Summary
⍰ Modulation of stomatal development may be an acclimation response to low water availability. However, stomatal development plasticity has been assessed in very few species.
⍰ We quantified leaf anatomy traits, including stomatal index (SI), density (SD), size (SS), and pore index (SPI), in response to water-deficit stress in river birch (Betula nigra L.), eastern redbud (Cercis canadensis L.), and silver maple (Acer saccharinum L.).
⍰ Birch and redbud, but not maple, had reduced SPI in response to water deficit. The mechanism by which SPI reduction occurred (via SD or SS) varied among species and with severity of water stress. Despite reduced SPI in birch and redbud, anatomical changes were relatively small and had a minor to no effect on the theoretical maximum stomatal conductance. Furthermore, gas-exchange rates were equivalent to well-watered plants following media re-irrigation.
⍰ In some tree species, stomatal development is downregulated in response to water deficit conditions. Stomatal development plasticity is facilitated by smaller or fewer stomata, depending upon the species and the intensity of the stress. Water-deficit-induced plasticity in stomatal development is species-specific, likely due to species adaptation to ecological niches.
Introduction
Understanding plant acclimation and adaptation to water deficit will become increasingly important as climate change increases the frequency and intensity of drought. Drought events are projected to increase in frequency globally, and in the US Midwest in particular, from once every five years to once every other year by 2050 (Jin et al., 2018). In natural ecosystems, drought events result in reduced ecosystem productivity (Wu et al., 2011) and plant mortality (Gitlin et al., 2006; Klos et al., 2009). Reduced water availability constrains CO2 assimilation and reduces carbon provisioning from shoots to roots (Ruehr et al., 2009). Drought events may decrease tree water potentials to below the zero-carbon assimilation point (Breshears et al., 2009), which can deplete tree carbon reserves and result in tree death (McDowell, 2011). Although there is evidence for large-scale ecosystem resilience to dynamic seasonal availability of water (Ponce-Campos et al., 2013), the role of leaf hydrological plasticity in achieving these acclimations is unclear.
The capacity of stomata to regulate water loss during periods of water deficit is critical to plant growth and survival. Stomatal closure is an important transitory response during drought events, but during chronic water deficit, modulation of stomatal anatomy may also be important for water conservation (Liu et al., 2018). Smaller stomata close more quickly in response to decreased leaf hydration status than larger stomata (Drake et al., 2013; Giday et al., 2013) and smaller guard cells are biomechanically optimized to open at lower turgor pressure (Spence et al., 1986) but without increased carbon costs (Raven, 2014) enabling maintenance of CO2 assimilation during water-stress events.
In dicotyledonous plants, stomatal development primarily occurs during early leaf emergence, when developing leaves are composed of dividing and differentiating stem cells (Rawson & Craven, 1975; Andriankaja et al., 2012), although the specific timing of development varies among species (Woodall et al., 1998). Stomatal differentiation can occur late in leaf development in some species (Ludlow, 1991), and there is some evidence for the persistence of meristemoids that can form new stomata late in leaf development (Geisler et al., 2000). Within the first two days of leaf emergence, leaf veins form (Kang & Dengler, 2004) and vein density is subsequently modulated by passive dilution during leaf expansion (Carins Murphy et al., 2012). Based on this developmental timing, environmental effects on the ultimate anatomy of a leaf are likely to occur early in the development of that leaf. A stress occurring during the expansionary phase of leaf development may result in changes in stomatal density due to altered cell turgor and therefore cell size, but the relative numbers of guard cells to pavement cells is unlikely to be significantly altered because cell identity is mostly fixed by this stage. This illustrates the difficulty in drawing conclusions on mechanistic responses to diverse environments from ecological experiments, such as those involving stands of trees that receive different precipitation levels (Reyer et al., 2013), since it is difficult to ascertain when a water deficit event occurs relative to leaf development.
Stomatal development is sensitive to environmental cues at the cellular level, particularly light (Casson et al., 2009; Kang et al., 2009) and CO2 (Woodward, 1987; McElwain & Chaloner, 1995; Franks & Beerling, 2009), suggesting that the modulation of stomatal development is an acclimation trait. The expression of several genes that regulate stomatal development are downregulated in response to osmotic stress (Kilian et al., 2007; Harb et al., 2010; Yoo et al., 2010; Baerenfaller et al., 2012; Kumari et al., 2014; Yoo et al., 2019), and have been linked to a concomitant suppression of stomatal development in Arabidopsis (Skirycz et al., 2011; Kumari et al., 2014; Yoo et al., 2019) and soybean (Tripathi et al., 2016). Similar gene expression patterns may exist in tree species, but very few data have been collected (Hamanishi et al., 2012; Viger et al., 2016).
Despite these observed links between water deficit and stomatal plasticity at the anatomical and molecular levels, plasticity of SI, stomatal density (SD), and stomatal size (SS) in response to water deficit varies widely among (de Silva et al., 2012; Hamanishi et al., 2012) and within (Pääkkönen et al., 1998; Hovenden & Vander Schoor, 2012) tree species. Between-species variation may be explained by different degrees and methods of water deficit imposed by different authors, such as withholding water over a period of time (Hamanishi et al., 2012) or different degrees of maintained media water content (de Silva et al., 2012; Aasamaa et al., 2001; Catoni et al., 2017). Within-species differences in stomatal development plasticity could be caused by different provenances having different stomatal patterning as an adaptation to local levels of moisture (Dunlap & Stettler, 2001; Pearce et al., 2006, McKown et al., 2014).
Stomatal pore index (SPI, Sack et al., 2003) and theoretical maximum stomatal conductance (gsmax, Franks & Farquhar, 2001) are derived traits that combine the frequency and dimensions of stomata to describe the effective potential water loss from the leaf interior, which effectively predicts leaf gas exchange (Dow et al., 2014; McElwain et al., 2016). These traits account for all the possible changes in stomatal anatomy that could occur in producing leaves optimized for water conservation. For example, simply reporting a lack of SD plasticity may omit the production of smaller stomata that leads to reduced SPI and/or gsmax. Since SPI and gsmax are frequently not reported, conclusions cannot yet be drawn about the role of stomatal development plasticity in facilitating water deficit tolerance in trees.
We hypothesized that perennial species exposed to persistent water-deficit stress acclimate via a reduction in stomatal development and overall stomatal coverage to minimize water loss. We examined three tree species: Betula nigra L., Acer saccharinum L., and Cercis canadensis L., chosen to represent the Betulaceae, Sapindaceae, and Fabaceae families, respectively. These species occupy a large range of urban, rural, and forested land across temperate North America.
Materials and methods
Plant materials and growth conditions
One-year-old bare-root river birch (Betula nigra L.) were planted in 8.5 L containers, and four-week-old redbud (Cercis canadensis L.) and silver maple (Acer saccharinum L.) seedlings were planted in 3.4 L containers in a BM8 Berger soilless substrate (Berger, Saint-Modeste, QC, Canada). Plants were maintained in a greenhouse in the Purdue Horticulture Plant Growth Facility from June to December 2018. A minimum 14 h photoperiod was provided with 100-watt high-pressure sodium lamps. Temperature and relative humidity were measured using two HOBO® data loggers (Onset Computer Corporation, Bourne, MA, USA), and the daily light integral was quantified with an external weather station (Fig. S1). Average day and night temperatures were 24.7 and 23.6°C, respectively, and average relative humidity (RH) was 70%. Vapor pressure deficit (VPD, kPa) was calculated as where SVP is the saturated vapor pressure at a given daily temperature, derived from standard tables of the two quantities.
Two experiments were performed, the first imposing a mild stress, and the second imposing a more severe stress. Media water content (MWC) was maintained by weighing plants and replacing water on a regular basis to 100 or 60% MWC (experiment 1), and 100 or 40% MWC (experiment 2) of the initial saturated weight (including initial plant biomass) and was calculated as where MW is the weight of the container system (consisting of container, media, and plant) on a given day, and MSW is the saturated weight of the container system at the beginning of the experiment. For the first experiment, plants were irrigated to media capacity until establishment. After 95 d post-planting, water was withheld for 18 d until the target MWC of 60% was reached in the water-stressed (WS) treatment plants. Control well-watered (WW) plants were irrigated every 3–4 d as necessary, and WS plants every 24 h to 60% of their initial saturated weight (Fig. S2). In the second experiment, plants were irrigated to media capacity until establishment and after 54 d, WS was initiated. WW plants were irrigated every 2 d, and WS plants were irrigated every day (redbuds) or 2 d (maples) to 40% of their initial saturated weight (Fig. S2). Birch was not included in the 40% MWC treatment.
Acidified water was supplemented with water-soluble fertilizer (ICL Specialty Fertilizers, Dublin, OH, USA) to provide the following (in mg L-1): 150 N, 9.8 P, 119 K, 12 Mg, 21 S, 1.5 Fe, 0.4 Mn and Zn, 0.2 Cu and B, and 0.1 Mo. Nitrate and ammonium sources of nitrogen were provided as 61 and 39% total N, respectively. Irrigation water was supplemented with 93% sulfuric acid (Brenntag, Reading, PA, USA) at 0.08 mL L-1 to reduce alkalinity to 100 mg L-1 and pH to a range of 5.8 to 6.2.
Measurements
Leaves that emerged after experimental days 20 and 34 in experiments 1 and 2, respectively, following the establishment of treatment MWC levels, were used for data collection (Fig. S3; Table S1). To quantify leaf development in experiment 1, leaf 2 was photographed next to a ruler with a digital camera c. every two days over 40 d and leaf area (LA) was quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA). A logistic curve was fit to the leaf area A at time t for each leaf (Fig. S4a-c): where α is the sigmoidal midpoint and κ is the logistic growth rate. The logistic curve was linearized and used to calculate the maximum leaf growth rate, experimental day on which the leaf reached 50% full expansion, and the number of days to full leaf expansion (Fig. S4d-f).
To confirm that leaves used for data collection were fully expanded, LA was assessed in leaves 2 (Fig. S4) and 4 in experiment 1. In experiment 2, the size of one leaf that had developed after the initiation of water deficit was measured over 7 days prior to harvest to ensure full expansion of this leaf. Final LA was quantified during the final destructive harvest, by removing the leaf and imaging against a white background.
Net CO2 assimilation (A, μmol CO2 m-2 s-1), stomatal conductance (gs, mol H2Om-2 s-1) and transpiration (E, mmol H2O m-2s-2) were measured using a portable gas-exchange analyzer (LI-6400XT; LI-COR Biosciences, Lincoln, NE, USA). Chamber conditions during measurements were 1800 μmol m-2 s-1 PAR, air temperature range of 29.0–32.8°C, and average VPD was 2.5 ± 0.4 kPa.
After scanning the leaf used for gas exchange measurements, a segment of leaf, including a portion of the midrib and petiole, was collected, weighed to obtain the fresh weight (FW) and placed into a 50 mL conical tube with 25 mL of water. After the leaf had been immersed for c. 8 h, the leaf was blotted dry and weighed to obtain the turgid weight (TW). The leaf was then dried at 45°C to a constant weight to obtain the dry weight (DW). The relative water content (RWC) was calculated as
This portion of the leaf was also scanned for LA, from which the specific leaf weight was calculated as
Osmotic potential was measured on the same leaves used for RWC. A leaf segment was removed, placed into an Eppendorf tube with a Costar Spin-X insert (Corning Incorporated, Corning, NY, USA), and immersed in liquid nitrogen. Samples were stored at −20°C. To extract cell sap, samples were thawed in sealed tubes for 5 min and then centrifuged for 5 min at 120 RPM x 100. Osmolality of a 10 μL of volume of cell sap was measured using a vapor pressure osmometer (VAPRO 5520; Wescor Inc., Logan, UT, USA). Osmolality was converted to osmotic potential (Ψπ) as where Cs is osmolality, R is the gas constant, and T is temperature. The osmotic potential at full turgor (Ψπ100) was calculated as
Epidermal traits were quantified from leaf impressions made on microscope slides using cyanoacrylate (Duro Super Glue; Henkel, Düsseldorf, Germany). Four images were taken from each impression with a DCM 900 microscope CMOS Camera (Oplenic Optronics, Hangzhou, China) of a 0.03 mm2 area, under 400X magnification using a light microscope (BH-2; Olympus, Tokyo, Japan). All three species are hypostomatous, so epidermal impressions were made only on the abaxial side of leaves. The number of stomata (SD) and pavement (PD) cells per unit area, and stomatal size (SS), equivalent to the area within the outer edges of the guard cell, was determined using ImageJ software. Only whole stomata and pavement cells bordering the top and right sides of each image were counted. Stomatal index (SI) was calculated as
Pavement cell size was estimated by dividing the number of pavement cells counted in a region and dividing by the size of the visual field (30 000 μm2).
Stomatal size was calculated using the formula for an ellipse:
Stomatal pore index (SPI) was calculated as (Liu et al., 2018)
Theoretical maximum conductance (gsmax) was calculated using the equation developed by Franks & Farquhar (2001) where d is the diffusivity of water in air (0.0000249 m2 s-1 at 25°C), v is the molar volume of air (0.0224 m3 mol-1), l is the stomatal pore depth equivalent to the width of a fully turgid guard cell, and amax is the maximum pore area, calculated using half the stomatal length as the pore length, as follows:
To quantify vein density (VD), the same leaf section from leaves 3–4 (experiment 1) and 1 (experiment 2) used for epidermal impressions was submerged in 50% sodium hypochlorite in a water bath at 50°C for 6 h. After the leaf was cleared, it was placed in a 0.0015% toluidine blue solution for 12 h, then observed under 40X magnification. Four images were taken per sample, each covering an area of 2.95 mm2 and avoiding the midvein of the leaf. Major and minor veins in the image were traced and the total length summed across all orders of veins. This sum was then divided by the area of the image to calculate VD as mm veins mm-2. Images were processed using ImageJ software.
Statistical analysis
Experiments were conducted using a completely randomized design. In the first experiment, several measurements were made on multiple leaves from the same plant, and it was therefore analyzed as a repeated measures ANOVA. The leaf-treatment interaction was used to determine which leaves could be pooled for a given trait. Data was pooled across leaves if the interaction of treatment groups and leaves was not significant (P > 0.05). Regardless of pooling, each WS group was compared against the comparable control group by one-way ANOVA. Data were transformed with a Box-Cox transformation if needed to fulfill the assumption of normality. To calculate plasticity of anatomical traits, the natural logarithm of the response ratio was determined as ln(average of treatment group/average of control group), with the standard deviation of WW and WS groups combined to calculate the 95% confidence interval. A single leaf per plant was examined in the second experiment. All analyses were conducted in Jamovi v1.
Results
Plant growth and water status were adversely affected by water deficit
To verify the effects of the imposed WS, we quantified plant growth. Under WS, all trees were shorter than WW trees (Figs S5a, S6) and all trees produced smaller leaves except redbud, in which leaf size was not affected by 60% MWC (Figs S5b, S7). The 60% MWC treatment resulted in a reduced maximum leaf growth rate in all species (Fig. S4d), whereas the length of leaf development was longer only in maple leaves (Fig. S4e, f).
To assess treatment effects on plant water content, we quantified several water relations traits. Restricting MWC to 60% resulted in lower leaf RWC during the stress period, but this quickly recovered to WW levels following re-saturation of the media (Fig. S8). However, in the 40% MWC treatment, reduced RWC was observed in maple, but not redbud leaves (Fig. S8b, c). The 60% MWC WS level did not induce a reduction in osmotic potential (Ψπ) in any species, whereas Ψπ was lower in maple and redbud leaves on trees grown under 40% MWC (Fig. S9a-c). In redbud leaves, this appears to be a passive effect, as Ψπ100 was not reduced (Fig. S9f).
Water deficit resulted in plasticity of stomatal anatomy and leaf physiology
In these experiments, we quantified changes in leaf anatomy and physiology of three temperate tree species in response to WS. The plasticity of the various traits depended on species and the intensity of the water deficit treatment. Here we describe these changes for each of the five treatment combinations, relative to WW plants.
Under WS, birch trees produced leaves that were 45% smaller than WW leaves (Figs S5b, S7a), with similar leaf thickness (Fig. S10). Stomatal density was unchanged (Figs 1a, S11a) despite the production of 22% fewer stomata relative to the total cell population (Figs 1b, S12a). This is because stomata in WS leaves were 20% smaller (Figs 1c, S13a), primarily due to a decrease in stomatal length (as opposed to width) (Fig. S14a, d). Pavement cells were 26% smaller (Fig. S15a, b), so cell density of both types was unchanged (Fig. S16a, b), leading to the same stomatal distribution per unit area in all leaves. However, the smaller stomata in WS leaves resulted in a 17% lower SPI (Figs 1d, S17a). Despite the 20% reduction in SS, birch did not exhibit plasticity for gsmax (Fig. 2a, b). There was no reduction in VD in WS birch leaves (Figs 3, S18a).
Under WS conditions, gs was reduced by 90% in leaf 3 (in which gsmax was reduced) and 76% in leaf 4 (in which gsmax was unchanged) (Fig. 4a). During the water deficit stress, A was also 99 and 68% lower (Fig. 4d) and transpiration rate was reduced by 84 and 67% (Fig. S19a) in birch leaves 3 and 4, respectively. Due to stomatal closure, reduced SPI and smaller leaf size, WS birch leaves lost 85% less water during the water-deficit period (Fig. S19d). gs and A eventually recovered to levels comparable to WW leaves seven days after media re-saturation (Figs 4a, d). Despite the smaller LA and SS, whole-leaf transpiration also recovered to WW levels in this time frame (Fig. S19d). It is possible that such a recovery could occur because in leaf 4 SPI was not different in WW and WS leaves (Fig. S17a), and the size of WS leaf 4 was not as reduced as previously developed leaves (Fig. S7a).
In response to 60% MWC, maple leaves were 48% smaller (Figs S5b, S7b) and 18% thinner (Fig. S10) than WW leaves. However, epidermal anatomy was mostly non-plastic, with no change in SS, SI, or SD (Figs 1, S11b, S12b, S13b). There was also no change in pavement cell size (Fig. S15a, c) or density (Fig. S16a, c). As a result, neither SPI nor gsmax were lower in WS leaves (Figs 1d, 2a, c, S17b). Maple leaf VD was also unchanged under 60% MWC (Figs 3, S18b).
Despite the lack of anatomical plasticity, gs and A were 83% lower overall in WS leaves, relative to WW leaves (Fig. 4b, e). Maple WS leaves 3 and 4 grown under 60% MWC had reduced gs, A, and E of 90 and 77% (Fig. 4b), 87 and 80% (Fig. 4e), and 78 and 72% (Fig. S19b), respectively. Due to stomatal closure and smaller leaf size, WS maple leaves lost 88% less water during the water-deficit period (Fig. S19e). A and E eventually recovered to levels comparable to WW leaves three days after media re-saturation (Figs 4e, S19b). Due to the smaller leaf size and stomatal closure, WS leaves lost 88% less water, relative to WW leaves, and whole-leaf transpiration was still depressed in WS leaves at the end of the recovery period (Fig. S19e).
Under the severe 40% MWC WS, maple leaves were 77% smaller (Figs S5, S7b) with no change in leaf thickness (Fig. S10). Due to a reduction in length (Fig. S14b), stomata in WS leaves were 13% smaller than those of WW leaves (Figs 1c, S13b). As in the 60% MWC treatment, no changes in stomatal anatomy occurred (Figs 1, 2, S11b, S12b, S17b). However, there was a 24% increase in VD (Figs 3, S18b). The 40% MWC WS resulted in a 50, 71, and 42% reduction in gs, A, and E, respectively (Fig. S20a-c). Overall, 40% MWC WS maple leaves lost 85% less water than WW leaves (Fig. S20d).
Redbud trees produced leaf types depending on the intensity of the WS treatment. Under the 60% MWC WS, redbud leaves were not smaller (Figs S5b, S7c), but they were 11% thinner (Fig. S10). There was no change in SS (Figs 1c, S13c, S14c, f), but SI was reduced by 10% (Figs 1b, S12c). Because epidermal cell size and density were unchanged (Figs S15a, d; S16a, d), this decrease in stomatal development resulted in a 20% decrease in SD (Figs 1a, S11c). With fewer stomata in redbud leaves, SPI and gsmax were reduced by 20% (Figs 1d, 2a, d, 17c). Vein density was not different between WS and WW leaves (Figs 3, S18c).
Under the 60% MWC WS, gs was reduced by 87% in leaf 3 and 72% in leaf 4 (Fig. 4c), as both leaves had reduced gsmax. A was also 86 and 61% lower during the water deficit period in these leaves (Fig. 4f). E was 85 and 67% lower (Fig. S19c). Although LA was unchanged, stomatal closure and lower SD resulted in WS redbud leaves losing 75% less water during the water-deficit period (Fig. S19f). Despite reduced SD and gsmax in leaf 4, gs, A, E, and whole-leaf transpiration eventually recovered to levels comparable to WW redbud leaves three days after root zone re-saturation (Figs 4c, f; S19c, f).
By contrast, the severe (40% MWC) WS resulted in smaller redbud leaves (Figs S5b, S7c), but SLW was similar in WW and WS leaves (Fig. S10). In this case, redbud stomata were 23% smaller (Figs 1c, S13c), because of both shorter and narrower guard cell dimensions (Fig. S14c, f). Despite SI being reduced by 12% in these leaves (Figs 1b, S12c), SD was unchanged (Figs 1a, 11c). Because of smaller stomata, SPI was reduced by 25% (Figs 1d, S17c), but this was not sufficient to reduce gsmax (Fig. S2a, d). However, gs and A were reduced (by 85 and 88%, respectively, Fig. S20a, b). Overall, the smaller LA and SS, as well as stomatal closure, meant that 40% MWC WS redbud leaves lost 93% less water than WW leaves (Fig. S20c, d).
Discussion
Effects of water stress on tree growth and development
Water-deficit treatments resulted in shorter trees in all species and smaller leaves in almost all species and treatments (Figs S5–S7). Similar to previous studies, birch LA was reduced with only a small effect on plant size under the mild stress we applied (Kleczewski et al., 2012). In maple and redbud, the severe 40% MWC treatment resulted in a more dramatic reduction in growth compared to the 60% MWC treatment (Fig. S5).
Increased leaf thickness under water deficit conditions may increase the diffusion path from the leaf interior to the environment, thereby minimizing water loss from leaves (Syvertsen et al., 1995; Sobrado, 2007). As noted in previous studies (Kleczewski et al., 2012), there was no change in leaf thickness of river birch leaves in response to drought stress. Both increases and decreases in water-deficit-induced leaf thickness have been reported in related species B. ermanii (Kitao et al., 2003; Tabata et al., 2010) and B. pendula (Possen et al., 2011; Aspelmeier & Leuschner, 2006). Genotypes of Cercis canadensis adapted to drier regions have thicker leaves than those adapted to wetter regions (Donselman & Flint, 1982; Abrams, 1988; Tipton & White, 1995; Fritsch et al., 2018), but there are no reports of temporal water-deficit events on leaf thickness in redbud. In the present study, redbud leaves were thinner in response to the 60% MWC and were unchanged in response to the 40% MWC treatment, and the same pattern existed in maple. Decreased water availability also resulted in thinner leaves in related species A. truncatum (Li et al., 2017) and A. davidii (Guo et al., 2019). Our data show that the reduction in leaf thickness that occurs under the mild stress is not enhanced by the more severe water deficit. In redbud and maple, thinner leaves may be produced in response to mild water deficit to allow for easier hydration of the leaf. It is unclear why this response would not exist under more severe stress.
Decreasing water availability results in decreased RWC in tree species (Reddy et al., 2004) and experimental treatments that drastically reduce water availability can result in extremely low RWC (Ma et al., 2015; Vieira et al., 2017). In this study, RWC was reduced by approximately 12% overall during the water deficit periods (Fig. S8). This relatively small decrease in RWC is likely due to the fact that although water was reduced in the WS treatments, small amounts of water were delivered on a regular basis, as opposed to a long-term dry-down (de Silva et al., 2012; Catoni et al., 2017). In all species, RWC increased to control levels shortly after irrigation of the WS plants, as has been observed previously when an imposed water deficit did not severely reduce leaf RWC (Tognetti et al., 1995). In this way, our water deficit treatment mimicked the type of stress encountered by trees in periods with reduced precipitation compared to wet periods (Kubiske & Abrams, 1991; Backes & Leuschner, 2000). This established that our experiments tested acclimation responses as they may arise in natural drought conditions, rather than leaf responses to rapid dehydration.
Some tree species are able to adjust osmotically in response to water deficit episodes (Ranney et al., 1991; Wang & Stutte, 1992), whereas other species show no evidence of osmotic adjustment (OA) under water deficit conditions (Tschaplinski et al., 1995). The observed maintenance of high RWC in WS leaves generally occurred without OA. In the 60% MWC treatment, there was no evidence for OA in any species (Fig. S9d–f). Redbud accumulates soluble carbohydrates in response to WS, but apparently do not osmotically adjust, as shown in the current and past (Griffin et al., 2004) experiments. The only evidence we found of OA was in 40% MWC maple (Fig. S9e). This OA under only severe WS has been observed in Fraxinus excelsior (Guicherd et al., 1997) and hybrid Populus genotypes (Gebre et al., 1998). However, this did not affect the water status or growth of leaves, as both species produced smaller leaves (Figs S5, S7), and maple leaves still had reduced RWC despite the OA (Fig. S8b). Altogether, the growth and water relations response to WS was similar across silver maple, river birch, and eastern redbud leaves.
Stomatal development plasticity in trees via different mechanisms to a common outcome
Stomatal closure in response to WS is a common response (Loewenstein & Pallardy, 1998; Bréda et al., 2006). However, the plasticity of stomatal anatomy in leaves that emerge under water-deficit conditions is far less studied, especially in a manner that integrates different stomatal traits to show the final overall change in stomatal anatomy across the leaf epidermis. It is likely that this anatomical plasticity plays some role in acclimation to water deficit, since the molecular basis for stomatal development plasticity in response to water deficit has been established in some tree species (Hamanishi et al., 2012; Viger et al., 2016).
Reduced SPI in response to water deficit has been demonstrated in some tree species (Gindel, 1969; Camposeo et al., 2011), but is not the typical response in trees (unpublished metaanalysis; Aasamaa et al., 2001; Luo et al., 2007; Eksteen et al. 2013) due to no anatomical changes occurring or the fact that the often-observed reduction in SS is not sufficient to have an impact on SPI (Aasamaa et al., 2001; Luo et al., 2007; Machado et al., 2010; de Silva et al., 2012; Hovenden et al., 2012; Eksteen et al., 2013; Catoni et al., 2017). However, it is clear from our study that some tree species do respond to water deficit stress via stomatal development plasticity. In birch and redbud, we observed a common outcome of reduced SPI, but the basis of SPI plasticity differed between the species and stress severity. Under the 60% MWC treatment, birch stomata were smaller, whereas redbud leaves had fewer stomata. Under the 40% MWC treatment, redbud leaves instead had smaller stomata. Water-deficit-induced reductions in SD (Pääkkönen et al., 1998; Silva, et al., 2009; Camposeo et al., 2011; Rajabpoor et al., 2014) and SS (Luo et al., 2007; Maes et al., 2009) have been noted in other tree species. Many of these studies were conducted in dry regions and/or dry-adapted species, but in this study we show that these mechanisms of stomatal development plasticity also exist in temperate North American species adapted to more mesic environments.
The response to WS by maple leaves is more reflective of the broader literature on stomatal anatomy plasticity. This lack of stomatal development plasticity may be because maintenance of existing stomatal (and thus gas-exchange) capacity is advantageous for postdrought recovery, and because transient responses to water deficit, such as stomatal closure or solute accumulation, can be easily reversed, whereas anatomical changes to leaves are permanent. Still, the fact that there is a molecular basis and empirical data for stomatal anatomy plasticity in tree species suggests a potential acclimation/adaptive role.
The basis of SPI variation differs among species even under WW conditions. In WW birch and redbud leaves, higher SPI was a product of higher SD and SS (Fig. S21 a, c), but in maple leaves, higher SPI was due to larger stomata (Fig. S21b). Because SPI was correlated with stomatal frequency and size in WW and WS leaves, it appears that different tree species allocate a similar epidermal allocation of stomatal pore area (similar range of SPI in all three species, Fig. S21) differently via alteration of SS and/or SD. Silver maple leaves appear to have evolved to maintain a minimal range of small stomata while maximizing SPI via SD (Franks et al., 2009). Maintaining small stomata as SPI variation is dependent on variation in SD would also minimize the cost associated with opening stomata (Spence et al., 1986; Raven et al., 2014), while maximizing the benefit (potential conductance or gsmax) obtained by increasing SD (de Boer et al., 2016). Redbud WS leaves also had a relatively tight range of small stomata, but exhibited SPI variation via a much broader range of SD (Fig S21c).
It appears that the components of SPI are subject to independent mechanisms of plasticity, resulting in different anatomical mechanisms to a common outcome (Fig. 5). Firstly, passive control of guard cell size is evident in birch WS leaves because both cell types were smaller in WS leaves (birch, Figs 1, S13a, S15a, b) and pavement cell size and SS were positively correlated (Fig. S22a, c). Passive control of stomatal size, whereby guard cell dimensions are a function of cell turgor, probably result in a mechanical advantage frequently ascribed to smaller stomata (Spence et al., 1986), and thus is favored under WS conditions. Under WW conditions, SS could be actively controlled to respond to other environmental factors, such as light.
Redbud WW and WS leaves exhibited differential coordination of stomatal and pavement cell size (Fig. S22c), raising the possibility of different mechanisms controlling stomatal size in response to changing environmental conditions. Previous work in Arabidopsis has demonstrated that pavement and guard cell size develop differently during leaf growth due to differential regulation of cell growth (Asl et al., 2011).
The WS-induced changes in stomatal traits appear to be elicited by different mechanisms in the different species. Specifically, we observed that decreases in birch SI, SS, and SPI were correlated with decreased leaf RWC (Fig. S23). Additionally, since guard and pavement cells were smaller in birch leaves (Figs 1c, S13a, S15a, b), stomatal trait plasticity may be primarily turgor-driven in this species. This response in birch leaves may be due to a reduced RWC-induced accumulation of abscisic acid in drought-stressed leaves (Sack et al., 2018). The sensitivity of stomatal and leaf development pathways to ABA accumulation may be more pronounced in certain species such as birch via variation in expression, copy number, or protein homology of the genes involved in ABA signaling and leaf cell development.
Vein density is often higher in more drought-tolerant species or genotypes, often coupled with reduced LA (Scoffoni et al., 2011; Nardini et al., 2012), which may enable leaf hydration during drought conditions, as embolized veins can be bypassed through additional venation (Sack et al., 2008). Although our VD data was similar to prior values for related species (Sellin et al., 2012; Uhl & Mosbrugger, 1999), VD was similar in WW and WS leaves (Figs 3, S18). In a variety of species, Aasamaa et al. (2001) also found VD to be similarly non-plastic. Maple leaves grown under 40% MWC were the only case in which VD increased in response to water deficit (Figs 3. S18b), possibly as part of the overall response in which there was no change in stomatal development or SPI. This may suggest that only a severe water deficit necessitates a change in water supply to leaves, but this response may itself be absent if leaf water demand is reduced by other anatomical plasticity such as lower SPI in redbud leaves. Fiorin et al. (2016) propose that the mean stomata-vein distance imposes a limit on the density of stomata that can be adequately supplied water by leaf venation. With this in mind, the reduction of leaf thickness (in the mild stress) and stomatal or vein plasticity may have been sufficient to keep leaf tissue and the associated stomata hydrated.
The coordination of stomatal and vein development has been described as the balance between water demand and supply in leaves (Brodribb & Jordan, 2011; Schneider et al., 2017). To achieve a high rate of gas exchange, a high SD must be matched with high VD (Fiorin et al., 2016), which has often been demonstrated in the correlation between SD and VD (Carins Murphy et al., 2012; Carins Murphy et al., 2016). In T. ciliata, this relationship was found to persist even after imposition of a low humidity treatment (Carins Murphy et al., 2014). Although maple leaves did not exhibit stomatal development plasticity in response to WS, SD was nevertheless coordinated with VD in the range of SD across WW and WS leaves (Fig. S24b). Vein density also increased in 40% MWC maple leaves (Fig. S18b), and it is unlikely that this was simply due to smaller leaves under this treatment, as SD did not similarly increase (Figs 1a, S11b). Thus, WS maple leaves may produce additional venation to maintain the positive SD-VD correlation (Carins Murphy et al., 2014). Furthermore, the SD-VD relationship was distinct in redbud and birch WS versus WW leaves (Fig. S24). Although SD-VD coordination has been proposed as a critical factor in the evolution of angiosperms (Boyce et al., 2009; Zhang et al., 2012), it is nonetheless absent in several woody angiosperm species (Torre et al., 2003; Zhao et al., 2016).
Impact of stomatal development plasticity on leaf physiology
Water deficit results in stomatal closure, and thus reduced gs and A in leaves of river birch (Ranney et al., 1991), eastern redbud (Abrams, 1988; Griffin et al., 2004) and related maple species (Bauerle et al., 2003). Stomatal closure is sometimes observed in conjunction with stomatal plasticity (Cavender-Bares et al., 2007; Eksteen et al., 2013). In a diverse panel of gymnosperms, ferns, and angiosperms, McElwain et al. (2016) showed that as the anatomical capacity for gas exchange increases, the operational rate of gas exchange also increases. However, in this study, gas exchange was not strongly dependent on stomatal anatomy. In birch and redbud, but not maple leaves, stomatal development was positively correlated with A and gs (Fig. S25). However, many other stomatal anatomical traits were not correlated with gas exchange, and especially critically, neither SPI nor gsmax were correlated with gs (data not shown). The lack of a clear link between stomatal anatomy and gas exchange during drought conditions has previously been reported in a variety of species (Pääkkönen et al., 1998; Carins Murphy et al., 2014; Vieira et al., 2017; Toscano et al., 2018) Based on these reports and the present data set, we propose that among tree species, stomatal traits are only partially responsible for leaf physiology during water deficit episodes.
Plasticity in SPI was frequently not accompanied by plasticity in gsmax (Figs 1a, 2), with only redbud leaves under 60% MWC exhibiting a decrease in both traits. Other instances of reduced SPI, such as birch leaves under 60% MWC and redbud leaves under 40% MWC, were not accompanied by a decrease in gsmax. Reducing leaf SPI but maintaining gsmax thus minimizes stomatal production and operating costs while maximizing carbon assimilation gains, which is especially important in water-stressed leaves. Moreover, maintenance of gsmax would enable rapid return to normal gs after restoration of water-sufficient conditions, especially if a similar overall anatomy of water-stressed leaves now comprises smaller stomata that can be opened more easily. Because most species, including those from the present dataset, show operational gs as a very low fraction of gsmax, (Fig. S26; McElwain et al., 2016), the reduction of gsmax in 60% MWC redbud leaves would have little to no impact on redbud recovery, and indeed this was observed in redbud following root-zone re-saturation (Fig. 4c, f).
Instead of a balance between stomatal costs and benefits, it may be the case that stomatal anatomy under water-deficit conditions is directed towards facilitating stomatal closure. None of the anatomical traits that control the dimensions and overall area of stomata for gas exchange (SD, SS, SPI, gsmax) were correlated with operational A or gs. Instead, in all three species, the area coverage of stomata was correlated with the degree of stomatal closure observed during the stress period (Fig. S27). Thus, lower SPI reduced the degree of stomatal closure necessary during the WS period. Although not typically discussed in these terms, we propose that stomatal developmental inhibition is aimed towards achieving minimum gs, without constraining maximum gs.
Conclusions
A common outcome of reduced SPI is achieved by some North American tree species via different mechanisms in response to water deficit (Fig. 5). These species and treatment-level differences illustrate the importance of reporting all stomatal traits in leaf anatomical plasticity studies. For instance, although SS or SD was reduced in some Eucalyptus grandis clones under certain water-deficit treatments, SPI calculated from these values was almost always unchanged (Eksteen et al., 2013). A similar effect was observed in certain Prunus dulcis ecotypes: reduced SPI was the result of either smaller or fewer stomata, so presenting these traits in isolation would have missed this phenotypic plasticity (Camposeo et al., 2011). In all three species across both treatment levels, examining stomatal frequency and size, as well as the combinations of these traits via SPI and gsmax, was critical to the conclusions drawn. Had SD or SS been examined in isolation, the different mechanisms of plasticity between birch and redbud at 60% MWC would have not been revealed. Similarly, the differential response in redbud leaves grown at 60% or 40% MWC would have also been missed. We could also deduce that reduced gs in WS maple (and the fourth leaf of WS birch) was due primarily to reduced stomatal aperture, since there were no changes in the total coverage of stomata in these leaves, despite the smaller stomata produced in these leaves under the 40% MWC treatment (Figs 1c, S13).
Author contributions
NAM and MVM designed the experiment; NAM and SFL performed all greenhouse work and data collection; NAM and MVM analyzed the data; NAM, SFL, and MVM wrote the manuscript.
Supporting information
Figure S1. Environmental conditions throughout the experiment.
Figure S2. Media water content throughout the experiment.
Figure S3. Schematic of leaves used for data collection in the 60% MWC treatment.
Table S1. Description of measurements made on different leaves during both experiments.
Figure S4. Leaf growth dynamics in the 60% MWC treatment.
Figure S5. Plasticity of tree height and leaf area in both experiments.
Figure S6. Height of trees after both treatments of WS.
Figure S7. Leaf area of individual leaves after both WS treatments.
Figure S8. Relative water content of individual leaves in both WS treatments.
Figure S9. Leaf osmotic potential and osmotic potential at full turgor after both WS treatments.
Figure S10. Specific leaf weight of leaves after both WS experiments.
Figure S11. Stomatal density of individual leaves after both experiments.
Figure S12. Stomatal index of individual leaves after both experiments.
Figure S13. Stomatal size of individual leaves after both experiments.
Figure S14. Stomatal length and width of individual leaves after both experiments.
Figure S15. Collected plasticity and individual leaf pavement cell size after both experiments.
Figure S16. Collected plasticity and individual leaf pavement cell density after both experiments.
Figure S17. Stomatal pore index of individual leaves after both WS experiments.
Figure S18. Vein density in leaves 3 and 4 measured after development under well-watered or water-stressed (WS, 60 or 40% MWC) conditions.
Figure S19. Transpiration rate and whole-leaf transpiration in leaves during and after the 60% MWC water deficit.
Figure S20. Stomatal conductance, net CO2 assimilation, transpiration rate, and whole-leaf transpiration during the 40% MWC water deficit.
Figure S21. The relationship between stomatal pore index and stomatal size, density, and index across both experiments.
Figure S22. Relationship between pavement cell size and stomatal size across both experiments.
Figure S23. Relationship between relative water content and stomatal traits across both experiments.
Figure S24. Relationship between stomatal and vein density across both experiments.
Figure S25. Relationship between stomatal index and gas exchange across both experiments.
Figure S26. Changes in stomatal conductance as a % of gsmax in the fourth leaf that developed under water-stressed conditions (60% MWC), along with well-watered controls over time.
Figure S27. Relationship between stomatal pore index and degree of stomatal closure in WS plants.
Acknowledgements
We thank James McKenna for plant material; Nathan Deppe and Dan Little for greenhouse assistance; Mike Gosney, Amanda Ávila Cardoso, and Huangai Bi for assistance with data collection; Robert Heath for assistance with leaf growth analysis; Scott McAdam and Chris Oakley for suggestions on methodology and manuscript advice; Peter Goldsbrough for helpful edits to the manuscript; Tom Kronewitter of Purdue University Agricultural Communication for the production of Figure 5; and the Purdue University Center for Plant Biology for funding.