a true continuum between all potential water sources. In our case, the establishment of a water potential gradient between wet leaves and dry soil was enough to induce a basipetal sap flow in D. brasiliensis.
Considering that FWU mitigates the deleterious effects of drought in plant performance, such as carbon starvation and hydraulic failure (McDowell et al., 2008), we believe that FWU of fog water might be a key trait explaining the ecological distribution of D. brasiliensis in high-altitude environments that are often subjected to seasonal droughts (Bertoncello et al., 2011). Global climatic models predict an increase in the average height of cloud formation in tropical cloud forests (Still et al., 1999), leading to a drier montane climate with fewer fog events. Atlantic cloud forests may experience a change in their composition and functioning in the future, because of the predicted higher mortality of fog-dependent species. More detailed studies on the relationships between fog, drought and vegetation functioning are necessary to better understand and predict the extent of these changes.
Our study, together with several others that have recently challenged the unidirectional SPAC model (Burgess & Dawson, 2004; Oliveira et al., 2005; Simonin et al., 2009; Nadezhdina et al., 2010), urges new parameterization of ecosystems and global climate models that consider soil water as the only water source used by vegetation for carbon fixation and evapotranspiration. Considering D. brasiliensis as a model tree for tropical cloud forests, and assuming that FWU and fog hydraulic redistribution are common physiological processes in these forests, we can assume that they play an important role in the hydrology and productivity of these ecosystems, just like hydraulic redistribution between soil layers in lowland tropical forests (Ryel et al., 2002; Lee et al., 2005).
The authors gratefully acknowledge the following: financial support by São Paulo Research Foundation (FAPESP) (Grant number 10/17204-0 to R.S.O. and Biota Gradiente Funcional 03/12595-7); National Council for Scientific and Technological Development (CNPq) and Higher Education Co-ordination Agency (CAPES) (scholarships to A.L.L. and C.B.E.); Graduate Program in Ecology and Plant Biology from University of Campinas (UNICAMP), Forestry Institute (COTEC -IF); staff of the CJSP and Umberto Bonini for logistic support; research support and facilities generously offered by the Plant Anatomy and Physiology Laboratories of UNICAMP (Profs. Sandra Guerreiro, Marilia Castro, Paulo Mazzafera and their students) and Federal University of Rio Grande do Sul (UFRGS) (Prof. Jorge Mariath, Alexandra Mastroberti and Carlos Widholzer); Isotope Ecology of Center for Nuclear Energy in Agriculture (CENA) (Prof. Plinio Camargo, Marcelo Moreira, Luiz Martinelli, Geraldo Arruda and Maria Antonia Perez); Leonardo Jorge for assistance with analysis; Stephen Burgess for helping us with the sap flow data; Graham Zemunik and Hans Lambers for providing comments on the manuscript.