The influence of species and growing conditions on the 18-O enrichment of leaf water and its impact on ‘effective path length’

Authors


Author for correspondence:
Ansgar Kahmen
Tel: 01 510 6431749
Email: akahmen@berkeley.edu

Summary

  •  The stable oxygen isotope ratio (δ18O) of plant material has been shown to contain essential information on water and carbon fluxes at the plant and ecosystem scales. However, the effective path length (Lm), a parameter introduced to leaf-water models still requires a comprehensive biological characterization to allow interpretation of δ18O values in plant material with confidence.
  •  Here, we tested the variability of Lm across and within three species that developed leaves in environments with different relative humidity. We also tested whether the Lm of fully developed leaves is affected by short-term fluctuations in relative humidity.
  •  We determined that significant differences in Lm exist among Phaseolus vulgaris, Rizinus communis and Helianthus annuus. Within a given species, however, Lm values did not differ significantly among individuals.
  •  These findings indicate that Lm is species specific and a relatively constant parameter and that Lm will not obscure the interpretation of δ18O values in plant material of a given species. We urge caution, however, because values for Lm are derived from fitting leaf-water models to measured values of δ18O, so care must be taken in assigning a ‘cause’ to values of Lm as they likely capture a combination of different biological leaf properties

Introduction

The stable oxygen and hydrogen isotope ratios (δ18O and δD, respectively) of plant materials have been shown to contain essential information for understanding plant and ecosystem water and carbon fluxes (Dawson et al., 2002; Barbour, 2007). Applications of δ18O and δD data range from the reconstruction of paleoclimates and neoclimates using tree rings (Schiegl, 1974Epstein & Yapp, 1977) or plant-derived organic compounds in lake sediments (Sachse et al., 2004, 2006), with the heavy isotope (i.e. 18O) to the analyses of carbon and water fluxes at the global (Farquhar & Lloyd, 1993) and ecosystem scales (Yakir & Wang, 1996; Moreira et al., 1997; Williams et al., 2004). In addition, δ18O and δD data collected at the leaf level can provide critical information on plant ecophysiological responses to environmental variability (Barbour et al., 2000a; Cernusak et al., 2007, 2008; Ripullone et al., 2008) and anthropogenic factors such as air pollution (Grams et al., 2007; Jaggi & Fuhrer, 2007; Bassin et al., 2009).

A primary cause of δ18O and δD variability in plant materials (e.g. water and organic matter) is the evaporative enrichment of leaf water with the heavy isotope (i.e. 18O). With the exception of halophytic (Lin & Sternberg, 1994) and a few extreme-ophylic plants (Ellsworth & Williams, 2007), no fractionation has been observed during the uptake of soil water by plant roots and its subsequent transport through the xylem to the leaves (White et al., 1985; Dawson & Ehleringer, 1991). Once water arrives at the leaf, enrichment through evaporative water loss in 18O can be substantial (Dongmann et al., 1974). Recent improvements in mechanistic leaf water models have significantly advanced our understanding of the physiological and environmental factors that lead to leaf water enrichment in 18O (Flanagan et al., 1991; Farquhar & Lloyd, 1993; Farquhar & Cernusak, 2005; Cuntz et al., 2007; Kahmen et al., 2008). However, even with these theoretical advances, the complexity of interactions among the different factors known to influence the magnitude of 18O enrichment of leaf water has hampered efforts to interpret data obtained from the δ18O of organic matter.

Of particular interest in this regard is the ‘effective path length’ (Lm), because it can substantially complicate the interpretation of oxygen isotope values in plant material (Kahmen et al., 2008). The Lm roughly describes the flow-path of water in the leaf lamina from the veinlets to the sites of evaporation. Despite the importance of Lm, the factors contributing to variation in Lm still require a comprehensive description. Recent efforts to understand biological controls over leaf water have revealed the significance of Lm as a driver for variation in leaf water δ18O across species (Wang et al., 1998; Barbour & Farquhar, 2003; Kahmen et al., 2008). As a result, plant δ18O signals compared across species are not only driven by environmental and ecophysiological parameters, but also by Lm, which makes the interpretation of δ18O signals across different species complicated. Furthermore, it has recently been proposed that Lm could not only vary across, but also within, any given species (Keitel et al., 2006; Barnard et al., 2007; Ferrio et al., in press). If true, Lm could therefore also obscure the environmental or ecophysiological interpretation of δ18O signals within individuals of a given species; this in turn would make using δ18O data in organic matter (e.g. in tree ring or breeding studies) particularly problematic.

Because a comprehensive investigation that addresses the presence and extent of variation in Lm within individuals of a given species is still missing, we specifically designed the investigation presented here to explore whether Lm varied within individuals of a given species in response to different environmental treatments. For three different plant species, we tested if (1) Lm varies as a result of different environmental conditions during leaf development or (2) as a result of short-term environmental fluctuations.

Materials and Methods

To evaluate the effects of different environmental conditions during leaf development on Lm, we grew three different plant species that varied in a number of leaf characteristics: common bean (Phaseolus vulgaris L.), castor bean (Ricinus communis L.) and sunflower (Helianthus annuus L.). Plants were grown from seed to mature, flowering plants in two adjacent glasshouses with contrasting environmental conditions that we refer to here as ‘long-term treatments’. Plants were either grown in ‘wet’ conditions defined by high relative humidity (RH; ranging from 70% to 100%) and high soil water availability, or ‘dry’ conditions defined by low RH (40–65%) and low soil water availability (Fig. 1). Soil water availability in the long-term wet treatment was kept at a constant level using drip irrigation, whereas plants in the long-term dry treatment were watered only every other day. In addition to the long-term treatments, we tested the effects of short-term environmental fluctuations on Lm of fully developed mature leaves by growing plants under the wet conditions described above and transferring these plants to an additional dry glasshouse 7 d before conducting our sampling. Our treatment results are therefore expressed as ‘dry’ (long-term dry growth conditions), ‘wet’ (long-term wet growth conditions) and ‘transfer’ (short-term transfer from wet to dry).

Figure 1.

 Average diurnal variability of air temperature and relative humidity in the three glasshouse chambers (8-wk average shown for dry (dashed–dotted line) and wet (grey line) glasshouse; 1-wk averages shown for transfer (black/grey dashed line) glasshouse).

We grew and sampled five replicate individuals for every species in each of the three treatments. Environmental conditions were constantly monitored with permanent climate sensors installed in each of the glasshouses. In addition, we installed four additional climate sensors (RH/TempLog Datalogger; Oakton Instruments, Vernon Hills, IL, USA) in each glasshouse during our intensive sample collection period to obtain a more precise estimate of the variation in relative humidity and air temperature that plants experienced for each treatment.

All plant samples were collected on 20 September 2007, 8 wk after seeds were planted. The diurnal changes in stomatal conductance (gs), transpiration (E) and leaf temperature were measured seven times each day for all three species in the three treatments using a LI-COR 6400 (LI-COR Biosciences, Lincoln NE, USA). Measurements started at sunrise and ended after sunset. For the gas-exchange measurements, we adjusted light and atmospheric conditions within the cuvette of the LI-COR 6400 to reflect the environmental conditions of the respective glasshouse. Nevertheless, environmental conditions within the cuvette always deviated slightly from ambient conditions in the glasshouse. To avoid cuvette artifacts on our estimates of E and leaf temperature, we did not use cuvette-derived values for E and leaf temperature in our analyses. Instead, we adjusted values of E and leaf temperature to ambient RH and air temperature using the Penman–Monteith equation (Monteith, 1965). In our calculations, we first solved the Penman–Monteith equation for leaf net radiation (Rn) using E, gs RH and air temperature from within chamber measurements. In a second step, we again used the Penman–Monteith equation to recalculate E with Rn and ambient values for RH and air temperature that we obtained from the four climate loggers in a glasshouse. In this second step, we used gs values obtained from chamber measurements in our calculations, assuming that the slight differences between chamber and ambient atmosphere RH and air temperature had little instantaneous effects on gs. The adjusted values for E represent the rate of transpiration that would be observed inside the leaf chamber at ambient RH and air temperature. In a final step, leaf temperature at ambient RH and air temperature was calculated with the Penman–Monteith corrected values for E by solving the leaf energy balance (Jones, 1992).

To test for the effect of growth conditions on plant and leaf morphology, we determined total plant height, leaf size (length, width and area), leaf weight and specific leaf area (SLA) for every species in the wet and the dry treatment. Leaf area was measured using a LI-COR 3100 leaf area meter. We determined leaf water concentration (mol H2O m−2) for leaves from all species in the three treatments by measuring FW, DW and leaf area twice a day (once in the morning shortly after sunrise and once in the evening just before sunset). To evaluate the effect of growth conditions on plant water status, we measured pre-dawn and midday water potentials using a pressure chamber (PMS Instruments, Albany OR, USA) on the day of the experiment. We also determined maximum leaf hydraulic conductance (Kl; mol s−1 m−2 MPa−1) for plants in the wet and in the dry treatment. Leaves were sampled for determination of maximum Kl within 2 d of the experiment. Maximum Kl was determined using the evaporative flux method described by Sack et al. (2002). Briefly, leaves were collected in the glasshouse by cutting the base of the petiole under water and these were then allowed to rehydrate by placing the cut ends in a 10 mm aqueous KCL solution overnight in the dark. The next morning, petioles were attached to an evaporative flux apparatus, which consisted of low resistance PVC tubing, filled with KCl solution running to a plastic cup on a balance (± 0.01 mg; Metler-Toledo AG245, Columbus, OH, USA). Mineral oil was used to eliminate evaporation from the KCl solution in the cup on the balance. A light source was suspended above the leaf, producing > 1200 μmol m−2 s−1 of photosynthetically active radiation at the leaf surface. Fans were placed around the leaf to minimize the boundary layer resistance. Leaves were allowed to transpire until a steady-state flow through the lamina was achieved over a 10-min interval. They were then removed from the tubing and placed in a pressure chamber until the balancing pressure was recorded (Soilmoisture Equipment Corp., Santa Barbara, CA, USA). Maximum Kl was calculated as the stable flow rate/balancing pressure for leaves with a balancing pressure ≥−0.65 MPa.

Since Lm cannot be directly measured and can only be determined by fitting leaf water models to measured values of leaf water δ18O, we determined leaf water δ18O of the three plant species in the three treatments on diurnal timescales. To do so, we collected one leaf from each replicate species seven times each day, starting at sunrise and ending after sunset. After the leaves were clipped from the plant, the primary veins were removed and the remaining leaf lamina sealed in 5 ml PVC vials and immediately frozen. Bulk leaf lamina water was extracted from the leaves using cryogenic vacuum distillation at the Center for Stable Isotope Biogeochemistry, UC Berkeley, USA, following the method described by West et al. (2006).

To determine the isotopic composition of the plant’s source water, we collected the water used for irrigation three times during the sampling day. To avoid isotopic enrichment of irrigation water by evaporation from the soil, all pots were covered with 3 cm quartz sand; this sand ‘cap’ decouples the soil from the atmosphere and prevents evaporation.

All measurements of leaf and plant morphology, as well as leaf water status and leaf water isotopic composition, were performed on five replicate individual plants per species and treatment. Repeated destructive sampling of leaves from the plants had no noticeable effects on the plants’ physiological performance.

We collected atmospheric water vapor five times during each sampling day in the dry and transfer treatments, and six times in the wet treatment. Vapor was trapped using polyethylene tubing that was looped three times, with the bottom two-thirds of the loops submerged in an ethanol–dry ice slurry (c. −80°C), and attached to a small diaphragm pump that pulled air through the traps. The airflow through the cryogenic traps was monitored by flow meters and set at 0.5 l min−1. While we were able to collect two replicate samples at each time-point from the wet treatment, our sampling system allowed for the collection of only one sample per point in time from the dry and the transfer treatment.

Bulk leaf water, source water and water vapor samples were analysed for δ18O by equilibration of a 50 μl sample of H2O with 2% CO2 for 48 h. The isotope composition of the CO2 gas was then determined using an isotope ratio mass spectrometer running in continuous flow mode (Finnigan MAT Delta Plus XL; Thermo Instruments Inc., Bremen, Germany) housed at the Center for Stable Isotope Biogeochemistry at UC Berkeley. Calibration standards were included every sixth position to account for instrument drift during a run. The long-term external precision for all of our water analyses was ± 0.14‰.

Leaf water models

We calculated effective path length using isotopic leaf water models. The development and precision of the models has been discussed in depth in the literature (Barbour et al., 2004; Cernusak et al., 2005; Cuntz et al., 2007; Ogee et al., 2007; Kahmen et al., 2008). Briefly, the models are based on equations that mechanistically describe the steady-state enrichment of leaf water in 18O at the sites of evaporation above the source water (Δ18Oe) as:

image( Eqn 1)

ε+ is the equilibrium fractionation between liquid water and vapor at the air-water interfaces (Bottinga & Craig, 1969); εk is the kinetic fractionation that occurs during water vapor diffusion from the leaf intercellular air space to the atmosphere (Farquhar et al., 1989; Cappa et al., 2003; Cernusak et al., 2003a); Δ18Ov is the isotopic value of water vapor in the atmosphere compared with source water; ea/ei is the ratio of ambient to intercellular vapor pressures (Craig & Gordon, 1965; Dongmann et al., 1974; Farquhar et al., 1989; Flanagan et al., 1991).

Equation 1 was originally developed to estimate the 18O enrichment of well-mixed surface waters of large water bodies such as lakes (Craig & Gordon, 1965) and has been shown to overestimate the evaporative enrichment of mean lamina mesophyll water (Flanagan et al., 1991; Wang & Yakir, 1995; Roden & Ehleringer, 1999b; Barbour & Farquhar, 2000; Cernusak et al., 2002). As such, Farquhar & Lloyd (1993) and Barbour et al. (2000b) have suggested that the discrepancy between the predicted leaf water enrichment based on Eqn 1 (i.e. the enrichment at the sites of evaporation) and the observed values of mean leaf water is caused by isotopic gradients within the leaf. These gradients may form as a result of the mixing of the transpirational stream of unenriched (source) water with enriched water moving backwards by diffusion in the opposing direction from the sites of water evaporation and therefore 18O enrichment. The ratio of transpirational flow over back-diffusion is described by the Péclet number (□; after Farquhar & Lloyd, 1993), which relates the mean lamina mesophyll leaf water isotopic enrichment over source water (Δ18OL) to Δ18Oe as

image( Eqn 2)

where the Péclet number is defined as,

image( Eqn 3)

In Eqn 3, E is transpiration rate (mol m−2 s−1), C is the molar concentration of water (mol m−3), D is the diffusivity of H2O in water (m2 s−1), and Lm is the effective path length for the transpirational flow of water from the xylem veinlets through the mesophyll (m) to the site of evaporation. Since the exact nature of Lm remains unclear, this parameter must be determined by iteratively by fitting the model to measured values of bulk leaf water Δ18O.

Equations 1 and 2 describe the enrichment of leaf water in 18O at steady state (i.e. under constant environmental conditions). Such conditions rarely occur in nature, where leaf water enrichment in 18O is subject to diurnally changing evaporative conditions. Dongmann et al. (1974), and more recently Farquhar & Cernusak (2005), therefore accounted for nonsteady-state enrichment of mean lamina mesophyll water in 18O. In their model, nonsteady-state leaf water enrichment (Δ18OLN) is expressed as

image( Eqn 4)

W is the water concentration of the leaf (mol m−2); wi is the mole fraction of water vapor in the leaf intercellular air spaces (mol mol−1). In essence, the deviation of leaf water enrichment in 18O from the steady state is accounted for in this model by emphasizing the one-way flux of water from the leaf to the atmosphere (gwi) (Farquhar & Cernusak, 2005). This nonsteady-state model has now been tested in several studies and has shown good agreement with measured bulk leaf water values of 18O (Cernusak et al., 2005; Barnard et al., 2007; Gessler et al., 2007; Kahmen et al., 2008).

To estimate values of Lm for the different species in the different treatments, we used the nonsteady-state leaf water model (Eqn 4). We fitted the model to measured diurnal values of bulk leaf water Δ18O of a given species and treatment by adjusting Lm until the sum of the differences between modeled Δ18OLN and measured Δ18OM values reached a minimum. For our estimates of Lm, we used the solver function in Microsoft Excel as suggested by Cernusak et al. (2005). As we fitted the nonsteady-state model to an entire set of diurnally measured bulk leaf water Δ18O values, we obtained a single value for Lm for each replicate plant (= 5) in each of the three treatments.

In a final evaluation of what parameters ultimately explain the enrichment of leaf water in 18O for a given species across the different treatments, we tested if the deviation of Δ18OM from Δ18Oe (i.e. the fraction of unenriched water in the leaf (f ) for a given species was correlated with E or Lm. The fraction of unenriched water in leaves, f, was determined for each plant as:

image( Eqn 5 )

δ18OM is the measured isotopic composition of bulk leaf water; δ18Oe is the isotopic composition at the site of evaporation calculated with Eqn 1; δ18Os is the isotopic composition of source water or xylem water (Leaney et al., 1985; Gan et al., 2002).

Statistical effects on plant and leaf morphological properties, leaf water potential, leaf water concentration and mean daily values of Δ18OM were tested using a one-way anova with an LSD post-hoc test.

Results

The climate in the three different treatments differed significantly and consistently for the entire duration of plant growth and development. Figure 1 shows the mean diurnal patterns of air temperature and RH for each of the three treatments. Temperatures and RH were comparable between the dry and the transfer treatments. The wet treatment, however, was on average 30% more humid than the dry and transfer treatments. Photosynthetic radiation did not differ across treatments and peaked at 1000 μmol m−2 s−1 on sunny days during midday. The climatic patterns on the day of the experiment were comparable to the mean values shown in Fig. 1.

The different environmental growth conditions in the wet and the dry treatments had significant effects on growth and morphology of the three species (Fig. 2). Plants grown in the wet treatment were all significantly taller by more than a factor of two. Also, plants developed significantly larger leaves in wet conditions compared with the dry growing conditions. Specific leaf area (SLA; m2 g−1) was significantly higher in the wet treatment for sunflowers. Beans also showed a trend towards higher SLA in the wet treatment, but this trend was nonsignificant. The SLA for castor bean did not differ between the two treatments. Maximum Kl differed across species, but did not differ for any species as a result of wet or dry growing conditions.

Figure 2.

 Effects of dry (closed bars) and wet (tinted bars) growing conditions in the two long-term treatments on leaf morphological and leaf hydraulic properties of the three species (bean, castor bean and sunflower) (n = 5). Significance: ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05, n.s., not significant. Error bars, 1 SE. SLA, specific leaf area; Kl, maximum leaf hydraulic conductance.

Pre-dawn leaf water potential was most negative for sunflower and least negative for castor bean (Fig. 3). For any given species, pre-dawn leaf water potentials did not differ across the three treatments. At midday, however, leaf water potentials in the dry and transfer treatments were more negative for a given species when compared with the wet treatment (Fig. 3). Unfortunately, data for midday leaf water potential of bean were lost during the analysis.

Figure 3.

 Pre-dawn (tinted bars) and midday (closed bars) leaf water potential for the three plant species (bean, castor bean and sunflower) in the different treatments (n = 5). Midday values for the dry and transfer treatment of bean were not available (n.a.). Significance: ***P ≤ 0.001; **≤ 0.01; *P ≤ 0.05; n.s., not significant. Different lower case letters indicate significantly different pre-dawn water potential values across treatments for a given species and different upper case letters indicate significantly different midday water potential values across treatments for a given species. Error bars, 1 SE.

Morning and evening values for leaf water concentration did not differ for any of the three plant species. We therefore used morning and evening values to calculate daily means. Castor bean showed the lowest leaf water concentration, while bean and sunflower had a higher but comparable leaf water concentration (Fig. 4). For any given species, leaf water concentration did not differ between wet and transfer treatment but was significantly higher in the dry treatment.

Figure 4.

 Leaf water concentration (LWC) for the three different plant species (bean, castor bean and sunflower) in the three different treatments (wet, light grey bars; transfer, dark tinted bars; dry, closed bars). No significant differences for LWC were found in leaves collected just after sunrise and leaves collected just before sunset. The values shown are therefore mean daily values. Different lower case letters indicate significantly different LWC concentrations across treatments for a given species (P ≤ 0.05). Error bars, 1 SE.

For all three species, diurnal stomatal conductance was substantially higher in the wet treatment compared with the dry and transfer treatments (Fig. 5). Across all species, in all treatments, sunflower grown in the wet treatment had the highest stomatal conductance. Diurnal patterns and the magnitude of stomatal conductance in the dry and in the transfer treatments were roughly comparable for bean and castor bean, but differed substantially for sunflower, where stomatal conductance was higher in the transfer treatment compared with the dry treatment. Transpiration rates for all three species showed similar diurnal pattern in the wet treatment, with maximum transpiration rates in the early afternoon (Fig. 5). Transpiration rates in the dry and in the transfer treatment showed very similar diurnal patterns and magnitudes for bean and castor bean. For both species in both the dry and the transfer treatment, transpiration rates peaked in the late morning and then gradually declined over the course of the day. By contrast, sunflower showed lower transpiration rates in the dry treatment and these were consistently low over the course of the day but were consistently high in the transfer treatment, with a substantial peak in the late afternoon.

Figure 5.

 Diurnal variability in stomatal conductance (gs), transpiration (E), and enrichment of bulk lamina leaf water in 18O above source water (LWΔ18OM) of the three plant species (bean, castor bean and sunflower) in the different treatments (wet, circles; transfer, squares; dry, triangles). Data shown represent the means of five replicates. Numbers in figures indicate mean daily standard error.

The different treatments also had substantial effects on bulk leaf water enrichment in 18O above source water (Δ18OM; Fig. 5). In the wet treatment, leaf water Δ18OM was comparable across species and significantly lower compared with the dry and the transfer treatments with minimal diurnal patterns. Statistical testing of mean daily Δ18OM values of a given species revealed no significant differences between Δ18OM values from the dry and the transfer treatment, but Δ18OM values from the wet treatment were significantly different from Δ18OM values from the dry and the transfer treatment. This pattern was consistent for all three species.

The isotopic composition of atmospheric water vapor was comparable in all three treatments (Fig. 6). The only variation was in the morning, when atmospheric vapor in the wet treatments was c. 2‰ less enriched compared with the dry and the transfer treatments. Overall, diurnal variability was less than 3‰ in either treatment.

Figure 6.

 Diurnal patterns of water vapor (WV) δ18O values in the three glasshouse chambers at the day of the experiment. Wet, solid grey line; transfer, dashed line; dry (dashed–dotted line).

Predictions by the isotopic leaf water models as described in Eqns 1–4 closely matched the diurnal patterns of isotopic leaf water enrichment for all three species in the different treatments (Fig. 7). As expected, the basic Craig and Gordon (CG) (Craig and Gordon, 1965) model overestimated leaf water enrichment in 18O except for the first three samples collected for castor bean in the wet treatment. The inclusion of a Péclet effect improved model performance. The best fit between the measured and modeled Δ18O values was achieved with the nonsteady-state model (Fig. 7).

Figure 7.

 Modeled and measured enrichment of bulk lamina leaf water in 18O above source water of the three plant species (bean, castor bean and sunflower) in the different treatments. Data shown represent the means of five replicates. LWΔ18OM is the ratio of measured leaf water, LWΔ18Oe is the enrichment in 18O at the sites of evaporation as predicted by (Eqn 1), LWΔ18OL is the enrichment in 18O of bulk lamina leaf water in the steady-state (Eqn 2) and LWΔ18OLN is the enrichment of bulk lamina leaf water in 18O in the non-steady-state (Eqn 4). Each individual value shown represents the mean of five replicate samples.

The effective path lengths obtained by fitting the nonsteady-state model to the measured values of leaf water Δ18O were significantly different for the three species (Fig. 8). However, no significant differences were observed for Lm within a species across different treatments (Fig. 8).

Figure 8.

 Effective path-length as predicted by the nonsteady-state leaf water model (Eqn 4) for the three species (bean, castor bean and sunflower) in the three treatments (dry, transfer and dry). The total mean for effective path-length including 1 SE is given for each species in the respective plot. Letters indicate significant differences across species and treatments. Error bars represent 1 SE.

Our analyses revealed that the mean daily ratio of atmospheric to leaf internal vapor pressure (ea/ei) was the key driver of both the mean leaf water enrichment in 18O at the site of evaporation (Δ18Oe) and Δ18OM (Fig. 9). In addition, transpiration also explained some of the variability in Δ18OM by driving the fraction of unenriched water in a leaf (f ). However, the relationship between f and transpiration was not significant for the daily mean values. As indicated earlier, daily transpiration rates for common bean and castor bean showed complementary diurnal patterns across treatments but did not differ in their average daily rates (Fig. 6). We therefore determined the relationship between E and f separately for mean morning and mean afternoon values. Except for castor bean in the morning, all species showed a significant relationship between E and f (Fig. 9).

Figure 9.

 The relationship between the ratio of atmospheric and leaf internal vapor pressure (ea/ei) and leaf water enrichment in 18O at the sites of evaporation (LWΔ18Oe, upper panels) and measured lamina leaf water (LWΔ18OM, middle panels) for the three species (bean, diamonds and dashed–dotted lines; castor bean, circles and dashed line; sunflower, triangles and dotted line) in the three treatments. The lower panels show the relationship between f, the fraction of unenriched source water in a leaf, to transpiration (E), where f = 1 − (Δ18OM18Oe). The relationships are shown for average values across the entire day (07:00–19:00 h) as well as separated in average morning values (07:00–13:00 h) and afternoon values (13:00–19:00 h). All regressions that are shown are significant. The relationship between f and transpiration for entire day averages was significant for only sunflower (r = 0.88), for morning average values it was significant for sunflower (r = 0.90) and bean (= 0.95) and for average afternoon values it was significant for sunflower (r = 0.99), bean (r = 0.95) and castor bean (r = 0.99). Effective path-length (Lm): bean, 17.35; castor bean, 24.67; sunflower, 10.61).

Discussion

Treatment effects and gas exchange

Greater height growth and larger leaf sizes for wet treatment plants illustrate that the different growth conditions in the wet and dry glasshouses significantly affected the morphology of the three plant species (Fig. 2). Pre-dawn water potentials did not differ across treatments for a given plant species (Fig. 3), even though plants in the wet glasshouse were subject to constant water supply via a drip irrigation system, while plants in the dry and transfer glasshouse were watered only every other day. In contrast to pre-dawn, the different growth conditions in the wet, dry and transfer glasshouses had significant effects on the midday leaf water potentials of two plant species (Fig. 3; data for bean were not available). Growing conditions also affected the leaf water concentrations of all three plant species (Fig. 4). In combination, these data suggest that differences in RH in the wet, dry and transfer glasshouse strongly not only affected the leaf morphology but also leaf water relations of the plant species investigated.

Stomatal conductance and transpiration showed a clear response to the different environmental treatments (Fig. 5). Notably, mature leaves from common bean and castor bean plants grown in a wet environment and transferred into a dry environment 1 wk before sampling (transfer treatment) showed almost identical patterns in their gas exchange to leaves that had experienced continuously dry growing conditions (dry treatment). This suggests that fully developed leaves from these two species were quickly able to adjust their gas exchange physiology to the dramatic changes in the environmental conditions that plants experience in dry environments. By contrast, sunflower leaves that fully developed in the wet environment and were then transferred to the dry environment had consistently higher stomatal conductance and transpiration when compared with leaves that had developed in the dry environment, suggesting an inability to adjust.

Diurnal patterns of Δ18OM also varied substantially across the different treatments (Fig. 5). Values for Δ18OM were significantly higher in the dry and transfer treatments for all three species compared with Δ18OM values from the wet treatment plants. Interestingly, the Δ18OM values for leaves in the dry and transfer treatment were roughly comparable. This again suggests that fully developed leaves were quickly able to acclimatize physiologically to the marked changes in the environmental conditions experienced when transferred from the wet to the dry treatment. The overall patterns in Δ18OM are similar to several previous studies where leaves in dry environments were more enriched in 18O compared with leaves in humid environments (Yakir et al., 1990; Flanagan et al., 1991; Roden & Ehleringer, 1999a; Barbour & Farquhar, 2000; Helliker & Ehleringer, 2002; Santrucek et al., 2007; Ripullone et al., 2008). The relative contribution of different environmental and ecophysiological drivers to the observed patterns of Δ18OM across different species and treatments is discussed below.

Modeling leaf water δ18O and estimating Lm

We estimated values for effective path length by fitting the nonsteady-state isotopic leaf water model to diurnally measured values of leaf water δ18O. The precision and uncertainties of the different models that we used to determine Lm have been discussed in depth in other literature and will therefore not be discussed here (Barbour et al., 2004; Cernusak et al., 2005; Cuntz et al., 2007; Ogee et al., 2007). The Lm values obtained by fitting the leaf water models were significantly different for the three species and were within the range of previously determined effective path lengths across a broad range of different plant species (Wang et al., 1998; Kahmen et al., 2008). However, we found no significant differences for Lm values for a given species across different treatment types (Fig. 8). In other words, for the species investigated, Lm appeared to be a constant factor that was independent of the long- or short-term environmental variability used in setting up our treatments. This result is surprising given that several previous studies, including our own, have speculated that Lm should be related to leaf anatomical/morphological properties (Barbour & Farquhar, 2003; Cuntz et al., 2007; Kahmen et al., 2008). The SLA, plant height, leaf area, midday leaf water potential and leaf water concentration all showed significant responses to the three treatments in this study. However, contrary to what we had expected, values for Lm remained constant across treatments for a given species (Fig. 8).

We were unable to detect a significant relationship between leaf morphological traits and Lm in the study we present here. The true morphological, anatomical or physiological nature of Lm therefore remains unclear. Barbour et al. (2004) tried to link effective path length values derived from 18O leaf water models to the physical distance between the main vein and the site of evaporation in wheat leaves. While this early study showed that physically measured and model-derived values for Lm were within an order of magnitude, the study was also limited to a single species and thus did not allow any correlative investigation relating leaf functional traits to Lm. Kahmen et al. (2008) specifically tested the relationship between Lm of and a range of different leaf traits such as SLA, leaf area and leaf size across a large number of different Eucalyptus species, but found no significant correlation. It has also been suggested that Lm should be related to aquaporin expression and activity and consequently to Kl (Keitel et al., 2006; Ferrio et al., in press). In a recent study, Ferrio et al. (in press) have shown that Lm in leaves of slightly water-stressed beach saplings was threefold higher than Lm in leaves of well-hydrated beach saplings. The authors speculate that reduced aquaporin expression and leaf-internal cavitation resulted in a loss of conductivity for water in the leaf and thus an increase in Lm. In our study, we specifically tested effects of leaf hydraulic conductance on Lm. We detected significant differences in maximum Kl among the three species investigated in this study, but maximum Kl of the three different species was not correlated with Lm. Interestingly, we did not find treatment effects on maximum Kl for any species in this study (Fig. 2). It could therefore be argued that Lm for a given species was not affected by our treatments since maximum Kl was not affected by the treatments. We wish to urge caution here, however, because maximum Kl was determined under standardized conditions for the different species and treatments. Previous research has shown that Kl can respond quickly to changes in environmental conditions such as light and water availability (Cochard et al., 2007). Therefore, the maximum Kl values that we measured may not reflect Kl values that plants experienced at the time when isotopic leaf water enrichment was determined. Our data therefore do not necessarily allow us to establish a mechanistic link between Kl and Lm across treatments for a given species.

The general properties of Lm

As indicated above, there have now been several attempts to link model-derived values for Lm to leaf morphological, anatomical or leaf hydraulic traits, yet, there is no clear indication that these types of leaf traits determine the nature of Lm. A reason for this could be hidden in the method employed for determining Lm. Since direct measurements of Lm are not possible, Lm must be estimated as a fitting parameter in isotopic leaf water models. As a result, values for Lm will not only capture effects of the ‘true’ effective path length, but also other biological characteristics that may influence the isotopic enrichment of leaf water that are yet to be identified and therefore are not accounted for in current leaf water enrichment models. What is more, Lm as the fitting parameter in isotopic leaf water models will not only contain the unexplained biological information, but will also capture the sum of all measurement errors in the input parameters that are used to fit the isotopic leaf water models. This adds further complications and could explain why Lm values that we publish for a given species here differ to some extent from Lm values that have been published for the same species in previous studies. For R. communis, for example, previous studies have published Lm values of 13.5 mm (Barbour et al., 2000b), 15.0 mm and 11.1 mm (Cernusak et al., 2003b). It is likely that different instrumentation, different calibration precision and different methodological routines of the researchers are all likely to introduce small measurement errors to the individual parameters used to model leaf water enrichment in 18O and to calculate Lm.

We tested the sensitivity of Lm to small variations (± 2.5%) in input parameters using the leaf water models that we parameterized with mean environmental and physiological data from the wet and the dry treatment. This sensitivity analyses revealed that a small variation of only ± 2.5% in some input parameters had substantial effects on estimates of Lm (Fig. 10). This shows that values for Lm that are derived from fitting leaf water models capture a combination of different biological leaf properties but also the measurement errors associated with the model input parameters. Such multidirectional influences on Lm could explain why no individual leaf morphological or hydraulic parameter has yet been identified to explain Lm.

Figure 10.

 The result of a sensitivity analysis testing the effects of small variations in input parameters (2.5%) on effective path length, Lm. The sensitivity analysis was performed with models that were parameterized with average environmental and physiological values for the wet treatment (densely hatched bars) and dry treatment (coarsely hatched bars). Δ18Ov, isotopic value of water vapor in the atmosphere compared with source water; δ18OM, measured isotopic composition of bulk leaf water; Tleaf, leaf temperature; gs, stomatal conductance; E, transpiration; rh, relative humidity; Tair, air temperature.

Despite the uncertainties involved with determining values for Lm, the data presented here strongly indicate that Lm is a constant factor for a given species. This finding is interesting, given that other studies have suggested that Lm should vary with environmental conditions as a result of changes in the leaves hydraulic properties (Keitel et al., 2006; Ripullone et al., 2008; Ferrio et al., in press). Despite substantial differences in long- and short-term growing conditions and the resulting effects on leaf morphology and leaf water relations, we found consistent values of Lm for individuals of three different species. This extensive and rigorous test gives us confidence that – at least for the species investigated here –Lm is a species-specific parameter and should therefore not obscure the response of a plant’s δ18O values to environmental or ecophysiological drivers. To test this assertion we evaluated, for a given species, the relationship between Δ18OM and major environmental and ecophysiological parameters used to influence Δ18O across the different treatments. As expected, Δ18Oe can be explained almost exclusively by the ratio of atmospheric to leaf internal vapor pressure (ea/ei) for all species across all three treatments (Fig. 9). Similarly, Δ18OM was also strongly related to ea/ei, but was substantially less enriched in 18O than Δ18Oe (Fig. 9), an observation that has largely been attributed to the Péclet effect in previous studies (Barbour et al., 2000b, 2004). The influence of the Péclet effect on isotopic leaf water enrichment depends on E and Lm (Eqn 3). As we have asserted that Lm is constant for a given species, E should then largely influence f, the fractional difference between Δ18OM and Δ18Oe for any given species (Fig. 9). We found that all species showed a significant relationship between E and f for mean morning and afternoon values with the exception of castor bean during the morning (Fig. 9). The strong relationship between E and f demonstrate that E, and not Lm, is the parameter that drives the variability of leaf water Δ18O and of the Péclet effect for a given species (Flanagan et al., 1991; Barbour et al., 2000b, 2004; Ripullone et al., 2008).

Conclusions

The data presented here show that Lm, or what has been suggested to be the effective path-length of water flux in the leaf lamina, differs significantly across species but not for individuals that are within a given species (even when subject to dramatically different environmental conditions). This finding is important because it facilitates the interpretation of δ18O values in plant material. For example, variation in δ18O values of plant material that has been generated under different climatic conditions but originates from the same species, a situation typically found in tree ring time series, will reflect the air to leaf vapor pressure ratio and transpiration but will not be obscured by variation in effective path length. Further, constant values for Lm will allow one to use the variability of δ18O values in plant material that originates from plants of the same species that have grown under the same climatic conditions as an indicator for transpiration. This will be particularly useful in studies where integrative measures of transpiration are needed to compare, for example, the ecophysiological performance of plants in common garden evaluations of different varieties of agriculturally important species (Barbour et al., 2000a; Morison et al., 2008), or in large-scale environmental experiments such as in ozone or CO2 (FACE) fumigation experiments (Grams et al., 2007; Jaggi & Fuhrer, 2007; Bassin et al., 2009).

Acknowledgements

The authors would like to thank Kim DeLong for growing and maintaining the plant specimens used in this study and David Barajas for his assistance with sample collection in the glasshouses and cryogenic leaf water extractions in the laboratory. We would also like to thank Paul Brooks for analysing water samples at the Center for Stable Isotope Biogeochemistry at UC Berkeley. A.K. was supported by an Erwin–Schroedinger Postdoctoral fellowship of the Austrian Science Foundation (FWF) and a Marie-Curie Outgoing International Fellowship of the EU.

Ancillary