We investigated how leaf hydraulic conductance (Kleaf) of loblolly pine trees is influenced by soil nitrogen amendment (N) in stands subjected to ambient or elevated CO2 concentrations (CO2a and CO2e, respectively). We also examined how Kleaf varies with changes in reference leaf water potential (Ψleaf-ref) and stomatal conductance (gs-ref) calculated at vapour pressure deficit, D of 1 kPa. We detected significant reductions in Kleaf caused by N and CO2e, but neither treatment affected pre-dawn or midday Ψleaf. We also detected a significant CO2e-induced reduction in gs-ref and Ψleaf-ref. Among treatments, the sensitivity of Kleaf to Ψleaf was directly related to a reference Kleaf (Kleaf-ref computed at Ψleaf-ref). This liquid-phase response was reflected in a similar gas-phase response, with gs sensitivity to D proportional to gs-ref. Because leaves represented a substantial component of the whole-tree conductance, reduction in Kleaf under CO2e affected whole-tree water use by inducing a decline in gs-ref. The consequences of the acclimation of leaves to the treatments were: (1) trees growing under CO2e controlled morning leaf water status less than CO2a trees resulting in a higher diurnal loss of Kleaf; (2) the effect of CO2e on gs-ref was manifested only during times of high soil moisture.
The magnitude of plant response to elevated atmospheric CO2 concentration (CO2e) often depends on the availability of other resources, such as nutrients and water (Monje & Bugbee 1998; Oren et al. 2001; Hyvönen et al. 2007). For example, although leaf level photosynthesis is generally stimulated under CO2e (Medlyn et al. 1999; Bernacchi et al. 2003), gas exchange over diurnal and repeated drying cycles may be often dominated by the nitrogen availability and the hydraulic regulation of the stomata (Katul, Leuning & Oren 2003). Understanding the effects of CO2e on gas exchange and on tree water relations over a range of nutrient and water availabilities is therefore critical to our ability to predict forest productivity as the climate changes (Wullschleger, Tschaplinski & Norby 2002). However, at present, the interaction effects between CO2, nitrogen (N), and water availabilities on plant hydraulics are poorly understood.
In isohydric species, stomata conservatively regulate plant water status by controlling the rate of water loss to the atmosphere such that it matches the capacity of the soil–plant hydraulic system to supply water to leaves. Should stomata fail to sense and respond to a lower capacity of the soil-plant system to supply water, xylem would embolize rapidly, increasing the risk of hydraulic dysfunction and dehydration of leaves (Maseda & Fernández 2006). However, stomatal regulation of transpiration rate imposes limits on photosynthesis. Thus, water availability is one of the most important factors limiting productivity and has likely been an important selective regime influencing the evolution of plant physiology (Schulze et al. 1987). Despite decades of research on the physiology of stomata, the specific mechanisms that permit coordination of stomatal conductance (gs) with plant water balance and hydraulic properties remain elusive (Meinzer 2002; Buckley 2005). Nevertheless, there does seem to be a general agreement that stomata sense leaf water potential (Ψleaf) somewhere within the leaf so that similarity exists in the response of both gs and leaf xylem cavitation to decreasing Ψleaf (Brodribb & Cochard 2009; Domec et al. 2009).
The diversity in form and longevity of leaves translates to large differences in water transport capacity and, thus, in gas exchange (Sack & Holbrook 2006). Water transport capacity can be quantified in terms of leaf hydraulic conductance (Kleaf, Aasamaa, Sober & Rahi 2001; Domec et al. 2009). Consistent with earlier works documenting coordination of gs with overall plant hydraulic conductance (Meinzer et al. 1995; Cochard et al. 2002), Kleaf and its relationship to stomatal control has recently been investigated, revealing that maximum gs is very sensitive to Kleaf within and among species (Brodribb et al. 2003; Woodruff et al. 2007). Leaves comprise the terminal portion of the liquid water transport pathway and their xylem is under greater tension than in stems, yet leaf xylem appears to be generally more vulnerable to embolism than that of the stems. This low resistance to embolism results in large decreases of Kleaf on a regular basis, even under non-extreme environmental conditions (Brodribb & Holbrook 2003; Bucci et al. 2003). Thus, although the hydraulic system of leaves represent less than 5% of the hydraulic pathway it constitutes a substantial (30–80%) and variable part of the resistance to water flow through plants (Sack & Holbrook 2006; Domec et al. 2009). These observations imply that the large diurnal declines of Kleaf may be an inherent component of the stomatal regulatory system acting as a signal rather than a catastrophe to be avoided.
Environmental changes that increase the availability of resources (either above- or below-ground) result in the long-term acclimation of a less conductive (per unit leaf area) hydraulic system because of anatomical acclimation (Mencuccini 2003). For example, loblolly pine trees growing under CO2e and N-fertilization exhibit some differences in wood properties such as shorter tracheids and lower wood density (Oren et al. 2001; McCarthy et al. 2006a). In addition, when soil nutrient availability is low, plants allocate more carbon to roots compared with leaves to enhance nutrient uptake (Gerbauer, Reynolds & Strain 1996; Palmroth et al. 2006). On the other hand, when nutrient limitation is relieved by fertilization, the pattern of biomass allocation may change, favouring the expansion of the leaf surface area to enhance CO2 capture at the expense of below-ground plant parts (King, Thomas & Strain 1997; McCarthy et al. 2006b). The effect of nutrient limitation on carbon partitioning might impact leaf hydraulic architecture and, consequently, leaf water relations. For example, increase in canopy leaf area with increasing nutrient availability and CO2e (McCarthy et al. 2007) should cause Kleaf to decrease if not accompanied by a proportional increase in sapwood area and root area (Ewers, Oren & Sperry 2000; Addington et al. 2006). This reduction in Kleaf may then impose a series of constraints on the water economy of the plant, with ultimate impacts on the carbon economy (Sperry et al. 2002; Bucci et al. 2006).
Under CO2e, gs in most species decreases (e.g. Medlyn et al. 2001), reducing transpiration per unit leaf area. Coniferous species generally have stomata that are less responsive to CO2e than broad-leaved species (Pataki, Oren & Tissue 1998; Ellsworth 1999; Maier, Palmroth & Ward 2008). However, it has also been proposed that gs might decrease under CO2e only in longer-term (>1 year) experiments, and that the seasonal timing of data collection was important (Medlyn et al. 2001; Wang et al. 2005). The magnitude of the response of stomata to CO2e also depends on the sensitivity of gs to soil moisture and vapour pressure deficit (Ainsworth & Rogers 2007). In loblolly pine trees, it has been shown that gs under CO2e is similar to that under ambient CO2 concentration (CO2a) when subjected to water stress, but lower without moisture limitation (Murthy et al. 1996). Therefore seasonal monitoring of stomatal response to CO2e may be necessary to understand stomatal response under varying soil moisture and evapourative demand.
Here we focused on the possible effects of CO2e and N-fertilization on leaf water transport traits and their consequence for long distance water transport and drought tolerance of Pinus taeda L. (loblolly pine) trees, species very sensitive to low soil moisture (Noormets et al. 2009). We specifically investigated the effects on gs regulation and possible compensatory adjustments. The 2007 severe summer drought in the Atlantic Southeast of the United States provided an opportunity to study whether: (1) N-fertilization and CO2e lower leaf and whole tree hydraulic conductance; (2) Kleaf and gs show coordinated acclimation to long term N-fertilization and CO2e; and (3) N-fertilization and CO2e result in a reduced sensitivity of gs to declining vapour pressure deficit and soil moisture. It was hypothesized that when compared with unfertilized trees under CO2a, water transport capacity would adjust to nitrogen availability and CO2e, with Kleaf and gs being lower in fertilized trees and under CO2e.
MATERIALS AND METHODS
The treatments have been administered in a loblolly pine plantation established in 1983 on low fertility, acidic clay loam of the Enon series, in the Blackwood Division of Duke University Forest, in Orange County, North Carolina (35°58′N, 79°08′W). The average height of pines was ∼19 m in 2007. Loblolly pine accounts for up to 90% of the basal area (McCarthy et al. 2007). The most prevailing co-dominant species is Liquidambar styraciflua, and the most common sub-canopy species are Acer rubrum, Ulmus alata and Cornus florida. Mean annual temperature is 15.8 °C and mean annual precipitation is 1150 mm, with usually an even distribution throughout the year. Summers are warm and humid with a growing season mean temperature of 22.1 °C. Further details about the site can be found in Oren et al. (1998a).
The experimental site consisted of four plots exposed to ambient CO2 (CO2a) and four plots targeted at +200 µmol mol−1 CO2 (CO2e) above current, with half of each plot fertilized with N (Oren et al. 2001; Schäfer et al. 2002). CO2 enrichment is implemented according to the free-air CO2 enrichment (FACE) protocol throughout the year whenever ambient temperature is above 5 °C and wind speed is below 5 m s−1 (Hendrey et al. 1999). In 1994, two 30-m-diameter plots were established: the FACE prototype plot (Plot 7) and its adjacent untreated reference plot (Plot 8). In 1996, six additional plots (replicated FACE; Plots 1 to 6) were established, three of which received CO2e. In 1998, the prototype plot and its reference plot were halved using a ditch and a barrier, and one-half of each has received annual nitrogen fertilization (N) of 11.2 g N m−2 (Oren et al. 2001); the same design was implemented in the rest of the plots in 2005.
Measurement of leaf hydraulic conductance (Kleaf)
Aiming to ensure long-term integrity of the Duke FACE experiment, only limited sampling is permitted. Thus, one terminal branch per tree was collected from four trees per plot (two trees from each half). Branches were collected at pre-dawn and enclosed in sealed plastic bags to prevent water loss. Kleaf was measured on single fascicles by assessing the rehydration kinetics of needles after detachment and determined as (Brodribb & Holbrook 2003):
where Kleaf (mmol m−2 MPa−1 s−1) is the leaf hydraulic conductance, Cleaf (mmol m−2 MPa−1) is the leaf capacitance, Ψleaf(o) (MPa) is the leaf water potential prior to rehydration, Ψleaf(t) is the leaf water potential after rehydration, and t (s) is the duration of rehydration of needles detached under water from the stem. Cleaf was determined from the slope of relative water content to Ψleaf obtained from pressure–volume curves. All capacitances values reported and used to calculate Kleaf corresponded to Cleaf determined before the water potential at turgor loss point (Ψtp) (Brodribb & Holbrook 2003).Values of Kleaf were corrected for a viscosity of water at a temperature of 20 °C.
Pressure–volume analyses (Tyree & Hammel 1972) were conducted on single fascicles taken on the same trees used to determine Kleaf, between the months of January and February 2007, when soil moisture and pre-dawn water potentials were high. These samples were excised early in the morning prior to significant water loss, sealed in plastic bags with moist paper to prevent desiccation, and then stored in a refrigerator within 1h of excision. Pressure–volume curves were initiated by first determining the fresh weight of the fascicle, and then measuring Ψleaf with a pressure chamber (PMS Instrument Company, Albany, OR). Alternate determinations of fresh weight and Ψleaf were repeated during slow dehydration on the laboratory bench until values of Ψleaf ranging from −4.0 MPa to −5.0 MPa were attained. The inverse of water potential was plotted against relative water content to establish a pressure–volume curve and determine Ψtp. For normalizing Cleaf on a leaf area basis, needle areas were obtained geometrically from dimensions measured using a digital caliper (series 500 Mitutoyo, Aurora, IL, USA) (Rundel & Yoder 1998).
Because differences may exist between Kleaf measured under field condition and Kleaf measured in the laboratory (Tyree et al. 2005; Sack & Holbrook 2006), we also estimated Kleaf based on field measurements on some of the same trees on which stomatal conductance (gs) and transpiration (E; see further discussion) were measured (Supporting Information Fig. S1). Field Kleaf was calculated as Kleaf = ν/νoE/(Ψstem − Ψleaf), where ν and νo are the kinematic viscosities of water at the measured leaf temperature and at 20 °C, respectively; E is the transpiration rate (mmol m−2 s−1) measured with a LI-6400 portable photosynthesis system (Li-Cor Inc., Lincoln, NE, USA), and Ψstem is the stem water potential estimated from non-transpiring covered shoots (Meinzer 2002). Although Kleaf estimated from the dehydration curves and Ψleaf was 12% lower than field Kleaf, the difference was not statistically significant (P = 0.11, paired T-test; Supporting Information Fig. S1).
Field leaf water potential (Ψleaf) and stomatal conductance (gs)
Measurements of Ψleaf were conducted with a pressure chamber, and gs with a LI-6400 portable photosynthesis system (Li-Cor Inc.). Measurements of Ψleaf and gs were conducted on detached fascicles taken from the same shoot simultaneously every 90 min on one non-fertilized and one fertilized tree per ring for a total of 16 trees during each sampling period. Previous studies at the site has shown that there were no differences between excised and attached needle gas exchange when measurements were restricted to less than 15 min after excision (Maier et al. 2008). Vapour pressure deficit (D) and average CO2 concentration, inside the chamber followed ambient conditions, and photosynthetically active radiation were matched with the average conditions over a 15 min period prior the measurement. Diurnal measurements commenced prior to dawn and continued until 1500 h on 22 and 23 May, 24 and 25 July, and between 19 and 23 September 2007. These dates encompassed large differences in climatic and soil conditions allowing us to study a wide range of liquid- and gas-phase values.
In two of the plots where leaf area data were available for each treatment (rings 7 and 8), mean crown canopy conductances (Gs) derived from basal sap flow were used to provide an independent assessment of stomatal control of transpiration. We used leaf area data measured in 2005 (McCarthy et al. 2007), which is thought to accurately represent leaf area of 2007 (McCarthy, personal communication). Sap flow was measured at breast height in at least five trees per treatment with 20 mm heat dissipation sensors installed at three depths in the sapwood. Sensor operation at the FACE site is detailed in Schäfer et al. (2002). The sensor signal was converted to sap flux density (Js in g m−2 s−1) according to Granier (1987) and accounted for the effects of non-zero night-time fluxes (Kim, Oren & Hinckley 2008; Oishi, Oren & Stoy 2008). Sap flux density was scaled and converted to a tree-scale average transpiration per unit leaf area (E, in mmol m−2 s−1; Oren et al. 1998b; Schäfer et al. 2002). Basal Js values were lagged by half an hour in order to take into account water use from capacitance and thus to more closely approximate the relationship between D and transpiration in the crown (Phillips et al. 1997). The half-hour lag time was derived from the observed time lag of Js with respect to D (Ewers & Oren 2000; Chuang et al. 2006). The sap-flux based Gs was calculated from E and D, using the simplification of the inversion of Penman–Monteith model (Ewers & Oren 2000). The simplified calculation was permitted because in all treatments D was close to the leaf-to-air vapour pressure deficit because of the high boundary-layer conductance. Indeed, given that >90% of the daytime mean wind velocity was >0.7 m s−1, and that leaf dimension never exceeded 0.0017 m, we estimated using Jones (1992) that the mean daytime boundary-layer conductance averaged 65 times Gs.
Analysis of the response of liquid and gas conductance to changes in driving force
Stomata of isohydric plants respond to D in a manner consistent with protection of the xylem integrity for water transport. The emergent behaviour is a decreasing gs with increasing D at a rate that is predictable and proportional to gs at low D (Oren et al. 1999). Thus, gs data from each treatment were analysed based on:
where b is gs at D = 1 kPa (hereafter designated as reference stomatal conductance, gs-ref) and m is the sensitivity of gs to D[−dgs/dlnD, in mmol m−2 s−1 (lnkPa)−1]. Based on the stated hydraulic consideration, −dgs/dlnD is proportional to gs-ref with the proportionality averaging ∼0.60, and varying predictably depending on the range of D used in the analysis (Oren et al. 1999; Ewers et al. 2007; Kim et al. 2008; Ward et al. 2008).
Similarly, we evaluated whether the sensitivity of xylem conductivity to Ψleaf is related to Kleaf by generating a slope of the reduction in Kleaf versus the natural logarithm of Ψleaf[dKleaf/dlnΨleaf in mmol m−2 s−1 (lnMPa)−1] (Ewers et al. 2000). To analyse the effect of Kleaf on whole tree hydraulic conductance (Kt), Kt was calculated from the slope of the relationship between E and Ψleaf (Loustau, Domec & Bosc 1998).
The effects of CO2e and N on Kleaf, Cleaf, Ψleaf, Ψtp and gs-ref were tested through analysis of variance (anova) based on a split-plot design. CO2 concentration and N were the main and split-plot effects, respectively, and individual plots were used as replicates. Measurements made on multiple dates were analysed by repeated measure anova. Statistical analyses were performed using SAS (version 9.1, Cary, NC, USA) and curve fits were performed using Sigmaplot (version 9.0, SPSS Inc., San Rafael, CA, USA).
Leaf hydraulic conductance (Kleaf) determined in the laboratory declined exponentially with declining Ψleaf (Fig. 1; R2 ranging from 0.74 to 0.89; P < 0.01). We obtained the treatment means of maximum Kleaf and the water potentials at which 50% of maximum Kleaf was lost (Ψ50) from curves fitted to data from each tree. CO2e and N significantly decreased maximum Kleaf by 21% and 13%, respectively (Table 1). No treatment effects were observed in Ψ50.
Table 1. Maximum leaf hydraulic conductance (max. Kleaf, mmol m−2 s−1 MPa−1), water potential that induces 50% loss of Kleaf (Ψ50, MPa); water potential at the turgor loss point (Ψtp, MPa), osmotic potential at full hydration (Π, MPa), leaf capacitance on a volume basis (Cleaf-RWC, %RWC MPa−1) and leaf capacitance on a leaf area basis (Cleaf, mmol m−2 MPa−1) for foliage samples obtained in the upper crowns of loblolly pine trees, growing under ambient carbon dioxide concentration, elevated and/or N-fertilized conditions
CO2 × N
Analysis of variance (anova) probability values for carbon dioxide concentration (CO2) and N-fertilization treatment (N) are also shown (the probability level P < 0.15 was considered to indicate a trend).
ns, not significant
6.38 ± 0.48
5.95 ± 0.51
5.48 ± 0.36
4.63 ± 0.27
−0.91 ± 0.014
−0.92 ± 0.17
−1.01 ± 0.19
−0.73 ± 0.11
−2.22 ± 0.02
−2.27 ± 0.03
−2.34 ± 0.03
−2.40 ± 0.03
0.96 ± 0.06
1.13 ± 0.07
1.13 ± 0.05
1.21 ± 0.08
10.1 ± 0.4
9.4 ± 0.7
8.5 ± 0.4
7.7 ± 0.6
771 ± 45
644 ± 59
641 ± 53
544 ± 35
Although CO2e decreased the water potential at the turgor loss point (Ψtp) by 0.12 MPa and Cleaf by 15%, no treatment effects were observed in the osmotic potential at full hydration, although a trend was discernable (Table 1). There was no CO2e × N interaction effects on any parameter measured in the laboratory (Table 1). Among the treatments, the maximum Kleaf and Cleaf decreased linearly with decreasing Ψtp (Fig. 2). Extrapolating the regression lines to zero Kleaf and zero Cleaf generated similar values of Ψtp (−2.99 MPa in Fig. 2a, and −3.08 MPa in Fig. 2b, respectively).
In 2007, climate along the eastern seaboard of the United States was characterized by a long summer drought with only traces of precipitation from July through September (Fig. 3). Because the rooting zone is very shallow (∼35 cm; Oren et al. 1998a), tree water uptake during a summer without significant rain events causes soil moisture to decline quickly. In 2007, extractable moisture from the upper 30 cm was nearly exhausted by early August, and remained so until sizable rain events resumed in late October (Fig. 3). The seasonal pre-dawn Ψleaf decreased by 1.0 MPa from its highest in winter to its lowest following two dry months (<35 mm), averaging −1.31 MPa between May and October (Table 2). The minimum midday Ψleaf, which averaged −2.34 MPa throughout the season (Table 2, Fig. 3), never fell below −2.6 MPa. At a given sampling date neither leaf water potential variable differed between treatments (P > 0.21). However, there was a date effect on Ψleaf (Table 3) because from May to September, pre-dawn and midday Ψleaf decreased by −0.33 MPa and −0.18 MPa, respectively.
Table 2. Mean seasonal (between May and October) leaf water potential (Ψleaf, MPa) at midday (minimum) and at pre-dawn, mean seasonal leaf water potential at reference stomatal conductance (Ψleaf-ref = Ψleaf at gs-ref), daily percent loss of leaf hydraulic conductance at minimum Ψleaf (PLKleaf), daily percent loss of leaf hydraulic conductance at Ψleaf-ref (PLKleaf-ref) and mean seasonal leaf time constant (τ, min) for foliage samples obtained in the upper crowns of loblolly pine trees, growing under CO2a (ambient), CO2e (elevated) and/or N-fertilized conditions
−1.32 ± 0.04
−1.32 ± 0.05
−1.33 ± 0.02
−1.27 ± 0.03
−2.33 ± 0.05
−2.26 ± 0.08
−2.39 ± 0.05
−2.36 ± 0.06
−1.41 ± 0.09
−1.50 ± 0.07
−1.53 ± 0.07
−1.59 ± 0.04
70 ± 4
71 ± 3
72 ± 3
77 ± 5
24 ± 2
33 ± 3
34 ± 1
42 ± 4
τ = Cleaf/Kleaf
7.0 ± 0.7
8.2 ± 1.6
9.9 ± 1.15
16.5 ± 1.3
Table 3. Analysis of variance probability values for carbon dioxide concentration (CO2), N-fertilization treatment (N) and date of measurements (date) on pre-dawn and minimum (midday) leaf water potential (Ψleaf), reference stomatal conductance (gs-ref = gs at D = 1 kPa), leaf water potential at reference stomatal conductance (Ψleaf-ref = Ψleaf at gs-ref), daily percent loss of leaf hydraulic conductance at minimum leaf water potential (PLKleaf), daily percent loss of leaf hydraulic conductance at Ψleaf-ref (PLKleaf-ref), and the leaf time constant (τ)
The probability level P < 0.15 was considered to indicate a trend.
ns, not significant; d.f., degrees of freedom.
CO2 × N
CO2 × date
N × date
CO2 × N × date
Diurnal measurements showed a typical pattern of gs increasing in the early morning once sunlight reached the foliage and then declining from late morning (Fig. 4). Similar trends were observed in both sap flow-based crown canopy conductance (Gs) and porometry-based gs. At the end of May, CO2e decreased the daily maximum gs by 31% (P = 0.018, Fig. 4) but, because in July and September CO2e had no effect on gs (P > 0.55), CO2e had no significant effect on gs (P = 0.19) when assessed over all measurement dates. There was no CO2e × N interaction (P = 0.85) on gs. In this study N did not significantly affect gs (P = 0.38). There was no consistent relationship between gs and Ψleaf. Unlike the peaking pattern described for gs, Ψleaf declined continuously during the day (Fig. 4).
Compared with trees growing under CO2a, trees growing under CO2e tended to have lower Kleaf at high Ψleaf and lower gs at low D and showed reduced hydraulic sensitivity to Ψleaf and reduced gs sensitivity to D (Fig. 5a,b). Using the diurnal relationships between Ψleaf and D, we were able to calculate Ψleaf at gs-ref taken as the Ψleaf at D = 1 kPa (hereafter, Ψleaf-ref). Similarly, Kleaf-ref (Kleaf at D = 1 kPa) was calculated from Ψleaf-ref and the relationships between Kleaf and Ψleaf (Fig. 1). There was a date effect on Ψleaf-ref (Table 3), generated by a −0.31 MPa drop from May to September. There was a marginal decrease in the Ψleaf-ref under CO2e (Table 3). Between treatments, the sensitivity of Kleaf to Ψleaf increased linearly with Kleaf-ref (Fig. 5c). We also tested whether the coefficients from Equation 2, used on gs in all measurement days, conformed to the theoretical expectations of the response of gs to D (Oren et al. 1999). In all treatments gs showed an expected sensitivity to D that was proportional to gs-ref. The slope was equal to 0.64, similar to the general slope of 0.60 (P = 0.13; Fig. 5d).
With rapidly decreasing pre-dawn Ψleaf and soil moisture over the growing season, a date effect was apparent with a sharp decrease in gs-ref (Table 3; Fig. 6). There was no N effect on gs-ref (Table 3). The decline in gs-ref associated with a decline in soil moisture was more pronounced (P = 0.02, slope analysis) in the CO2a plots (50%) than in the CO2e plots (33%).
Values of field Kleaf were estimated using the relationships between Ψleaf and Kleaf established from laboratory measurements (Fig. 1) and Ψleaf measured in the field (Fig. 4). The percent loss of needle hydraulic conductance at the lowest measured Ψleaf (PLKleaf) and at Ψleaf-ref (PLKleaf-ref) were calculated by comparing Kleaf at pre-dawn to either Kleaf at midday or to Kleaf-ref, respectively. Averaged across all days, PLKleaf-ref increased by ∼10% under CO2e, whereas PLKleaf at minimum Ψleaf increased marginally by ∼5% (Tables 2 & 3). N marginally increased PLKleaf-ref (∼8%), and there was no CO2e × N interaction on PLKleaf or on PLKleaf-ref. Based on these field Kleaf and the calculated capacitance (Table 1), we estimated the time constant (τ = Cleaf/Kleaf) of water flow in the leaf. The time constant represents the time required for Ψleaf to reach 63% of its steady state value after a step change in transpiration. Mean τ varied from 7 min in non-fertilized trees growing under CO2a to 16 min in fertilized trees growing under CO2e (Tables 2 & 3) and there was a significant CO2e × N interaction on τ (Table 3).
The gs-ref obtained from the four treatments in the three measurement periods increased linearly and proportionally with Kleaf-ref (Fig. 7a). Similarly, gs-ref increased with treatment and seasonally based increases in whole tree hydraulic conductance (Kt, calculated from the relationship between leaf transpiration and Ψleaf; Fig. 7b), but the reduction in gs-ref was less than proportional to the reduction in Kt (i.e. changes in gs-ref were more sensitive to changes in Kleaf-ref). Between May and September, Kt declined by 56% in trees growing under CO2a, and by 41% in trees growing under CO2e. There was a linear relationship between leaf hydraulic resistance (1/Kleaf-ref) and whole tree hydraulic resistance (1/Kt) (Fig. 7c). The contribution of 1/Kleaf-ref to 1/Kt decreased with 1/Kleaf-ref, which was affected by treatments and changed seasonally (Fig. 7c). Therefore, Kleaf-ref exerted a greater constraint on whole-plant water transport under CO2a than under CO2e, and mostly in May, when Kleaf-ref was high (or 1/Kleaf-ref was low), than in July or September. As a consequence, the contribution of the woody parts (root-to-branch hydraulic resistance) to the whole tree hydraulic resistance increased from ∼50% in May to ∼70% in July–September (P = 0.007).
Across treatments, Ψleaf-ref was higher (less negative) but strongly correlated with Ψtp (P < 0.001; Fig. 8), indicating that stomata tended to begin closing before the leaf reached its turgor loss point. The minimum measured field Ψleaf on the other hand were similar to Ψtp (P = 0.21).
There have been no studies on the effects of CO2e and N availability on leaf hydraulics and their downstream effect on gas exchange. Our study revealed that CO2e and N had a significant effect on maximum Kleaf, and that both treatment-induced seasonal variation in Kleaf-ref translated to similar variation in gs-ref and, thus, gas exchange. The similarity of trends in liquid- and gas-phase conductances implies that functional adjustments under CO2e and N contributed to homeostasis in the operation of the hydraulic regulatory systems that were driven by Kleaf-ref and the point of turgor loss. The results of this study show that, in loblolly pine, it required a structural change of the hydraulic pathway to produce stomatal closure under CO2e. This contrasts with the current view that CO2e directly causes stomatal closure (Ainsworth & Rogers 2007).
Effect of elevated CO2 and N-fertilization on Kleaf, Cleaf and Ψtp
As shown in a study on needle ‘vulnerability curves’ of Pseudotsuga menziesii (Woodruff et al. 2007), Kleaf was highly sensitive to Ψleaf (Fig. 1). However, compared with trees growing under CO2e, trees growing under CO2a showed higher maximum Kleaf and higher sensitivity of Kleaf to Ψleaf. The differences observed in maximum Kleaf among treatments was not associated with differences in Ψ50, in contrast to the pattern observed in other tree organs (Tyree & Zimmermann 2002; Domec et al. 2008). This lack of trade-off between hydraulic efficiency and hydraulic safety points to the more complex role Kleaf plays in water transport, and to the refilling dynamics of leaves. The mechanism explaining the reduction in Kleaf with Ψleaf likely involves cavitation-induced embolism (Johnson et al. 2009), although we can not rule out that needle xylem might have collapsed at lower Ψleaf (Cochard et al. 2004; Brodribb & Holbrook 2005). The decrease in maximum Kleaf, Cleaf, and their lower sensitivities to Ψleaf in trees growing under CO2e and higher soil fertility, as hypothesized (first hypothesis), may be partly related to the development of conducting tissue with different hydraulic characteristics (Centritto et al. 1999). Treatment-induced structural and anatomical adjustments, such as a decrease in the length and diameter of leaf tracheids (Prior et al. 1997; Woodruff, Meinzer & Lachenbruch 2008), and probably in the size of connecting pit membranes between adjacent conduits, would reduce Kleaf. Such structural modifications may also explain the lower Ψtp because thick and stiff cell walls have been shown to reduce turgor loss (Marshall & Dumbroff 1999).
Coordination between liquid- and gas-phase leaf conductances
Following our second hypothesis, treatment-induced reductions in gs-ref were consistent with those in Kleaf-ref (Table 3). However, and in contrast to another study on loblolly pine (Ewers et al. 2000), N did not affect gs-ref, although a tendency for reduced gs was observed in May (Fig. 4). The reason for lack of statistical difference may simply be the power of the experiment, owing to the low number of replicates (n = 4) and the split-plot N factor. Furthermore, stomata of trees growing under CO2e and high soil fertility were less sensitive to D than those of trees growing under CO2a, unfertilized conditions (Fig. 5). These results also showed that the sensitivity of gs to D was related to the variation in gs-ref, meaning that CO2e and N-fertilization did not affect the relative sensitivity of gs to D, a behaviour consistent with an isohydric regulation of water potential (McNaughton & Jarvis 1991; Oren et al. 1999). These results are consistent with the third hypothesis.
Mirroring the stomatal behaviour, the sensitivity of Kleaf to Ψleaf was related to Kleaf-ref, as has been found in roots of the same species (Ewers et al. 2000). Furthermore, the common patterns in the sensitivity of Kleaf to Ψleaf and gs to D were additional evidence of the strong coordination between liquid- and gas- phase conductances (Fig. 5). Because of this coordinated sensitivity between the liquid- and gas-phase fluxes, treatment-induced differences in Kleaf-ref and gs-ref have implications to gas exchange on both the short (diurnal) and long (drying cycle) time scales (Domec et al. 2009).
Diurnally, the reduction in gs with D is proportional to gs-ref, meaning that on an absolute basis, treatments with higher gs-ref experienced a greater loss of CO2 uptake than those with lower gs-ref. The diurnal patterns in Kleaf, Ψleaf and gs provide insight to a possible mechanism involved in the coordination between the liquid- and gas-phase water flows. The patterns indicate that Kleaf does not directly control gs. For example, Kleaf begins to decline immediately after dawn whereas gs is increasing with light, generating a fast decrease in Ψleaf even though D is not very high at this time. This sets up the leaf hydraulic system to a state in which gs will respond quickly to increasing D later in the morning, thus regulating minimum Ψleaf. Indeed, later, gs decreases sharply in response to D whereas Kleaf decreases slowly, mirroring the slow decrease in Ψleaf. The balance between xylem tension and the loss of conductivity has caused several authors to suggest a functional role for cavitation as part of a feedback mechanism linking stomatal regulation to hydraulic conductance and plant water status (Sperry 2000; Meinzer 2002). The diurnal patterns described earlier suggest that hypersensitive leaf xylem serves to protect the integrity of the upstream woody portion of the hydraulic pathway by causing a rapid stomatal closure before water potential drops in the woody xylem, consistent with the hydraulic segmentation hypothesis (Sperry 1986).
During drying cycles, or seasons in areas where soil moisture becomes progressively limiting, the coordinated sensitivity between the liquid- and gas-phase fluxes coupled with treatment-induced differences in Kleaf-ref and gs-ref have different implications to gas exchange. Lower values of gs-ref at the beginning of the growing season, and the lower seasonal reduction in gs-ref in trees growing under CO2e and N were related to lower reductions in Kleaf-ref and Kt (Fig. 7). Reduced Kt has seldom been explicitly reported in studies comparing trees growing under CO2a and CO2e (Bunce & Ziska 1998; Kupper et al. 2006). In our study, treatment-induced differences in Kt were to a great extent explained by differences in Kleaf-ref because the short portion of the pathway through the needles constituted a major part (30–50%) of the whole tree hydraulic resistance to water flow (1/Kt). Seasonally, the effect of Kleaf-ref on Kt decreased in all treatments (Fig. 7c). In May, Kleaf-ref dominated Kt, whereas in July and September, Kt became less limited by Kleaf-ref, probably because of a large decrease in root and stem hydraulic conductances as soil moisture declined (Hacke et al. 2000; Domec et al. 2009).
The treatment-induced differences in the sensitivity of gs to D remained proportional to gs-ref on both diurnal and seasonal time scales. However, treatment effects on gs-ref changed over the season as the soil dried (Fig. 6). As a result of a greater sensitivity of gs-ref to soil moisture under CO2a, the differences in gs observed early in the season among treatments disappeared by the end of the season even though soil moisture was similar in all treatments. Indicative of a severe soil water depletion, values pre-dawn Ψleaf fell below −1.5 MPa, which has already been reported for this site during a previous dry year (Ellsworth 1999). Although pre-dawn Ψleaf was similar in all treatments during the growing season (Fig. 3), this might not reflect access to the same amount of water as the soil dries (Donovan, Richards & Linton 2003). The lower sensitivity to soil drying of gs-ref under CO2e might reflect deeper penetration of fine roots (King et al. 2001; Pritchard et al. 2008), accessing soil moisture beyond the measurement depth, as reflected in a lesser seasonal decrease in Kt of CO2e trees (Fig. 7b). Regardless of whether CO2e trees had access to more moisture at the end of the year, the differences in gs observed early in the season but not late, might also explain differences among studies on the effect of CO2e on gs.
Consequences of adjustments in turgor loss, Kleaf and time constant on tree water economy
The minimum Ψleaf observed in field measurements corresponded across treatments to the Ψleaf at turgor loss (Fig. 8). Leaves of trees in treatments (ambient and unfertilized) with higher (less negative) Ψtp began regulating stomata at a higher Ψleaf. It is possible that the strong linear correlation across treatment between Ψtp and Ψleaf-ref means that declining Kleaf was related to declining cell turgor in the leaf prior to the onset of cavitation in the needle xylem (Brodribb & Holbrook 2005; Woodruff et al. 2007). As result, the rate of decrease in Ψleaf is greater under CO2a and without fertilization between the time in which stomata begin to respond to D and the time in which Ψleaf reaches its minimum value. Moreover, in trees growing under CO2e, more negative Ψtp allowed stomata to remain open at lower Ψleaf, which decreased the effect of drought on gs-ref. However, lower Kleaf and lower sensitivity to Ψleaf under CO2e induced a steeper drop in Ψleaf at gs-ref relative to the decrease in turgor loss points, and therefore increased the percent loss of needle hydraulic conductance (Table 2).
Another consequence of the differences in Cleaf and maximum Kleaf among treatments was reflected in the time constants (Table 2). The time constant of the needles growing under ambient conditions represented around half the time constant of the soil–trunk–leaf compartment calculated at the same site (Phillips et al. 1997), implying that woody tissue and leafy tissue both exhibit a similar time constant. An important consequence of the increase in time constant in trees growing under CO2e and N is that, under dry conditions, these trees would have less time to restore their equilibrium water content and could experience a greater degree of leaf dehydration.
The results from this study are useful in a number of ways. Firstly, we show that in species such as loblolly pine, CO2e and N-fertilization alter the hydraulic pathway, most likely structurally, affecting the liquid phase transport, thus reducing stomatal conductance. This contrasts with previous studies focusing on a direct stomatal response to CO2e. We also show evidence that the hydraulic changes allow plants growing under CO2e to reduce stomatal conductance less under drought than plants growing under current CO2. These conclusions will inform models employed to predict ecosystem responses to climate change, and biosphere-atmosphere interactions under current and future climates. Secondly, the results show that clear effects of CO2 and N-fertilization treatments on gs can be observed only under certain conditions. This may explain some of the contradictions in published responses. Furthermore, stomatal response to elevated CO2e has been investigated in two types of studies. One type exposes leaves grown under CO2a to CO2e, detecting clear stomatal closure in most species but conifers (Murthy et al. 1996; Saxe, Ellsworth & Heath 1998). The other type grows plants under CO2e and compares their gs with that of plants grown under CO2a. In broadleaved species, studies of this type find similar responses to the first type, but the responses in conifers are more variable, with a tendency for the longer studies to find an effect (Medlyn et al. 2001; Wang et al. 2005). If the response we describe is particularly important in conifers, the length of exposure to CO2e necessary to produce a reduction in gs would depend on the time it takes the species to replace a large proportion of pretreatment foliage with foliage produced under CO2e.
The Duke FACE research site was supported by the United States Department of Energy (DOE) through the Office of Biological and Environmental Research (BER) Terrestrial Carbon Processes (TCP) programme (DE-FG02-95ER62083). Support for S. Palmroth also came from the National Science Foundation (NSF-EAR 0628342).