Rapid and long-term effects of water deficit on gas exchange and hydraulic conductance of silver birch trees grown under varying atmospheric humidity

Effects of air humidification and rapidly-imposed water deficit

The air humidification caused a decrease of up to 10% in atmospheric water vapour pressure deficit (VPD) during the misting application (Figure 1). ANCOVA revealed that the humidification treatment influenced (P < 0.05) most of the studied characteristics (Table 1). The strongest effects were observed for leaf conductance to water vapour (g_L) and leaf water potential (Ψ_L), whereas leaf temperature (T_L), ratio of intercellular to ambient CO₂ concentrations (C_i/C_a), net photosynthetic rate (A_n) and intrinsic water-use efficiency (IWUE) remained unaffected by the manipulation. The rapidly-induced water deficit, quantified by leaf (Ψ_L) or branch water potential (Ψ_B), had a highly significant (P < 0.001) effect on all studied parameters. Except for leaf hydraulic conductance, K_L, inclusion of Ψ_B into the analysis model resulted in a considerably better fit than inclusion of Ψ_L. An analysis of sensitivity of the physiological parameters to changes in plant water status (dx/dΨ_B), estimated by slopes of the corresponding linear regressions, revealed that almost all variables of trees grown under elevated atmospheric humidity (H treatment) tended to respond more sensitively to water deficit. However, in only two cases the corresponding slopes differed significantly between the treatments (Figure 2): g_L (P < 0.05) and K_L (P < 0.01). In order to compare the g_L and K_L responses to each other, we normalised the absolute values with corresponding means and analysed sensitivity of the normalised g_L and K_L (values of both characteristics below or above 1) to developing water deficit. K_L declined 2.3 times (P < 0.01) and g_L 1.4 times (P < 0.05) faster in humidity-treated trees compared to the control with decreasing Ψ_B.

Net photosynthetic rates were strongly correlated with stomatal conductance (g_S; R² = 0.970, P < 0.001) across a wide range of stomatal openness for both treatments combined (Figure 3A). At first IWUE increased in response to the rapidly-induced water deficit and attained a maximum of >70 μmol mol^-1, corresponding to g_S ~0.06 mol m^-2 s^-1 (Figure 3B). When g_S fell below this value (at Ψ_B < -1.0 MPa), IWUE declined very steeply as A_n decreased more rapidly than g_S. Two characteristics - T_L and C_i/C_a - demonstrated opposite trends with increasing water deficit. None of the characteristics differed significantly among the treatments under severe water deficit (Ψ_L<-1.55 MPa; Table 2).

Long-term effects of water deficit

Long-term water deficit was imposed by reducing soil water availability due to a moderate drought that developed in July (Table 3; Figure 4). Although the misting application decreased transpirational water loss, bulk soil water potential (Ψ_S) in H plots also underwent substantial decline (dropped to -180 kPa) in July. Inclusion of Ψ_S as an index of soil water availability into the analysis models changed the outcome radically: the effect of the humidification treatment became – with one exception – insignificant for all gas exchange and water relations characteristics (Table 4). Only g_L depended simultaneously on the treatment (P = 0.036), rapidly-induced water deficit (Ψ_B; P < 0.001) as well as soil water availability (Ψ_S; P < 0.001). Consequently, the effects of humidification manipulation were to a great extent realized through changes in soil water status. Four characteristics [g_L, E, soil-to-branch hydraulic conductance (K_S-B) and whole-tree hydraulic conductance (K_T)] were 2.1–2.3 times greater in humidified trees than in control trees.

Month	Year
	2008	2009	2010
June	79	152	110
July	64	90	33

In fact, the differences in physiological characteristics between the treatments recorded on intact branches in the morning (Table 2) reflect co-effects of the air humidification and long-term soil water deficit. The responses of g_L and K_L to variation in Ψ_B were analysed also separately for the data obtained before and after cutting branches, and for moister (Ψ_S>-218 kPa) and drier soil conditions (Ψ_S≤-218 kPa). Before cutting, neither of the response slopes differed between the treatments; after cutting, both slopes differed significantly between the treatments (g_L, P < 0.05; K_L, P < 0.01). dg_L/dΨ_B and dK_L/dΨ_B showed no difference within treatments between the different soil moisture ranges.

Liquid versus gaseous phase conductance

Changes in K_L were co-ordinated with those in both stomatal conductance and net photosynthesis, while the relationships were substantially stronger for humidified trees. Specifically, R² in C treatment was 0.264 and 0.293 for g_S and A_n, respectively. In H treatment the respective R² values were 0.583 and 0.601 (for all cases P < 0.001). g_S and A_n were associated considerably more strongly with K_S-B (R² = 0.75-0.85) and K_T (R² = 0.80-0.85; Figure 5). IWUE in intact branches declined with increasing hydraulic capacity: with K_L (R² = 0.204, P < 0.05), K_S-B (R² = 0.356, P < 0.01) as well as K_T (R² = 0.356, P < 0.01). There was no statistical relationship between K_L and IWUE across the whole data sets (i.e., throughout the whole range of water deficit). The reliability of gasometric measurements was proved by an excellent accord among the readings obtained with different instruments: although g_S and total leaf conductance (g_L) were measured on different leaves and under different conditions (controlled versus ambient conditions, respectively), the two characteristics exhibited a near perfect concordance (R² = 0.944 for C trees, R² = 0.901 for H trees, for both P < 0.001).

General responses to water deficit

The rapidly-induced water deficit had highly significant (P < 0.001) effect on all parameters measured at the leaf level (Table 1). Under moderate water deficit (Ψ_L≥-1.55 MPa) leaf conductance to water vapour, transpiration rate and leaf hydraulic conductance were significantly (P < 0.05) higher in trees grown at elevated air humidity than in control trees. These differences are attributable to higher initial values (a result of long-term effects) and probably also to larger branch internal water storage in H treatment under moderate drought, although statistically not proven by the Ψ_B data. Leaf temperature, on the contrary, was higher (P < 0.05) in C trees due to the diminished transpiration. Under severe water deficit (Ψ_L<-1.55 MPa) the treatments showed no difference in any of the characteristics (Table 2).

Two characteristics - g_L and K_L - exhibited a significantly steeper decline with increasing water deficit in H treatment than in the control, indicating higher susceptibility to weather fluctuations of trees grown under increased RH. The observed stomatal responses are primarily associated with impact of rapidly-induced water deficit and obviously driven by hydraulic signals, because dg_L/dΨ_B did not differ between the treatments in intact branches and did not depend on soil water status if the data was analysed separately in subsets. Thus, the effect of soil drying is secondary. Various mechanisms are suggested as signalling cues to initiate or enhance ABA biosynthesis, including hydraulic signals [36]. The priority of hydraulic versus metabolic stimuli is considered fundamentally important in preventing plant desiccation and is maintained in stomatal control through vascular plant phylogeny [37, 38]. However, the apparent change in stomatal sensitivity to branch water status induced by the humidity manipulation could be due to differences in the leaf-borne ABA levels, as previous reports describe that endogenous ABA concentrations in leaves grown for a long time under high humidity are lower than under moderate humidity [11, 12]. Also fast de novo synthesis or conversion of inactive conjugates of ABA [15, 16] in shoot vascular tissues triggered by branch dehydration cannot be dismissed. Although studies on Arabidopsis thaliana provide crucial information on stomatal responses, species-specific differences exist, especially when the plants are exposed to simultaneously changing environmental factors [39].

Thus, our experiment supports the first hypothesis that trees acclimated to higher humidity exhibit greater sensitivity to rapidly-induced water deficit with respect to two functional traits. However, these changes have different consequences on plant water status. The reduction of g_L helps to limit water loss, slows down further Ψ_L falling and prevents runaway xylem embolism. The impact of decreasing K_L is opposite – leaf water supply declines causing Ψ_L to fall. Birch trees showed differential changes in these two fundamental traits due to the experimental manipulation: the humidity-treated trees exhibited substantially faster water deficit-driven reduction in K_L than in g_L if compared to the control. Thus, greater risk of leaf dehydration and xylem dysfunction is probably imposed on the trees grown under higher atmospheric humidity in case of sudden weather extremes, because strict stomatal control over water loss is a crucial factor in preventing water deficit-induced xylem cavitation [13]. Plant hydraulic conductance does not limit stomatal openness under moist weather conditions, but it could become crucial in climate extremes (severe drought, disastrous heat wave), which are scarcely predictable and yet will become more frequent in the future [8]. Among ecosystems, forests are particularly sensitive to climate change, because the long life-span and conservative structure of the water-conducting system of trees do not allow rapid acclimation to environmental fluctuations [3].

The air humidification manipulation affected most of the studied characteristics, but not IWUE (Table 1), unlike the soil humidity manipulation reported by Possen et al. [40]. In some species even long-term soil drought does not affect IWUE if A_n and g_S decrease with equal rates [41]. The inclusion of Ψ_S in the analysis models excluded the treatment effect (Table 4), suggesting that the impact of experimental manipulation in droughty summer (Table 3) is realized largely through changes in soil water status (i.e. due to higher Ψ_S in H treatment). Only leaf conductance to water vapour (g_L) depended simultaneously on the treatment (P = 0.036), rapidly-induced water deficit (P < 0.001) and soil water availability (P < 0.001). However, g_L in H trees demonstrated higher sensitivity to water deficit, i.e. an opposite trend to that observed by Fanourakis et al. [4]. Weak stomatal control could be a consequence of the low transpiration in plants grown continuously under high RH (>85%). The degree of stomatal acclimation depends on both the duration and timing of exposure to high RH during leaf development, while determinative is just a stage of leaf expansion completion [4]. In silver birch, elevated atmospheric humidity had the widest consequences on stomatal regulation, as the effects extended beyond that of soil water availability. This is an important point in view of climate change: Roelfsema and Hedrich [20] argue that stomata will play an essential role in the adaptation of plants to climate change, because of their interrelated roles in CO₂ uptake and release of water. As for g_S, we observed less pronounced response (compare Tables 1 and 4), obviously because of its being measured under artificial conditions (constant irradiance, temperature and air humidity).

Changes in plant hydraulic traits

The air humidity manipulation led to higher soil water availability (Figure 4) in H treatment due to reduced transpirational water loss [5, 7] under low VPD during the misting application (Figure 1). This resulted in higher hydraulic capacity of the trees grown in more humid environment, i.e. a long-term effect (Table 2). This response was observed under the moderate drought in July 2010 (Table 3). By contrast, we did not observe unequivocal shifts in hydraulic traits in the rainy summer of 2009: K_L decreased, while hydraulic conductance of root systems (K_R) and leaf-specific conductivity of stem-wood increased in response to elevated RH [42]. The present study revealed some alleviating effect of elevated RH under moderate drought, and the plant response to increased air humidity seems to differ depending on prevailing weather conditions. Nor can we dismiss increased xylem vulnerability and possible hydraulic dysfunction under unexpected severe drought, although on average the climate will become more humid at high latitudes [2, 3].

The differences in K_S-B and K_T observed on intact trees in 2010 likely ensued from xylem cavitation in response to differential soil drying (i.e. a long-term effect) in the treatments. The differences in K_L resulted from rapidly-imposed water deficit rather than soil water availability, because K_L measured on intact branches showed no significant difference between the treatments (Table 2) and dK_L/dΨ_B was invariant of soil water status. K_S-B demonstrated a greater intertreatment variation compared to K_L - by a factor of 2.3 versus 1.6, respectively. This is attributable to greater susceptibility of root xylem than of shoot xylem to water stress-induced embolism [33, 43]. Domec et al. [44] reported that K_R declines faster than K_L as soil dries. The increasing resistance between soil and trunk has been shown to be the main cause of K_T decline and has also the highest weight in the stomatal control [45]. We cannot exclude also concurrent mechanisms responsible for the differences in the decline of K_S-B versus K_L, such as that associated with contribution of apoplastic versus cell-to-cell route to liquid water transport under water deficit. When transpiration stream is attenuated, plasma membrane AQPs are upregulated, the membrane water permeability increases and transcellular water flux becomes much more significant [46]. One must consider that the soil-to-branch pathway represents predominantly an apoplastic route, while in leaves the contributions of the two routes to the total hydraulic resistance are of the same magnitude [47]. Nevertheless, Johnson et al. [48, 49] measured K_L concurrently with ultrasonic acoustic emissions in dehydrating leaves of several woody species and presented reliable evidence that xylem embolism is a primary factor in dehydration-induced declines in leaf hydraulic conductance. Findings of Nardini et al. [50] highlight the role of regulation of K_L in plant acclimation suggesting that leaf resistance to drought-induced hydraulic dysfunction is a key to plant survival and competition even over limited geographical ranges.

Co-ordination between gas exchange and hydraulic traits

Net photosynthetic rate (A_n) and stomatal conductance (g_S) in silver birch were positively correlated with plant hydraulic characteristics (Figure 5), whereas gas exchange parameters were considerably more strongly associated with K_S-B or K_T than with K_L. This result confirms that maximum g_S and A_n depend on hydraulic conductance of the whole soil-to-leaf pathway (expresses potential capability for leaf water supply) rather than solely on that of the leaf [45, 51, 52].

The rapidly-imposed water deficit affected (P < 0.001) all parameters measured at the leaf level, showing substantially stronger association with Ψ_B than with Ψ_L (Table 1). Thus, the gas exchange and stomatal conductance of silver birch are determined by direct water availability to the leaf, estimated by Ψ_B in the petiole insertion point, rather than by the current leaf water status (Ψ_L) itself. The relationship between gas exchange and Ψ_B is probably mediated by stem hydraulic capacitance, because the internal water storage in trees plays a role in mitigating diurnal fluctuations in plant water status caused by transpirational water losses [14, 53]. So, plants with a great capacity to avoid high stem water deficits during periods of high transpiration tend to have a relatively risky stomatal strategy and maintain higher midday g_S[17]. On the other hand, our results support the idea that stomatal openness is regulated in a way to prevent primarily dysfunction of stem xylem, as proposed by Meinzer et al. [14]. This is likely a general trait for broad-leaved trees, as recently reported for a number of subtropical tree species [17].

Leaf gaseous phase conductance began to decrease simultaneously with K_L in response to the rapidly-imposed water deficit, i.e. with no threshold level in the water potential range experienced in the present study. This result coincides with that obtained on leaves of Quercus, Pinus and Pseudotsuga species [54]. Although the field measurements under uncontrolled conditions did not allow construction of vulnerability curves, our data imply narrow hydraulic safety margin existing in silver birch (the 50% decline of K_L was observed at about -1.2 MPa), a characteristic of angiosperm species [55]. Blackman et al. [56] sampled 20 phylogenetically disparate woody angiosperms and found that the greater the water potential inducing a 50% loss in K_L, the narrower the safety margin. This trait suits well with general life strategy of a fast-growing pioneer species, such as B. pendula. In this context the present result is consistent with our previous findings: stomatal sensitivity of sun leaves of B. pendula to atmospheric VPD (80 mmol m^-2 s^-1 ln(kPa)^-1[57]) exceeds the corresponding mean of angiosperms (73 mmol m^-2 s^-1 ln(kPa)^-1[55]). Contrary to the paradigm that isohydric species avoid cavitation, it has been revealed that relatively isohydric species tend to experience far greater cavitation and refilling of xylem on a daily basis than anisohydric species, the benefit of which is enhanced capacitance for use in transpiration [58].

Silver birch has been reported to be able for efficient acclimation to lack of water, including adjustment of WUE [40]. Thus, the drought developed in Estonia in summer 2010 was not severe enough to induce significant changes in photosynthetic water-use efficiency (IWUE; Table 4). Our earlier studies [57, 59] performed on large birch trees growing in a natural forest stand revealed the opposing height-related trends in IWUE and soil-to-leaf hydraulic conductance (K_T) within tree crowns at sufficient light intensities, suggesting a trade-off between water transport and use efficiencies. The inverse relationships between hydraulic characteristics and IWUE found in this study suggest that the respective trade-off between hydraulic capacity and WUE occurs in silver birch both at the leaf (K_L) and whole-plant levels (K_T). The trade-off reflects co-ordinated adjustment of plant gas exchange and hydraulic system to long-term water deficit, but not a response to rapidly-imposed interference; therefore, the converse relation was discovered only in intact branches. Hence, it is always necessary to consider time scales when analysing trends in plant WUE. Abril and Hanano [19] indicated that WUE in Mediterranean woody species reduces during the day by water stress, but it increases as seasonal drought proceeds.

Study area and environmental variables

The studies were carried out on 5-year-old silver birch (B. pendula) trees in an experimental forest plantation at the Free Air Humidity Manipulation (FAHM) site, situated in Rõka village (58°14′N, 27°17′E, 40–48 m ASL), Eastern Estonia, representing a hemiboreal vegetation zone. The long-term average annual precipitation in the region is 650 mm and the average temperature is 17.0°C in July and -6.7°C in January. The growing season lasts 175–180 days from mid-April to October. The soil is a fertile Endogenic Mollic Planosol (WRB) with an A-horizon thickness of 27 cm. Total nitrogen content is 0.11-0.14%, C/N ratio is 11.4, and pH is 5.7–6.3.

Three sample plots served as control areas (C treatment) and three plots were humidified (H treatment) using the computer-operated FAHM system. The system integrates two different technologies - a misting technique to atomize/vaporise water and a FACE-like technology to mix humidified air inside the plots, enabling relative humidity of the air (RH) to be increased by up to 18% over the ambient level during humidification treatment, depending on the wind speed inside the experimental stand. The humidification was applied in daytime 6 days a week throughout the growing period if ambient RH was <75% and mean wind speed <4 m s^-1. As a long-term average, RH was increased by 7–8%. A detailed description of the FAHM site and technical setup has been presented by Kupper et al. [5]. The manipulation was started in June of 2008; gas exchange and hydraulic measurements were performed on 15 H and 15 C trees in June and July of 2010. Environmental variables measured continuously were air temperature (T_A) and relative humidity (RH) with HMP45A humidity and temperature probes (Vaisala, Helsinki, Finland), precipitation with TR-4 tipping bucket rain gauges (Texas Electronics, Dallas, TX), bulk soil water potential (Ψ_S) with EQ2 equitensiometers (Delta-T Devices, Burwell, UK) at depths of 15 and 30 cm. The readings of the sensors were stored as 10 minute average values with a DL2e data logger (Delta-T Devices).

Gasometric and hydraulic measurements

One sample branch (mean height above the ground 140±9.3 cm for C trees and 138±8.4 cm for H trees) per tree from the middle third of the crown was selected for gasometric and hydraulic measurements. Two branches, one from C and another form H treatment, were sampled simultaneously using two instruments. Net photosynthetic rate (A_n), stomatal conductance to water vapour (g_S) and ratio of intercellular to ambient CO₂ concentrations (C_i/C_a) were measured with a LCpro+ portable photosynthesis system (ADC BioScientific, Hoddesdon, UK) on four or five leaves per branch at a saturating photosynthetic photon flux density (1196 μmol m^-2 s^-1) applying constant CO₂ concentration (C_a = 360 μmol mol^-1), air humidity (water vapour pressure 15 mbar) and temperature (25ºC). Leaf conductance to water vapour (i.e. total gaseous phase conductance, g_L), transpiration rate (E) and leaf temperature (T_L) were measured on six leaves per branch with a LI-1600M steady-state diffusion porometer (Li-Cor, Lincoln, NE) at ambient conditions. Intrinsic water-use efficiency (IWUE) was calculated as the ratio of A_n to g_S[41, 60]. Bulk leaf water potential (Ψ_L) was determined in four detached leaves by the balancing pressure technique using a Scholander-type pressure chamber simultaneously with gas exchange measurements. Xylem water potential of the branches (Ψ_B) was estimated by applying the bagged leaves technique, sampling two leaves per branch at each measurement time, prepared the previous evening. Water potential of the non-transpiring (bagged) leaves, presumed to have equilibrated with the xylem water potential of the branch proximal to the petiole, was taken as an estimate of Ψ_B. The first measurement series was performed on intact branches in the morning immediately before branch cutting. Then the sample branches were cut off and allowed to dehydrate in open-air conditions in order to generate a rapidly-imposed water deficit. The next four measurement series were conducted within ~3 h after cutting. All measurements were done on dry leaves under non-misting conditions: on intact branches in the morning before misting started and after that outside the experimental plots.

where E is the evaporative flux. As E is expressed per unit leaf area, values of K_L have been scaled by leaf area. K_L was standardized for the dynamic viscosity of water at 28ºC. Soil-to-branch (K_S-B) and whole-tree hydraulic conductance (K_T) were calculated analogically based on water potential drops across the corresponding segments (Ψ_S-Ψ_B and Ψ_S-Ψ_L, respectively). K_S-B and K_T were left unstandardized, because of variable temperature along these long transport pathways.

Data analysis

where SS_effect is the sum of squares for given effect and SS_error is the sum of squares for the respective error term.