Stomatal responses to humidity and temperature in darkness

Authors


K. A. Mott. Fax: +1 435 797 1575; e-mail: keith.mott@usu.edu

ABSTRACT

Stomatal responses to leaf temperature (Tl) and to the mole fractions of water vapour in the ambient air (wa) and the leaf intercellular air spaces (wi) were determined in darkness to remove the potential effects of changes in photosynthesis and intercellular CO2 concentration. Both the steady-state and kinetic responses of stomatal conductance (gs) to wa in darkness were found to be indistinguishable from those in the light. gs showed a steep response to the difference (Δw) between wa and wi when wa was varied. The response was much less steep when wi was varied. Although stomatal apertures responded steeply to Tl when Δw was held constant at 17 mmol mol−1, the response was much less steep when Δw was held constant at about zero. Similar results were obtained in the light for Δw = 15 mmol mol−1 and Δw ≈ 0 mmol mol−1. These results are discussed in the context of mechanisms for the stomatal response to humidity.

INTRODUCTION

Stomata respond to both air humidity and leaf temperature (Tl), but there is no commonly accepted mechanism for either response. This deficiency is important because responses to humidity and temperature represent much of the variation in stomatal conductance (gs) for leaves in a natural environment (Schulze et al. 1972, 1973; Fredeen & Sage 1999). In addition, water loss through stomata provides an essential, but poorly understood, flux in general circulation models (Sellers et al. 1997; Bernacchi et al. 2007) and has been linked to climate change (Sellers et al. 1996; Betts et al. 1997) and river run-off (Gedney et al. 2006). Attempts to account for stomata in global fluxes have been hindered by the lack of an accurate mechanistic understanding of stomatal responses to environmental factors (Randall et al. 1996).

The stomatal response to air humidity has generally been assumed to be a response to the water loss rate from the leaf rather than a direct response to air humidity. This idea is supported by data showing that stomata close when normal nitrogen–oxygen air (nitox) is replaced by helium–oxygen air (helox) (Mott & Parkhurst 1991). Water molecules diffuse 2.3 times faster in helox than in nitox, and this causes transpiration to increase by a factor of 2.3 with no change in air humidity. Because stomata are assumed to respond to transpiration rate, stomatal responses to humidity are commonly plotted versus the driving gradient for transpiration, that is, the difference in water vapour mole fraction (Δw) between the intercellular air spaces (wi) and the ambient air (wa). This creates a complicated interaction between responses to air humidity and responses to Tl. Most studies on stomatal responses to humidity use changes in wa to effect changes in Δw at a constant Tl. In these studies, stomata generally close as wa decreases and Δw increases, and this has led to the commonly accepted idea that stomata close as Δw increases. It is possible, however, to vary Δw by changing the mole fraction of air in the intercellular spaces (wi) while keeping wa constant. In these experiments Tl is varied to change wi, and wa is kept constant. There are very few examples of this type of experiment in the literature, which is surprising because in a natural environment, most of the change in Δw is likely to be the result of changes in Tl rather than changes in wa. Typically, stomata show only a very small closing response as Δw is increased by increasing wi and Tl (Ball, Woodrow & Berry 1987; Grantz 1990; Fredeen & Sage 1999), and this contrasts sharply with the steep response of gs to Δw observed when wa is varied at constant Tl.

This difference between the responses to wa and wi is difficult to reconcile mechanistically if one assumes that stomata actually sense transpiration rate. To understand why this is the case, one must first recognize that there are two types of mechanisms that have been proposed to explain how stomata might respond to transpiration. In the first type, more water is lost from the guard cells than from the surrounding epidermal cells. This reduces the water potential of the guard cells in proportion to the humidity of the air and causes stomata to close as humidity is decreased. In some of these proposed mechanisms, water loss to the atmosphere from the guard cells occurs through the cuticle (Lange et al. 1971; Maier-Maercker 1983); in others, water loss occurs from the inside surface of the guard cells (Dewar 2002). The second general type of mechanism relies on a closing signal that is generated in the mesophyll or epidermis in response to low water potential or low turgor pressure. This signal therefore tends to close stomata as leaf water potential decreases at high transpiration rates (Buckley, Mott & Farquhar 2003). Mathematical models for the stomatal response to humidity have been constructed for both of these general mechanisms (Dewar 2002; Buckley et al. 2003), and both types of model can accurately predict stomatal responses to Δw when wa is varied at a constant Tl. However, both types of model predict that the response of gs to Δw when wi is varied should be the same as when wa is varied.

One way to explain the response to wi using these models is to postulate that stomata show a direct opening response to increasing Tl. If this is the case, then increasing Tl has two counteracting effects; it increases Δw, which tends to close stomata, and it increases Tl, which tends to open stomata. The result is a very small apparent response to Δw when wi is varied. Stomatal apertures in isolated epidermes have been shown to be strongly temperature dependent with a peak aperture at about 35 °C (Willmer & Mansfield 1970), and some ion channels show temperature dependence (Ilan, Moran & Schwartz 1995). However, there have been very few studies addressing guard cell temperature dependence, and there is no consensus on a mechanism by which guard cell turgor could vary with temperature.

It is also possible to partially explain the response to wi using these models by accounting for the decrease in the viscosity of water with increasing temperature. Hydraulic conductivity increases as viscosity decreases, so increases in temperature will cause increases in the hydraulic conductivity for water movement to the evaporating site in the leaf. This will decrease the magnitude of the water potential drop to the evaporating site or to the guard cells for a given rate of water loss. Thus, as Tl increases, Δw and transpiration rate increase, which causes water potential in the guard cells or other cells in the leaf (depending on the model) to decrease and close stomata. At the same time, though, the hydraulic conductivity increases so the drop in water potential, and the amount of stomatal closure, is not as large as if Δw were changed at a constant Tl (Fredeen & Sage 1999; Matzner & Comstock 2001; Sack, Streeter & Holbrook 2004; Sack & Holbrook 2006). About half of the apparent increase in hydraulic conductivity can be explained by the decrease in the viscosity of water with temperature and the remaining increase has been attributed to the temperature dependence of flow outside the xylem, presumably through cell membranes (Fredeen & Sage 1999; Matzner & Comstock 2001; Sack et al. 2004). Although this is an attractive hypothesis, data for stomatal responses to humidity in at least two species of amphistomatous leaves seem inconsistent with it because they suggest that essentially all of the hydraulic resistance is located outside the xylem. When wa was changed for one surface of an amphistomatous leaf and maintained constant for the other surface, stomata on the surface for which wa was changed responded normally, but stomata on the surface for which wa was constant showed no response (Mott 2007). Furthermore, when wa was lowered for only one surface, epidermal turgor decreased immediately for that surface but remained constant for the other surface. These results suggest that, at least for the two species tested (one homobaric and one heterobaric), there was almost no change in water potential in the xylem as transpiration increased and that all the decrease in water potential occurred between the xylem and the epidermis. This is inconsistent with data suggesting that about half of the temperature effect on leaf hydraulic conductivity is due to resistance in the xylem (Sack et al. 2004).

Another explanation for the difference between the responses to wa and wi is that stomata respond to the relative humidity at the leaf surface rather than to Δw (Ball et al. 1987). This suggestion is interesting because it does a better job of predicting stomatal responses to wa and wi. However, in the model proposed by Ball et al. (1987), it is necessary to include a stomatal response to photosynthetic rate (A) to completely account for stomatal responses to Tl. Furthermore, a response to relative humidity is inconsistent with data that suggest that the response to humidity is actually a response to the rate of water loss (Mott & Parkhurst 1991), and no mechanism by which stomata might respond to relative humidity has been proposed.

The entire situation is made more complicated by the fact that it is difficult or impossible to study stomatal responses to Δw and Tl in isolation. The response of gs to wi is complicated by the fact that photosynthesis (A) is affected by Tl. It is therefore impossible to vary Tl without changing A and ci, and although it is possible to maintain ci or A constant by varying ambient CO2 concentration (ca), it is not possible to hold both ci and A constant as Tl varies. Stomata are known to respond to ci, and there are data to suggest that stomata may also respond to A or some component of A (Messinger, Buckley & Mott 2006). Since the exact nature of the stomatal responses to ci and A is controversial, it is impossible to compensate for these responses, and it is therefore impossible to unequivocally determine the response of gs to wi. This problem also exists for stomatal responses to wa at a constant Tl because A and ci change slightly as stomata respond to wa (Hall & Kaufmann 1975). In this case, A or ci can be held constant by adjusting ca, but this is rarely done in practice, and A and ci vary slightly for most of the published stomatal responses to wa.

Because the quest for a mechanistic understanding of stomatal responses to humidity and temperature depends critically on how these responses interact with other responses, the study reported here was initiated to determine the stomatal response to wa, wi and Tl in the absence of the potential confounding effects of A and ci. To do this, we studied stomatal responses to these factors in darkness using Tradescantia pallida. Recent studies have shown that stomata can be open in darkness in some species under some conditions (Snyder, Richards & Donovan 2003), and that stomata respond to Δw under these circumstances (Barbour & Buckley 2007). In this study, we take advantage of dark opening of stomata and their response to humidity to examine the stomatal responses to Δw and Tl that are independent of the responses to ci and A.

METHODS AND MATERIALS

T. pallida plants were grown from cuttings in a controlled environment greenhouse. Growth conditions were as described previously (Shope, Peak & Mott 2008). Gas exchange measurements were performed as described previously (West et al. 2005; Mott 2007). Briefly, a leaf was placed in a clamp-on-type chamber that was made of nickel-plated aluminium and enclosed 6.45 cm2 of the leaf area. The top and bottom of the chamber were made of glass, which allowed light (>710 nm in dark experiments, see below) to be delivered to the leaf from the top while imaging stomata from the bottom. The chamber air was circulated through a heat exchanger by miniature fans, and the temperature of the heat exchanger was maintained by circulating water through it from a temperature-controlled water bath. Tl was measured with a fine-wire (0.13 mm) chromel–constantan thermocouple pressed to the lower surface. Gas of known composition was mixed from N2, O2 and 1% CO2 in air using needle valves, two-stage gas regulators and mass flow controllers. The mole fraction of water vapour in the mixture was controlled by bubbling a variable portion of the total stream through CO2-free water before the 1% CO2 was added. The CO2 concentration and dew point of the final mixed gas were determined with an infrared gas analyser set in the absolute mode (ADC Instruments, MkIII, Hoddesdon, UK) and a dew point hygrometer (General Eastern, Dew-10, Watertown, MA, USA). The difference in CO2 and water vapour concentrations before and after the chamber were determined with an infrared gas analyser set in the differential mode (LiCor, model 6262, Lincoln, NE, USA). All experiments were conducted at 100 µmol mol−1 CO2 and 21 mmol mol−1 O2.

To determine stomatal responses in the light, plants were taken from the greenhouse in the morning and a leaf was placed in the gas exchange chamber. To determine stomatal responses in darkness, plants were taken from the greenhouse late in the day prior to the experiment and left in darkness overnight before measurements were made the next day. When placed in the gas exchange chamber the next morning, these plants had gs values that were comparable to those measured for plants illuminated with saturating light. For experiments in both light and darkness, gs and aperture declined slowly throughout the day despite constant conditions. To compensate for this problem, we determined the response of gs or aperture to a given variable in both directions, that is, starting at high values and decreasing the variable and vice versa. For example, to determine the response of gs to Δw at constant Tl, two sets of experiments were performed. In the first, the plant was initially brought to steady state at a Δw of 10 mmol mol−1 and Δw was progressively increased by 5 mmol mol−1 by changing wa until it reached 25 mmol mol−1. In the second, the plant was initially brought to steady state at a Δw of 25 mmol mol−1 and Δw was progressively decreased by 5 mmol mol−1 until it reached 10 mmol mol−1. Approximately 1 h was allowed for stomata to respond to each change in Δw.

Stomatal apertures were observed in the chamber with an inverted microscope (Olympus, CK2, Tokyo, Japan) and photographed with a peltier-cooled CCD digital camera (Roper Scientific CoolSnap HQ, Tucson, AZ, USA) attached to a personal computer. Digital imaging software (MediaCybernetics ImagePro, Bethesda, MD, USA) was used to measure apertures. For measurements in the light, the leaf was illuminated from the top using a fibre optic illuminator, and for measurements in the dark, a 710 nm long-pass filter (Schott, RG 9, San Jose, CA, USA) was used. Although no visible light reached the leaf in the latter case, the camera was sufficiently sensitive to wavelengths above 710 nm to produce high-quality images.

RESULTS

Stomatal responses in darkness to Δw at a constant Tl were determined by varying wa. Because there was variation among experiments in the value of gs at a given Δw value, it was necessary to normalize data relative to the value of gs at Δw = 10 mmol mol−1 for each experiment. Although there was variation among experiments in the values of gs, there was only a small difference between these values in the dark and light. For example, for the data shown in Fig. 1, the average (±one standard deviation) maximum values of gs (observed at Δ= 10 mmol mol−1) in the dark and light were 0.073 ± 0.018 and 0.081 ± 0.017 mol m−2 s−1, respectively.

Figure 1.

Stomatal responses to Δw in darkness at constant Tl. Leaf temperature = 25 °C, [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1. A total of 6 experiments on six different leaves are shown. All data are normalized to the value for Δw = 10 mmol mol−1. Open circles show data for three leaves for which the starting value of Δw was 10 mmol mol−1 and Δw was progressively increased by 5 mmol mol−1. Closed circles show data for three leaves for which the starting value of Δw was 25 mmol mol−1 and Δw was progressively decreased by 5 mmol mol−1. The lines show linear regressions of the data.

The value of gs showed a steep dependence on Δw when Δw was started at a low value and increased progressively throughout the experiment by decreasing wa at a constant Tl (open symbols, Fig. 1). In contrast, there was almost no dependence of gs on Δw when Δw was started at a high value and then progressively lowered by increasing wa (closed symbols, Fig. 1). We interpreted this result to mean that gs was declining steadily throughout the experiment independent of environmental changes, and this was verified by experiments in which humidity was held constant throughout the experiment (data not shown). To correct for this problem, we performed three experiments with increasing Δw values and three with decreasing Δw values. In each experiment, we used the same Δw values and allowed approximately the same time interval between each measurement (Fig. 1).

To correct for the decrease in conductance over time, we assumed that conductance declined throughout the experiment as a constant percentage of its initial value. The data for each experiment were then normalized to the value at Δw = 10 mmol mol−1 and linearly regressed. The percentage of the initial value by which conductance declined per unit time was assumed to be the same for all experiments of a particular type and was adjusted for all data simultaneously until the two regression lines (for increasing and decreasing Δw) gave the same slope. The open symbols in Fig. 2 show the data from Fig. 1 after they have been corrected using this protocol, and the dashed line shows a linear regression of the data. These data can then be compared to the response of gs to Δw in the light (photon flux = 700 µmol photons m−2 s−1), which was normalized and corrected in the same manner, and is shown by the solid points in Fig. 2.

Figure 2.

Stomatal responses to Δw at constant Tl in darkness and light. Leaf temperature = 25 °C, [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1 for all data. All data have been corrected for the decline in conductance with time and normalized to the value at Δ= 10 mmol mol−1. The open symbols show the data from Fig. 1 (taken in darkness), and the dashed line is a linear regression of the data. The closed symbols show the response in the light (photon flux = 700 µmol photons m−2 s−1), and the solid line shows a linear regression of the data. For the light data, each point represents an average of six experiments on six different leaves, and the error bars show one standard deviation around the mean.

To further compare the stomatal responses to Δw at constant Tl in darkness and light, the kinetics of the stomatal response to an increase in Δw from 15 to 20 mmol mol−1 were determined by decreasing wa from 21.8 to 16.8 mmol mol−1 (Fig. 3). Both dark and light data show an initial, rapid increase in conductance followed by a slower decline to a conductance value that is approximately 80% of the initial value. As with data shown in Fig. 2, these data showed substantial plant-to-plant variation, but the conductance at Δw = 20 mmol mol−1 was between 73 and 87% of the value at Δw = 15 mmol mol−1 in all four experiments. The data taken in the light appear to have a slightly longer lag before the slow decline in conductance than the data taken in the dark, but the general shapes and magnitudes of the responses are similar.

Figure 3.

Kinetics of stomatal responses to Δw at constant Tl in darkness and light. Leaf temperature = 25 °C, [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1. At time zero, Δw was increased from 15 to 20 mmol mol−1 by decreasing wa from 21.8 to 16.8 mmol mol−1. Data, immediately after the change in wa, were omitted to eliminate artefacts caused by the response time of the gas exchange system. wa was maintained constant as stomata responded by controlling the humidity of the air entering the chamber. Each trace represents data from one leaf.

The response of gs to Δw at constant wa in darkness was determined by varying Tl (and therefore wi) between 20 and 30 °C (Fig. 4). Tl was varied by changing air temperature, and wa was maintained constant at 17 mmol mol−1 by changing the humidity of the air entering the chamber. These data were collected and analysed in the same way as those in Figs 1 and 2, that is, data were taken in both directions, corrected for the decline in conductance and then normalized to one value (in this case, to the value at Δw = 7 mmol mol−1). The data show that there was a small decline in gs as Δw increased (Fig. 4), and gs is also plotted versus Tl (Fig. 4, inset) for reference.

Figure 4.

Response of gs to Δw in darkness at a constant wa of 17 mmol mol−1. Leaf temperature was varied to control Δw. [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1. Data were taken as described in the text and corrected for temporal drift as described in the text and illustrated in Figs 1 and 2. Three leaves were measured with increasing Δw and three with decreasing Δw, and the data were normalized to the value at Δ= 7 mmol mol−1. The inset shows the same data plotted against Tl and normalized to the value at Tl = 20 °C.

Next, the response of stomatal aperture or gs to Tl was determined in darkness at two constant values of Δw: 17 mmol mol−1 and approximately zero. Δw was maintained constant as Tl varied by changing the humidity of the air entering the chamber. Because these experiments were conducted in the dark, there was essentially no radiation load on the leaf. In addition, there was also very little or no transpiration, so there was little evaporative cooling. These conditions made it possible to keep the leaf and air very close to isothermal and it was therefore possible to keep Δw very close to zero. These experiments were conducted as described above and were normalized to the value at 30 °C. They were corrected for the decline in gs over time in darkness as described above. At a Δw value of 17 mmol mol−1, there was a steep positive dependence of gs and aperture on Tl (Fig. 5). At a Δw value of close to zero, there was only a very shallow positive dependence of aperture on Tl (It was impossible to measure gs at Δw ≈ 0 because there was no transpiration). The slopes of the regression lines for Δw = 17 mmol mol−1 and for Δw ≈ 0 mmol mol−1 were significantly different (P < 0.001).

Figure 5.

Stomatal conductance or aperture in darkness as a function of Tl at a constant Δw. [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1. Data were taken for both increasing and decreasing Tl and corrected for temporal drift as described in the text. Solid and open circles represent a mean of six experiments (three with increasing Tl, and three with decreasing Tl), and data were normalized to the value at Tl = 30 °C. Error bars show one standard deviation on either side of the mean, and the regression lines for Δw = 17 mmol mol−1 and for Δw ≈ 0 mmol mol−1 are significantly different (P < 0.001). Open squares show individual aperture measurements from two separate experiments.

To verify that this difference in temperature dependency was present in an illuminated leaf, we conducted a similar experiment in the light (photon flux = 700 µmol photons m−2 s−1). In these experiments, we determined stomatal apertures as a function of Tl in 1 °C intervals with 30 min allowed for each steady state. The radiation load caused the Tl to be a few tenths of a degree above air temperature meaning that Δw was never exactly zero, but the value of Δw was, however, never greater than 1.0 mmol mol−1. The data were corrected for the decrease in conductance with time and normalized as described above. They show a steeper relationship between aperture and Tl at Δw = 15 mmol mol−1 than at Δw ≈ 0 mmol mol−1, and the slopes of the regression lines are significantly different (< 0.001).

To determine the effect of helox in darkness, we performed eight experiments, four in which nitox was replaced with helox at a constant Δw of 10 mmol mol−1, and four in which Δw was changed from 10 to 23 mmol mol−1 in nitox. In each experiment, gs was followed until a new steady state was achieved (usually about an hour) or until stable oscillations were observed. In the latter case, the steady state value for gs was taken as the midpoint of the oscillations. When helox was substituted for nitox at Δw = 10 mmol mol−1, the new steady-state conductance was 59% ± 6% of the starting conductance. When Δw was increased from 10 to 23 mmol mol−1 in nitox, the new steady-state conductance was 52% ± 18% of the original.

DISCUSSION

In this study, stomatal responses to wa, Tl and wi were determined in darkness to avoid the confounding effects of changes in A and ci. We used T. pallida because we found that it showed high values of gs in the dark, and because it has large stomata for which stomatal aperture can be easily measured. The latter point is necessary for determining stomatal responses when Δw is near zero, since gs cannot be accurately measured if there is no transpiration. Darkness clearly removes any potential effect of A on stomata, and because respiration rates in these leaves were low, changes in ci with gs or Tl were always less than 10 µmol mol−1. Because this change is too small to have a measureable effect on gs, we conclude that the responses to gs and temperature observed in this study were independent of either A or ci. Responses of gs to wa or wi were, however, complicated by the fact that gs declined throughout the day even if conditions were constant. We corrected for this problem by assuming that gs or aperture declined by a constant percentage of its starting value over a given period of time. Although this protocol appeared effective because data for increasing values could be reconciled with data for decreasing values, it remains possible that the hysteresis observed in the data could have been caused by differences in the preceding conditions rather than a slow decline with time.

Our results confirm a previous study (Barbour & Buckley 2007) showing that stomata respond to atmospheric humidity in darkness. In that study, however, stomatal responses to humidity were studied during the night, while in our study, they were studied in darkness during the day. Our data contrast with the previous study in several ways. Firstly, we found that values of gs in the darkness were similar to those in the light for a given wa and Tl, whereas in their study, gs values were much smaller in the darkness than in the light for most experiments. Secondly, we found that the responses of gs to wa and Tl in darkness were indistinguishable from those in the light, whereas they found that the response to wa in the dark was either more sensitive or less sensitive than in the light, depending on growth conditions. Plants that were well watered and grown at low wa were more sensitive to humidity in darkness than in light. The opposite was true for well-watered plants grown at high wa and plants that were droughted. In this study, plants were grown at a moderate wa and were not droughted, so it is possible that the conditions fell between those in the previous study and therefore yielded similar results in darkness and light.

This study shows for the first time that (1) the steady-state (Fig. 2) and transient (Fig. 3) responses of gs to wa in darkness are similar to those in light; and (2) the responses of stomatal aperture to Tl in darkness are similar to those in light. These results suggest that the mechanisms responsible for the stomatal response to wa and Tl are similar in darkness and light. It also suggests that in the light, the responses of gs to ci or A did not significantly affect the response of gs to wa or Tl in our experiments. This conclusion is reasonable because photosynthesis in these plants was not very sensitive to temperature between 20 and 30 °C (data not shown). While this was true in our study, it may not hold for plants for which photosynthesis is highly sensitive to temperature over the temperature range of interest or for plants with a very high stomatal sensitivity to ci.

As has been reported for illuminated leaves (Ball et al. 1987; Grantz 1990; Fredeen & Sage 1999), we found that the stomatal response to Δw was much smaller when Δw was changed with wi than when it was changed with wa (compare Figs 2 & 4). As discussed in the Introduction, this result has been explained in several different ways in previous studies. In mechanisms that rely on water potential drawdown of the guard cells or adjacent cells (Dewar 1995; Buckley et al. 2003), this result can be explained by some other temperature-responsive process that affects stomatal aperture and that counteracts the closing response to Δw. This idea is supported by the response of gs and aperture to Tl at a constant Δw value of 17 mmol mol−1 in the dark (Fig. 5) and by the response of aperture to Tl at a constant Δw value of 15 mmol mol−1 in the light (Fig. 6). Because Δw and transpiration rate (E) were held constant (by varying wa) as wi and Tl were changed, it seems plausible that the steep response of gs is a direct response to temperature rather than a response to Δw or E. It is worth noting that there was an increase in steady-state E as gs increased with Tl, but this was the result, rather than the cause, of the increase gs.

Figure 6.

Stomatal aperture in light as a function of Tl at Δ= 15 mmol mol−1 and Δ≈ 0. Photon flux = 700 µmol photons m−2 s−1, [CO2] = 100 µmol mol−1 and [O2] = 21 mmol mol−1. Data were taken for both increasing and decreasing Tl and corrected for temporal drift as described in the text. Each point represents an average of four experiments, and data were normalized to the value at 30 °C. Error bars are omitted for clarity, but the regression lines for Δw = 15 mmol mol−1 and for Δw ≈ 0 mmol mol−1 are significantly different (P < 0.001).

When the above experiment was repeated in darkness with Δw ≈ 0 (Fig. 5), however, the response of aperture to Tl was significantly smaller than when Δw was 17 mmol mol−1. Similarly, in the light, the response of aperture to Tl was steeper when Δw = 15 mmol mol−1 than when Δw ≈ 0 mmol mol−1 (Fig. 6) (Because there was no measureable transpiration under these conditions, it was impossible to calculate gs, so only apertures were measured). The difference between the response of aperture to Tl for non-zero and near-zero values of Δw is significant for at least two reasons. Firstly, it is not consistent with a temperature-dependent process that independently affects stomatal aperture, which makes it unlikely that this is responsible for the difference between the response to Δw when wa is varied and the response when wi is varied. Secondly, it suggests that the temperature response of stomatal aperture is dependent on the value of Δw, which is unexplainable by models that rely on water potential drawdown to explain stomatal responses to humidity.

Another proposed explanation for the difference between the responses to wa and wi is a temperature effect on the hydraulic conductivity to the evaporating site (Fredeen & Sage 1999; Matzner & Comstock 2001; Sack et al. 2004; Sack & Holbrook 2006). The difference between the response of aperture to Tl at near-zero Δw and that at non-zero Δw is consistent with this idea since any effect of hydraulic conductivity would disappear in the absence of transpiration. However, as discussed in the Introduction, stomatal responses to humidity for the independent surfaces of amphistomatous leaves suggest that this explanation is unlikely, at least in some leaves.

Interestingly, our results are consistent with a stomatal response to relative humidity as suggested by Ball, Woodrow and Berry (Ball et al. 1987). If most of the temperature response arises from an unknown effect of relative humidity on stomata rather than from an effect on leaf or guard cell water potential, then there should be little or no temperature response when Δw = 0 because changing temperature would not change the relative humidity. A simple response to relative humidity is, however, not consistent with data that show stomatal closure when leaves are switched from nitox to helox at a constant wa.

In conclusion, our data show that stomatal responses to humidity are similar in darkness and light, which suggests that the mechanism for stomatal responses to humidity does not depend on processes that cause stomatal opening in light. We confirm that stomatal responses to wa and wi are not consistent with a simple response to Δw, and our data show that this inconsistency is not caused by effects of A or ci. Furthermore, we show that the apparent temperature response of stomata at constant Δw is greatly diminished when Δw is zero, which is inconsistent with a direct response of guard cells to temperature or with any other temperature-dependent process that affects stomatal aperture that has been proposed to explain differences in stomatal response to wa and wi. A single mechanism for stomatal responses to humidity and temperature that applies to all species and is consistent with all published data remains elusive.

ACKNOWLEDGMENTS

We thank Rand Hooper and Erik Sibbernsen for excellent technical assistance.

Ancillary